Datasets:

TaylorAI

/

pubmed_noncommercial

like 5

PMC010xxxxxx/PMC10353660.txt

==== Front Belitung Nurs J Belitung Nurs J BNJ Belitung Nursing Journal 2528-181X 2477-4073 Belitung Raya Foundation BNJ-9-1-079 10.33546/bnj.2380 Original Research Factors associated with the completion of antenatal care in Indonesia: A cross-sectional data analysis based on the 2018 Indonesian Basic Health Survey https://orcid.org/0000-0002-3483-6717 Idris Haerawati * Sari Indah Faculty of Public Health, Sriwijaya University, Indralaya, Ogan Ilir, South Sumatera 30662, Indonesia * Corresponding author: Dr. Haerawati Idris, SKM.M, Kes, Faculty of Public Health, Sriwijaya University Indralaya, Ogan Ilir, South Sumatera 30662, Indonesia. Email: haera@fkm.unsri.ac.id Cite this article as: Idris, H., & Sari, I. (2023). Factors associated with the completion of antenatal care in Indonesia: A cross-sectional data analysis based on the 2018 Indonesian Basic Health Survey. Belitung Nursing Journal, 9(1), 79-85. https://doi.org/10.33546/bnj.2380 12 2 2023 2023 9 1 7985 21 10 2022 23 11 2022 05 2 2023 © The Author(s) 2023 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License, which allows others to remix, tweak, and build upon the work non-commercially as long as the original work is properly cited. The new creations are not necessarily licensed under the identical terms. Background The global incidence of maternal mortality remains high, including in Indonesia, and the utilization of antenatal care services can help reduce these rates. Despite numerous studies examining factors affecting antenatal care utilization, there has been limited focus on identifying factors related to the completion of these services. Objective This study aimed to analyze factors associated with the completion of antenatal care in Indonesia. Methods The study used a cross-sectional analysis of secondary data from the Basic Health Research conducted by the Indonesian Ministry of Health in March 2018. The sample consisted of 65,929 pregnant women aged 15 to 49. Descriptive statistics, chi-square tests, and multiple logistic regression were used for data analysis. Results The majority of respondents (75.2%) completed antenatal care. Factors significantly correlated with antenatal care completion were education level, occupation status, health insurance ownership, place of antenatal care services, travel time to health facilities, area of residence, history of pregnancy, parity, desired pregnancy, and pregnancy complications (p <0.05). The multiple logistic regression test showed that education level was the most dominant factor associated with antenatal care completion (p <0.001, OR = 2.023, 95% CI = 1.839-2.225). Conclusion Completion of antenatal care is influenced by various factors, including education, job status, health insurance ownership, antenatal care services’ location, travel time to health facilities, residence area, previous pregnancy history, number of children, desired pregnancy, and pregnancy complications. However, education is crucial in determining a mother’s understanding and approach toward using these services. The Indonesian government should enhance public education and awareness initiatives to increase utilization. Healthcare professionals, particularly nurses and midwives, play a vital role in educating pregnant women about the significance of utilizing prenatal care services consistently and facilitating their access to these services efficiently. antenatal care cross-sectional studies female Indonesia pregnancy nurses midwives ==== Body pmcBackground A nation’s public health is evaluated by various factors, including the maternal mortality rate, which the 2030 Sustainable Development Goals aim to decrease to 70 deaths per 100,000 live births. However, according to a prediction by the World Health Organization (WHO) in 2017, the maternal mortality rate was expected to be 211 deaths per 100,000 live births (World Health Organization, 2020). In 2015, Indonesia’s maternal mortality rate was 305 deaths per 100,000 live births, four times higher than the target set in the Sustainable Development Goals (Indonesian Ministry of Health, 2020a). In addition, WHO reports that HIV infection and malaria contribute to morbidity and death in pregnant women (World Health Organization, 2016). In Indonesia, the leading causes of maternal death are bleeding (1,280 cases), hypertension (1,066 cases), and infections (207 cases) (Indonesian Ministry of Health, 2020a). The Indonesian Ministry of Health is implementing policies to decrease maternal mortality rapidly, such as improving cross-sectoral collaboration, increasing professional coordination, optimizing the use of national health insurance, and activating standby villages (Indonesian Ministry of Health, 2018b). Antenatal Care (ANC) is provided to ensure all mothers have access to adequate maternal health services, as stated in Regulation of the Indonesian Ministry of Health No. 97 of 2014 on maternal health services, including antenatal care services. Antenatal care involves pregnancy examinations to enhance physical and mental health, prepare for childbirth and postpartum, promote exclusive breastfeeding and restore reproductive health (Indonesian Ministry of Health, 2018b). Using antenatal care services can lower maternal mortality (Idris, 2022). In 2002, the World Health Organization introduced the Focused Antenatal Care (FANC) program (World Health Organization, 2016), which has become a standard for high-quality and comprehensive care for pregnant women, requiring at least four visits during the first trimester (Nurlaili, 2019). Antenatal care has several benefits for pregnant women, such as reducing the risk of adverse pregnancy outcomes, perinatal and infant mortality, and morbidity (Brown et al., 2008). It also raises the chances of receiving skilled birth assistance and postnatal care (Ml et al., 2012). The number of visits is often used as an indicator of antenatal care use, but other factors, such as the timing of visits, services received, and care provider type, may also be considered (Ataguba, 2018). In Indonesia, a middle-income country with a large population of 262 million spread across 17,774 islands, pregnant women can receive antenatal health services at primary care facilities, hospitals, or clinics (Mahendradhata et al., 2017). The recommended standards are one visit during the first trimester (0-12 weeks of gestation), one visit during the second trimester (12-24 weeks of gestation), and two visits during the third trimester (24 weeks of gestation up to delivery) to ensure optimal health and safety, as well as to prevent and manage complications from pregnancy (Indonesian Ministry of Health, 2019). The Indonesian antenatal care standard includes 11 procedures to be completed by nurse-midwives, with guidelines for implementation in each public health center. These procedures are as follows: 1) Weighing, 2) Measuring upper arm circumference, 3) Taking blood pressure, 4) Measuring fundal height, 5) Monitoring fetal heart rate, 6) Identifying fetal presentation, 7) Administering tetanus toxoid vaccine, 8) Providing iron tablets, 9) Conducting laboratory tests, 10) Making appropriate referrals, and 11) Providing health education (Indonesian Ministry of Health, 2020b). To properly implement these procedures, each public health center must create a technical strategy or guideline that outlines how nurse midwives should carry out these procedures promptly. In addition, the communication and clinical skills of nurse-midwives guarantee the effective management of pregnant women during prenatal care (Widyawati et al., 2015). A systematic review by Simkhada et al. (2008) discovered multiple factors that impact the use of prenatal care in developing countries, including maternal education, husband’s education, marital status, accessibility, cost, household income, women’s employment, media exposure, and history of obstetric complications. Cultural beliefs and perceptions about pregnancy also play a role, with a significant effect of parity on adequate attendance observed. Women with higher parity tend to use prenatal care less often, but this may vary based on age and religion (Simkhada et al., 2008). Previous studies have shown that various factors can affect the utilization of complete prenatal care services. For example, in Ethiopia, a study found links between ethnicity, place of residence, education level, and household economic status with the use of prenatal care (Tiruaynet & Muchie, 2019). Similarly, research in Rwanda revealed that older women, single women or divorcees, and those with insufficient social support were associated with inadequate utilization of prenatal care services (Rurangirwa et al., 2017). However, despite numerous studies exploring factors that influence prenatal care coverage, there is limited research focused on factors related to the completion of prenatal care. Therefore, this study aimed to identify and analyze the factors associated with the completion of antenatal care in Indonesia. Methods Study Design This study utilized a cross-sectional analysis of secondary data obtained from the Basic Health Research (RISKESDAS) conducted by the Indonesian Ministry of Health in March 2018 (Indonesian Ministry of Health, 2018a). Samples/Participants A multi-stage, systematic random sampling approach was used for data collection that was based on the RISKESDAS. In the first stage, census block groups were chosen as the primary sampling units, and then in the second stage, census blocks were selected based on the probability proportional to their enrolment size. In the third stage, ten census buildings were selected through systematic random sampling from each census block, and finally, one household was randomly chosen from each census building. The data analyzed in this study was gathered every three years starting in 2007 by the National Institute of Health Research and Development and the Indonesian Ministry of Health (Indonesian Ministry of Health, 2018a). The sample size of this current study was 65,929 pregnant women aged 15 to 49. Measures The dependent variable in this study was the completion of antenatal care, divided into two categories: complete and incomplete. A woman was considered to have complete antenatal care if she received at least one prenatal care visit during the first trimester, one visit during the second trimester, and two visits during the third trimester. Any other scenario was categorized as incomplete. The study had 12 independent variables: education level, occupation status, health insurance ownership, place of antenatal care services, travel time to health facilities, area of residence, history of pregnancy, parity, history of abortion, multiple pregnancies, desired pregnancy, and pregnancy complications. The education level was divided into three categories: college, secondary education, and primary education. The occupation status was categorized as either working or not working. Health insurance ownership was divided into two categories: yes and no. The place of antenatal care services was either at health facilities or home. Travel time to health facilities was divided into ≤15 minutes and >15 minutes. The area of residence was either urban or rural. The history of pregnancy was divided into two categories: once and more than once. The parity was grouped into two categories: two times and more than two times. The history of abortion was divided into two categories: ever and never. Multiple pregnancies were divided into two categories: yes and no. The desired pregnancy was divided into two categories: yes and no. Finally, the pregnancy complications were divided into two categories: yes and no. Data Analysis The data were analyzed using descriptive statistics, chi-square, and multiple logistic regression. The analysis was performed using SPSS 23.0 for Windows. Ethical Considerations This study was approved by the Ethics Review Center of the Faculty of Public Health, Sriwijaya University, with the Letter of Ethical Qualification No: 042/UN9.FKM/TU.KKE/2021. Results Table 1 shows that 75.2% of the participants completed their antenatal care. The majority of the participants had completed their education at the secondary level (60.3%). Of the participants, 40.8% were employed, and 39.2% had health insurance. Nearly all the participants (99.1%) sought ANC services at health facilities. The time to reach these facilities was ≤ 15 minutes for 44.1% of the participants. The majority of the participants (54.3%) lived in urban areas, and 31.8% had a history of one pregnancy. The number of times they had been pregnant was two or fewer (73.6%) in most cases. 14.4% had a history of abortion, and 0.8% had multiple pregnancies. The majority (91.6%) of the participants had desired pregnancy, and 24.8% reported pregnancy complications. Table 1 Respondents’ characteristics (N = 65,929) Variables n % Antenatal care completion Complete 49,584 75.2 Incomplete 16,345 24.8 Education level High 8,131 12.3 Secondary education 39,746 60.3 Primary education 18,052 27.4 Occupation status Employed 26,896 40.8 Unemployed 39,033 59.2 Health insurance ownership Yes 25,832 39.2 No 40,097 60.8 Place of ANC services Health facilities 65,366 99.1 Home 563 0.9 Travel time to health facilities ≤ 15 minutes 29,100 44.1 > 15 minutes 36,829 55.9 Area of residence Urban 35,772 54.3 Rural 30,157 45.7 History of pregnancy Once 20,982 31.8 > 1 time 44,947 68.2 Parity ≤ 2 times 48,521 73.6 > 2 times 17,408 26.4 History of abortion Ever 9,480 14.4 Never 56,449 85.6 Multiple pregnancies Yes 519 0.8 No 65,410 99.2 Desired pregnancy Yes 60,371 91.6 No 5,558 8.4 Pregnancy Yes 16,347 24.8 No 49,582 75.2 The results of the bivariate analysis using the chi-square test are presented in Table 2. The variables of occupation, education, health insurance ownership, place of ANC services, travel time to health facilities, area of residence, history of pregnancy, parity, desired pregnancy, and pregnancy complications showed a significant relationship with the completion of antenatal care (p <0.05). In contrast, the variables of history of abortion and multiple pregnancies were not associated with the completion of antenatal care (p >0.05). Table 2 Bivariate analysis of antenatal care completion and its related factors (chi-square) Variables Antenatal care completion p OR (95% CI) Complete Incomplete n % n % Education level High 6,871 84.5 1,260 15.5 <0.001 2.650 (2.418-2.904) Secondary education 30,563 76.9 9,183 23.1 <0.001 1.617 (1.517-1.712) Primary education 12,150 67.3 5,903 32.7 Occupation status Employed 21,219 78.9 5,678 21.1 0.006 1.075 (1.021-1.132) Unemployed 30,261 77.5 8,772 22.5 Health insurance ownership Yes 19,928 77.1 5,904 22.9 <0.001 1.188 (1.128-1.252) No 29,657 74 10,440 26 Place of ANC services Health facilities 49,276 75.4 16,090 24.6 <0.001 2.526 (2.042-3.125) Home 309 54.8 254 45.2 Travel time to health facilities ≤ 15 minutes 23,192 79.7 5,908 20.3 <0.001 1.552 (1.468-1.642) > 15 minutes 26,392 71.7 10,437 28.3 Area of residence Urban 28,410 79.4 7,362 20.6 <0.001 1.637 (1.551-1.728) Rural 21,174 70.2 8,983 29.8 History of pregnancy Once 16,351 77.9 4,631 22.1 <0.001 1.245 (1.173-1.320) > 1 time 33,233 73.9 11,714 26.1 Parity ≤ 2 times 37,845 78 10,676 22 <0.001 1.712 (1.624-1.804) > 2 times 11,739 67.4 5,669 32.6 History of abortion Ever 7,194 75.9 2,286 24.1 0.253 1.043 (0.970-1.122) Never 42,390 75.1 15,059 24.9 Multiple pregnancies Yes 393 75.6 126 24.4 0.870 1.023 (0.779-1.344) No 49,192 75.2 16,218 24.8 Desired pregnancy Yes 46,051 76.3 14,320 23.7 <0.001 1.844 (1.697-2.002) No 3,533 63.6 2,025 36.4 Complication of pregnancy Yes 12,621 77.2 3,726 22.8 <0.001 1.156 (1.086-1.231) No 36,963 74.6 12,619 25.4 Table 3 further demonstrates the significant relationship between education level, health insurance ownership, place of ANC services at health facilities, travel time to health facilities, the urban area of residence, desired pregnancy, and pregnancy complications with the completion of antenatal care. The most influential variable was education level, with the highest adjusted odds ratio (OR) of 2.023 (95% confidence interval (CI) = 1.839-2.225). The results indicate that respondents with higher levels of education are 2.023 times more likely to complete antenatal care than those with a primary education level, even when controlled for other variables (95% CI =1.839-2.225). Table 3 Multiple logistic regression analysis results Variables p Adjusted OR Education level High <0.001 2.023 (1.839-2.225) Secondary education <0.001 1.356 (1.279-1.439) Primary education Ref Health insurance ownership Yes <0.001 1.114 (1.056-1.172) No Ref Place of ANC services Health facilities <0.001 1.926 (1.553-2.389) Home Ref Travel time to health facilities ≤ 15 minutes <0.001 1.241 (1.164-1.324) > 15 minutes Ref Area of residence Urban <0.001 1.399 (1.315-1.488) Rural Ref Parity ≤ 2 times <0.001 1.500 (1.419-1.586) > 2 times Ref Desired pregnancy Yes <0.001 1.703 (1.559-1.860) No Ref Pregnancy complications Yes <0.001 1.143 (1.073-1.218) No Ref Note: Ref – Reference category Discussion This study analyzed the factors influencing antenatal care completion in Indonesia and found that 75.2% of the respondents completed their antenatal care. In addition, education level, occupation status, health insurance ownership, place of ANC services, travel time to health facilities, area of residence, history of pregnancy, parity, desired pregnancy, and pregnancy complications were found to have significant relationships with antenatal care completion. Education level was found to play a significant role in the completion of antenatal care. Respondents with a high educational degree were 2.65 times more likely to complete their antenatal care than those with primary education. This finding is consistent with previous research, showing that education level statistically correlates with maternal compliance in antenatal care visits. Mothers with higher levels of education are 3.383 times more likely to attend antenatal care visits than those with primary education (Wulandatika, 2017). To increase knowledge, mothers should be encouraged to have consultations with health workers, attend pregnancy classes, and learn about pregnancy through various media. Health workers should also conduct counseling and outreach to residential homes to raise public awareness about the importance of utilizing health facilities for antenatal care, especially among expecting mothers. Junga et al. (2017) also found a correlation between education and the regularity of antenatal care visits. A relationship between education and the utilization of antenatal care services (OR = 7.286) was also found by Humokor et al. (2019), meaning that mothers with higher education were 7.286 times more likely to utilize these services than those with lower education. This study also found a relationship between occupation status and antenatal care completion. Mothers employed had a 1.075 higher chance of completing their antenatal care than non-working mothers. This finding aligns with the research conducted by Dengo and Mohamad (2019), which demonstrates a significant relationship between maternal work and antenatal care visits. This may be due to the fact that working mothers tend to have a higher educational level, thereby having more knowledge about the importance of regular antenatal care. In addition, working mothers allocate time to visiting health facilities and receiving care at home for their pregnancy. However, this study contradicts the results of research conducted by Ariestanti et al. (2020) and Putri and Hastutik (2021), which found no relationship between work status and maternal behavior during pregnancy. This study found that health insurance ownership played a role in antenatal care completion. Mothers with health insurance had a 1.188 higher likelihood of completing antenatal care than those without health insurance. Research in Mongolia has established a connection between health insurance ownership and health service utilization (Gan-Yadam et al., 2013). Similarly, Kurniawan and Intiasari (2012) found a correlation between health insurance ownership and health service utilization. Health insurance makes individuals more likely to use health services, as they do not have to pay for them. Health insurance ownership can improve access to health services. The study also determined that the place of ANC services impacted antenatal care completion. Mothers who received ANC services at health facilities were 2.526 times more likely to complete antenatal care than those who received services at home. Research supports this finding by showing a correlation between service availability and visits for antenatal care. The presence of health facilities and health workers plays a crucial role in increasing visits for antenatal care. However, inadequate health facilities and health workers may contribute to the low frequency of pregnancy check-up visits. To address this issue, mothers need to be motivated to carry out antenatal care services (Supliyani, 2017). A study conducted by Fitrina et al. (2020) also revealed a relationship between access to health facilities and antenatal care visits. High-risk pregnant women with easy access to health facilities were nine times more likely to complete antenatal care visits than those without easy access. The proximity to health facilities affects travel time. This study found a correlation between travel time to health facilities and the completion of antenatal care services. Mothers who had a travel time of less than or equal to 15 minutes to health facilities had a 1.552 higher chance of completing antenatal care compared to mothers who had a travel time of more than 15 minutes to health facilities. Research supports the finding that travel time to health facilities is significantly linked to antenatal care visits. Mothers who spent more than 25 minutes traveling to health facilities had 1.789 times the chance of having less than four pregnancy check-ups (Supliyani, 2017). Pregnant women who can reach health care facilities faster are more likely to utilize more frequent antenatal care services than those who take a long time to get the facilities. Therefore, having a health facility close to one’s home can improve access to health care. This finding is supported by research by Gamelia et al. (2013), who showed a relationship between travel time to health services and behavior toward pregnancy care. This study highlights the correlation between the area of residence and access to health services. Mothers residing in urban areas have a higher chance (1.637 times) of completing antenatal care services than those in rural areas. Similarly, Fatali and Budyanra (2020) affirm this by stating that the location of residence significantly impacts pregnancy visits. Expectant mothers in urban areas have 2.065 times more chances to undergo pregnancy checks than those in rural areas. The proximity of the location of residence to such facilities determines the ease of access to health services. Housing with adequate infrastructure, such as proper roads and transportation, facilitates access to health services in urban areas (Fatali & Budyanra, 2020). As such, the government should ensure that health facilities are developed evenly in both rural and urban areas. This study also found a significant relationship between a history of pregnancy and the completion of antenatal care. Mothers with at least one previous pregnancy had a 1.245 higher chance of completing antenatal care than mothers with multiple pregnancies. This aligns with the research of Saptarini and Suparmi (2016), who found a connection between pregnancy history and antenatal care visits. Mothers with three or more pregnancies were less likely to complete visits for antenatal care and pregnancy check-ups, as they felt they had adequate knowledge and experience. However, mothers with two pregnancies felt inexperienced and needed information and help. According to Kusuma (2018), 70% of mothers had multiple pregnancies. These multigravida mothers had some experience with visits to public health centers, hospitals, clinics, and others during their previous pregnancies. Parity is the number of children who were born live to mothers. This study found that parity is related to the completion of antenatal care. Mothers with parity of two or less had a 1.712 times higher chance of completing antenatal care than mothers with parity greater than two. This is consistent with Junga et al. (2017), which revealed a connection between maternal parity and the regularity of antenatal care examinations. Mothers with a higher parity feel more experienced in pregnancy and childbirth, making them less concerned during subsequent pregnancies. In addition, pregnant women with fewer children tend to be more diligent in checking their pregnancy compared to mothers with more children (Choirunissa & Syahputri, 2018). Sari (2015) also found a correlation between parity and adherence to standard antenatal care visits, with lower parity resulting in more routine visits. This may be because high parity pregnancies are often unplanned. Dengo and Mohamad (2019) also found a substantial relationship between pregnancy parity and antenatal care visits. This study found that desired pregnancy is related to the completion of antenatal care, with mothers who had desired pregnancies having a 1.844 times greater likelihood of completing antenatal care compared to mothers who had unwanted pregnancies. This finding is supported by research by Dumilah (2019), which found that desired pregnancies receive 5.1 times more regular check-ups compared to unwanted pregnancies. Conversely, Dini et al. (2016) found that mothers with unwanted pregnancies had a 1.79 times greater chance of not receiving standard prenatal care than those with desired pregnancies. Therefore, mothers with unwanted pregnancies should seek information about pregnancy health earlier to prevent late detection and treatment of pregnancy complications. The study found that pregnancy complications are linked to the completion of antenatal care. Women who experience complications during pregnancy are more likely to complete antenatal care than those without. This is supported by research conducted by Jusniany et al. (2016), which found a significant relationship between pregnancy complications and the utilization of antenatal care services. Women who experience complications are more likely to be aware of their health and seek antenatal care services. However, some women with mild complaints may not seek a pregnancy check-up. The level of knowledge and maternal awareness can be improved by participating in regular antenatal care during pregnancy, as complaints related to the disease are likely to result in the utilization of antenatal care services (Indrastuti & Mardiana, 2019). Limitations This study, which employed a cross-sectional survey approach, was unable to establish causality between the factors due to its limitations. Furthermore, secondary data analysis revealed the exclusion of certain variables, such as geographical, economic, and social-cultural factors, owing to limitations in the available data. Implications of the Study for Nursing Practice The results of this study found relationships between completion of antenatal care and various factors such as education level, occupation status, health insurance ownership, place of ANC services, travel time to health facilities, area of residence, pregnancy history, parity, desired pregnancy, and pregnancy complications, can serve as valuable information for nurses and other healthcare professionals in developing effective healthcare interventions. The role of nurses in promoting the use of antenatal care services is crucial. They play a significant part in educating and providing health information to pregnant women. Nurses should take steps to improve health education and raise community awareness to increase utilization of these services, which can lead to better health outcomes for expectant mothers. Conclusion Several variables have been found to significantly influence the completion of antenatal care, including education level, health insurance ownership, location of antenatal care services within health facilities, travel time to health facilities, urban residency, desired pregnancy, and pregnancy complications. Education level is essential in shaping a mother’s knowledge and attitudes toward using antenatal care services at health facilities. To enhance the utilization of these services, the Indonesian government should undertake increased socialization efforts to raise community awareness and knowledge of the importance of regular antenatal care. Healthcare workers, especially nurses and midwives, should also provide increased health education to pregnant women to help them access these services. Acknowledgment The authors would like to thank the Agency of Health Research and Development Indonesia for giving access to the raw data of Basic Health Research 2018. Declaration of Conflicting Interest None to declare. Funding None. Authors’ Contributions IS conceptualized the study design and acquired the raw data for analysis. HI conceptualized the article and prepared the original draft of the manuscript. All authors were accountable in each step of this current study and approved the final version of the manuscript to be published. Authors’ Biographies Dr. Haerawati Idris, SKM.M, Kes is a Lecturer in the Faculty of Public Health, Sriwijaya University, Indonesia. Indah Sari is an alumnus of Faculty of Public Health, Sriwijaya University, Indonesia. Data Availability The Agency of Health Research and Development Indonesia gives access to data on Basic Health Research by applying to their website [https://www.litbang.kemkes.go.id/layanan-permintaan-data-riset/]. ==== Refs References Ariestanti, Y., Widayati, T., & Sulistyowati, Y. (2020). Determinan perilaku ibu hamil melakukan pemeriksaan kehamilan (antenatal care) pada masa pandemi Covid-19 [Determinants of the behavior of pregnant women carrying out antenatal care during the Covid-19 pandemic]. Jurnal Bidang Ilmu Kesehatan, 10 (2 ), 203-216. 10.52643/jbik.v10i2.1107 Ataguba, J. E.-O. (2018). A reassessment of global antenatal care coverage for improving maternal health using sub-Saharan Africa as a case study. PloS One, 13 (10 ), e0204822. 10.1371/journal.pone.0204822 30289886 Brown, C. A., Sohani, S. B., Khan, K., Lilford, R., & Mukhwana, W. (2008). Antenatal care and perinatal outcomes in Kwale district, Kenya. BMC Pregnancy and Childbirth, 8 , 1-11. 10.1186/1471-2393-8-2 18179721 Choirunissa, R., & Syahputri, N. D. (2018). Analisis faktor yang berhubungan dengan pemeriksaan K4 pada ibu hamil di Puskesmas Bakung Provinsi Lampung tahun 2017 [Analysis of factors related to K4 examination in pregnant women at the Bakung Health Center in Lampung Province in 2017]. Jurnal Akademi Keperawatan Husada Karya Jaya, 4 (1 ), 72-93. Dengo, M. R., & Mohamad, I. (2019). Faktor berhubungan dengan rendahnya kunjungan antenatal pada kontak pertama pemeriksaan ibu hamil (K-1) [Factors related to the low number of antenatal visits at the first contact of the examination of pregnant women (K-1)]. Gorontalo Journal of Public Health, 2 (2 ), 162-169. 10.32662/gjph.v2i2.746 Dini, L. I., Riono, P., & Sulistiyowati, N. (2016). Pengaruh status kehamilan tidak diinginkan terhadap perilaku ibu selama kehamilan dan setelah kelahiran di Indonesia (analisis data SDKI 2012) [The effect of unwanted pregnancy status on maternal behavior during pregnancy and after birth in Indonesia: analysis of IDHS data 2012]. Jurnal Kesehatan Reproduksi, 7 (2 ), 119-133. 10.22435/kespro.v7i2.5226.119-133 Dumilah, R. (2019). Umur, interval kehamilan, kehamilan yang diinginkan dan perilaku pemeriksaan kehamilan [Age, pregnancy interval, desired pregnancy and pregnancy examination behavior]. Jurnal Penelitian Kesehatan" SUARA FORIKES"(Journal of Health Research" Forikes Voice"), 10 (2 ), 73-79. 10.33846/sf10201 Fatali, A. M. A., & Budyanra, B. (2020, 2020). Variabel-variabel yang memengaruhi status kunjungan pemeriksaan kehamilan (antenatal care) di Provinsi Papua tahun 2017 (analisis regresi logistik biner) [Variables affecting the status of antenatal care visits in Papua Province in 2017 : Binary logistic regression analysis]. Seminar Nasional Official Statistics, Jakarta. Fitrina, Y. R., Kamil, H., & Agustina, A. (2020). Hubungan ibu hamil risiko tinggi dengan kelengkapan kunjungan Antenatal Care (ANC) [Relationship between high risk pregnant women and the completeness of Antenatal Care (ANC) visits]. Jurnal Aceh Medika, 4 (2 ), 150-161. Gamelia, E., Sistiarani, C., & Masfiah, S. (2013). Determinan perilaku perawatan kehamilan [Determinants of pregnancy care behavior]. Kesmas: Jurnal Kesehatan Masyarakat Nasional (National Public Health Journal), 8 (3 ), 133-138. 10.21109/kesmas.v8i3.358 Gan-Yadam, A., Shinohara, R., Sugisawa, Y., Tanaka, E., Watanabe, T., Hirano, M., Tomisaki, E., Morita, K., Onda, Y., & Tokutake, K. (2013). Factors associated with health service utilization in Ulaanbaatar, Mongolia: A population-based survey. Journal of Epidemiology, 23 (5 ), 320-328. 10.2188/jea.JE20120123 23831715 Humokor, A. C., Rumayar, A. A., & Wowor, R. E. (2019). Hubungan antara pendidikan dan pendapatan keluarga dengan pemanfaatan pelayanan antenatal care di wilayah kerja Puskesmas Tuminting Kota Manado [The relationship between education and family income with the use of antenatal care services in the working area of the Tuminting Health Center Manado City]. Kesmas, 8 (7 ), 208-213. Idris, H. (2022). Factors associated with the choice of delivery place: A cross-sectional study in rural areas of Indonesia. Belitung Nursing Journal, 8 (4 ), 311-315. 10.33546/bnj.2095 Indonesian Ministry of Health. (2018a). Basic Health Research 2018 report (Riskesdas) [in Bahasa]. https://www.litbang.kemkes.go.id/laporan-riset-kesehatan-dasar-riskesdas/ Indonesian Ministry of Health. (2018b). The importance of antenatal care in health facilities [in Bahasa]. Direktorat Promosi Kesehatan dan Pemberdayaan Masyarakat, Kementerian Kesehatan RI. https://promkes.kemkes.go.id/pentingnya-pemeriksaan-kehamilan-anc-di-fasilitas-kesehatan Indonesian Ministry of Health. (2019). Indonesia Health Profile 2018 [in Bahasa]. https://www.kemkes.go.id/article/view/19070400001/profil-kesehatan-indonesia-tahun-2018.html Indonesian Ministry of Health. (2020a). Indonesia Health Profile 2019 [in Bahasa]. https://www.kemkes.go.id/folder/view/01/structure-publikasi-pusdatin-profil-kesehatan.html Indonesian Ministry of Health. (2020b). Pedoman pelayanan antenatal terpadu [Guidelines for integrated antenatal care]. https://perpustakaan.kemkes.go.id/inlislite3/opac/detail-opac?id=11594 Indrastuti, A. N., & Mardiana, M. (2019). Pemanfataan pelayanan antenatal care di puskesmas [Utilization of antenatal care services at puskesmas]. HIGEIA (Journal of Public Health Research and Development), 3 (3 ), 369-381. 10.15294/higeia.v3i3.26952 Junga, M. R., Pondaag, L., & Kundre, R. (2017). Faktor-faktor yang berhubungan dengan keteraturan pemeriksaan antenatal care (ANC) ibu hamil trimester III di Puskesmas Ranotana Weru Kota Manado [Factors associated with regularity of antenatal care examinations for third trimester pregnant women at Ranotana Weru Health Center Manado]. Jurnal Keperawatan, 5 (1 ). 10.35790/jkp.v5i1.14690 Jusniany, M., Mutahar, R., & Utama, F. (2016). Determinan pemanfaatan pelayanan antenatal yang adekuat di Indonesia (Analisis data SDKI 2012) [Determinants of adequate utilization of antenatal care in Indonesia (2012 IDHS data analysis)]. Jurnal Ilmu Kesehatan Masyarakat, 7 (3 ), 174-181. 10.26553/jikm.2016.7.3.174-181 Kurniawan, A., & Intiasari, A. D. (2012). Kebutuhan jaminan kesehatan masyarakat di wilayah perdesaan [The need for public health insurance in rural areas]. Kesmas: Jurnal Kesehatan Masyarakat Nasional (National Public Health Journal), 7 (1 ), 3-7. 10.21109/kesmas.v7i1.69 Kusuma, R. (2018). Hubungan pengetahuan dan sikap ibu hamil tentang antenatal care dengan kunjungan K4 [The correlation of knowledge and attitudes of pregnant women about antenatal care with K4 visit]. Jurnal Psikologi Jambi, 3 (1 ), 24-24. 10.22437/jpj.v3i1.6370 Mahendradhata, Y., Trisnantoro, L., Listyadewi, S., Soewondo, P., Marthias, T., Harimurti, P., & Prawira, J. (2017). The Republic of Indonesia health system review (Vol. 7 ). New Delhi: WHO Regional Office for South-East Asia. https://apps.who.int/iris/handle/10665/254716 Ml, A. N., Dramaix-Wilmet, M., & Donnen, P. (2012). Determinants of maternal health services utilization in urban settings of the Democratic Republic of Congo–a case study of Lubumbashi City. BMC Pregnancy and Childbirth, 12 , 1-13. 10.1186/1471-2393-12-66 22230245 Nurlaili, H. (2019). Determinan penggunaan pelayanan ANC di negara berkembang: Tinjauan pustaka tradisional [Determinants of antenatal care services use in developing countries: a traditional literature review]. PLACENTUM: Jurnal Ilmiah Kesehatan dan Aplikasinya, 7 (2 ), 1-7. 10.20961/placentum.v7i2.29718 Putri, N. K. S. E., & Hastutik. (2021). Analisis pekerjaan dengan perilaku ibu hamil untuk melakukan kunjungan antenatal care [Analysis of job and behavior of pregnant women to conduct antenatal care visits]. Jurnal Stethoscope, 1 (2 ), 106-113. 10.54877/stethoscope.v1i2.810 Rurangirwa, A. A., Mogren, I., Nyirazinyoye, L., Ntaganira, J., & Krantz, G. (2017). Determinants of poor utilization of antenatal care services among recently delivered women in Rwanda; a population based study. BMC Pregnancy and Childbirth, 17 , 1-10. 10.1186/s12884-017-1328-2 28049520 Saptarini, I. I., & Suparmi, S. (2016). Pemanfaatan dan kelengkapan pelayanan antenatal care di Kelurahan Kebon Kalapa, Kota Bogor Tahun 2014 [Utilization and completeness of antenatal care services in Kebon Kelapa Village, Bogor City in 2014]. Indonesian Bulletin of Health Research, 44 (3 ), 173-180. Sari, R. D. (2015). Hubungan antara karakteristik ibu hamil dengan kepatuhan ibu terhadap standar kunjungan antenatal care di Bps “X” Cikarang tahun 2014 [The relationship between the characteristics of pregnant women and the adherence of mothers to standard antenatal care visits at Bps “X” Cikarang in 2014]. Jurnal Bidang ilmu kesehatan, 5 (1 ), 211-217. 10.52643/jbik.v5i1.111 Simkhada, B., Teijlingen, E. R. v., Porter, M., & Simkhada, P. (2008). Factors affecting the utilization of antenatal care in developing countries: Systematic review of the literature. Journal of Advanced Nursing, 61 (3 ), 244-260. 10.1111/j.1365-2648.2007.04532.x 18197860 Supliyani, E. (2017). Jarak, waktu tempuh, ketersediaan pelayanan dan kunjungan pemeriksaan kehamilan di puskesmas [Distance, travel time, availability of services and visits for antenatal care at the puskesmas]. Jurnal Informasi Kesehatan Indonesia, 3 (1 ), 14-22. Tiruaynet, K., & Muchie, K. F. (2019). Determinants of utilization of antenatal care services in Benishangul Gumuz Region, Western Ethiopia: A study based on demographic and health survey. BMC Pregnancy and Childbirth, 19 (1 ), 1-5. 10.1186/s12884-019-2259-x 30606156 Widyawati, W., Jans, S., Utomo, S., van Dillen, J., & Janssen, A. L. M. (2015). A qualitative study on barriers in the prevention of anaemia during pregnancy in public health centres: Perceptions of Indonesian nurse-midwives. BMC Pregnancy and Childbirth, 15 (1 ), 1-8. 10.1186/s12884-015-0478-3 25591791 World Health Organization. (2016). WHO recommendations on antenatal care for a positive pregnancy experience. Geneva: World Health Organization. https://www.who.int/publications/i/item/9789241549912 World Health Organization. (2020). World health statistics 2020: Monitoring health for the SDGs, sustainable development goals. Geneva: World Health Organization. https://apps.who.int/iris/handle/10665/332070 Wulandatika, D. (2017). Faktor-faktor yang berhubungan dengan kepatuhan ibu dalam melakukan kunjungan antenatal care di wilayah kerja Puskesmas Gambut Kabupaten Banjar, Kalimantan Selatan tahun 2013 [Factors associated with maternal compliance in antenatal care visits in the work area of Gambur Health Center, Banjar Regency, South Kalimantan 2013]. Jurnal Ilmu Keperawatan dan Kebidanan, 8 (2 ), 8-18. 10.26751/jikk.v8i2.269

PMC010xxxxxx/PMC10353661.txt

==== Front J Pharm Bioallied Sci J Pharm Bioallied Sci JPBS J Pharm Bioall Sci Journal of Pharmacy & Bioallied Sciences 0976-4879 0975-7406 Wolters Kluwer - Medknow India JPBS-15-88 10.4103/jpbs.jpbs_647_21 Original Article Nicotine Free Herbal Composition for Smoking De-Addiction - A Placebo Controlled, Double Blind, Randomized, Multicentric Clinical Study Tamoli Sanjay Motilal 1 Harit Mahesh Kumar 2 Mundhe Narendra 3 Pande Shishir Purushottam 4 Damle Neena 5 Chavan Sheetal 6 Kamde Rahul 3 Pawar Vinay Ankush 5 Mahadik Swapnali 1 1 Target Institute of Medical Education and Research, Jaswanti Allied Business Center, Malad West, Mumbai, Maharashtra, India 2 Department of Maullik Siddhant, Dr. D.Y. Patil Ayurved College and Hospital, Nerul, Navi Mumbai, Maharashtra, India 3 KVTR College Hospital Boradi Shirpur, Maharashtra, India 4 Department of Rasashastra and BK Ayurved Seva Sangh Ayurved Mahavidyalaya Nashik, Maharashtra, India 5 D. Y. Patil School of Ayurveda, Nerul, Navi Mumbai, India 6 Ayurved Seva Sangh, Ayurveda Research Centre, Ganeshwadi, Panchvati, Nashik, Maharashtra, India Address for correspondence: Dr. Sanjay Motilal Tamoli, Target Institute of Medical Education and Research, 402/A-B-C, Jaswanti Allied Business Center, Ramchandra Lane, Off Link Road, Malad (West), Mumbai, Maharashtra, India. E-mail: sanjaytamoli@hotmail.com Apr-Jun 2023 08 6 2023 15 2 8894 22 10 2021 21 1 2023 21 1 2023 Copyright: © 2023 Journal of Pharmacy and Bioallied Sciences 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. ABSTRACT Background: Smoking is a major predisposing factor for many health problems including cancers, vascular disorders, etc., To quit smoking is the only solution to prevent them. Various medicinal and non-medicinal methods are used worldwide for the same. The present study evaluates the effect of a nicotine free herbal formulation containing ingredients like Mucuna pruriens, Withania somnifera, Bacopa monnieri, etc., for cessation of smoking and its effects on other health parameters related to smoking. Materials and Methods: The present study was a placebo controlled, double blind, randomized, and multi-centric clinical study conducted at three clinical sites in India. After ethical approval and informed consent, all participants were given Smotect Tablets or Placebo tablets in a dose of 2 tablets twice daily for 90 days. A total of 103 participants (52 in trial group and 51 in placebo group) completed the study. Evaluation of cessation of smoking was done along with other parameters like measurement of lung capacity, clinical assessment, and laboratory investigations before and after the study. Results: A significant reduction in smoking as well as in the alveolar Carbon monoxide (p < 0.05) and Carboxyhemoglobin levels (p < 0.05) were observed with the use of Smotect tablets as compared to placebo over a period of 90 days. Significant improvement was also observed in quality of life, energy and stamina levels, and reduction of stress level. Smotect tablets were found to be safe without causing any adverse effects. Conclusion: Smotect Tablets is an effective and safe remedy for cessation of smoking and reducing other effects related to smoking. KEYWORDS: CO COHb de-addiction Smotect tablets ==== Body pmcINTRODUCTION All over the world, tobacco has been used, in chewable or smoking forms for euphoria, enjoyment, stimulation, or pleasure.[1,2] India is the second largest manufacturer and consumer of tobacco after China.[3] Tobacco smoking is among the major preventable causes of premature deaths worldwide.[4,5] Smoking is known to be injurious to health causing several health hazards including various respiratory tract diseases, cardiovascular diseases, GI tract diseases, liver diseases, neurological diseases, stroke, and cancer.[6-9] Smoking primarily impacts the respiratory tract causing many respiratory tract diseases, including acute and chronic cough, emphysema, COPD, and bronchitis.[8] Nicotine present in smoke causes physical and psychological dependency and is the major factor for tobacco dependence and addiction.[6,7] As the avoidance of etiological factors is said to be the first line of treatment for any disease, quitting tobacco smoking is the best way to significantly reduce the risk associated with smoking.[7] However, due to the sudden stoppage/quitting smokers can suffer from nicotine withdrawal symptoms like anxiety, irritability, increased eating, dysphoria, hedonic dysregulation, etc.[7] Presently behavioral treatment, Nicotine Replacement Therapy (NRT) and pharmacological treatment are used for smoking cessation.[10,11] Nicotine patches, gums, inhalers, and sprays are some of the options used currently. Some of the quitters also try e-cigarettes that are battery operated devices converting liquid nicotine to vapor. However, NRT has shown to be not a successful option by majority of the smokers. The pharmacological approaches have a beneficial role in alleviating withdrawal symptoms, but they are expensive and possess numerous potential side effects.[12,13] Some quitters also select to go for hypnosis, acupuncture, and counseling. Considering limitations of the treatment options, there is therefore a need to develop a anti-smoking formulation that is free from the existing drawbacks of the anti-smoking formulations without nicotine. Need is also observed to develop formulation that not only helps in de-addiction of smoking but also treats the ailments and/or side effects caused by smoking. Smotect Tablets is an herbal formulation comprising of standardized herbal extracts mainly Mucuna pruriens, Withenia somnifera, etc., These ingredients can be supportive in cessation of smoking and reducing other effects related to smoking. Looking at the activities of the ingredients, the present clinical study was planned. The objective of the study was to evaluate the effect of Smotect Tablets, a Nicotine free herbal formulation for cessation of smoking and its effects on other health parameters related to smoking. MATERIALS AND METHODS Study design, sites – The present study was a placebo controlled, double blind, randomized, multi-centric clinical study conducted at three clinical sites in India, viz; Ayurved Sanshodhan Vibhag, Ayurved Seva Sangh Hospital, Ganeshwadi, Panchvati, Nashik; KVTR Ayurvedic College, Boradi, Dhule-425 428, and D.Y. Patil School of Ayurveda, Sector 7 Nerul, Navi Mumbai, Maharashtra. Ethical considerations- Ethical approvals from Institutional ethics committees of all study centers were obtained. The study was registered on Clinical Trials Registry India (CTRI) vide registration number CTRI/2017/06/008790, dated 08/06/2017. Enrolment of participants- Participants having a history of smoking a minimum of 5 cigarettes daily for at least 3 years and showing a high level of alveolar CO levels attending out-patient department of the study centers were considered for the study. The study was carried out and reported adhering to CONSORT statement [Figure 1]. Study duration & Visits: The total duration of treatment was 3 months (90 days). Patients were asked to visit study site every 30th day for 3 months (90 days). Primary and secondary Outcomes: The primary outcome of the study was to evaluate the efficacy of Smotect Tablets in smokers by assessing cessation of smoking (reduction or complete giving up). Also, changes in Lung capacity were assessed on spirometer. The secondary outcomes of study were to evaluate the efficacy of Smotect Tablets in smokers by assessing changes in levels of alveolar CO and COHb, Quality of life (QOL) on WHO-QOL Questionnaire, changes in the Cardiac risk markers (Apolipoprotein A1 and B), serum cortisol level, change in stress level and anxiety on Hamilton Anxiety Rating Scale (HAM-A scale), change in level of energy, stamina, and physical strength on a 7 point scale, global assessment for overall change by the subject and investigator at the end of the study. Safety and tolerability of study drug were assessed by any occurrence of adverse events (AEs) and adverse drug reactions and change in laboratory parameters such as Liver Function Tests, Renal Function Tests, lipid profile, CBC, ESR, hemoglobin, and urine examinations. Figure 1 CONSORT Figure Selection of participants Male and female participants between 18 to 70 years of age (both inclusive) having a history of smoking a minimum of 5 cigarettes daily for at least 3 years and showing a high level of alveolar CO levels and who gave written consent and were ready to abide to trial procedures included in the study. Participants suffering from any major illnesses, uncontrolled hypertension and diabetes, hepatic or renal impairment, central nervous system disorders, and known hypersensitivity to any ingredient of the study drug were excluded from the study. Participants with continuing history of alcohol and/or drug abuse were excluded from the study. Sample size Sample size calculation assumed that a sample size of minimum 100 evaluable cases (divided into two groups of minimum 50 participants each in placebo and study drug group) would provide an 80% power to estimate cessation of smoking and change in lung capacity at 5% level of significance at the end of the study. Treatment Groups- After fulfilling the eligibility criteria participants were randomized to either trial group or placebo as per computer generated randomization list. Subjects were advised to consume given medication in a dose of two Tablets twice daily orally after meals. Study drug The study was double blinded study and therefore the study product and Placebo Tablets were manufactured and packed in such a way to ensure that the subject and Investigator were blinded from the same. Table 1 provides details of composition of the study product. Unblinding was done at the completion of the study. Placebo Tablets were made using inert materials IP grade—MCC, HPMC, and Talc. Table 1 Composition of Smotect Tablet (each film coated tablet contains) Ingredient Scientific Name Quantity Ashvagandha extract Withania somnifera 100 mg Amla extract Emblica officinalis 50 mg Gokshura extract Tribulus terrestris 50 mg Bramhi extract Bacopa monnieri 50 mg Yashtimadhu extract Glycyrhizza glabra 100 mg Shirisha extract Albezzia lebbek 25 mg Shunthi extract Zingiber officinale 25 mg Lavang extract Syzygium aromaticum 15 mg Kapikacchu extract Mucuna pruriens 250 mg Tulsi extract Ocimum sanctum 50 mg Haridraextract Curcuma longa 50 mg Assessment parameters On baseline visit, participant’s alveolar CO and COHb levels were measured. Participant’s lung capacity was measured using spirometer. After an overnight fasting (10-12 hours), blood samples were collected for laboratory tests viz. CBC, ESR, BSL-Fasting, LFT, RFT, Lipid Profile, Cardiac risk markers (Apo-lipoprotein A1 and B), Total testosterone level and HIV II, urine routine & microscopic & I. Subject’s chest expansion was measured by measuring chest circumference (Axilla and Xiphoid process) and chest diameter (AP and ML). Participants were asked for their average daily consumption of cigarettes on every 15 days interval. Spirometer test and CO and COHb levels were done to evaluate their lung capacity every month. On monthly basis, subject’s Quality of life on WHO QOL, level of energy, level of stamina, physical strength was evaluated on a 7-point scale. Also, level of stress on VAS and level of anxiety on HAMA scale were evaluated and Subject’s chest expansion was measured. All the participants were closely monitored for any adverse event, starting with the baseline visit till the end of the study visit. Plan for statistical analysis All baseline and demographic data were summarized descriptively. All continuous variables were summarized using mean, standard deviation, and standard error of mean and median. All categorical variables were summarized using frequency and percentages. The primary population for this study was Per-Protocol population. GraphPad InStat Version 3.6 (www.graphpad.com) software was used for statistical analysis of data. Comparison of variables representing categorical data was performed using Chi-square test. All other secondary outcomes were analyzed by applying appropriate statistical methods like paired and unpaired t test, etc., All P values were reported based on two-sided significance test and all the statistical tests were interpreted at least up to 5% level of significance. RESULTS Out of 119 recruited participants, 103 participants (52 in trial group and 51 in placebo group) completed the study, while 16 participants dropped out prematurely due to loss to follow-ups [see Figure 1]. All the participants who took even a single dose of study drug were considered for safety evaluation. The average age of participants in trial group was 40.58 ± 14.64 years while the average age of the participants in placebo group was 39.58 ± 13.79 years. There was no significant difference (p > 0.05) in the age between the two groups. Assessment of primary outcome parameters Number of cigarettes smoked per day by the participants: At baseline visit, the average numbers of cigarettes smoked per day in trial group were 9.53 ± 6.42 which reduced significantly (p < 0.05) to 6.67 ± 5.54 at the end of 30 days. At the end of 60 days, the same was 5.25 ± 4.10 and further to 4.00 ± 3.43 at the end of 90 days. In the placebo group, the average number of cigarettes smoked per day reduced from 10.04 ± 6.20 to 8.73 ± 5.97, i.e., insignificant reduction (p > 0.05) at the end of 30 days. At the end of 60 days, the reduction was significant and was observed to be 8.00 ± 5.55 and further to 7.09 ± 4.83 at the end of 90 days. On analysis between the groups, the reduction was significantly more (p < 0.05) in trial group as compared to placebo. The details are given in Table 2. At the end of the study, complete cessation of smoking was observed in 12 (23.08%) participants in trial group and 02 (3.92%) participants in placebo group. The details are given in Table 3. Measurement of lung capacity on Spirometry: Measurement of lung capacity on Spirometry observed that there were no significant changes in FVC, FEV1, and FEV1/FVC in trial group on day 30, 60, and 90 while in the placebo group a significant decrease (p < 0.05) in FVC and FEV1 was observed on day 30, 60, and 90. No significant change was observed on FEV1/FVC value on day 30, 60, and 90 in placebo group. The difference between both the two groups was statistically insignificant (p > 0.05) at all the follow up visits till the end of the study. The details on changes in mean FVC, FEV1, FEV1/FVC ratio are shown in Table 4. There was a significant increase in FEF50% and FEF25-75% in trial group from baseline to the end the study, i.e., 90 days. However, there was no change on these parameters in placebo group from baseline to the end of the study. The difference between the groups was statistically insignificant (p > 0.05) at the end of the study. Table 2 Assessment of Number of Cigarettes Smoked per Day (Cessation of Smoking) Groups Baseline Day 30 Day 60 Day 90 TRIAL Group 9.53±6.42 6.67±5.54, P<0.05 5.25±4.10, P<0.05 4.00±3.43, P<0.05 Placebo Group 10.04±6.20 8.73±5.97, P>0.05 8.00±5.55, P<0.05 7.09±4.83, P<0.05 Between Group Analysis P>0.05 P<0.05 P<0.05 P<0.05 Table 3 Distribution of Participants for Cessation of Smoking % Range for Smoking Cessation TRIAL Group Placebo Group Number of Participants (Total 52) Percentage of Participants Number of Participants (Total 51) Percentage of Participants 100% 12 23.08 02 03.92 81-100% 02 03.83 00 00.00 61-80% 18 34.61 06 11.76 41-60% 10 19.23 08 15.69 21-40% 06 11.53 06 11.76 0-20% 00 00.00 09 17.65 Remained Same 03 05.76 12 23.53 Increased 01 01.92 08 15.69 Total 52 100% 51 100% Table 4 Assessment of Changes in Spirometer Values (FVC, FEV1, FEV1/FVC ratio) Group Parameter Baseline 30 Days 60 Days 90 Days TRIAL Group FVC (in L) 3.40±0.97 3.09±1.40 P>0.05 3.07±1.40 P>0.05 3.15±0.76 P>0.05 FEV1 (in L) 2.43±0.81 2.28±1.19 P>0.05 2.34±1.20 P>0.05 2.44±0.90 P>0.05 FEV1/FVC (in %) 73.86±17.31 74.55±34.41 P>0.05 77.47±33.92 P>0.05 76.13±20.98 P>0.05 Placebo Group FVC (in L) 3.27±0.85 3.01±0.86 P<0.05 2.95±1.01 P<0.05 3.01±0.87 P<0.05 FEV1 (in L) 2.40±0.81 2.19±0.84 P<0.05 2.26±0.89 P<0.05 2.16±0.91 P<0.05 FEV1/FVC (in %) 74.11±16.77 73.75±20.07 P>0.05 78.51±19.21 P>0.05 72.73±21.94 P>0.05 Between group analysis FVC P>0.05 P>0.05 P>0.05 P>0.05 Between group analysis FEV1 P>0.05 P>0.05 P>0.05 P<0.05 Between group analysis FEV1/FVC P>0.05 P>0.05 P>0.05 P>0.05 Assessment of secondary outcome parameters Alveolar CO levels: A significant reduction in the CO levels was observed in trial group from a baseline of 13.33 ± 7.76 ppm to 12.20 ± 9.88 ppm at the end of 30 days. There was a further reduction (p < 0.05) to 10.76 ± 6.90 ppm at the end of 60 days and 10.63 ± 7.70 ppm at the end of 90 days. There was no significant (p > 0.05) change in the CO levels in placebo group from baseline of 14.00 ± 7.51 ppm to 14.84 ± 8.45 at the end of 30 days. Further on day 60 and day 90 as well the change was found to be non-significant (p > 0.05) as the score was 13.8 ± 7.42 and 13.53 ± 7.50, respectively. On analysis between the groups, it was observed that the reduction in CO levels was significantly higher (p < 0.05) in trial Group as compared to placebo. See Graph 1. Alveolar COHb levels: The COHb levels in trial group showed a significant reduction from 2.80 ± 1.25% at baseline levels to 2.37 ± 1.27% at the end of 90 days. In the placebo group, the reduction was non-significant from baseline 2.87 ± 1.19 to the end of the study 2.84 ± 1.21. Between groups analysis showed that consumption of trial significantly reduced COHb levels as compared to placebo. See Graph 2. Quality of life (QOL) on WHO QOL Questionnaire: Assessment of quality of life showed that there was a significant improvement on the physical health score from 19.28 ± 2.44 at baseline visit to 20.49 ± 1.97 on day 90. In placebo group, the mean physical health score showed non-significant change (p > 0.05) from 20.14 ± 2.50 at baseline visit to 20.37 ± 2.29 on day 90. The improvement in trial group was found to be significantly higher as compared to placebo. Also, in trial group, the mean psychological health domain score improved significantly (p < 0.05) from 19.06 ± 2.65 at baseline visit to 19.86 ± 2.37 on day 90, whereas in placebo group, the mean psychological health domain score showed non-significant change (p > 0.05) from 19.33 ± 2.67 at baseline visit to 19.98 ± 2.40 at day 90. The improvement in trial group was found to be significantly higher as compared to placebo. Changes in the Cardiac risk markers (Apolipoprotein A1 and B): There was no significant change in the levels of cardiac risk markers from baseline to the end of the study in both the study groups. The levels remained in the normal physiological range at both the visits. Serum cortisol and total testosterone levels: The mean serum cortisol (μg/dl) and total testosterone (mg/dl) levels were found within normal range in all participants at baseline visit and at the end of the study in both the groups and did not show any significant change from baseline to the end of the study. Change in stress level and anxiety on HAMA scale: The mean stress score assessed on VAS scale reduced significantly (p < 0.05) from 40.58 ± 17.79 at baseline visit to 29.90 ± 15.12 at the end of 90 days in trial group while the score reduced from 42.16 ± 18.01 to 35.32 ± 17.43 in placebo group which was significant (p < 0.05). Though the reduction in stress levels was better in trial as compared to placebo it was found to be non-significant (p > 0.05). Similarly, HAMA score to assess anxiety showed significant reduction in both the groups after 90 days of treatment. However, there was no significant difference between the two groups. There were no significant changes on various parameters to measure chest circumference in both the groups. Change in level of energy, stamina, and physical strength on a 7-point scale: The levels of energy, stamina, and physical strength showed a significant improvement (p < 0.05) from baseline to monthly follow-ups in trial group while the change on these parameters was insignificant (p > 0.05) in placebo group. The improvement on energy, stamina, and physical strength was found to be significant in trial group as compared to placebo. Global assessment for overall change by the subject and investigator at the end of the study treatment: Global assessment for overall change by the physician on CGI-I scale showed that most participants in trial group showed very much to minimal improvement as compared to placebo Also a higher percentage of participants in placebo group either showed no change or worsening of their condition. Graph 1 Assessment of Changes in CO levels (in ppm) Graph 2 Assessment of Changes in COHb levels (in %) Assessment of safety parameters No significant (p > 0.05) changes were observed in laboratory parameters such as CBC, ESR, Hb%, LFTs, RFTs, lipid profile, blood sugar level, and urine examination when compared between baseline visit and day 90 visit in both the study groups. All the laboratory values were within normal range at baseline visit and at the end of the study. No clinically significant change in vitals such as pulse rate, temperature, respiration rate, and blood pressure (systolic and diastolic pressure) was observed from baseline visit to every follow up visits and at the end of the study in both the study groups. AEs including abdominal discomfort, abdominal pain, fever, cough, backache, and body ache were noted during the trial. In the trial group, 19 participants reported 26 AEs and in placebo group 24 participants reported 31 AEs. None of the AE was found to be related to the study product or procedure. No treatment or procedure or interruption was required in both the study groups to resolve these episodes. Almost all the participants showed excellent to good tolerability to the investigational products. DISCUSSION Smoking is considered as the single greatest risk factor that plays role in the incidence of major diseases that cause death due to heart diseases, peripheral vascular diseases, hypertension, lung cancer, diabetes, cancer, etc., There is a wide range of treatment options that have proved effective, including behavioral and pharmacological therapies. The therapies vary widely in their efficacy, their acceptability and it is their cost-effectiveness. Smotect Tablets is an herbal formulation comprising of standardized herbal extracts mainly Mucuna pruriens, Withenia somnifera, etc., which can be supportive in cessation of smoking and reducing other effects related to smoking. It was observed that ninety days treatment with Smotect tablets significantly reduced the mean number of cigarettes smoked per day as compared to placebo. While 23.08% completely gave up smoking in trial group only 4.0% did so in the placebo group. At the end of the study, only 1.92% of participants in trial group showed increase in smoking frequency compared to 16.00% participants in placebo group. Also, treatment with Smotect tablets showed significant reduction in mean CO level (ppm) and COHb (%) levels. The findings suggest that Smotect tablets may help to reduce dependence on smoking and reduce toxic residues in lungs and restore normal function of respiratory system. Apart from reporting a significant reduction in craving for cigarette smoking, participants also reported to reduction in stress level (on VAS), anxiety level (as per HAMA scale) and other symptoms such as insomnia, irritability, nervousness, difficulty concentrating and restlessness and improved quality of life and energy, stamina, and physical strength levels compared to placebo. These findings suggest that Smotect tablets were not only helpful to quit smoking but also helped in alleviating the withdrawal symptoms of tobacco smoking. Smotect tablets is poly herbal combination of 11 standardized herbal extracts, which are helpful to smokers to quit or reduce smoking and to reduce the ill effects and complications of smoking. Mucuna pruriens seed extract contains a high concentration of L-dopa a vital source of dopamine. Neuro-protective action of seeds of Mucuna pruriens has shown to help in restoration of the endogenous monoamine contents including dopamine in the substantia nigra of the brain indicating its dopaminergic action.[14] This dopaminergic action helps to boost energy, elevate mood, and reduce depression. This helps to overcome the urge of smoking over a period and thus helps in cessation of smoking. Withania somnifera (2% withanoloids) is used in patients with nervous exhaustion, insomnia, and debility due to stress.[15] Thus, it helps in withdrawal effects of smoking cessation. Other ingredients possess anti-anxiety, anti-inflammatory, analgesic, immunomodulator, anti-oxidant, rejuvenator, brain tonic, and stimulant properties. Most of the ingredients are useful in various respiratory diseases due to their bronchodilator, anti-inflammatory, anti-allergic, anti-tussive, and mucolytic actions.[16-25] The synergistic action of these ingredients could have helped in cessation of smoking and to overcome other effects related to smoking. Most of the AEs reported in both the groups were unrelated to the study drug. The mean values of almost all laboratory parameters were within normal limits at the end of the study. No significant change in any of the vitals parameters was observed during and at the end of the study. Taken together these observations demonstrated that Smotect tablets are safe to use in smokers. CONCLUSION Three months of treatment with Smotect tablets helps in cessation of smoking. Smotect tablets helps to improve quality of life of smokers along with improvement in the levels of energy, stamina and physical strength, and reduction in levels of stress. Smotect tablets, a nicotine free herbal composition, can be recommended as a safe and effective remedy for de-addiction of smoking along with improving health-related problems arising due to smoking. Further studies with larger sample size are warranted to establish the efficacy of Smotect tablets especially on various lung functions. Financial support and sponsorship The study was sponsored and funded by Project Happiness (Mr. Gurseet Singh). Conflicts of interest There are no conflicts of interest. ==== Refs REFERENCES 1 Bruijnzeel AW Reward processing and smoking Nicotine Tob Res 2017 19 661 2 28486714 2 Kishore K Monograph of tobacco (Nicotiana tobacum) Indian J Drugs 2014 2 5 23 3 Semwal DK Mishra SP Chauhan A Semwal RB Adverse health effects of tobacco and role of Ayurveda in their reduction J Med Sci 2015 15 139 46 4 Mishra S Joseph RA Gupta PC Pezzeck B Ram F Sinha DN Trends in bidi and cigarette smoking in India from 1998 to 2015, by age, gender and education BMJ Glob Health 2016 1 e000005 5 West R Tobacco smoking: Health impact, prevalence, correlates and interventions Psychol Health 2017 32 1018 36 28553727 6 Jindal SK Aggarwal AN Chaudhry K Chhabra SK D’Souza GA Gupta D Tobacco smoking in India: Prevalence, quit-rates and respiratory morbidity Indian J Chest Dis Allied Sci 2006 48 37 42 16482950 7 Benowitz NL Neurobiology of nicotine addiction: Implications for smoking cessation treatment Am J Med 2008 121 S3 10 8 Onor IO Stirling DL Williams SR Bediako D Borghol A Harris MB Clinical effects of cigarette smoking: Epidemiologic impact and review of pharmacotherapy options Int J Environ Res Public Health 2017 14 1147 28956852 9 Saha SP Bhalla DK Whayne TF Jr Gairola C Cigarette smoke and adverse health effects: An overview of research trends and future needs Int J Angiol 2007 16 77 83 22477297 10 Hughes JR Motivating and helping smokers to stop smoking J Gen Intern Med 2003 18 1053 7 14687265 11 Wadgave U Nagesh L Nicotine replacement therapy: An overview Int J Health Sci (Qassim) 2016 10 425 35 27610066 12 Buddhadev SG Buddhadev SS A review article on phytochemical properties of Tamraparna and its traditional uses Int J Herbal Med 2014 2 39 41 13 Sood A Ebbert JO Prasad K Croghan IT Bauer B Schroeder DR A randomized clinical trial of St. John's Wort for smoking cessation J Altern Complement Med 2010 16 761 7 20590478 14 Rana DG Galani VJ Dopamine mediated antidepressant effect of Mucuna pruriens seeds in various experimental models of depression Ayu 2014 35 90 7 25364207 15 Verma SK Kumar A Therapeutic uses of Withania somnifera (Ashwagandha) with a note on withanolides and its pharmacological action Asian J Pharm Clin Res 2011 4 1 4 16 Uddin Q Samiulla L Singh VK Jamil SS Phytochemical and pharmacological profile of withania somnifera dunal: A review J Appl Pharm Sci 2012 02 170 5 17 Dasaroju S Gottumukkala KM Current trends in the research of emblica officinalis (Amla): A pharmacological perspective Int J Pharm Sci Rev Res 2014 24 150 9 18 Prajapati SM Patel BR Phyto-pharmacological perspective of Yashtimadhu (Glycyrrhiza Glabra LINN.) – A review Int J Pharm Biol Arch 2013 4 833 41 19 Verma SC Vashishth E Singh R Kumaril A Meena AK Pant P A review on parts of Albizia lebbeck (L.) Benth. used as ayurvedic drugs Res J Pharm Tech 2013 6 1307 13 20 Chhatre S Nesari T Somani G Kanchan D Sathaye S Phytopharmacological overview of Tribulus terrestris Pharmacogn Rev 2014 8 45 51 24600195 21 Bhakat A Saha S Review on the traditional and contemporary use of Sunthi (Zingiber officinale Rosc.) and its medical importance in Ayurveda Int Ayurvedic Med J 2017 5 3075 81 22 Yadav KD Reddy K Critical review on pharmacological properties of Brahmi Int J Ayurvedic Med 2013 4 92 9 23 Pandey G Madhuri S Pharmacological activities pharmacological activities of ocimum sanctum (Tulsi): A Review Int J Pharm Sci Rev Res 2014 5 61 6 24 Krup V Prakash LH Harini A Pharmacological activities of turmeric (Curcuma longa linn): A review J Homeop Ayurv Med 2013 2 1167 206 25 Kharat RS Lad MD Review of pharmacological activities of haridra (Curcuma longa L.) World J Pharma Res 2016 3 412 23

PMC010xxxxxx/PMC10353662.txt

==== Front J Pharm Bioallied Sci J Pharm Bioallied Sci JPBS J Pharm Bioall Sci Journal of Pharmacy & Bioallied Sciences 0976-4879 0975-7406 Wolters Kluwer - Medknow India JPBS-15-81 10.4103/jpbs.jpbs_646_22 Original Article 3-Aminophenylboronic Acid Conjugation on Responsive Polymer and Gold Nanoparticles for Qualitative Bacterial Detection Wikantyasning Erindyah Retno 1 Da’i Muhammad 1 Cholisoh Zakky 1 Kalsum Ummi 2 1 Faculty of Pharmacy, Universitas Muhammadiyah Surakarta, Jl. A. Yani 157, Sukoharjo, Indonesia 2 Study Program of Pharmacy, STIKES Telogorejo, Jl. Yos Sudarso, Semarang, Jawa Tengah, Indonesia Address for correspondence: Dr. Erindyah Retno Wikantyasning, Faculty Pharmacy, Universitas Muhammadiyah Surakarta, Jl. A. Yani 157, Pabelan, Kartasura, Sukoharjo, 57169, Indonesia. E-mail: erindyah.rw@ums.ac.id Apr-Jun 2023 08 6 2023 15 2 8187 13 12 2022 07 3 2023 08 4 2023 Copyright: © 2023 Journal of Pharmacy and Bioallied Sciences 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. ABSTRACT Background: Because of their sensitive and selective responses to a wide variety of analytes, colorimetric sensors have gained widespread acceptance in recent years. Gold nanoparticles (AuNPs) are widely employed in visual sensor strategies due to their high stability and ease of use. Combining AuNPs with a responsive polymer can result in distinct surface plasmon resonance (SPR) changes that can be utilized as colorimetric biosensors. Objectives: The purpose of this research is to develop a colorimetric-based sensor through the utilization of the optical properties of gold nanoparticles (AuNPs) crosslinked with pH-responsive polymers poly (acrylic acid) (PAA) conjugated to 3-aminophenyl boronic acid (APBA). Methods: The polymer (PAA) was synthesized via RAFT polymerization. The inversed Turkevic method was used to produce AuNPs, which were subsequently used in a self-assembly process using poly (acrylic acid)-aminophenyl boronic acid (PAA-APBA) to create the self-assembled AuNPs-APBA-PAA. The particle size, zeta potential, and reversibility of the polymer-modified gold nanoparticles were determined using a transmission electron microscope (TEM), a particle size analyzer (PSA), and an Ultraviolet-Visible spectrophotometer (UV-Vis spectrophotometer). Visual, UV-Vis spectrophotometer and TEM observations confirmed the system’s ability to identify bacteria. Statistical analysis was performed using a one-way analysis of variance using Excel software. Results: Using UV-Vis spectrophotometry, the particle size of AuNPs was determined to be 25.7 nm, and the maximum absorbance occurred at 530 nm. AuNPs PAA APBA colloid exhibited an absorbance maximum of 532 nm, a zeta potential of -41.53, and a pH transition point between 4 and 5. At E. coli concentrations of 4.5 x 107 CFU/mL, the color of the system sensors changed from red to blue after 15 hours of incubation, whereas at S. aureus concentrations of 1.2 x 109 CFU/mL, the color changed to purple immediately after mixing. The TEM confirmed that the detection mechanism is based on the boronate-polyol bonding of saccharides on the outer membranes of Escherichia coli and Staphylococcus aureus. Conclusions: The use of APBA in conjunction with pH-responsive PAA polymers containing AuNPs to detect E. coli and S. aureus bacteria induces a maximum wavelength transition, followed by a color change from red to blue. By the process of de-swelling of the responsive polymer, which induces the aggregation of the AuNPs, the established sensor system is able to alter the color. The conjugated polymer and gold nanoparticle-based sensor system demonstrated a promising method for bacterial detection. KEYWORDS: Bacteria boronic acid colorimetric sensors gold nanoparticles sensitive polymer ==== Body pmcINTRODUCTION Bacterial infectious diseases account for roughly one-third of all deaths worldwide. Accurate and timely detection of bacterial pathogens can result in immediate and effective measures to prevent or control outbreaks of the pathogens. Many of the current methods for identifying pathogenic bacteria, however, require at least 18 hours between sampling and results. This lengthy response time necessitates the development of rapid and reliable methods for detecting bacteria in environmental, food, and water samples.[1,2] Nanotechnology is now playing an important role in the development of biosensors. Nanotechnology-based biosensors are being developed for use in the health industry to protect against pathogenic contamination.[3] Because of their high stability and ease of use, gold nanoparticles (AuNPs) are widely used in visual sensor strategies. An optical signal and color transition can be produced by combining metal nanoparticles and a responsive polymer. Surface plasmon resonance (SPR), a unique property of metal nanoparticles, can show different changes in SPR by increasing particle size, composition, distance, or surrounding media, followed by visible color changes.[4,5] Colorimetric sensors based on nanoparticles and pH-responsive polymers have received a lot of attention. Previous research reported a pH-responsive polymer poly (4-vinyl pyridine) functionalized with graphene oxide and AuNPs to induce absorption and detection of negatively charged analytes at low pH, allowing it to be used as a sensor.[6] In addition, AuNPs can be used as sensors for the detection of Escherichia coli bacteria by conjugating cysteine with polymers that are pH-responsive poly (acrylic acid).[7] Colorimetric detection with AuNPs colloids can be accomplished through the conjugate binding on polymers triggered by a target to increase sensor sensitivity. The ability of phenylboronic acids to form covalent bonds with diol derivatives such as polysaccharides, dopamine, sugar, nucleosides, antibodies, and glycoproteins is well known.[8] Many bacterial surfaces contain diol molecules and boronic acids containing lipopolysaccharides (LPS), which have been used as identification molecules for bacterial detection. The majority of the boronic acid can be absorbed and aggregated on the surface of bacterial cells. AuNPs-conjugated APBA can be absorbed on the surface of E. coli at a maximum wavelength of 652 nm and increases with increasing pH changes, but decreases at pH more than 5, and the optimal pH value for absorption is 4.5.[9,10] The isoelectric point value of E. coli is 4.4.[11] Another study reported that APBA can be conjugated with pH-responsive polymers by using EDC/NHS reagents.[12] The conjugation of boronic acid against poly (lysine) and alginate, which coat AuNPs, can be exploited to detect diol groups from glucose. The sensor system can detect glucose with a sensitivity of 0.5 μg/mL, where a poly (lysine) bond occurs with the phenylboronic acid group through boronate-diol interactions.[8] We investigated the use of colorimetric sensors based on AuNPs and pH-responsive polymers, poly (acrylic acid), conjugated with 3-aminophenyl boronic acid (APBA), to detect E. coli and Staphylococcus aureus bacteria. MATERIALS AND METHODS 1. Materials All of the materials were purchased from Sigma-Aldrich: hydrogen tetrachloroaurate (III) hydrate, acrylic acid, tribasic sodium citrate, tetrahydrofuran, hydrochloric acid, azobisisobutyronitrile (12 wt. % in acetone), 3-aminophenyl boronic acid, and sodium hydroxide. All materials were used as purchased. 2. Synthesis of Gold Nanoparticles (AuNPs) As previously stated,[12] AuNPs were synthesized via chemical reduction. In 24 mL of aqua pro injection, 1% sodium citrate (0.18 mM) was dissolved. Then, using a stirrer, 1 mL of 6,5 mM HAuCl4.xH2O was added to the boiling sodium citrate solution while stirring at 500 r/min. The solution turned bright red, and stirring was continued for 15 minutes or until the color was stable. To characterize the optical properties and morphology of AuNPs, UV-Vis spectrophotometry (Genesys 10S) and Transmission Electron Microscopy (JEOL JEM-1400) were used. 3. Synthesis of APBA-conjugated PAA The polymer (PAA) was synthesized using the previously synthesized the reversible addition-fragmentation chain (RAFT) agents[7] via the RAFT polymerization method. The polymer was then dialyzed against water for 2 days and lyophilized for 24 hours. 1H-Nuclear Magnetic Resonance (NMR) spectroscopy was utilized for the characterization of the polymer (Jeol Resonance 400 MHz). PAA-APBA was synthesized using the method previously described. Around 15 mg of PAA was dissolved in Phosphate Buffer Saline (PBS) buffer with a pH of 7.4 (5 mL; 10 mM). The PAA solution is then placed in an ice bath. While stirring for 1 hour, EDC (25.76 mM; 20 mg) and Sulfo-NHS (9.21 mM; 10 mg) were added to the solution. APBA (25.8 mM; 20 mg) was then added to the solution, which was stirred for 2 hours in an ice bath before being left at room temperature for 22 hours. The resulting solution was centrifuged for 5 minutes at 5000 rpm. The obtained supernatant was mixed with 15 mL of absolute ethanol. A few minutes were allowed to pass before the mixture was centrifuged at 10,000 rpm for 5 minutes. The resultant PAA-APBA was washed three times with absolute ethanol and then vacuum-dried.[13] Fourier transform infrared (FT-IR) spectroscopy confirmed the successful conjugation of APBA onto PAA (ATR Perkin Elmer). 4. Preparation of sensor system of AuNPs-PAA-APBA Around 25–125 μL of a 1% w/v PAA-APBA solution were added to 1 mL of the AuNPs solution. The solution was then characterized with a UV-Vis spectrophotometer (Genesys 10) for its visible spectra, pH transition point, and color shift when exposed to different pHs. Particle Size Analyzer was also used to measure the zeta potential of the combined solution (Horriba Scientific SZ 100). 5. Qualitative detection of E. coli and S. aureus bacteria in vitro Bacterial concentrations were prepared in the range of 9 × 103–4.5 × 108 and 1.2 × 102–1.2 × 108 for E. coli and S. aureus, respectively. The bacteria solution was added to the AuNPs-PAA-APBA solution and incubated at room temperature. Using a UV-Vis spectrophotometer, the visible spectra were recorded as the system’s color changed (Genesys 10), and the binding of AuNPs-PAA-APBA to the surface of bacteria was observed using TEM (JEOL JEM-1400) right after the color change. RESULTS AND DISCUSSION 1. Synthesis and characterization of AuNPs and conjugated of PAA-APBA Figure 1 depicts the characteristics of the AuNPs synthesized using the inversed Turkevic method. As the plasmon band is correlated with the size and shape of nanoparticles, UV-Vis spectroscopy has been widely used for the initial characterization of the optical and electrical properties of AuNPs.[12] The red hue of AuNPs was due to their 530 nm SPR band [Figure 1a]. TEM analysis reveals that the reduction of gold chloride solution by trisodium citrate solution yielded spherical AuNPs with average diameters of 25.7 ± 2.68 nm [Figure 1b]. Due to the ease of manufacturing, spherical AuNPs in the size range of 13–20 nm with an absorbance peak around 520 nm have been utilized most frequently in biosensors.[14] In this research, we synthesized bigger particle sizes in order to increase the sensitivity of the colorimetric sensor. Figure 1 (a) The spectra of AuNPs solution showed an SPR peak at 530 nm (b) The image of AuNPS recorded using TEM Using APBA-conjugated PAA, a sensor was developed for the detection of the bacterial outer membrane. APBA was responsible for binding the diol component of bacterial outer membranes.[13] This study utilized the RAFT method to produce a pH-responsive polymer from PAA. This polymer has thiol groups on both sides, which, due to their high affinity for the surface of noble metal nanoparticles, can be used as a capping agent. Figure 2 presents the result of 1H-NMR of PAA, which revealed that polymers peak (δ) at 2.5–3 ppm while monomers peak (δ) at 5.8–6.3 ppm.[15] Figure 2 1H-NMR of poly (acrylic acid) in CDCl2 PAA-APBA was produced by conjugating APBA and PAA with EDC/NHS in pH 7.4 PBS buffer saline. Figure 3 depicts the successful conjugation of APBA to PAA as determined via FT-IR. Strong bands at 3307.18 cm-1 and 1554.18 cm-1 in the FTIR spectrum of PAA-APBA demonstrated the formation of an amide bond. The –B (OH)2 band at 1338.21 cm-1 and the para-substituted benzene ring proton at 796.20 cm-1 in the PAA spectrum and 797.54 cm-1 in the PAA-APBA spectrum confirmed the presence of APBA in PAA-APBA.[16] Figure 3 FT-IR spectra of PAA and PAA-APBA 2. Optical Properties and Characterization of AuNPs-PAA-APBA The AuNPs-PAA-APBA solution was created by combining 1 mL of AuNPs colloids with a 1% polymer PAA-APBA. Due to the affinity of the thiol group (-SH) in PAA, AuNPs crosslinked to PAA-APBA were formed by self-assembly in the trithiocarbonate groups on both sides of the polymer.[17] System solution exhibits a property, which is the solution’s color change in response to changes in pH. In an acidic state, the solution appeared blue; in an alkaline state, the color changed to red. The color change was clarified by the swelling and deswelling phenomenon of PAA, which can cause the aggregation and disaggregation of AuNP in response to environmental conditions.[18] Boronic acids have been widely used in the field of biomaterials due to their ability to bind with biologically significant 1,2- and 1,3-diols, such as saccharides and peptidoglycans, or with polyols to prepare a complex covalent or sensitive system. APBA is an amino-substituted aryl boronic acid that binds to the diol group in a bacterial saccharide. The presence of OH groups in the APBA conjugate influences the increasing negative charge in the Au-PAA-APBA system.[19,20] The negative charge of Au-PAA-APBA can interact with the outer membrane of bacteria by forming bonds. As shown in Table 1, AuNPs-PAA-APBA was characterized by measuring the zeta potential of each system. Table 1 Zeta potential of the different systems System Zeta Potential (mV) AuNPs -31.40±4.51 Au-PAA -31.53±0.21 Au-PAA-APBA -41.53±0.96 The system solutions for AuNPs-PAA-APBA were synthesized. Figure 4 shows the spectra of the synthesized solution. The absorbance maximum of the PAA-APBA spectrum crosslinked by AuNPs in solution was observed at 532 nm. Figure 4 Spectra of the different systems The color transition behavior of AuNPs-PAA-APBA was observed with a visible spectrometer and the naked eye [Figure 5]. The advantage of the colorimetric approach is that its results can be detected qualitatively or semi-qualitatively with the naked eye, which is ideal for point-of-care testing.[21] The polymer’s pH transition point is between pH 4 and 5. As the pH increased, the polymer began unionizing and undergoing a phase change from a swollen to a contracted state, resulting in the aggregation of AuNPs. This transition can be observed as the solution changes from red to blue in color. At pH 12, the AuNPs-PAA-APBA spectrum exhibited a sharp plasmon band with a peak at 539 nm, whereas a decrease in pH resulted in a redshift at 700 nm due to the coupling of the plasmon band in nearby particles. The pKa transition of PAA can explain the observed color transition point at pH 5.[22] Below the pKa of PAA, the average inter-particle distance of AuNPs in solution is several times smaller than the diameter of AuNPs. Because of the SPR of aggregated particles, the solution appeared blue. Upon increasing the pH above the pKa of the PAA, the polymer swells, causing the AuNPs to become more distinguishable from a distance and causing the color to change from blue to red.[23,24] Figure 5 pH Transition Point of AuNPs-PAA-APBA The reversibility of the AuNPs-PAA-APBA solution was investigated using UV-visible spectroscopy and naked-eye observation at a given pH [Figure 6]. The solution’s color change from red to blue is reversible and repeatable up to five times. These color modifications were a result of the swelling and shrinking of the responsive polymer that crosslinked the AuNPs and modulated the disaggregation of the AuNPs when exposed to varying pH levels. Figure 6 The reversibility of AuNPs-PAA-APBA solution in pH 2 and 12 (5 cycles, three replication) 3. Qualitative Detection of E. coli and S. aureus using Sensor System The sensors of the AuNPs-PAA-APBA system were then used to detect E. coli and S. aureus bacteria [Figure 7]. At the concentrations of 4.5 × 107; 9 × 107 and 2.25 × 108 CFU/mL of E. coli, the color of the system sensors changed from red to blue after 15 hours of incubation, whereas at the concentration of 12 × 108 CFU/mL of S. aureus, the color changed to purple in 0 hour or immediately after mixing. After incubation with E. coli and S. aureus bacteria, spectrophotometry revealed a 2–5 nm shift in the SPR peaks of the AuNPs-PAA-APBA system [Figure 8]. The color change of the sensor system is a qualitative technique to confirm the presence of bacteria. Previous studies reported that the reaction of thiol groups on the surface of phenylboronic acid-modified AuNPs with target analytes could result in the aggregation of AuNPs accompanied by color changes.[25] Figure 7 Schematic representation of the color change of AuNPs-PAA-APBA in the presence of E. coli and S. aureus bacteria Figure 8 The SPR peaks of the AuNPs-PAA-APBA system after incubation with different concentrations of (a) E. coli and (b) S. aureus TEM was used to observe the interaction between the AuNPs-PAA-APBA and the bacteria [Figure 9]. The images revealed an accumulation of nanoparticles surrounding the bacteria, which may have caused the solution’s color change. Cell walls exhibit aggregate formation. The color change from red to blue in the sensor system is caused by the aggregation of Au-PAA-APBA on the bacterial cell wall. Boronic acid can bind to different sugars in gram-negative and gram-positive bacteria’s LPS and peptidoglycan.[26] Figure 9 The morphological observations of the AuNPs-PAA-APBA aggregation system for S. aureus bacteria on the magnification of 5000 (a) and 20,000X (b), and E. coli on the magnification of 5000 (c) and 20,000X (d). Aggregation can be seen in cell walls and open walls of bacterial cells A number of polyols in sugars are essential components of Gram-negative bacteria. LPS are found in the outer membrane of Gram-negative bacteria.[27] LPS consists of a lipid, a polysaccharide composed of O-antigen, and an outer core and inner core joined by a covalent bond. Before the sensor solution can diffuse onto the surface of a microbial cell, the LPS and the outer membrane must be traversed. The O-antigen is bound to the central oligosaccharide and contains the outermost domain of the LPS molecule. Before reaching the plasma membrane of Gram-positive bacteria, the sensor solution must first penetrate peptidoglycan. Peptidoglycan is a polymer composed of sugars and amino acids that forms the cell wall beyond the plasma membrane of the majority of bacteria. Gram-positive bacteria have a significantly thicker peptidoglycan layer than Gram-negative bacteria. In addition, various sugar groups are associated with d-alanine peptides containing positively charged amine groups in S. aureus peptidoglycan, causing boronates to bind more strongly to S. aureus peptidoglycans. The difference in charge between these bacteria and the negatively charged Au-PAA-APBA sensor permits differences in the time detection system. It demonstrates that the system is more sensitive to gram-positive bacteria like S. aureus.[28,29] CONCLUSIONS In order to detect E. coli and S. aureus bacteria, APBA is used in conjunction with pH-responsive PAA polymers crosslinked AuNPs to cause a maximum wavelength transition, followed by a color change from red to blue. The established sensor system can modify color through the process of deswelling the responsive polymer, which induces the aggregation of the AuNPs. Conjugated polymer and gold nanoparticle-based sensor systems have shown promise for detecting bacteria. Financial support and sponsorship RISTEKDIKTI, the Indonesia Government. Conflicts of interest There are no conflicts of interest. Acknowledgement The authors acknowledge the support of the Indonesian Ministry of Research, Technology and Higher Education (Grant No. 133.4/A.3-III/LPPM/IV/2020). ==== Refs REFERENCES 1 Wardani DL Setiyaningrum Z Identification of escherichia coli bacteria in sauce of snacks sold around the campus of Universitas Muhammadiyah Surakarta Journal of Health 2019 12 91 101 2 Putri A Kurnia P Identification of coliform bacteria and the total mikrobes in dung-dung ice around Universitas Muhammadiyah Surakarta Campus National Nutrition Journal 2018 13 41 8 3 Dewan N Ahmed P Chowdhury G Pandit S Dasgupta D Nanotechnology based biosensors and its application Pharma Innov 2016 5 18 25 4 Kvasnička P Homola J Optical sensors based on spectroscopy of localized surface plasmons on metallic nanoparticles: Sensitivity considerations Biointerphases 2008 3 FD4 11 20408699 5 Sener G Uzun L Denizli A Colorimetric sensor array based on gold nanoparticles and amino acids for identification of toxic metal ions in water ACS Appl Mater Interfaces 2014 6 18395 400 25330256 6 Yao A Fu Q Xu L Xu Y Jiang W Wang D Synthesis of pH-responsive nanocomposites of gold nanoparticles and graphene oxide and their applications in SERS and catalysis RSC Adv 2017 7 56519 27 7 Wikantyasning ER Mutmainnah M Cholisoh Z Hairunisa I Bakar MFA Da’i M Preparation of hydrogel nanocomposite containing gold nanoparticles with unique swelling/deswelling properties Rasayan J Chem 2019 12 1857 63 8 Mansour O Peker T Hamadi S Belbekhouche S Glucose-responsive capsules based on (phenylboronic-modified poly (lysine)/alginate) system Eur Polym J 2019 120 109248 doi: 10.1016/j.eurpolymj.2019.109248 9 Zheng L Qi P Zhang D A simple, rapid and cost-effective colorimetric assay based on the 4-mercaptophenylboronic acid functionalized silver nanoparticles for bacteria monitoring Sens Actuators B Chem 2018 260 983 9 10 Liu L Xia N Liu H Kang X Liu X Xue C Highly sensitive and label-free electrochemical detection of microRNAs based on triple signal amplification of multifunctional gold nanoparticles, enzymes and redox-cycling reaction Biosens Bioelectronic 2013 53 399 405 11 Su H Zhao H Qiao F Chen L Duan R Ai S Colorimetric detection of Escherichia coli O157: H7 using functionalized Au@Pt nanoparticles as peroxidase mimetics Analyst 2013 138 3026 31 23577341 12 Afrapoli ZB Majidi RF Negahdari B Tavoosidana G ‘Inversed Turkevich’ method for tuning the size of Gold nanoparticles: Evaluation the effect of concentration and temperature Nanomed Res J 2018 3 190 6 13 Sang L Wang H Aminophenylboronic-acid-conjugated polyacrylic acid-Mn-doped ZnS quantum dot for highly sensitive discrimination of glycoproteins Anal Chem 2014 86 5706 12 24854708 14 Verma M Rogowski J Jones L Gu F Colorimetric biosensing of pathogens using gold nanoparticles Biotechnol Adv 2015 33 666 80 25792228 15 Zeinali E Haddadi-Asl V Roghani-Mamaqani H Nanocrystalline cellulose grafted random copolymers of N-isopropylacrylamide and acrylic acid synthesized by RAFT polymerization: Effect of different acrylic acid contents on LCST behavior RSC advances 2014 4 31428 42 16 Liu Y Zhang Y Zhao Y Yu J Phenylboronic acid polymer brush-enabled oriented and high density antibody immobilization for sensitive microarray immunoassay Colloids Surf B Biointerfaces 2014 121 21 6 24929524 17 Zhang H Nayak S Wang W Mallapragada S Vaknin D Interfacial self-assembly of polyelectrolyte-capped gold nanoparticles Langmuir 2017 33 12227 34 28985464 18 Liu X-Y Cheng F Liu Y Li WG Chen Y Pan H Thermoresponsive gold nanoparticles with adjustable lower critical solution temperature as colorimetric sensors for temperature, pH and salt concentration J Mater Chem 2010 20 278 84 19 Dordovic V Vojtová J Jana S Uchman M Charge reversal and swelling in saccharide binding polyzwitterionic phenylboronic acid-modified poly (4-vinylpyridine) nanoparticles Polym Chem 2019 10 5522 33 20 Wang Y Chai Z Ma L Shi C Shen T Song J Fabrication of boronic acid-functionalized nanoparticles via boronic acid-diol complexation for drug delivery RSC Adv 2014 4 53877 84 21 Wei S Su Z Bu X Shi X Pang B Zhang L On-site colorimetric detection of Salmonella typhimurium NPJ Sci Food 2022 6 1 8 doi: 10.1038/s41538-022-00164-0 35017542 22 Swift T Swanson L Geoghegan M Rimmer S The pH-responsive behaviour of poly (acrylic acid) in aqueous solution is dependent on molar mass Soft Matter 2016 12 2542 9 26822456 23 Chen Z Zhang C Tan Y Zhou T Ma H Wan C Chitosan-functionalized gold nanoparticles for colorimetric detection of mercury ions based on chelation-induced aggregation Microchim Acta 2015 182 611 6 24 Mortezaei M Dadmehr M Korouzhdehi B Hakimi M Ramshini H Colorimetric and label free detection of gelatinase positive bacteria and gelatinase activity based on aggregation and dissolution of gold nanoparticles J Microbiol Methods 2021 191 106349 doi: 10.1016/j.mimet.2021.106349 34699865 25 Li R Gu X Liang X Hou S Hu D Aggregation of gold nanoparticles caused in two different ways involved in 4-mercaptophenylboronic acid and hydrogen peroxide Materials 2019 12 1802 doi: 10.3390/ma12111802 31163635 26 Yang P Bam M Pageni P Zhu T Chen YP Nagarkatti M Trio act of boronolectin with antibiotic-metal complexed macromolecules toward broad-spectrum antimicrobial efficacy ACS Infect Dis 2017 3 845 53 28976179 27 Qiao S Luo Q Zhao Y Zhang X Huang Y Structural basis for lipopolysaccharide insertion in the bacterial outer membrane Nature 2014 511 108 11 24990751 28 Abbaszadegan A Ghahramani Y Gholami A Hemmateenejad B Dorostkar S Nabavizadeh M The effect of charge at the surface of silver nanoparticles on antimicrobial activity against gram-positive and gram-negative bacteria: A preliminary study J Nanomater 2015 16 53 29 Gross M Cramton SE Friedrich G Peschel A Key role of teichoic acid net charge in Staphylococcus aureus colonization of artificial surfaces Infect Immun 2001 69 3423 6 11292767

PMC010xxxxxx/PMC10353663.txt

==== Front J Pharm Bioallied Sci J Pharm Bioallied Sci JPBS J Pharm Bioall Sci Journal of Pharmacy & Bioallied Sciences 0976-4879 0975-7406 Wolters Kluwer - Medknow India JPBS-15-68 10.4103/jpbs.jpbs_96_23 Review Article Past, Present, and Likely Future of Nutraceuticals in India: Evolving Role of Pharmaceutical Physicians Malve Harshad 1 Bhalerao Pramod 2 1 Freelance Medical Affairs and Clinical Research Consultant, Nashik, Maharashtra, India 2 Department of Pharmacology, D. Y. Patil Medical College Kolhapur, Maharashtra, India Address for correspondence: Dr. Harshad Malve, 6/2, Aabhish Nagar, Shivaji Nagar, Opposite Inox, Nashik - 422 006, Maharashtra, India. E-mail: dr.harshad.malve@gmail.com Apr-Jun 2023 08 6 2023 15 2 6874 01 2 2023 22 2 2023 04 3 2023 Copyright: © 2023 Journal of Pharmacy and Bioallied Sciences 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. ABSTRACT Nutraceuticals are gaining importance owing to the current pandemic situation and increasing focus on overall health. Nutraceuticals include products, which help in maintaining immunity and prevent diseases. It also includes products that support the optimal functioning of the human body. Poor nutrition plays an important role in lifestyle-related disorders as well. Various nutraceuticals have exhibited therapeutic potential, hence gaining popularity. Nutraceuticals are mainly categorized into dietary supplements and functional foods. With multiple factors contributing to the growth of nutraceutical industry in India, we are marching toward global leadership in nutraceuticals. Food Safety and Standards Authority of India (FSSAI) is solely responsible for regulating the approvals, promotions, and labeling standards for health supplements and nutraceuticals. As the understanding of these nutraceuticals is improving, the regulations are becoming stricter and there is a pressing need to monitor the usage of such products regularly and stringently. Nutrivigilance and phytovigilance are relatively new concepts in our country; however, regulatory authorities need to proactively observe the adverse effects and issues related to substandard and counterfeit nutraceutical products. Healthcare professionals including pharmaceutical physicians can play an important role in safeguarding the population by advocating the rational use of nutraceuticals, food supplements, and consumer health products. KEYWORDS: Consumer health dietary supplements food safety functional foods medical affairs nutrivigilance pharmaceutical physician public health ==== Body pmcINTRODUCTION “Let food be thy medicine and medicine be thy food” was quoted by Hippocrates in 400 BC, which holds true in today’s world as well. The importance of food and nutrition in disease pathophysiology and prevention was never in doubt. Significance of right nutrition is ever increasing especially with the increasing trend of lifestyle-related diseases and novel infectious diseases these days. Focus of public health is moving from “curative” intent to “preventive” one. Although patients are pondering over the importance of food in both causing and relieving their problems, clinicians’ knowledge of nutrition is still limited. Most feel much more comfortable with drugs than foods, and the “food as medicine” philosophy of Hippocrates has been largely neglected. However, that attitude is changing over the last decade and the focus on “nutraceuticals” is ever-rising. The word nutraceuticals contains “nutrition” and “pharmaceuticals.” The term “nutraceutical” was initially devised by Dr. Stephen DeFelice, originator of the Foundation of Innovation Medicine, Crawford, New Jersey, in 1989.[1] Now, the old proverb “an apple day keeps a doctor away” has become “a nutraceutical a day keeps the doctor away.” The idea of nutraceuticals has evolved in the last three decades.[2] Considering the growing importance of nutraceuticals, we present this narrative review of its usage, rationale, growth potential, regulatory scenario, and future challenges with the ever-increasing use of nutraceuticals in India. We also explored the expanding responsibility of pharmaceutical physicians to regulate the rational use of nutraceuticals, its promotion, and vigilance for adverse events due to such products. Nutraceuticals and rationale for its use Dietary components assume critical importance in health and disease pathophysiology. Most of the lifestyle-related disorders are related to diet, for example, obesity, hypertension, and diabetes. Poor nutritional status leads to poor immunity making the individual susceptible to various infections. Nutritional deficiencies are also common in developing countries such as India, which has led to the development of national healthcare programs for diseases such as anemia, goiter, and blindness. Fortification of table salt with iodine and wheat flour with folic acid has been utilized to counter iodine deficiency and iron deficiency for a long time in our country. The results of these healthcare programs highlighted the importance of fortified foods, nutraceuticals, and their role in well-being. With advances in science and innovative technologies, new ways of enriching food have evolved. Now, our understanding of nutraceuticals and their role as preventive and therapeutic treatments has much improved.[3,4] The government of India has focused on improving the nutritional status of its citizen. The coronavirus disease 2019 (COVID-19) pandemic situation has highlighted the interdependence of nutritional status and economic status of our country.[5] The lessons from the pandemic are likely to enhance the attention and importance of nutrition in public health programs, healthcare strategies, and schemes than ever before to prevent disease. This pandemic has also led to a shift in consumer behavior with the focus on health and well-being. The focus on enhancing one’s immunity has increased significantly. It augurs well coupled with rich heritage of herbal and Ayurveda medicine that our country has. With agriculture being one of the biggest industries in India, there is abundant availability of raw materials and ingredients for nutraceuticals and food supplements.[5] Furthermore, there are other innovative sources such as marine environment being explored for nutraceuticals and health supplements.[6] There is enough skilled manpower, manufacturing expertise, and excellence, which can fast-track the growth of nutraceutical industry in India. Indian nutraceutical market — A breakthrough success story The nutraceutical market in India is evolving and is estimated to reach USD 18 billion by the end of 2025 as compared to USD 4 billion by end of 2020. So it is more than four times growth in a span of five years.[7,8] The interest of foreign investors in the nutraceutical market has gone up significantly due to the opening of 100% Foreign Direct Investment (FDI) in the nutraceutical and food supplement manufacturing sector, and such companies can also sell their products through wholesale, retail, or e-commerce platform. Thus, the FDI has increased from USD 131.4 million in 2012 to USD 584.7 million in 2019.[7,8] With such phenomenal growth, India is poised to be a global leader in nutraceuticals. Regulation of nutraceuticals in India Currently, the Food Safety and Standards Authority of India (FSSAI) regulates the standards for health supplements and nutraceuticals. The FSSAI defined regulatory guidelines for the approval of nutraceuticals in India. These guidelines cover eight categories of functional foods, as given in Figure 1. To avoid confusion between categories of nutraceuticals, FSSAI issued a guidance note on August 28, 2020. It clarifies that health supplements are intended to supplement the normal diet of a person with one or more nutrients with known benefits. Foods for special dietary use (FSDU) are specially processed or formulated to satisfy dietary requirements for specific ailments. Foods for special medical purposes (FSMP) are intended for exclusive or partial feed to people with digestive issues.[9] Nutraceuticals are naturally occurring ingredients that are extracted, isolated, and purified from food or nonfood sources, which when consumed provide physiological benefits and maintain the good health of the recipients. Health supplements and nutraceuticals are targeted for the healthy population with age above five years and above two years, respectively. However, the FSDU and FSMP are for people with specific requirements with age above two years.[9] Figure 1 Classification of food products The Food Safety and Standards Act (FSSA) of 2006 consists of 21 chapters, and in that, the fourth article of the act regulates nutraceuticals, dietary supplements and various functional foods, and their production, manufacturing, marketing, sale, distribution, and import. For a considerable length of time, Food and Drug Administration (FDA) controlled dietary supplements as sustenance to guarantee that they were protected and that their marking was honest and not deluding. In 2006, the Indian government passed FSSA to incorporate and streamline the numerous controls covering nutraceuticals, fortified food, and dietary supplements.[10] FSSAI act was passed by the parliament in 2006, but FSSAI was implemented in 2008. On September 5, 2008, the government of India advised the foundation of the FSSAI as a top administrative expert, comprising a chairperson and 22 individuals.[11] In 2011, FSSAI released the food safety and standard (FSS) rules. These rules included the regulations for licensing and registration of food products, food business, packaging and labeling methods, standards for food products, and additives used in the food product. The FSS rules were put into practice in August 2011. FSS regulations in 2015 provided guidelines regarding nutraceuticals along with food/health supplements, foods with special dietary purposes, medicinal purposes, and novel and functional foods. As per these regulations, nutraceuticals shall contain any of the ingredients specified in the “Food Act Schedule” as given in Table 1.[12] As per these rules, the following information should be included in any claimed novel foods: chemical composition of the engineered food, surface modification/surface chemistry, primary particle size, solubility, digestibility, amount of nanomaterial if any in the food product, and specific claim, if applicable.[12] Table 1 Schedule wise ingredients as per FSSA 2016 Schedule Ingredients I Vitamins and minerals II Essential amino acids, Non-essential amino acids and Nucleotides III Values for vitamins, minerals and trace elements allowed to be used in food for special dietary use and food for special medical purpose IV Plants and botanical ingredients V Food additives VI Nutraceuticals VII Probiotics VIII Prebiotics In 1990, Nutrition Labeling and Education Act (NLEA) defined how food is labeled, including nutrition labeling, in accordance with definitions established by FDA, that was updated with “Food Safety and Standards (Packaging and Labeling) Regulation, 2011.” These regulations allowed greater legal security and more predictable environment. It supports innovation and prevents unfair competition from manufacturers using false or misleading claims. If positive claims cannot be made, the regulation does not oblige anyone to make negative claims about the product.[12,13] In 2018, Food Safety and Standards (Advertising and Claims) Regulations came into place. It was mainly to regulate the claims and advertisements by food business operators in respect of their food products. General principles for advertising and claims of nutraceuticals from these regulations are summarized in Table 2.[14] All important regulations and amendments can be accessed from the FSSAI website (https://www.fssai.gov.in/cms/food-safety-and-standards-regulations.php). Table 2 General principles for advertising and claims of Nutraceuticals Principle • Claims must be truthful, unambiguous, meaningful, not misleading and help consumers to comprehend the information provided. • Claims shall not encourage or condone excess consumption of a particular food. • Claims shall not state, suggest or imply that a balanced and varied diet cannot provide appropriate quantities of nutrients as required by the body. • Where the claimed benefit is related to or dependent on the method of preparation of the food the same shall be provided on the label. • Claims shall specify the number of servings of the food per day for the claimed benefit. • The claim that a food has certain nutritional or health attributes shall be scientifically substantiated by validated methods of characterizing or quantifying the ingredient or substance that is the basis for the claim. • Where the meaning of a trade mark, brand name or fancy name containing adjectives such as “natural”, “fresh”, “pure”, “original”, “traditional”, “authentic”, “genuine”, “real”, etc., appearing in the labeling, presentation or advertising of a food is such that it is likely to mislead consumer as to the nature of the food, in such cases a disclaimer in not less than 3 mm size shall be given at appropriate place on the label stating that –“This is only a brand name or trade mark and does not represent its true nature”. • All disclaimers related to a claim shall be conspicuous and legible. • Notwithstanding the mandatory declaration of Food Safety and Standards Authority of India logo and license number as per Food Safety and Standards (Packaging and Labeling) Regulations, 2011, no claim or promotion of sale, supply, use and consumption of articles of foods shall be made using Food Safety and Standards Authority of India logo and license number. • Advertisements shall also not undermine the importance of healthy lifestyles. • Advertisements for food or beverages shall not be promoted or portrayed as a meal replacement unless otherwise specifically permitted as a meal replacement under any other Regulations made under Food Safety and Standards Act, 2006 • Claims in advertisements shall be consistent with information on the label of the food or beverage. • No advertisement shall be made for food products which is deceptive to the consumers. • Every declaration which is required to be made on advertisements under these regulations shall be conspicuous and legible. Current stepwise process to register nutraceuticals in India is summarized in Figure 2[15] Figure 2 Steps to register nutritional products in India Important nutraceutical products in India and their uses Nutraceuticals have been used for centuries. Their role is highlighted in various chronic diseases such as type 2 diabetes, inflammatory bowel disease, various enteropathies, and malabsorption syndromes leading to nutritional deficiencies.[16-25] Fish oil, vitamin B6, vitamin B12, and flaxseed oil reduce the risk of preterm labor, influence steroidal output, and regulate the menstrual cycle.[20] Dietary habits play a significant role in hypertension, coronary heart disease, heart failure, peripheral vascular disease, and cerebrovascular disease. Vitamin D, coenzyme Q 10, folic acid, omega-3, and polyphenols may reduce arterial disease by altering cellular metabolism. Flavonoids are present in vegetables such as onions and fruits such as grapes, apples, and cherries. Ginger has antioxidant and anti-inflammatory properties and is used to treat hypertension and palpitations. Green and yellow vegetables, rich in phytosterol, reduce the risk of cardiovascular diseases by blocking the uptake of cholesterol.[20] Nutraceuticals such as curcumin, lutein, lycopene, turmeric, and beta-carotene may exert positive effects on specific diseases such as neurodegenerative diseases—Alzheimer’s disease and Parkinson’s disease by combating oxidative stress. Jujube is shown to be effective in regaining memory in Alzheimer’s disease patients.[26,27] Various herbal products containing Amla, Guduchi, and Tulsi have also shown beneficial effects in improving learning and memory.[27-31] Important nutraceuticals with uses are summarized in Table 3, while important herbs and their properties are listed in Table 4.[16-31] Table 3 Important nutritional products with their uses Nutrient Uses Calcium Reduces risk of osteoporosis, Iron Helps in Hemoglobin formation and used for management of anemia Magnesium Diet high in magnesium, potassium, and calcium, and low in sodium and fat can decrease blood pressure, May reduce the risk of postmenopausal osteoporosis Zinc Increases circulating T cells and increased killing capacity of lymphocytes, Reduction in the incidence of common cold Selenium Neurological deficits due to iodine deficiency may improve with selenium, Antioxidant effects delayed progression of HIV Chromium With the help of insulin converts carbohydrates and fats into energy, Deficiency is associated with weight loss, neuropathy and glucose intolerance Polyphenols, flavonoids Present in fruits, vegetables, nuts. Flavonoids include proanthocyanidins, quercetin, and epicatechin, found mainly in chocolate, tea, and wine and possess antioxidant properties. It Inhibits LDL oxidation and platelet aggregation and enhances production of NO, thus leading to cardio-protection Fish oils Contains N-3 polyunsaturated fatty acids, Biological effects: inhibition of hepatic synthesis and secretion of triacylglycerol and VLDL and reduced postprandial lipemia, increased circulating HDL, inhibition of platelet aggregation and prevention of cardiac arrhythmias. Diets which include cold water fish are associated with reduced incidence of heart ailments Probiotics Live microorganisms when administered in adequate amounts give out to health benefit on the host e.g. yogurts. Physiological effects: reduction of gut pH, production of some digestive enzymes and vitamins, production of antibacterial substances, e.g., organic acids, bacteriocins, reconstruction of microbiomes in diarrhea, antibiotic therapy, reduction of cholesterol level in the blood, suppression of bacterial infections, removal of carcinogens Prebiotics Non-digestible or low-digestible food ingredients that benefit the host organism by selectively stimulating the growth or activity of one or a limited number of probiotic bacteria in the colon. Lactulose, galacto-oligosaccharides, fructo-oligosaccharides, inulin and its hydrolysates, malto-oligosaccharides, and resistant starch. Food sources: onion, garlic, asparagus, artichoke, leek, bananas, tomatoes, chicory. Advantage: Can be stable for longer period of time Antioxidants Onions, Broccoli, Soybean, Tomato, Carrot, Brussels sprouts, Kale, Cabbage, Green tea, Cauliflower, Red beets, Garlic, Cocoa, Blackberry, Cranberries Blueberry, Red grapes, Prunes, Citrus fruits are rich in antioxidants. Table 4 Important herbs/herbal products with their properties Herb Properties Emblica officinalis / Phyllanthus Emblica (Amla) A potent Rasayana with antimicrobial, antipyretic, antioxidant, cardio- and hepatoprotective, anticancer, anti-diabetic, analgesic, immunomodulatory effects Withania somnifera (Ashwagandha) Used for maintaining vitality and longevity. A popular Rasayana with historic uses in cases of lack of libido, chronic illnesses including mental illness and fatigue. Possesses anti-microbial, antioxidant, immunomodulatory, cardiovascular and hepatic protection and anti-aging effects Bacopa monnieri (Bramhi) Medhya Rasayana; Nervous system and a memory enhancer known to have anamnestic, nootropic, neuroprotective, immune-stimulatory and antifertility effects. Tinospora cordifolia (Guduchi) Rich in Vitamin C and trace elements with antibacterial and immune-boosting effects. Known to generate immune resistance, boosting memory, digestive health, longevity Ocimum Sanctum (Tulsi) Leaves are rich in essential oil while seeds contain fixed oil and mucilage. Eugenol is the major component of Tulsi’s essential oil and has wide importance as a nutraceutical. Secondary metabolites of Tulsi have antioxidant, anti-cancer, anti-inflammatory, anti-microbial, anti-stress, immunomodulatory and radiation protective activities. Bhavana with Ocimum Sanctum augment the actions of other herbal products. Brahma rasayana Boosts vitality, youth, longevity, memory, physical strength and concentration. Also known to be an antioxidant and immunity stimulator. Chyawanaprasha “Foremost of all Rasayana” and “elixir of life”. Known to boost intellect, memory, concentration, longevity, physical and mental strength, immunity against diseases, digestion and complexion. It is antitussive and anti-asthmatic. Examples of marketed nutraceutical products, include Revital, Complan, Horlicks, Ensure, Peptamen, NutriMix, and capsules containing fish liver oils—Nutrova, Curcumin Boost, Chyawanprash, etc. Example for nutraceutical are as follows: use of nutraceuticals in clinical practice, a randomized, double-blinded, placebo-controlled, clinical study of the effects of a nutraceutical combination (LEVELIP DUO®) on LDL cholesterol levels and lipid pattern in subjects with suboptimal blood cholesterol levels (NATCOL Study). Nutraceutical-related ongoing clinical studies in India are also available at https://clinicaltrials.gov/. Nutrivigilance, phytovigilance, and its importance Due to our limited knowledge of nutraceuticals and herbal products, we often end up misusing them. Irrational use of such products is common in our country. There is a rampant use of nutraceuticals, herbal products, and health supplements with no clear pathway to report the associated adverse events. It poses a major challenge of identifying such adverse events and taking the corrective actions for the same.[32-34] FSSAI needs to adopt proactive role for nutrivigilance. Stricter regulations are needed for the approval of nutraceuticals and health supplements. There must be a plan to monitor the long-term effect of these products. The lack of robust clinical studies with such products is a major limitation in a current regulatory scenario. Healthcare professionals can play an important role in nutrivigilance by proactively identifying the use of such products and any adverse events. It is essential to impart education at the primary care level about the rational use of nutraceuticals and the possible limitations associated with the products. Special populations need to use such products carefully; for example, sport persons and athletes should be careful about the ingredients of health supplements or sport drinks they are taking to avoid inadvertent doping.[35,36] The use of digital tools to educate the general population about the possible impacts of irrational use of nutraceuticals should be initiated by FSSAI. The time has come to specify the indications for complex nutraceutical products as it is specified for pharmaceutical products. It is also essential to curb unsubstantiated and false claims and their advertisement at the mass level. Need to strictly regulate the promotional materials for nutraceuticals. Evolving role of medical affairs/pharmaceutical physician for nutraceuticals and consumer health products The medical affairs function is a connecting link between commercial teams such as sales and marketing with research and development and with external stakeholders such as healthcare professionals and public. Hence, it plays an integral role in disseminating important and scientifically valid information on nutraceutical products to commercial teams and healthcare professionals, that is, to internal and external stakeholders. Medical affairs need to interact with key opinion leaders or key medical experts (KME) to understand the unmet needs and to impart key scientific messages related to the products. As regulatory norms are getting stricter for the promotion and use of nutraceutical products, the role of medical affairs further becomes indispensable. There is increasing scrutiny of promotional materials and product claims for nutraceuticals, which needs to be thoroughly backed by scientific evidence. Medical affairs can support the claim substantiation with evidence generation from proof of concept and clinical studies for nutritional products as part of evidence-based practice. Real-world data on product usage at large population can be generated. The value proposition of the products with health economics and outcomes research (HEOR) data can be justified.[37-39] Understanding the unmet needs and guiding the future development of novel and innovative products are also key roles of medical affairs. The innovation cycle with such nutraceutical products is small, and there is a need to continue evaluating new formulations, innovative ingredients, and/or delivery mechanisms with strong scientific rationale. Disseminating the evidence-based data on nutraceutical products in ethical and scientific manner forms a core for the medical affairs function. There is an increasing responsibility for the medical affairs team to promote the rationale use of nutraceuticals and nutrivigilance. This adds to the ever-expanding role of pharmaceutical physicians.[37-39] as summarized in Figure 3. Figure 3 Role of medical affairs in consumer health To summarize, nutraceuticals are of importance and its significance in maintaining a healthy lifestyle is increasing. However, it is important to use them judiciously. The market is flooded with irrational combinations of nutraceutical products; hence, regulatory authorities need to take stringent actions for the same. Nutrivigilance and phytovigilance are the need of hour, and healthcare professionals will have to play the role of a torch bearer. Pharmaceutical physicians need to play their part with their multifaceted role in medical affairs in nutraceuticals and consumer health industry. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. ==== Refs REFERENCES 1 Kalra EK Nutraceutical–definition and introduction AAPS Pharm Sci 2003 5 E25 doi: 10.1208/ps050325 2 Zeisel SH Regulation of “nutraceuticals.” Science 1999 285 1853 5 doi: 10.1126/science.285.5435.1853 10515789 3 Whitman M Understanding the perceived need for complementary and alternative nutraceuticals: Lifestyle issues Clin J Oncol Nurs 2001 5 190 4 11899764 4 Temple NJ Gladwin KK Fruit, vegetables, and the prevention of cancer: research challenges Nutrition 2003 19 467 70 doi: 10.1016/s0899-9007(02)01037-7 12714102 5 Nair L Six reasons why India is poised to be a global leader in nutraceuticals: An overview of India Nutra Inc's comparative advantages, growth drivers and development potential Express Pharma; India [Online] 2020 Available from: https://www.expresspharma.in/market-pharma/six-reasons-why-india-is-poised-to-be-a-global-leader-in-nutraceuticals/ [Last accessed on 2022 Apr 24]] 6 Malve H Exploring the ocean for new drug developments: Marine pharmacology J Pharm Bioallied Sci 2016 8 83 91 doi: 10.4103/0975-7406.171700 27134458 7 VanHorn B Chatterjee R India Nutraceuticals industry. Market intelligence. International Trade Administration; USA [Online] 2020 Available from: https://www.trade.gov/marketintelligence/india-nutraceuticals-industry#:~:text=India's%20nutraceutical%20industry%20is%20expected,by%20the%20end%20of%202025 [Last accessed on 2022 Apr 24] 8 India Nutraceuticals Market Outlook to 2021 - High Prevalence of Lifestyle Diseases coupled with Rising Awareness to Foster Future Growth. Research and Markets; USA [Online] 2017 Available from: https://www.researchandmarkets.com/reports/4412695/india-nutraceuticals-market-outlook-to-2021 [Last accessed on 2022 Apr 24] 9 Guidance Note on Food for Special Medical Purpose (FSMP). Food Safety and Standards Authority of India; India [Online] 2020 Available from: https://fssai.gov.in/upload/uploadfiles/files/Guidance_Note_FSMP_28_08_2020.pdf [Last accessed on 2022 Apr 25] 10 Madhavan M Sanyal K Legislative Brief: The Food Safety and Bill for Nutraceuticals, 2005 Parliamentary Research Service, Chanakyapuri, Authority of India; India [Online 2006] Available from: http://www.commonlii.org/in/other/INPRSLS/tfsasb2005lb419/ [Last accessed on 2022 Apr 25] 11 Food Safety and Standards Act. Food Safety and Standards Authority of India; India [Online] 2006 Available from: https://www.fssai.gov.in/cms/food-safety-and-standards-act-2006.php [Last accessed on 2022 Apr 24] 12 Putta S FSSAI guidance and notification on nutraceuticals –An insight FnBNews.com; India [Online] 2020 Available from: https://www.fssai.gov.in/upload/media/FSSAI_News_Guidance_FNB_09_06_2020.pdf [Last accessed on 2022 Apr 25] 13 Food Safety and Standards (Packaging and Labeling) Regulation Food Safety and Standards Authority of India; India [Online] 2011 Available from: https://www.fssai.gov.in/cms/food-safety-and-standards-regulations.php [Last accessed on 2022 Apr 25] 14 Food Safety and Standards (Advertising and Claims) Regulations Food Safety and Standards Authority of India;India [Online] 2018 Available from: https://www.fssai.gov.in/upload/uploadfiles/files/Gazette_Notification_Advertising_Claims_27_11_2018.pdf [Last accessed on 2022 Apr 25] 15 Verma B Popli H Regulations of nutraceuticals in India and US Pharma Innovation 2018 7 811 6 16 Pandey M Verma RK Saraf SA Nutraceuticals: New era of medicine and health Asian J Pharm Clin Res 2010 3 11 5 Available from: https://innovareacademics.in/journal/ajpcr/|vVol3Issu|ne1/265.pdf [Last accessed on 2021 Apr 25] 17 Derosa G Limas CP Macías PC Estrella A Maffioli P Dietary and nutraceutical approach to type 2 diabetes Arch Med Sci 2014 10 336 44 doi: 10.5114/aoms.2014.42587 24904670 18 Dutta S Ali KM Dash SK Giri B Role of nutraceuticals on health promotion and disease prevention: A review J Drug Deliv and Therap 2018 8 42 7 doi: https://doi.org/10.22270/jddt.v8i4.1759. 19 Adefegha SA Functional foods and nutraceuticals as dietary intervention in chronic diseases;Novel perspectives for health promotion and disease prevention J Diet Suppl 2018 15 977 1009 doi: 10.1080/19390211.2017.1401573 29281341 20 Sachdeva V Roy A Bharadvaja N Current prospects of nutraceuticals: A review Curr Pharm Biotechnol 2020 21 884 96 doi: 10.2174/1389201021666200130113441 32000642 21 Pushpaanjali G Geetha RV Rani L Lakshmi T Recent advances in nutraceutical and functional food Indian J Fore Med and Tox 2020 14 5238 43 22 Dang R Nutraceuticals for Healthy Life Indian J Pharm Edu and Res 2017 51 S148 51 23 Tahilani P Banweer J Chatterjee DP Goyanar G Sharma A Nutraceuticals and Fortified Foods Supplements in India: Challenges and Opportunities - A Comprehensive Review Indian J Pharm Drugs Stud [Internet]. 2021 1 1 5 Available from: https://mansapublishers.com/IJPDS/article/view/2438 [Last accessed on 2021 Apr 29] 24 Thalkari AB Karwa PN Thorat VM Jadhav SK Overview of nutraceuticals Research J Pharm and Pharmacody 2020 12 130 2 DOI: 10.5958/2321-5836.2020.00023.3 Available from: https://rjppd.org/HTMLPaper.aspx?Journal=Research%20Journal%20of%20Pharmacology%20and%20Pharmacodynamics;PID=2020-12-3-4 [Last accessed on 2022 Apr 25] 25 Williamson EM Liu X Izzo AA Trends in use, pharmacology, and clinical applications of emerging herbal nutraceuticals Br J Pharmacol 2020 177 1227 40 doi: 10.1111/bph.14943 31799702 26 Dietary Supplement Fact Sheets National Institutes of Health: Office of Dietary Supplements. UK [Online 2020] Available from: http://ods.od.nih.gov/factsheets [Last accessed on 2022 Apr 26] 27 Arun Raj GR Shailaja U Prasanna N Rao Parikshit Debnath Nutraceuticals and Functional foods in Ayurvedic perspective Govil JN Pathak M Recent Progress in Medicinal Plants 1st ed 42 Houston-Texas, USA Studium Press LLC 2016 172 99 28 Chauhan B Kumar G Kalam N Ansari SH Current concepts and prospects of herbal nutraceutical: A review J Adv Pharm Technol Res 2013 4 4 8 doi: 10.4103/2231-4040.107494 23662276 29 Malve HO Exploring Bhavana samskara using Tinospora cordifolia and Phyllanthus emblica combination for learning and memory in mice J Ayurveda Integr Med 2015 6 233 40 doi: 10.4103/0975-9476.157953 26834422 30 Malve HO Raut SB Marathe PA Rege NN Effect of combination of Phyllanthus emblica, Tinospora cordifolia, and Ocimum sanctum on spatial learning and memory in rats J Ayurveda Integr Med 2014 5 209 15 doi: 10.4103/0975-9476.146564 25624694 31 Malve HO Management of Alzheimer's Disease: Role of existing therapies, traditional medicines and new treatment targets Indian J Pharm Sci 2017 79 2 15 32 Resu NR Manju MS Kondaveti S Kumar SB Nutraceuticals and Nutrivigilance-Present Scenario in India Int J Food Biosci 2019 2 35 40 33 Malve H Fernandes M Nutrivigilance –The need of the hour Indian J Pharmacol 2023 55 62 3 36960523 34 Lehmann H Pabst JY La phytovigilance: Impératif médical et obligation légale [Phytovigilance: A medical requirement and a legal obligation Ann Pharm Fr 2016 74 49 60 French. doi: 10.1016/j.pharma.2015.06.004 26210820 35 Malve HO Sports pharmacology: A medical pharmacologist's perspective J Pharm Bioallied Sci 2018 10 126 36 doi: 10.4103/jpbs. JPBS_229_17 30237683 36 Malve HO Forensic pharmacology: An important and evolving subspecialty needs recognition in India J Pharm Bioallied Sci 2016 8 92 7 27134459 37 Setia S Ryan NJ Nair PS Ching E Subramaniam K Evolving role of pharmaceutical physicians in medical evidence and education Adv Med Educ Pract 2018 9 777 90 doi: 10.2147/AMEP.S175683 30464675 38 Crowley-Nowick P Smith J The Role of Medical Affairs in Moving from R&D to Commercialization BioProcess International; UK [Online 2013] Available from: https://bioprocessintl.com/analytical/qa-qc/the-role-of-medical-affairs-in-moving-from-randd-to-commercialization-341871/ [Last accessed on 2022 April 24] 39 Evers M Ghatak A Suresh B Westra A A vision for medical affairs in 2025 McKinsey & Company USA Online 2019 Available from: https://www.mckinsey.com/industries/pharmaceuticals-and-medical-products/our-insights/a-vision-for-medical-affairs-in-2025#. [Last accessed on 2022 April 24]

PMC010xxxxxx/PMC10353664.txt

==== Front J Pharm Bioallied Sci J Pharm Bioallied Sci JPBS J Pharm Bioall Sci Journal of Pharmacy & Bioallied Sciences 0976-4879 0975-7406 Wolters Kluwer - Medknow India JPBS-15-63 10.4103/jpbs.jpbs_364_22 Review Article Black Seeds (Nigella sativa) Medical Application and Pharmaceutical Perspectives Ferizi Rrahman 1 Ramadan Mohamed F. 2 Maxhuni Qenan 3 1 Premedical Department, Faculty of Medicine, University of Prishtina, 10000 Prishtine, Kosovo, Albania 2 Department of Clinical Nutrition, Faculty of Applied Medical Sciences, Umm Al-Qura University, Makkah 21955, Saudi Arabia 3 Department of Pharmacy, College of Medical Sciences Rezonanca, 10000 Prishtine, Kosovo, Albania Address for correspondence: Dr. Qenan Maxhuni, College of Medical Sciences Rezonanca, Prishtine - 10000, Kosovo, Albania. E-mail: qenan.maxhuni@rezonanca-rks.com Apr-Jun 2023 08 6 2023 15 2 6367 08 8 2022 29 9 2022 27 3 2023 Copyright: © 2023 Journal of Pharmacy and Bioallied Sciences 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. ABSTRACT Among the various medicinal plants, the black seed is emerging as a miracle herb with a rich historical background, as much research has revealed its wide spectrum of pharmacological potential. In this collection of literature, we have encountered and presented the preclinical treatment, as alternative medicine of Nigella sativa in the prevention and treatment of various diseases, as well as those that continue to be discovered by contemporary actual scientific data. Research to date has confirmed the pharmacological potential of the seed of Nigella sativa, its oil and extracts of some of its bioactive constituents, which possess remarkable pharmacological activity, in vitro and in vivo against a large spectrum of diseases, and it has been found that the use of black seed is relatively safe. Black Seed has been extensively studied for its biological activity and therapeutic potential and has been found to possess a broad spectrum of activities. Clinical trial investigations into the therapeutic effects of Nigella sativa affect the hypoglycemic, hypolipidemic, and bronchodilator effects and have passed clinical trials and received the green light to allow the next stage of clinical trials toward therapeutic drug design. However, there is still room and multidimensional research needed for prospective clinical trials in certain groups of animals before they can be applied to humans as pharmaceutical therapies. KEYWORDS: Black Seed clinical trials disease nigella sativa quality ==== Body pmcINTRODUCTION Medicinal herbs have been used to cure diseases for many centuries in various forms of folk medicine application. Furthermore, medicinal herbs are used in the preparation of herbal medicines as they are considered to be safer than modern medicines. Many researchers have focused on medicinal plants as only a few plant species have been thoroughly researched for potential medicinal properties, potential, mechanism of action, efficacy assessment, and toxicological studies. Among the various medicinal plants, the black seed (Family Ranunculaceae) is emerging as a miracle herb with a rich historical and religious background, as much research has revealed its wide spectrum of pharmacological potential. Black Seed and its oil have been used for centuries in the treatment of various diseases worldwide. It is an important plant in traditional fields of medicine known as Unani and Ayurveda. Among Muslims, it is considered one of the greatest forms of medical healing available because it is mentioned in sayings of the Prophet Muhammad that Black Seed is a cure for all diseases.[1] Black Seed has been extensively studied for its biological activity and therapeutic potential and has been found to possess a broad spectrum of activities such as antioxidant[1] anti-cough,[2] gastroprotective,[3] anti-anxiety,[4] anti-ulcer,[5] anti-asthmatic,[6] anti-cancer,[7] anti-inflammatory and immuno-modulatory,[8] antitumor,[9] hepatoprotective,[10] cures gastric ulcers,[11] slows tumor growth,[12] improves memory,[13] stimulates milk production,[14] has antibacterial activity,[15] etc. The use of herbal medicines as alternative medicine is spreading and gaining wide popularity worldwide. Many medicines are extracted directly from plants, while others are naturally occurring chemically modified (synthetic) products. Research date has confirmed the pharmacological potential of the seed of Nigella sativa, its oil and extracts of some of its bioactive constituents, particularly thymoquinone and alpha-hederine, which possess remarkable pharmacological activity, in vitro and in vivo against a large spectrum of diseases, and it has been found that the use of Black Seed is relatively safe.[16] The anatomy of the plant The black seed is an annual herbaceous plant from the Ranunculaceae family, is indigenous to southern Europe, North Africa, and Southwest Asia, and is cultivated in many countries around the world such as India, Pakistan, Turkey, Ethiopia, Yemen, and Saudi Arabia. It is a common plant whose stalk grows from 20 to 90 cm. This plant has well-separated leaves and flowers usually white, yellow, pink, pale blue, or pale purple, with five–ten petals. The flowers grow at the ends of the branches, while the leaves grow in pairs against each other on the other side of the stalk. When its flower dries, the Black Seed ripens. Black Seed is a plant that reproduces itself and forms a fruit capsule with three–seven fused follicles composed of many white seeds. When the fruit capsule is ripe, it opens and the seeds inside are exposed to the air, turning black. Seeds are angular, generally small, size (1–3 mm), and gray or dark black. The seeds have a slightly bitter and sour taste. History The use of black seed has its origins more than 2500 years ago. It is mentioned in the Greek medical literature, adding to the fact that it played an important role in ancient practices in the field of medicine applied even by the Egyptians themselves. The black seed was discovered in the tomb of Pharaoh Tutankhamun leaving it to be understood that it played an important role in the ancient practices of Egypt.[17] The earliest past references to the Black Seed are found in the book of Isaiah in the Old Testament.[18] Easston’s Bible Dictionary explains that the Hebrew word “ketsah” undoubtedly refers to the plant Nigella sativa. It is referred to as “Melanthion” by Hippocrates and Dioscorides[17] who have recorded that this seed has been used to treat headaches, flu, toothaches, and intestinal parasites. They have also been used as a urinary stimulant, stimulant, and milk supplement in breastfeeding mothers. The scientist Ibn Sina-AVICENA (980-1037), in his well-known book “The Encyclopedia of Medicine” (considered the most famous book in the history of medicine, is considered to have been used as the basic literature in medicine European until the seventeenth century) said: “Black seed acts as a means of extracting phlegm, stimulates bodily energy and helps in rehabilitation after fatigue and lethargy.”[18] Arabic medicine As narrated that the Prophet Muhammad said: “There is no disease for which Nigella seed does not provide remedy” and “in black seed there is healing for every disease,” then, “Use this black seed. For indeed it contains a cure for every disease.”[19] Muslims all over the world have used historically and promoted the use of the Black Seed for hundreds of years, and a lot of articles have been written about it. However, many Muslims, today, do not know that Black Seed is not only a medicinal plant, but it also holds a special place in the medicine of the Prophet Muhammad. Although, many natural herbs and cures in the Qur’an and Prophet sayings are mentioned briefly leaving scholars to explore their benefits. The Black Seed is a unique relic that is said to have “a cure for all diseases except death” and is also said to “Abstain.” which means continuity. Its peculiarity is that it was not used extensively before the Prophet Muhammad and he made it very popular, and it is one of the few plants that is described in detail with recipes and instructions. Although, there were more than 400 plants in use before the Prophet Muhammad recorded by Galen and Hipokrat. Black seed was not one of the most popular cures of the time. Due to the spread of Islam in various countries, the use and popularity of the Black Seed have spread and it is known worldwide as the “Medicine of the Muhammad Messenger.” In fact, since it became popular in the seventh century, there has never been a period in Islamic history when its use was ever banned. All the while, the Black Seed has been used for healing but also with the belief in the benefits that will flow from the traditional practice of the Prophet Muhammad. In addition to the above sayings, there are records in the history books of the Prophet Muhammad that show that the Prophet Muhammad used the Black Seed with honey regularly. Chemical composition Numerous studies have been done to identify the composition of Black Seed (Nigella sativa). Ingredients of Nigella sativa seeds include oils, proteins, carbohydrates, alkaloids, saponins, and essential oils. Fixed oil [non-volatile] (32–40%) contains unsaturated fatty acids which include arachidonic acid, eicosadienoic, linoleic, oleic, palmitic, stearic, and myristic acid, as well as beta-sitosterol, cycloeucalenol, cycloartenol, sterols, esters, and sterol glucosides.[20] Volatile oil (0.4–2.45%) contains saturated fatty acids, which include Nigellone (which is the only component of the carbonyl fraction), thymoquinone (TQ), thymohydroquinone (THQ), ditimoquinone, thymol, karva-krol, α and β-pinene, d-limonene, d-citronellol, and p-cimene and also contains t-anethole, 4-terpineol, and longifoline.[21] Black Seed has two different forms of alkaloids: isoquinoline alkaloids which include nigellicimine, nigellici-mine n-oxide, and pyrazole alkaloids which include nigellidine and nigellicine.[22] Nutritional compositions of Black Seed are vitamins, carbohydrates, minerals, fats, and proteins (containing eight or nine essential amino acids). Black Seed also has saponin and alpha hederine of course not found in traces of lemons and citronellol, just as it should be possible with vitamins and minerals that are different such as Fe, Ca, K, Zn, P, and Cu.[23] The greatest pharmacological effects are due to the component of the substance quinone, of which thymoquinone (TQ) predominates. It was found that thymoquinone possesses these activities: - anticonvulsant,[24] - antioxidant,[25] - anti-inflammatory,[26] - anti-carcinogen,[27] - antibacterial,[28] and - antifungal.[29] The benefits of Black Seed Raw black seed, in its full form, aids in the body’s natural healing process, in overcoming disease or maintaining health. It acts on those parts of the body that are considered to be affected by the disease without disturbing the natural balance. The effect of the combined nutritional and medicinal value of Black Seed is that it not only helps to alleviate the current condition, but also helps the body to build further resistance against future diseases (acquired immunity). Although historical evidence shows the use of Black Seed for a wide variety of diseases, we have limited the descriptions of the main healing properties in the latest scientific research findings on Black Seed. USE Black seed as a daily health supplement Most drugs work best when given the chance to perform their full function. Based on the essential nutritional components, as well as the specific medical content, the ‘body’s ability to maintain health and maintain a stable and natural balance is significantly enhanced through regular use of Black Seed. Black Seed as a source of energy Ibn Sina-Avicena (980-1037) Description of Black Seed: “Black seed acts as a means of extracting phlegm, stimulates bodily energy and helps in rehabilitation after fatigue and lethargy” is still successful for the practitioners of Tibb Nebevij (Medicine of the Messenger) today. The high nutritional value of Black Seed as described by scientific analysis also points to it as a great source of energy. From the perspective of Tibb Nebevij, Black Seed has the ability to store and restore the body’s energy. Western diets, made mainly from cold foods (ice in our drinks, yogurt, pizza, cheese, etc.), all reduce the body’s internal energy, which is necessary for the body to function optimally. Tibb Nebevij considers that a reduced rate of metabolism is the cause of most diseases. The body, with the loss of energy, loses the ability to fight external toxins, increasing the chances of disease. Black Seed and medicines Black seed can be used in combination with other medicines, whether natural or conventional. It is not recommended that black seed be used exclusively for the treatment of serious health complaints which require immediate action. For example, some cases of acute bronchitis require conventional antibiotics to prevent the condition from becoming more severe. However, Black Seed can be used as a therapeutic aid to combat the side effects from the use of antibiotics or other strong drugs that are chemically based [Table 1]. Table 1 Scientific names of plants mentioned in the examples Name Scientific name Anise Pimpinella anisum Basil Ocimum basilicum Green tea Camellia sinensis Linseed, flaxseed Linum usitatissimum Fennel seed Foeniculum vulgare Black Seed, black cumin Nigella sativa Dates Phoenix dactylifera Chamomile Matricaria chamomilla Cinnamon Cinnamomum cassia Clove Syzygium aromaticum Cardamom Elettaria cardamomum Peppermint Mentha piperita Oregano Origanum vulgare Beetroot Beta vulgaris Black pepper Piper nigrum Rosemary Rosmarinus officinalis Radish Raphanus sativus Sage Salvia officinalis Castor oil Ricinus communis Coconut oil Cocos nucifera Ginger Zingiber officinale Pregnancy and lactation Black seed is not recommended during pregnancy, but is recommended during lactation. This is a great form of nutrition for the growing mother and baby, while the boosting immune system properties of Black Seed serve as a natural and safe way to build disease resistance. In addition, studies have shown that Black Seed helps increase milk production. Babies and children In addition to its many nutrients, black seed contains the protein keratin in our body, and this protein is converted into Vitamin A, which is one of the main nutrients of the body and helps for better health, which performs various functions related to immunity and cell proliferation of the body. Black Seed gives children all the energy they need for this active phase of life. Regular use of Black Seed, which increases its effect on strengthening immunity in the body, will reduce the length and impact of common diseases of children, especially during the winter, when children are more susceptible to colds and influenza. Children are given half the recommended dose for parents. It is not given to babies at all. The elderly With its rich nutritional and energy value combined with strengthening the immune system, Black Seed is an ideal health supplement for the elderly. Soap and shampoo from Black Seed. Black Seed soap and shampoo is an amalgam containing more than 100 valuable ingredients. It is an excellent source of vitamins and minerals, so it has powerful effects on the skin and hair. It therapeutically formulated to remove blemishes from the skin and soften irritated skin as a result of painful injuries. It preserves the skin and cleanses it from impurities leaving a smooth appearance from hydration and ingredients and helps control sweat production by the sweat glands. Its antiseptic and antibiotic properties make it necessary as a means of treating small cuts, treats minor skin problems, and helps prevent scar formation. How to use In the morning and evening, cleanse the face with soap and lukewarm water for a few minutes, while the hair is foamed and then rinsed as usually. Examples of the use of Black Seed and its oil for various diseases Careful Black seed is not a medicine or substitute. Any reader who is concerned about his or her health should consult a pharmacist, physician, or other healthcare professional before applying any of the examples. Examples of the use of Black Seed and its oil for various diseases and problems that we have brought are taken from books or websites that deal with folk medicine and as such do not possess references—science, so we have not brought references to them either. It is not recommended for use by pregnant women (can only be used for external use). It is not used by persons with transplanted organs (can only be used externally). For children, half is the recommended dose to increase, while babies are not given at all. The use of Black Seed or oil in larger quantities does not increase the effect, the maximum daily dose for each disease is not recommended to exceed three teaspoons per day (15 g of seeds or 15 ml of oil). Financial support and sponsorship Nil. Conflicts of interest The authors have no conflicts of interest regarding this investigation. ==== Refs REFERENCES 1 Hosseinzadeh H Moghim FF Mansouri SMT Effect of Nigella sativa seed extracts on ischemia reperfusion in rat skeletal muscle Pharmacologyonline 2007 2 326 35 2 Hosseinzadeh H Eskandari M Ziaee T Antitussive effect of thymoquinone, a constituent of Nigella sativa seeds, in guinea pigs Pharmacologyonline 2008 2 480 4 3 Forouzanfar F Bazzaz BSF Hosseinzadeh H Black cumin (Nigella sativa) and its constituent (thymoquinone): A review on antimicrobial effects Iran J Basic Med Sci 2014 17 929 38 25859296 4 Sayeed MSB Shams T Hossain SF Rahman MR Mostofa AGM Kadir MF Nigella sativa L. seeds modulate mood, anxiety and cognition in healthy adolescent males J Ethnopharmacol 2014 152 156 62 24412554 5 Ahmed MW SR Mahammed NL Nasir MAM Anti ulcer effect of Nigella sativa linn. Against gastric ulcers in rats Int J Res Dev Pharm Life Sci 2016 5 2006 9 6 Boskabady MH Mohsenpoor N Takaloo L Antiasthmatic effect of Nigella sativa in airways of asthmatic patients Phytomedicine 2010 17 707 13 20149611 7 Zhang M Du H Huang Z Zhang P Yue Y Wang W Thymoquinone induces apoptosis in bladder cancer cell via endoplasmic reticulum stress dependent mitochondrial pathway Chem Biol Interactions 2018 292 65 75 8 Salem ML Immunomodulatory and therapeutic properties of the Nigella sativa L. seed Int Immunopharmacol 2005 5 1749 70 16275613 9 Majdalawieh AF Hmaidan R Carr RI Nigella sativa modulates splenocyte proliferation, Th1/Th2 cytokine profile, macrophage function and NK anti tumor activity J Ethnopharmacol 2010 131 268 75 20600757 10 Khan MA Chemical composition and medicinal properties of Nigella sativa Linn Inflammopharmacology 1999 7 15 35 17657444 11 Bukhari MH Khalil J Qamar S Qamar Z Zahid M Ansari N Comparative gastroprotective effects of natural honey, Nigella sativa and cimetidine against acetylsalicylic acid induced gastric ulcer in albino rats J Coll Physicians Surg Pak 2011 21 151 6 21419021 12 Salim EI Cancer chemopreventive potential of volatile oil from black cumin seeds, Nigella sativa L., in a rat multi organ carcinogenesis bioassay Oncol Lett 2010 1 913 24 22966405 13 Hosseini M Mohammadpour T Karami R Rajaei Z Sadeghnia HR Soukhtanloo M Effects of the hydro alcoholic extract of Nigella sativa on scopolamine induced spatial memory impairment in rats and its possible mechanism Chin J Integr Med 2015 21 438 44 24584756 14 Hosseinzadeh H Tafaghodi M Mosavi MJ Taghiabadi E Effect of aqueous and ethanolic extracts of Nigella sativa seeds on milk production in rats J Acupunct Meridian Stud 2013 6 18 23 23433051 15 Hosseinzadeh H Fazly Bazzaz BS Haghi MM Antibacterial activity of total extracts and essential oil of Nigella sativa L. seeds in mice Pharmacologyonline 2007 2 429 35 16 Ahmad A Husain A Mujeeb M Khan SA Najmi AK Siddique NA A review on therapeutic potential of Nigella sativa: A miracle herb Asian Pac J Trop Biomed 2013 3 337 52 23646296 17 Padhye S Banerjee S Ahmad A Mohammad R Sarkar FH From here to eternity the secret of pharaohs: Therapeutic potential of black cumin seeds and beyond Cancer Ther 2008 6 495 510 19018291 18 Ibn Sina H Kitab al Qanoun fi Al Toubb (The book of the canon of medicine) Beirut American University of Beirut 2007 19 Siddiqui AH Sahih Muslim Peace Vision 1976 20 Tembhurne SV Feroz S More BH Sakarkar DM A review on therapeutic potential of Nigella sativa (kalonji) seeds J Med Plant Res 2014 8 167 77 21 Enomoto S Asano R Iwahori Y Narui T Okada Y Singab AN Hematological studies on black cumin oil from the seeds of Nigella sativa L Biol Pharm Bull 2001 24 307 10 11256491 22 Ahmad A Husain A Mujeeb M Khan SA Najmi AK Siddique NA A review on therapeutic potential of Nigella sativa: A miracle herb Asian Pac J Trop Biomed 2013 3 337 52 23646296 23 Hosseinzadeh H Parvardeh S Nassiri Asl M Mansouri MT Intracerebroventricular administration of thymoquinone, the major constituent of Nigella sativa seeds, suppresses epileptic seizures in rats Med Sci Monit 2005 11 BR106 10 15795687 24 Hosseinzadeh H Taiari S Nassiri Asl M Effect of thymoquinone, a constituent of Nigella sativa L., on ischemia–reperfusion in rat skeletal muscle Naunyn Schmiedebergs Arch Pharmacol 2012 385 503 8 22271000 25 El Gazzar M El Mezayen R Marecki JC Nicolls MR Canastar A Dreskin SC Anti inflammatory effect of thymoquinone in a mouse model of allergic lung inflammation Int Immunopharmacol 2006 6 1135 42 16714217 26 Gali-Muhtasib H Ocker M Kuester D Krueger S El-Hajj Z Diestel A Thymoquinone reduces mouse colon tumor cell invasion and inhibits tumor growth in murine colon cancer models J Cell Mol Med 2008 12 330 42 18366456 27 Halawani E Antibacterial activity of thymoquinone and thymohydroquinone of Nigella sativa L. and their interaction with some antibiotics Adv Biol Res 2009 3 148 52 28 Abdel Azeiz AZ Saad AH Darweesh MF Efficacy of thymoquinone against vaginal candidiasis in prednisolone induced immunosuppressed mice J Am Sci 2013 9 155 9 29 Deepak Suri S Sikender M Garg V Samim M Entrapment of seed extract of Nigella sativa into thermosensitive (NIPAAm–Co–VP) co polymeric micelles and its antibacterial activity International Journal of Pharmaceutical Sciences and Drug Research 2011 3 246 52

PMC010xxxxxx/PMC10353665.txt

==== Front J Pharm Bioallied Sci J Pharm Bioallied Sci JPBS J Pharm Bioall Sci Journal of Pharmacy & Bioallied Sciences 0976-4879 0975-7406 Wolters Kluwer - Medknow India JPBS-15-95 10.4103/jpbs.jpbs_420_22 Original Article Gentamicin in Neonates with Hemodynamically Significant Patent Ductus Arteriosus Sridharan Kannan 1 Madhoob Abdulraoof Al 2 Jufairi Muna Al 23 Ansari Eman Al 2 Marzooq Reem Al 2 Hubail Zakariya 4 Hasan Sadiq Jaafar 4 1 Department of Pharmacology and Therapeutics, College of Medicine and Medical Sciences, Arabian Gulf University, Manama, Kingdom of Bahrain 2 Neonatology Intensive Care Unit, Department of Pediatrics, Salmaniya Medical Complex, Manama, Kingdom of Bahrain 3 Department of Pediatrics, College of Medicine and Medical Sciences, Arabian Gulf University, Manama, Kingdom of Bahrain 4 Department of Cardiology, Salmaniya Medical Complex, Manama, Kingdom of Bahrain Address for correspondence: Dr. Kannan Sridharan, Department of Pharmacology and Therapeutics, College of Medicine and Medical Sciences, Arabian Gulf University, Manama, Kingdom of Bahrain. E-mail: skannandr@gmail.com Apr-Jun 2023 08 6 2023 15 2 95100 25 9 2022 20 11 2022 21 11 2022 Copyright: © 2023 Journal of Pharmacy and Bioallied Sciences 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. ABSTRACT Background: Gentamicin has been shown to cause vasodilation in preclinical studies. Hemodynamically significant patent ductus arteriosus (hsPDA) is a commonly observed congenital heart disorder in preterm neonates. Concomitant gentamicin theoretically shall delay the closure/result in nonclosure of ductus arteriosus (DA). Similarly, hsPDA can alter the pharmacokinetics of gentamicin and so trough gentamicin concentrations. We carried out the present study to evaluate the association between gentamicin use and closure of hsPDA (treated with acetaminophen) as well as the effect of hsPDA on trough concentrations. Methods: This study was a prospective, observational study that included 60 neonates diagnosed with hsPDA by echocardiography and 102 neonates without hsPDA. Demographic details, size of DA as per echocardiography at the end of treatment with acetaminophen, gentamicin-dosing regimen, and trough concentrations were collected. Standard definitions were adhered in classifying the gestational age, birth weights, and size of DA. The numerical values are reported in median (range). Results: Neonates with hsPDA had significantly lower daily doses of gentamicin [4.5 (2.5–10), 7 (3.2–13) mg; P < 0.001] but longer duration of therapy [8 (3–14), 5 (3–7) days; P < 0.001] than those without hsPDA in very preterm neonates. No significant differences were observed in the trough concentrations of gentamicin between the groups. No association was observed between gentamicin use and closure of DA. However, those with successful closure of DA received gentamicin for a longer duration [6 (3–10), 4 (3–14) days; P < 0.05] that was independent of acetaminophen duration and had received higher cumulative doses of gentamicin. Conclusion: In conclusion, we observed a significantly longer duration of gentamicin therapy in neonates with hsPDA compared to those without hsPDA. No significant differences were observed in the rates of closure of DA with concomitant gentamicin administration and gentamicin trough concentrations. KEYWORDS: Aminoglycosides cardiovascular effects gentamicin PDA ==== Body pmcINTRODUCTION Patent ductus arteriosus (PDA) accounts for about a tenth of congenital heart disorders in neonates.[1] At birth, several physiological processes lead to closure of ductus arteriosus (DA) such as high oxygen tension, low levels of circulating prostaglandins, increase in endothelin, rise in intracellular calcium, inhibition of potassium channels, reduction in isoprostanes, and release of angiotensin II.[2] Physiological closure of DA usually occurs within 24–48 h of birth in term neonates; however, the risk of PDA is more in preterm neonates.[3] Nonclosure of DA will result in hemodynamically significant patent ductus arteriosus (hsPDA) that causes heart failure, necrotizing enterocolitis, intraventricular hemorrhage, and hampers the growth of neonates.[4] Aminoglycosides have been shown to block L-type of calcium channel in the arterial smooth muscles resulting in vasodilation.[5] Intraarterial administration of gentamicin has been shown to result in coronary vasodilation.[6] In-vitro tests reveal that gentamicin causes a decrease in the calcium influx in cardiac smooth muscles, thereby results in negative inotropic effect.[7] Further, inhibition of phospholipase C (through which calcium entry is mediated) has been observed with gentamicin.[8] Gentamicin has been observed with a relatively higher potency in dilating arterial blood vessels compared to other aminoglycosides as observed by the effective concentrations (EC50s) which was 0.53 + 0.08 mM (n = 12), 1.6 + 0.3 mM (n = 7), and 3.9 + 0.5 mM for gentamicin, streptomycin, and kanamycin, respectively.[7] Gentamicin further reduced the contractile response to norepinephrine, and histamine in isolated rabbit aortic strips.[9] Gentamicin is one of the most used aminoglycosides in neonatal intensive care unit.[10] Most often gentamicin is used in our critical care unit as a part of empirical regimen in suspicion of neonatal sepsis. Preterm neonates are at increased risk of neonatal sepsis due to diminished resistance and increased exposure to microorganisms.[11] Hence, it is inevitable that gentamicin is administered in preterm neonates that are at high risk of hsPDA. A previous multicentric study from USA concluded that gentamicin exposure increases the risk of hsPDA.[12] However, there is dearth of data regarding the association of gentamicin use in the outcome following treatment of hsPDA. Further, hsPDA has been argued to alter volume of distribution of water-soluble drugs such as gentamicin, thus may affect the trough-based therapeutic drug monitoring carried out in preterm neonates.[13] We carried out a study evaluating therapeutic effect of intravenous acetaminophen in preterm neonates with hsPDA and have evaluated the pharmacodynamic interaction with frusemide in the same population.[14,15] Hence, we envisaged the present study to evaluate the effect of gentamicin use in preterm neonates diagnosed with hsPDA that are treated with acetaminophen as first-line drug. METHODS Study design and ethics The present study was a prospective, observational study carried out between April 2018 and June 2021 in the largest neonatal intensive care unit in the Kingdom after obtaining approval from the Institutional Ethics Committee (E001-PI-04/18) and Ministry of Health, Kingdom of Bahrain (AURS/108/2019). The neonates with hsPDA were recruited as a part of pharmacogenomics study related to acetaminophen from April 2019 to June 2021, while those without hsPDA belonged to an observational study related to excipient use that was carried out between April 2018 and June 2021. Consent was obtained from parents of the recruited neonates. Study participants and procedure The criteria of study participants were mentioned in our previous report.[14] We recruited only preterm neonates with gestational age <37 weeks. Neonates with hsPDA were diagnosed with echocardiographic criteria as follows: size of DA larger than 1.8 mm or >60% left pulmonary artery (LPA) with flow left to right and at least one of left atrial aortic root ratio (LA: AO) >1.5/diastolic flow reversal on descending aorta/left ventricle or left atrial dilatation/end-diastolic anterograde flow in LPA >0.2 m/s. The other groups of neonates were those that have not been identified with hsPDA. Following details were captured: gestational age; birth weight; length; Appearance, Pulse, Grimace, Activity, and Respiration (APGAR) scores; duration of stay in the intensive care unit; size of DA at baseline; status of DA at the end of acetaminophen treatment (closed/failure to close as per echocardiographic assessment); gentamicin dosing regimen (dose, frequency, and duration); and gentamicin trough concentrations. Gentamicin dosing regimen in our unit follows Micromedex NeoFax recommendations 2020. Statistical analysis Descriptive statistics was used for representing the demographic variables. We used the following criteria for classifying gestational age of the study participants: extremely preterm (<28 weeks), very preterm (28 to <32 weeks), and late preterm (32 to <37 completed weeks of gestation).[16] Birth weights were grouped as follows: 1.5 to <2.5 kg—low, 1 to <1.5 kg—very low, and <1 kg—extremely low birth weights. The baseline size of DA was categorized as small (<1.5 mm), moderate (1.5–3 mm), and large (>3 mm).[17] Distribution of the numerical variables was tested by Kolmogorov–Smirnov test and accordingly Mann–Whitney U test was chosen as they were not normally distributed. Cumulative dose of gentamicin was estimated by the cross-product of daily doses of gentamicin with duration of therapy. Binary logistic regression analysis was carried out for successful closure of DA with the following independent variables: whether received gentamicin after being diagnosed with hsPDA, gentamicin duration after diagnosis of hsPDA, gestational age, and baseline size of DA. Receiver operating characteristic (ROC) curve was plotted between the duration of gentamicin therapy in neonates with hsPDA compared to those without. Area under the curve with 95% confidence intervals was estimated from ROC curve. A P value of ≤0.05 was considered significant. SPSS (IBM Corp. Released 2020. IBM SPSS Statistics for Windows, Version 27.0. Armonk, NY: IBM Corp.) was used for statistical analysis. RESULTS Demographic characteristics Sixty neonates with hsPDA were recruited with 102 neonates without hsPDA and their demographic characteristics are listed in Table 1. More numbers of extremely and very preterm neonates were observed in the hsPDA group like the distribution of birth weight categories. Amongst those with hsPDA, following concomitant conditions were observed: respiratory distress syndrome (n = 51), suspected sepsis (n = 34), metabolic acidosis (n = 15), and coagulopathy (n = 12); three each had congenital heart disease and pulmonary hypertension, two each had thrombocytopenia and intraventricular hemorrhage, and one had birth asphyxia. In the other group, the following concomitant diagnoses were observed: suspected sepsis (n = 15), respiratory distress syndrome (n = 49), congenital diseases (n = 5), metabolic acidosis (n = 4), pneumonia (n = 4), thrombocytopenia (n = 3), pulmonary hypertension (n = 2), retinopathy of prematurity (n = 2), hypoxic–ischemic encephalopathy (n = 2), diaphragmatic hernia (n = 1), and birth asphyxia (n = 1). Table 1 Demographic characteristics of the study participants (n=162) Variables Values hsPDA (n=60) No hsPDA (n=102) Gestational age (weeks)a 27.3 (23-34.7) 32 (25-36.7) Gestational age categories (n)  Extremely preterm 35 20  Very preterm 20 29  Late preterm 5 53 Birth weight (kg)a 0.94 (0.5-2.5) 1.5 (0.7-3.3) Birth weight categories (n)  Extremely low 35 19  Very low 20 30  Low 5 44 Normal None 9 Length (cm)a 34 (22-46) 40 (29-53) Male:Female (n) 29:31 46:56 APGAR scores  First minute 6 (1-9) 8 (0-10)  Fifth minute 9 (5-10) 10 (2-10)  Tenth minute 10 (7-10) 10 (5-10) Duration of ICU stay (days) 48.5 (3-106) 21.5 (3-126) aThe values are represented in median (range) Acetaminophen-dosing regimen and the baseline size of DA in the hsPDA group The distributions of acetaminophen dose and the baseline size of DA in each of the gestational age categories are depicted in Figure 1. Fifty-three (extremely preterm—33, very preterm—16, and late preterm—4) were moderate-sized DA and seven (extremely preterm—2, very preterm—3, and late preterm—1) were large-sized and no significant association was observed between gestational age category and ductus size (P > 0.5). Figure 1 Acetaminophen dose and size of ductus arteriosus. The boxplots represent the median and ranges of acetaminophen doses (a) and the size of ductus arteriosus (b) in each gestational age category Gentamicin dosing regimen and trough concentrations Forty-three neonates with hsPDA received gentamicin during acetaminophen therapy. Figure 2 summarizes the dose, frequency, and duration of gentamicin regimen in both the groups. Median (range) of gentamicin daily doses in the hsPDA group was significantly lower than those without hsPDA [4.5 (2.5–10), 7 (3.2–13) mg; P < 0.001]. Significantly higher gentamicin daily doses were observed in the very preterm neonates without hsPDA compared to those with hsPDA [6 (4–8), 5.9 (2.7–6) mg; P < 0.05]. On the contrary, significantly longer duration of therapy was observed in those with hsPDA compared to those without hsPDA [8 (3–14), 5 (3–7) days; P < 0.001]. No significant differences were observed in gentamicin doses/duration of therapy either in the extremely or late preterm categories. ROC curve comparing the duration of therapy between hsPDA and those without hsPDA revealed a significant area under the curve (0.69: 0.58–0.79) [Figure 3]. Figure 2 Gentamicin dosing regimen in the study population. The boxplots represent the median and range of gentamicin doses (a) and duration (b) in each gestational age category. ***Statistically significant (P < 0.001) Figure 3 ROC curve for duration of gentamicin therapy. The green line at 45° represents the diagonal reference line and the blue line indicates the ROC curve for duration of gentamicin therapy Twenty-eight neonates with hsPDA and all those without hsPDA had trough gentamicin concentrations available and no significant differences were observed between the groups [Figure 4]. Figure 4 Gentamicin trough concentrations. The boxplot depicts the gentamicin trough concentrations in both the groups Gentamicin administration and DA closure Forty-eight (80%) neonates had successful closure of hsPDA. Thirty-six (78%) neonates with successful closure of DA received gentamicin compared to seven (58.3%) with patent ductus, and it was not statistically significant (P > 0.05). Significantly longer duration of gentamicin therapy was observed in those with successful closure of hsPDA compared to those failed [6 (3–10), 4 (3–14) days; P < 0.05]. However, no significant differences were observed in the duration of acetaminophen therapy between those with successful closure and those that failed [5 (2–11), 4 (3–8) days, P > 0.5]. Extremely preterm neonates received significantly lower cumulative doses of gentamicin compared to late preterm neonates [extremely preterm: 10.9 (4.5–30), very preterm: 22 (9–36), and late preterm: 22.75 (9–50) mg; P < 0.01]. Similarly, those with successful closure had received significantly more cumulative doses of gentamicin compared to that failed [14 (4.5–50), 9 (6.75–17.5) mg; P < 0.05]. Logistic regression analysis did not reveal any significant association between gentamicin administration or not, gentamicin duration, gestational age, and baseline size of DA categories with closure of hsPDA. DISCUSSION We evaluated the association between gentamicin use in 60 neonates with hsPDA and compared with those without hsPDA (n = 102). Those with hsPDA had significantly lower daily doses of gentamicin but longer duration of therapy than those without hsPDA in very preterm neonates. No significant differences were observed in the trough concentrations of gentamicin between the groups. No association was observed between gentamicin use and closure of DA. However, those with successful closure of DA received gentamicin for a longer duration and had higher cumulative doses that were independent of acetaminophen duration. Despite a plethora of preclinical and in-vitro studies evaluating the association between aminoglycoside use and its effect on arterial blood vessels, limited evidence exists from clinical studies. A recent study from eight neonatal intensive care units in Europe revealed that the use of gentamicin and tobramycin in neonates with hsPDA increases the risk of surgical intervention for PDA closure with an odds ratio of 1.6 and 1.3, respectively.[18] However, the treatment for hsPDA varied between the neonatal units, including the criteria for surgical intervention that has not been considered by the authors in that study. Cakir et al.[19] reported that neonates with hsPDA were more likely to receive gentamicin compared to others; and that the duration of gentamicin was significantly longer than those without hsPDA (7 days vs. 9 days, P < 0.001) and gentamicin therapy more than 7 days is highly predictive of hsPDA. Although we observed significantly lower daily doses of gentamicin in our neonates with hsPDA, this is attributable to the lower body weight as most of the neonates in this group were extremely preterm compared to those without hsPDA. Similarly, the prolonged duration of gentamicin therapy amongst those with hsPDA can be attributed to presence of other concomitant diseases such as sepsis. We did not observe any significant differences in the trough concentrations between those with hsPDA compared to those without. However, Watterberg et al.[20] in a small group of neonates (n = 24) observed significantly greater apparent volumes of distribution [0.64 (0.20), 0.41 (0.08) L/kg; P < 0.001] and serum half-life [8.49 (2.69), 6.23 (1.92) h; P < 0.01] without any changes in total clearance. Similarly, Williams et al.[21] reported reduced clearance in addition to increased volume of distribution for gentamicin in neonates with PDA. However, the neonates in that study received loading gentamicin dose followed by maintenance doses; thus, a traditional dosing regimen was followed. On the contrary, Touw et al.[22] recently reported that only small differences that are clinically insignificant exist between the neonates with PDA and those with closed DA. Routine monitoring of trough gentamicin concentrations has not been shown to predict the risk of adverse drug reactions in neonates with lower risk factors.[23] However, in the absence of concrete reports attributing the pharmacokinetic changes of gentamicin in PDA, we recommend continuing the therapeutic drug monitoring with trough gentamicin concentrations. The strength of the present study is that it is prospective as the previous studies that explored similar hypotheses were retrospective; and this is the first report with the use of acetaminophen for treating hsPDA, while the previous reports were related mainly to either indomethacin or ibuprofen. However, a matched case-control study where potential confounders such as concomitant diseases, body weight, baseline DA size, and gestational age could be the best observational study design (although cumbersome) is preferable. The present study is limited by the following: we could not estimate the pharmacokinetic parameters due to limited sampling data points; and we did not consider concomitant diseases, particularly sepsis, as a factor due to unavailability of data. In conclusion, we observed a significantly longer duration of gentamicin therapy in neonates with hsPDA compared to those without hsPDA. No significant differences were observed in the rates of closure of DA with concomitant gentamicin administration. Although we did not observe any significant differences in the gentamicin trough concentrations between those with hsPDA and those without, we recommend continuing therapeutic drug monitoring in this subpopulation considering the uncertainty about the changes in the pharmacokinetic parameters. Data availability statement The data will be provided by the corresponding author with a reasonable request. What is already known Gentamicin dilates arterial blood vessels in preclinical studies. Patent ductus arteriosus may alter the pharmacokinetics of gentamicin owing to changes in the volume of distribution. There are hardly any prospective studies evaluating the effect of gentamicin in neonates with hemodynamically significant patent ductus arteriosus. What this study adds This is the first prospective study evaluating the effect of gentamicin in neonates with hemodynamically significant patent ductus arteriosus. We did not observe any significant differences in the rates of closure of ductus arteriosus with gentamicin, and the trough concentrations compared to those without patent ductus arteriosus. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. Acknowledgement We thank the Institutional Ethics Committees for proving approval for carrying out this study. ==== Refs REFERENCES 1 Dice JE Bhatia J Patent ductus arteriosus: An overview J Pediatr Pharmacol Ther 2007 12 138 46 23055849 2 Crockett SL Berger CD Shelton EL Reese J Molecular and mechanical factors contributing to ductus arteriosus patency and closure Congenit Heart Dis 2019 14 15 20 30468303 3 Schneider DJ Moore JW Patent ductus arteriosus Circulation 2006 114 1873 82 17060397 4 Koch J Hensley G Roy L Prevalence of spontaneous closure of the ductus arteriosus in neonates at a birth weight of 1000 grams or less Pediatrics 2006 117 1113 21 16585305 5 Miller AL Langton PD Streptomycin inhibition of myogenic tone, K+-induced force and block of L-type calcium current in rat cerebral arteries J Physiol 1998 508 793 800 9518733 6 Gotanda K Yanagisawa T Satoh K Taira N Are the cardiovascular effects of gentamicin similar to those of calcium antagonists? Jpn J Pharmacol 1988 47 217 27 3221528 7 Belus A White E Effects of antibiotics on the contractility and Ca2+transients of rat cardiac myocytes Eur J Pharmacol 2001 412 121 6 11165223 8 Gergawy M Vollrath B Cook D The mechanism by which aminoglycoside antibiotics cause vasodilation of canine cerebral arteries Br J Pharmacol 1998 125 1150 7 9863641 9 Adams HR Goodman FR Weiss GB Alteration of contractile function and calcium ion movements in vascular smooth muscle by gentamicin and other aminoglycoside antibiotics Antimicrob Agents Chemother 1974 5 640 6 15825418 10 Sridharan K Hasan H Al Jufairi M Al Daylami A Abdul Azeez Pasha S Al Ansari E Drug utilisation in adult, paediatric and neonatal intensive care units, with an emphasis on systemic antimicrobials Anaesthesiol Intensive Ther 2021 53 18 24 33625820 11 Martius JA Roos T Gora B Oehler MK Schrod L Papadopoulos T Risk factors associated with early-onset sepsis in premature infants Eur J Obstet Gynecol Reprod Biol 1999 85 151 8 10584628 12 Vucovich MM Cotton RB Shelton EL Goettel JA Ehinger NJ Poole SD Aminoglycoside-mediated relaxation of the ductus arteriosus in sepsis-associated PDA Am J Physiol Heart Circ Physiol 2014 307 H732 40 24993047 13 Gal P Gilman JT Drug disposition in neonates with patent ductus arteriosus Ann Pharmacother 1993 27 1383 8 8286815 14 Sridharan K Al Jufairi M Al Ansari E Al Marzooq R Hubail Z Hasan SJR Intravenous acetaminophen (at 15 mg/kg/dose every 6 hours) in critically ill preterm neonates with patent ductus arteriosus: A prospective study J Clin Pharm Ther 2021 doi: 10.1111/jcpt.13384 15 Sridharan K Al Madhoob A Al Jufairi M Al Ansari E Al Marzooq R Intravenous frusemide does not interact pharmacodynamically with acetaminophen in critically ill preterm neonates with patent ductus arteriosus Eur Rev Med Pharmacol Sci 2021 25 1612 5 33629330 16 Quinn JA Munoz FM Gonik B Frau L Cutland C Mallett-Moore T Preterm birth: Case definition &guidelines for data collection, analysis, and presentation of immunization safety data Vaccine 2016 34 6047 56 27743648 17 Arlettaz R Echocardiographic evaluation of patent ductus arteriosus in preterm infants Front Pediatr 2017 5 147 28680875 18 Marissen J Erdmann H Böckenholt K Hoppenz M Rausch TK Härtel C Aminoglycosides were associated with higher rates of surgical patent ductus arteriosus closure in preterm infants Acta Paediatr 2021 110 826 832 32810301 19 Cakir U Tayman C Relationship between gentamicin administration and ductal patency in very low birth weight infants Curr Clin Pharmacol 2021 doi: 10.2174/1574884716666210603110412 20 Watterberg KL Kelly HW Johnson JD Aldrich M Angelus P Effect of patent ductus arteriosus on gentamicin pharmacokinetics in very low birth weight (less than 1,500 g) babies Dev Pharmacol Ther 1987 10 107 17 3608741 21 Williams BS Ransom JL Gal P Carlos RQ Smith M Schall SA Gentamicin pharmacokinetics in neonates with patent ductus arteriosus Crit Care Med 1997 25 273 5 9034263 22 Touw DJ Proost JH Stevens R Lafeber HN van Weissenbruch MM Gentamicin pharmacokinetics in preterm infants with a patent and a closed ductus arteriosus Pharm World Sci 2001 23 200 4 11721679 23 Ibrahim J Maffei D El-Chaar G Islam S Ponnaiya S Nayak A Should gentamicin trough levels be routinely obtained in term neonates? J Perinatol 2016 36 962 5 27537855

PMC010xxxxxx/PMC10353666.txt

==== Front J Pharm Bioallied Sci J Pharm Bioallied Sci JPBS J Pharm Bioall Sci Journal of Pharmacy & Bioallied Sciences 0976-4879 0975-7406 Wolters Kluwer - Medknow India JPBS-15-75 10.4103/jpbs.jpbs_917_21 Original Article Antidiabetic Activity of Combination of Binahong (Anredera cordifolia Ten. Steenis), Cherry (Muntingia calabura L.) and Brotowali (Tinospora crispa L.) Extracts Kusriani Herni Susilawati Elis Nurafipah Lytia Nurkholifah Department of Biology Pharmacy, Faculty of Pharmacy, Bhakti Kencana University, Bandung, Indonesia Address for correspondence: Dr. Herni Kusriani, Faculty of Pharmacy, Bhakti Kencana University, Bandung, Indonesia. E-mail: herni.kusriani@bku.ac.id Apr-Jun 2023 08 6 2023 15 2 7580 31 12 2021 30 12 2022 17 1 2023 Copyright: © 2023 Journal of Pharmacy and Bioallied Sciences 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. ABSTRACT Background: Diabetes mellitus (DM) is a group of metabolic disorders characterized by hyperglycemia. Diabetes mellitus is a silent killer because sufferers are often not aware of it when it is realized, complications have occurred. Treatment for this disease must be done for life to control blood sugar in the body; however, oral antidiabetic drugs often produce unwanted side effects such as bloating, diarrhea, and stomach cramps. One of the treatments for diabetes is to find sources of treatment using natural ingredients that are relatively safe, including using plants as medicines. Based on several studies, Binahong leaves (Anredera cordifolia Ten. Steenis), brotowali (Tinospora crispa L.), and cherry (Muntingia calabura L.) are medicinal plants that can be used to reduce blood sugar levels. This study aims to test the antidiabetic activity using in vivo and in vitro testing methods of extracts of binahong leaves, cherry leaves, brotowali stems and their combinations. Methods: In vivo method uses animal modeling of insulin deficiency, whereas in vitro method with alpha glycosidase inhibition activity assay. The administration of extracts was repeated every day for 14 days and blood glucose levels were measured on the 0, 7th, and 14th days. Then surgery was performed on the pancreas and calculated the area of the islets of Langerhans, and the number of alpha and beta cells in the pancreas. The inhibitory activity of the alpha-glucosidase enzyme with the IC50 value of each extract and its combination was determined, with acarbose used as a standard. Result: The combination of binahong leaves (Anredera cordifolia Ten.Steenis) and cherry leaves (Muntingia calabura L.) and the combination of brotowali stems (Tinospora crispa (L.) and binahong leaves showed in vivo antidiabetic activity with insulin deficiency method. The combination of these extracts was able to reduce blood sugar levels until the observation on day 14. In in vitro testing by inhibiting alpha-glucosidase enzymes, binahong leaves extract, brotowali stems, and cherry leaves were able to inhibit alpha-glucosidase enzymes at IC50 of 35.07 ± 2.35; 29.42 ± 1.40; and 26.63 ± 1.15, respectively. Conclusion: The best combination of extracts by in vitro and in vivo methods was shown in the combination of binahong leaf and brotowal stem extract binahong leaves, brotowali stems (2:1). KEYWORDS: A combination extract antidiabetic binahong leaves (Anredera cordifolia Ten. Steenis) brotowali stems (Tinospora crispa (L.) cherry leaves (Muntingia calabura L.) the inhibitory activity of alpha-glucosidase insulin deficiency ==== Body pmcINTRODUCTION Diabetes mellitus (DM) is a group of metabolic disorders characterized by high blood glucose levels (hyperglycemia) and abnormalities in carbohydrate, fat, and protein metabolism. Due to disturbances or defects in insulin secretion, insulin sensitivity, or both.[1] Diabetes mellitus is also known as the silent killer because sufferers are often unaware and when it is realized, complications have occurred. According to the 2019 International Diabetes Federation data, it is estimated that there are 223 million women aged 20–79 years living with diabetes. By 2045, the number is expected to increase to 343 million people with diabetes aged 20–79 years.[2] According to the World Health Agency, the World Health Organization (WHO) predicts an increase in the number of people with diabetes mellitus, which is one of the global health threats. Treatment of diabetes can be done by several mechanisms such as suppressing hepatic glucose production (biguanides), stimulating insulin secretion (sulfonylureas and glinides), delaying the digestion and absorption of intestinal carbohydrates to maintain postprandial glucose levels (α-glucosidase and -amylase inhibitors), increasing receptor sensitivity. insulin and peripheral glucose uptake (thiazolidinedione and metformin) or insulin.[1] Various efforts to prevent and treat it continue to be made because most oral antidiabetic drugs provide unwanted side effects such as bloating, diarrhea, and stomach cramps, thus experts conduct research including developing various types of treatment. One of the treatments for diabetes is to find sources of treatment using natural ingredients that are relatively safe, including using plants as medicines.[3] One of the plants that can be used to lower blood sugar levels in the community is binahong leaves (Anredera cordifolia Ten.Steenis). Binahong is a vine plant that comes from the Basellaceae family. People usually use the leaves of binahong as a medicine for skin wounds and post-operative wounds, ulcers, hypertension, inflammation, and gout.[4-7] Binahong leaves contain secondary metabolites including saponins, alkaloids, flavonoids, phenols, triterpenoids, and sterols.[8] In addition to binahong leaves, there are stems of brotowali, which can also be used to lower blood sugar levels. Brotowali is a medicinal plant belonging to the Menispermaceae family. Brotowali is also a vine plant that lives in the tropics and is evenly distributed in Indonesian forests. Brotowali is known to contain alkaloids, flavonoids, saponins, and tannins.[9,10] Furthermore, the cherry leaf is also one of the Indonesian plant known as kersen, which contains several secondary metabolites including flavonoids, triterpenoids, tannins, saponins, and glycosides. Some research on this plant showed antidiabetic activity, gout, hypertension, cough laxative, flu, headache, antiseptic, antioxidant, and anti-inflammatory activities.[11] Based on empirical information and several studies support that binahong leaf extract (Anredera cordifolia Ten.Steenis), brotowali stem (Tinospora crispa L.), and cherry leaves (Muntingia calabura L.) can lower blood glucose levels with various active doses. In this study, we examined the activity of the three extracts and their combination in therapeutic use for the treatment of diabetes mellitus with in vitro and in vivo methods of lowering blood sugar. MATERIALS AND METHODS The study was conducted in several stages, including preparation of materials, extraction, in vitro alfa glucosidase inhibitory assay, deficiency insulin assay, and histology pancreas. Chemicals and reagents The materials used in this experiment were glucosidase enzymes, acarbose, dilute hydrochloric acid LP, 96% ethanol, 2% HCl, NaOH, dimethyl sulfoxide, p-nitrophenyl-α-D-glucopyranoside, Na2CO3, Aqua Dest, 0.1 M phosphate buffer solution (pH 6.8). Plants Binahong leaves (Anredera cordifolia Ten.Steenis), brotowali stems (Tinospora crispa L.), and cherry leaves (Muntingia calabura L.) [Figure 1] were planted in the local garden in Bandung City, West Java, Indonesia. This plant was identified and confirmed by Herbarium Jatinangor, Biology Department of Padjadjaran University, Indonesia. Figure 1 Anredera cordifolia Ten. Steenis (a) Muntingia calabura L. (b) Tinospora crispa L. (c) Sample preparation Binahong leaves (Anredera cordifolia Ten.Steenis), brotowali stems (Tinospora crispa L.), and cherry leaves (Muntingia calabura L.) were washed, cleaned in running tap water, cut into small pieces, and dried to constant weight using an oven at 40°C. The sample was pounded into small size using a grinder machine; it was stored in an air-tight container until when needed. Extraction Binahong leaf powder, cherry leaf powder, and brotowali stem powder that has been sifted were weighed as much as 200 g each, then extracted using 96% ethanol. Powder of each simplicia as much as 200 g was macerated with 2000 mL of ethanol 96% ratio (1:10) for 3 × 24 h, then filtered. The macerate was then concentrated with a rotary evaporator and weighed so that the extract yield was known. Experimental animal White Wistar rats aged 2–3 months and weighing 150–250 g (University of Padjajaran) were used in this study. Before animals were tested, they were adapted to new cage environments including feeding. Animal experiments were conducted according to the Commission of the Ethics of Health Research Faculty of Medicine, the University of Padjajaran Bandung (613/UN6.KEP/EC/2021). Insulin deficiency This method was carried out curatively using Swiss Webster mice as test animals. An animal model of insulin resistance was formed using the induction of Alloxan 55–60 mg/kg BW orally for 14 days with the parameter observed as the value of blood sugar levels. The acclimatized mice fasted for 6–8 h, and then their initial blood glucose levels were measured and then randomly divided into six groups, namely the positive control group, the negative control group, the Glibenclamide group 65 mg/Kg BB, sample extract 75, 150, and 300 mg/kg BW. After that, alloxan was used to induce mice in all groups except the negative group (−). After 3 days of induction, therapy was given for 14 days orally. On day 3rd (t0), 7th (t7), and 14th (t14). days the parameters blood glucose levels were measured.[12,13] Alfa glucosidase inhibitory activity assay The α-glucosidase enzyme plays a role in the process of converting carbohydrates into glucose; therefore, if there is inhibition of the activity of α-glucosidase it can lower blood sugar. Enzyme inhibitory activity was tested using 50 μL of phosphate buffer solution, added 25 μL of test sample/acarbose solution, added 25 μL of 0.1 M phosphate buffer (pH 6.8) containing glucosidase solution (0.6 U/mL) incubated in 96-well plates at 37°C for 15 min. After pre-incubation, 25 μL of 15 mM p-nitrophenyl-α-D-glucopyranoside (pNPG) in phosphate buffer (pH 6.8) was added to each well and incubated at 37°C for 15 min. Then, the reaction was stopped by adding 150 μL of 0.2 M NaCO3 to each well, and the absorbance reading was recorded at a wavelength of 405 nm with a microplate reader. RESULTS AND DISCUSSION In this study, we tested alfa glucosidase inhibitory activity and insulin resistance from Binahong leaf, cherry leaf, and brotowali stem extracts and their combinations. The powdered simplicia of binahong leaves, cherry leaves, and brotowali stems were each weighed as much as 200 g and then extracted by maceration with a ratio of 1:10. The principle of maceration is that the liquid will penetrate the cell wall and enter the cell. The extraction method chosen was maceration because it is easy and uses simple tools and can avoid damage to compounds that are not heat resistant. Maceration was carried out for 3 × 24 h using 96% ethanol as solvent. Liquid extracts then were concentrated by evaporation of the solvent until the extract became thick. Binahong leaves, cherry leaves, and brotowali stem thick extract, respectively, were obtained at 16.705 g, 31.78 g, and 27.82 g. Table 1 shows the extract yield. Table 1 Extract rendemen Simplicia Extract rendemen (%) Binahong leaves 8.35 Cherry leaves 15.89 Brotowali stem 13.91 Insulin deficiency This test was carried out curatively where the test animals were induced first and then treated. The comparison drug used was glibenclamide with its mechanism of increasing insulin release from the pancreas. In this method, alloxan was used at a dose of 55–60 mg/kg BW to destroy the pancreatic cells in the insulin-deficient animal model.[12] Alloxan was administered intravenously to the tails of mice in all groups except the negative control group (−). After 3 days of alloxan induction, fasting blood sugar levels were checked in mice using the Easy Touch® glucometer. Animals used for research included animals with blood glucose levels reaching 200 mg/dL (t0). The test animals to be studied were divided into several groups, namely group I as a negative control, group II as a positive control, group III as a comparison using glibenclamide 0.65 mg/Kg BW, and group IV–VI combination extracts. For treatment, it was given orally for 14 days. During therapy, fasting blood glucose levels will be checked on days 3(t0), 7(t7), 11(t11), and 14(t14). Blood sampling was carried out by piercing the venous blood flow using a lancet on the tail of the mice. In vivo antidiabetic test results method of insulin deficiency are shown in Table 2. The results showed a decrease in levels of blood glucose that varied enough to give a fairly large standard deviation. In general, the whole test group showed a decrease in glucose levels in blood on the 14th day compared to the condition beginning, except in combination 3. Based on the control group’s positive data, the test animal still showed hyperglycemia until the end of day 14; so, this alloxan diabetes method is suitable for use as an animal model for testing antidiabetic activity. Table 2 Insulin deficiency Group Blood sugar level (mg/dL) at days 0 7 14 Normal control 119.5±19.94 137.5±8.34 141±23.51 Positive control 458±92.77 477.87±89.92 500.20±106.62 Standard 252.54±122.79 248.48±101.92 242.60±80.14 Combination 1 (Binahong: kersen) 343.5±94.62 262±81.79 186.75±95.36 Combination 2 (Binahong: brotowali) 381.25±140.04 266±148.70 196±102.31 Combination 3 (Kersen: brotowali) 436.75±125.55 310.75±125.93 202.5±144.96 In the results of the test using the insulin deficiency method, ethanol extract of cherry leaves, binahong leaves, and stem brotowali had good antidiabetic activity, which showed a decrease in blood glucose in mice on 14th day as well as the administration of glibenclamide as a standard. Animals that have been tested for 19 days were sacrificed, two from each group, then the pancreas was isolated and Gomori staining was performed using victoria blue dye and floxin.[13] Gomori staining results are shown in Figure 2. Figure 2 Pancreas histology image. Negative control (a) Positive control (b) Standard (c) Combination1 (d) Combination 2 (e) Combination 3 (f) Parameters observed to assess the success of therapy can be seen from the average area islets of Langerhans, the number of alpha cells, and the number of beta cells in the islets of Langerhans in units of the same area [Figure 2]. Alpha cells have a portion of about 20%, whereas beta cells have a portion of about 75% of the islets of Langerhans. Alpha cells produce the hormone glucagon, whereas beta cells produce insulin. There is improvement in therapy indicated by the high number of beta cells, on the contrary, the high number of alpha cells will aggravate the condition of diabetes mellitus. The results of histologic observations can be seen in the following image: Inhibitory α glucosidase activity The inhibitory activity of the alpha-glucosidase enzyme was tested using the acarbose standard. It aims to compare the IC50 value of acarbose with the test sample. Acarbose was chosen as a comparison because it is an antidiabetic drug that works by inhibiting glucosidase. In addition, in terms of structure, acarbose has a similar structure to the substrate p-nitrophenyl-α-D-glucopyranoside (p-NPG).[14] This research was conducted at pH 6.8 and a temperature of 37°C, adjusted to the optimum reaction temperature enzymatic. Reduction of enzyme activity can be caused by storage factors after production, which do not meet the criteria listed on the enzyme label, such as exposed sunlight, storage temperature too high, and other things that can trigger loss of enzyme activity. Enzyme activity determined in international units, namely the number of enzymes that catalyze the formation of 1.0 mol D-glucose from p-nitrophenyl-α-D glucosidase at pH 6.8 and 37°C for 1 min. Incubation was carried out in two stages, first incubation for 20 min to give time for the test solution to reach a temperature of 37°C, and the second stage of incubation for 20 min, which is incubation for enzymatic reactions. Then, the reaction was stopped by adding natrium carbonate. The product of the reaction between glucosidase and p-nitrophenyl-α-D-glucosidase is p-nitrophenyl, which is yellow; thus, this product can be measured with a microreader at 425 nm wavelength. To correct the absorption generated, the measurement was carried out by blank absorption by replacing the position enzymes with buffers. It is intended to see the absorption given by the test sample without an enzymatic reaction. The magnitude of the inhibition of enzyme activity was seen from the decrease in p-nitrophenyl, which formed when compared to the activity of early enzymes. The magnitude of the inhibitory activity of each concentration range is seen from the percentage of the resulting inhibition; the greater the percentage of inhibition produced, the more inhibited many enzyme activities.[15,16] In this test, sample absorbance and blank absorbance were measured. The treatment for measuring the absorbance of the sample and blank was the same, using sample control and blank control as a correction factor for the sample and blank absorbance values because the color of the extract can also provide absorption at that wavelength.[15] The absorbance of p-nitrophenol formed was measured at a wavelength of 405 nm. The enzyme inhibitory activity test was carried out on the ethanol extract of binahong leaves, cherry leaves, and brotowali stems and their combinations. Each extract was weighed as much as 10 mg and added DMSO (dimethyl sulfoxide), which serves to help dissolve the extract as much as 100 L in the micropipette and made up to 10 mL with phosphate buffer pH 6.8 to obtain a concentration of 1000 g/mL. Then, the dilution was carried out with phosphate buffer pH 6.8 to obtain a concentration of 400 g/mL; 200 g/mL; 100 g/mL; 50 g/mL; 25 g/mL. The results of the glucosidase activity inhibition test from the sample and acarbose are shown in Figure 3. Figure 3 Inhibition α glucosidase activity extract After obtaining the IC50 value of each extract [Table 3], further testing was carried out on the combination of extracts, where the concentration of each extract in the combination was 15 g/mL (equivalent to an average of half the IC50 value of each extract). Table 3 Inhibitory α-glucosidase activity Sample IC50±SD (µg/mL) Cherry leaves 26.63±1.15 Binahong leaves 35.07±2.35 Brotowali stem 29.42±1.40 Standard acarbose 47.093±6.26 From the combination of extracts result, it was found that the combination of extracts of cherry: binahong 1:1, 1:2, and 2:1, binahong: brotowali 1:1, 1:2, and 2:1 can inhibit alpha-glucosidase enzymes successively in 85.11; 88.58; 79.48; 85.29; 83.49, and 91.29% [Figure 4]. Figure 4 Inhibition α glucosidase activity combination extract The best combination in inhibiting the alpha-glucosidase enzyme is a combination of extracts binahong: brotowali 2:1 with the highest inhibition of the alpha-glucosidase enzyme at 91.29%. CONCLUSION The combination of binahong leaves (Anredera cordifolia Ten.Steenis) and cherry leaves (Muntingia calabura L. and the combination of brotowali stems (Tinospora crispa (L.) and binahong leaves showed in vivo antidiabetic activity with insulin deficiency method. The combination of these extracts was able to reduce blood sugar levels until the observation on day 14. Meanwhile, the combination of brotowali stems and cherry leaves could not reduce blood sugar levels in in vitro tests. In vitro testing by inhibiting alpha-glucosidase enzymes, both binahong leaves extract, brotowali stems, and cherry leaves were able to inhibit alpha-glucosidase enzymes at IC50, respectively, that is, 35.07 ± 2.35; 29.42 ± 1.40; and 26.63 ± 1.15 µg/mL. The best combination of extracts by in vitro and in vivo methods was shown in the combination of binahong leaf and brotowal stem extract binahong leaves, brotowali stems (2:1). This research needs to be continued with making the preparations containing binahong leaf-brotowali stem extract and binahong leaves-brotowali stems extract (2:1). Financial support and sponsorship Bhakti Kencana University Research Funding. Conflicts of interest There are no conflicts of interest. ==== Refs REFERENCES 1 Dipiro JT Yee GC Posey LM Haines ST Nolin TD dan Ellingrod V Pharmacotherapy: A Pathophysiologic Approach 11th Edition 2020 McGraw-Hill Companies 2 IDF IDF Diabetes Atlas 6th ed S. L. International Diabetes Federation 2019 3 Modak M Dixn P Londir J Ghaskadai S dan Devasagayam T Indian herbs and herbal drugs used for the treatment of diabetes J Clin Biochem Nutr 2007 40 163 73 18398493 4 Octavia DR Uji aktivitas penangkap radika lekstrak petroleum eter, etil asetat dan etanol daunbinahong (Anredera cordifolia (Tenore) Steen) dengan metode DPPH (2, 2-difenil-1-pikrihidrazil) Doctoral Dissertation, Univerversitas Muhammadiyah Surakarta) 2009 5 Manoi F Binahong Anrederacordifolia (Tenore Steen) sebagai Obat J Warta Penelitian dan Pengembangan Tanaman Industri 2009 153 5 6 Djamil R Winarti W Zaidan S Abdillah S Antidiabetic activity of flavonoid from binahong leaves (Anredera cordifolia) extract in alloxan induced mice JPharmacogn Nat Prod 2017 03 2 5 7 Elya B Basah K Mun'im A Yuliastuti W Bangun A Septiana E Screening of α-glucosidase inhibitory activity from some plants of Apocynaceae, Clusiaceae, Euphorbiaceae, and Rubiaceae J Biomed Biotechnol 2012 281078 22187534 8 Restykania Suratman Pitoyo A Suranto, Morphology and isozyme variation among madeira vine (Anredera cordifolia) accessions from the southeastern part of Central Java, Indonesia Biodiversitas 2019 20 3024 32 9 Tarukbua YSF De Queljoe E Bodhi W Phytochemical screening and toxicity with Brine Shrimp Lethality Test (BSLT) of Brotowali Tinosporacrispa (L.) Pharmacon Jurnal Ilmiah Farmasi 2018 7 3 330 337 10 Manaharan T Teng LL Appleton D Ming CH Masilamani T Palanisamy U Potensiantioksidan dan antiglikemik bagian tumbuhan Peltophorum pterocarpum Kimia Makanan 2011 129 1355 61 11 Shinde J Taldone T Barletta M Kunaparaju N Hu B Kumar S Aktivitas penghambatan -Glucosidase dari Syzygium cumini (Linn.). Skeels seed kernel in vitro dan pada tikus Goto-Kakizaki (GK) Penelit Karbohidrat 2008 343 1278 81 12 Etuk EU Animals models for studying diabetes mellitus Agric Biol JN Am 2010 1 130 4 13 Susilawati E. Adnyana I.K Fisheri N Studies on The Antidiabetic Activity of Ethanol Exract and Its Fractions of Singawalang (Petiveria alliace L) Leaves in Mice Pharmacy Journal 2016 13 182 191 14 Gomori G A differential stain for cell types in the pancreatic islets Am J Pathol 1939 15 497 9 IDF. IDF Diabetes Atlas 9th ed Belgium International Diabetes Federation; 2019 19970466 15 Kaskoos AR In-vitro α-glucosidase inhibition and antioxidant activity of methanolic extract of Centaurea calcitrapa from Iraq Am J Essent OilsNatur Prod 2013 1 122 5 16 Standards of medical care in diabetes—2010 Diabetes Care 2010 33 Suppl 1 S11 61 20042772

PMC010xxxxxx/PMC10353667.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-471 10.4103/ijcm.ijcm_782_22 Original Article Assessment of Home-Based Care for Young Child (HBYC) Program in Aspirational Districts of Madhya Pradesh, India: A Cross-Sectional Study Das Priyanka Singh Manish Sakalle Shailesh K. Bhargava Saurabh 1 Khanna Rajat 1 Ganvir Dipak R. 1 Singh Ravindra 1 Goel Nimisha 1 Yadav Vivek 1 Bhat Ashfaq A. 1 National Health Mission, Madhya Pradesh, India 1 Norway India Partnership Initiative, India Address for correspondence: Dr. Rajat Khanna, B-84, Defence Colony, New Delhi - 110 024, India. E-mail: rkhanna27@gmail.com May-Jun 2023 30 5 2023 48 3 471477 16 9 2022 20 4 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Context: In 2018, Government of India initiated Home-Based Care for Young Child (HBYC) program having five quarterly structured home visits for children age 3 to 15 months to promote early childhood development. Assess knowledge and practices of Accredited Social Health Activist (ASHAs), other health functionaries, and mothers related to HBYC. Cross-sectional evaluation design with ASHAs, AWWs, ANMs, ASHA, and mothers of 3 to 15 month’s children as participants. Material and Methods: Knowledge and practices of 801 ASHAs, 200 other health functionaries, and 787 mothers were assessed on exclusive breastfeeding, complementary feeding, hand washing, iron folic acid (IFA) and oral rehydration solution (ORS) supplementation, danger referral signs in eight aspirational districts of Madhya Pradesh. Results: 88% ASHAs had correct knowledge on ORS, 85% on complementary feeding, 85% on adequacy of IFA, and 47% on danger signs which required child referral. Similarly, 85% of mothers had knowledge on exclusive breastfeeding, 40% mothers knew about complementary feeding, and only 18% knew correct ORS preparation. Statistically significant association was observed between ASHAs home visits and availability of ORS with mothers and their knowledge on correct Initiation of IFA (p < 0.001). Conclusion: Findings of study confirmed majority of health functionaries were aware about roles, responsibilities, and key tasks under HBYC. However, there observed a gap in knowledge transfer by health functionaries and thus inadequate translation of knowledge into practices among mothers on HBYC. This necessitates the need of appropriate actions from health system strengthening to capacity building to accelerate uptake of HBYC program. ASHA Child health Feeding practices HBYC Home visits ==== Body pmcINTRODUCTION Globally, past three decades have witnessed remarkable progress in child heath with millions of children having a better chance of survival. However, childhood mortality still remains high and an area of concern in developing countries.[1] India contributes to 20% of global under-five mortality.[2,3] According to findings of National Family Health Survey (NFHS 5), less than 12% of children (aged 6–23 months) receive adequate diet; around 36% of children (aged 12 months) receive complete immunization, and more than 67% of children below five years are anemic.[3-5] The country has observed a slow decline in child mortality rates[2] largely due to socio-demographic variance, malnutrition, poor sanitation, inaccessible healthcare services, and illiteracy.[3] India’s National Health Policy (NHP) 2017 envisages attainment of the highest possible level of health and well-being for all, including young children.[6] This includes provision of home-based health services by health functionaries to promote growth and development of newborn and children.[7] In 2011, Government of India (GoI) launched the Home-Based Newborn Care (HBNC) program to ensure optimal health for newborns and young children.[8] Under the program, Accredited Social Health Activist (ASHAs) conducted home visitations till 42 days.[8] However, a loss of contact still existed between newborns and healthcare system after the newborn period of 42 days.[8] The intervention: HBYC program In 2018, GoI initiated Home-Based Care for Young Child (HBYC) program to overcome programmatic gaps of HBNC and addressing disconnect between health system and young children (aged three months to two years).[9] The strategic goal of the program is to reduce childhood mortality and morbidity by promoting child nutrition, immunization, and hygiene practices and minimize common childhood illnesses.[9] The key services provisioned under HBYC program are presented below: Child development Age appropriate play and communication for children. Nutrition Exclusive breastfeeding Adequate complementary feeding Breastfeeding up to two years of age IFA supplementation Promote use of fortified food. Child health Full immunization for children Regular growth monitoring Appropriate use of ORS during diarrhea Early care seeking during sickness. Wash Appropriate hand washing practices. In Madhya Pradesh, HBYC program has been implemented in eight districts, namely Badwani, Damoh, Chhatarpur, Guna, Khandwa, Rajgarh, Singrauli, and Vidisha. All these districts are identified as “aspirational districts” by NITI Ayog, Government of India, due to poor health and development indices.[10] HBYC consists of a package of five structured home visits at third, sixth, ninth, 12th, and 15th months by ASHAs to all children attaining the age of three months. Under the program, ASHAs provide counseling of mothers on exclusive breastfeeding; hand hygiene practices; monitoring of immunization status; weighing and plotting growth charts; identification of danger signs1, and referral of sick newborns to health facilities.[11] For young child between three months up to two years, additional tasks are being performed including timely initiation of complimentary feeding; distribution of oral rehydration solution (ORS), iron folic acid (IFA) syrups; age appropriate play and communication as well as identification and referral of underweight children. The core tasks being performed by health functionaries under HBYC program are enlisted below: ASHA at third month visit Support for exclusive breastfeeding Counsel on hand washing practices Appropriate play and communication Check immunization status Monitor weight recording. ASHA at ninth, 12th, and 15th month visit. All the above activities plus the following: Counsel on initiation of complementary feeding and continued breastfeeding Age appropriate and adequate complementary feeding for children Age appropriate play and communication Ensure full immunization Distribution and counseling on prophylactic IFA and ORS. AWW at third month visit Monthly weighing of infants Plotting on growth chart Detect underweight children and take further action Counsel mother for exclusive breastfeeding. AWW at ninth, 12th, and 15th month visit Distribution on “Take Home Ration” Nutritional counseling to mother Counseling on complementary feeding Growth monitoring and weight recording on growth chart Identification of underweight children Counsel for deworming of children above 1 year of age. This study presents findings of an assessment conducted with health functionaries and mothers of children aged 3 to 15 months to determine their knowledge and practices with respect to HBYC. MATERIAL AND METHODS Study design This study employed a cross-sectional evaluation design. The data collection was conducted in the months of February and March 2021. Study participants The study participants were ASHAs, AWWs, ANMs, ASHA facilitators as well as eligible mothers with children aged 3 to 15 months. Sample size and sampling The data collection was carried out in all eight aspirational districts of Madhya Pradesh. The primary sampling unit (PSU) for this assessment was a village. Hence, two random blocks within each district and four villages with the highest population from each selected blocks were included in the sample. It means a total of eight villages per district and 64 villages across all eight aspirational district were included. For ASHAs, sample size required was calculated using evaluated knowledge level of ASHAs (80.3%) drawn from previous studies[12]; 95% confidence interval; 10% non-response rate; design effect of 1.5, and relative desired precision of ±10%. Similarly, the sample size for mothers was also calculated using immunization coverage (83.3%) among children as per NFHS 4,[13] thus achieving a total sample of 102 ASHAs and 103 mothers per intervention district. This was rounded off to 100 for each category and divided equally among two randomly selected blocks per district considering block as a cluster. For remaining three categories of health functionaries, that is, ANMs, AWWs, and ASHA facilitators, four participants per category per block were conveniently included in the sample considering timeframe and operational feasibility of data collection. A multistage cluster sampling method was employed to select the study sites and participants. Firstly, two blocks in each of the districts were selected through simple random sampling, and then four villages from each of the blocks having the highest population were conveniently selected. The participants for study were selected using cluster sampling method. The total sample size of ASHAs and other health workers were divided equally in the two blocks, whereas sample size of mothers was divided equally into eight villages per district (four villages per block). Fifty ASHAs per block were selected from primary health centers (PHCs) close to block headquarters, whereas ANM, AWW, and ASHA facilitators, that is, four participants from each category per block were selected through convenient sampling. The interviews of 100 randomly selected mothers were conducted at their respective homes. Data collection tools and process A structured knowledge and practices assessment tool for each category of participants was developed by a team of newborn and child health experts. Post-development, pilot testing, and validation of the tools were conducted at a non-intervention site, and necessary changes were made before the actual data collection. For health functionaries (ASHAs, AWW, ANM, and ASHA facilitators), relevant questions were included considering their roles and responsibilities provisioned under HBYC. Similarly, availability of Mother and Child Protection (MCP) cards, ORS packets, and availability of IFA syrup with mothers as well as their knowledge and practices on hand washing, immunization, growth monitoring, and danger signs were assessed. The data collection was carried out by four teams (each team consists of four independent investigators) selected by state National Health Mission (NHM) administration. Interviews of ASHAs of each block were scheduled at the block office, whereas other health functionaries were interviewed at their respective place of posting. The interviews of mothers were conducted at their respective homes. A prior informed consent was obtained from each participant, and their privacy was ensured throughout the data collection. This assessment was carried out when entire health system was facing the challenges of the COVID-19 pandemic. The majority of health functionaries during this period were supporting COVID-19 vaccination drive. However, in January 2021, the state NHM issued directions to revive Maternal Newborn and Child Health (MNCH) services at the community level which also include HBYC home visits by health functionaries. Hence, we were able to complete data collection within stipulated timeframe with due compliance to COVID-19 infection prevention protocols. RESULTS A total of 8098 ASHAs, 60 ANMs, and 480 ASHA facilitators were found trained across eight aspirational districts of Madhya Pradesh. However, knowledge and practices of 801 ASHAs, 60 ANMs, 68 AWWs, 72 ASHA facilitators, and 787 mothers were assessed under this study. Knowledge and practices of ASHAs The study found that 79% of ASHAs were conducting structured home visits. Around 82% and 77% of ASHAs were found to have adequate stock of ORS and IFA syrups with 88% and 85% of them having correct knowledge about dosage and frequency related to ORS and IFA supplementation, respectively. About 85% of ASHAs were found to have adequate knowledge on initiation of complementary feeding at six months of age, whereas only 34% were knowledgeable about correct frequency of meals in a day. Less than half (47%) of ASHAs had knowledge of at least three danger signs that require referral of a child, and merely 14% ASHAs actually had referred sick children to a healthcare facility in last three months [Table 1]. Table 1 Knowledge and practices of ASHAs under HBYC Key Indicators on ASHAs Knowledge and Practices Frequency (n=801) Percentage (%) Conducting structured home visits as per HBYC protocol 634 79 Receiving quarterly supervision visits from ASHA facilitators 657 82 Correct knowledge of correct ORS preparation method 705 88 ASHAs with availability of ORS 657 82 Correct knowledge on timely initiation of complementary feeding 681 85 Correct knowledge of correct frequency of meals during each day 272 34 Correct knowledge on timely initiation of pediatric IFA syrup 681 85 ASHAs with availability of pediatric IFA syrup 617 77 Knowledge of at least three danger signs that require child referral to facility 352 44 ASHAs referred any sick child to a health facility in the last three months 112 14 Knowledge and practices of mothers Around 69% mothers reported to have received age appropriate structured home visits from ASHAs. We found that 89% mothers had MCP cards available with them and nearly 53% and 39% of eligible mothers were having ORS and IFA syrups with them, respectively. However, only 18% mothers were having correct knowledge of ORS preparation. We also found that 85% infants are receiving exclusive breastfeeding; however, only 40% and 28% receive timely initiation and correct frequency of complementary feeding, respectively. Over 40% of mothers were reported washing their hands at all critical times of the day (after defecation, before cooking, before feeding the child and after washing the baby’s bottom). Similarly, status of immunization was also found to be low (48%), and only 34% mothers had knowledge of no less than two danger signs which need child to refer [Table 2]. Table 2 Knowledge and practices of mothers on HBYC Key Indicators Frequency (n=787) Percentage (%) Mothers having MCP card 700 89 Mothers having ORS packets 417 53 Mothers correctly preparing ORS 142 18 Mothers having IFA syrup and giving it to eligible infants at home 307 39 Mothers washing hands at all critical times of the day 323 41 Knowledge about exclusive breastfeeding 669 85 Knowledge about timely initiation of complementary feeding 314 40 Infants received age appropriate vaccination 378 48 Mothers with knowledge of at least two danger signs that require referral 268 34 Mothers adequately playing and communicating with infants (ECD) 614 78 Chi-square test was performed to determine association between mothers’ knowledge on key child health practices and home visits by ASHAs. The findings suggest a significant association (p < 0.001) with availability of ORS and mother’s correct knowledge on initiation of pediatric IFA syrup [Table 3]. Table 3 Association between mothers’ knowledge on key child health practices and HBYC home visits by ASHAs Indicators HBYC Home visits by ASHAs Received Not Received NR/NA$ Total P Correct use of ORS# 94 17 - 111 P=0.064 Incorrect use of ORS# 43 16 - 59 Total 137 33 617 787 ORS available with Mother## 340 17 - 357 P<0.001* ORS not available with mother## 189 74 - 263 Total 529 91 167 787 Correct initiation of pediatric IFA## 236 28 - 264 P<0.001* Incorrect initiation of pediatric IFA## 224 70 - 294 Total 460 98 229 787 Correct initiation of complementary feeding## 257 67 - 324 P=0.166 Incorrect initiation of complementary feeding## 227 44 - 271 Total 484 111 192 787 #Mothers of infants having diarrhea in last two weeks. ##Mothers with infants above six months of age. $ NR (non-respondents) and NA (not applicable)—some of the questions were not responded by mothers, whereas some of the questions were not applicable to mothers according to the age of their child. *Statistically highly significant at 95% confidence interval (P<0.001). **Statistically significant at 95% confidence interval (P<0.05) Knowledge and practices of other health functionaries ANMs: All ANMs that were interviewed had adequate knowledge on timely initiation of complementary feeding (85%), correct preparation of ORS (90%), and correct dosage of zinc tablet (77%). Majority of ANMs had knowledge on timely initiation (82%) of pediatric IFA and its correct dose (80%). However, knowledge of at least three danger signs that require a child’s referral was found to be grossly low (27%). AWWs: 85% AWWs were maintaining records, and 79% were monitoring length and height of children. 80% AWWs had correct knowledge of growth chart, and 85% of them were able to detect severe malnutrition. Their knowledge of timely initiation of complementary feeding and correct frequency of complementary feeding was 90% and 72%, respectively. However, merely 47% AWWs could mention correct quantity required during each meal. ASHA Facilitators: We found that though 71% ASHA facilitators were trained on HBYC, only 57% were completing all required tasks under supportive supervision visits. ASHA facilitators were aware regarding IFA supplementation including its timely initiation, frequency, and correct dosage of pediatric IFA syrup at 93%, 97%, and 94%, respectively. Similarly, around 85% ASHA facilitators were familiar with timely initiation of complementary feeding. However, just 53% knew its correct frequency and quantity (38%) [Table 4]. Table 4 Knowledge and practices of ANM, AWW, and ASHA facilitators on HBYC Knowledge of ANMs on HBYC Frequency (n=60) Percentage (%) Timely initiation of complementary feeding 51 85 Correct ORS preparation 54 90 Dosage of zinc tablet 46 77 Timely initiation of pediatric IFA syrup 49 82 Correct dose of IFA syrup 48 80 A least three danger signs for infant referral 16 27 Knowledge and Practices of AWWs on HBYC Frequency (n=68) Percentage (%) Maintaining records of infants (0–2 years) in their area 58 85 Knowledge of plotting weight on growth chart in MCP card 54 80 Maintaining record of age wise length and height of infants 54 79 Knowledge of detecting child with severe malnutrition 58 85 Knowledge of timely initiation of complementary feeding 61 90 Knowledge on correct frequency of complementary feeding 49 72 Knowledge on correct quantity of complementary feeding 32 47 Knowledge and Practices of ASHA Facilitators on HBYC Frequency (n=72) Percentage (%) Knowledge on timely initiation of IFA syrup 67 93 Knowledge of correct frequency of IFA syrup 70 97 Knowledge of correct dose of IFA syrup 68 94 Knowledge on ORS preparation 70 97 Knowledge on timely initiation of complementary feeding 61 85 Knowledge of correct frequency of complementary feeding 38 53 Knowledge of adequate quantity of complementary feeding 27 38 Complete all necessary tasks during supportive supervision visits 41 57 DISCUSSION The HBYC program design strengthens contact by the health system with newborns and infants during the first most critical 1000 days of their lives to improve nutrition by promoting infant and young child development (IYCD) practices and ensuring adequacy of diet.[14,15] The findings revealed that only 52% ASHAs in the eight aspirational districts have been trained against the training target set by state NHM in FY 2019–20 and 2020–21.[16] Training of ASHAs on HBYC protocol is central to overall program design. Therefore, it is imperative to reformulate strategies and expedite the process of ASHA trainings for effective implementation of HBYC program. Our results show that 82% ASHAs who interviewed during the assessment were found to be trained on HBYC, whereas 79% ASHAs were conducting structured home visits at the 3, 6, 9, 12, and 15 months, as per HBYC operational guidelines. However, in 2019, a similar evaluation conducted in Bihar, Rajasthan, and Jammu and Kashmir revealed that merely 6% of ASHAs were conducting the structured visits.[17] Though these proportions are much high for Madhya Pradesh, a mechanism to ensure every eligible mother–baby dyad receive all scheduled visits by ASHAs is still needed to meet the program objectives. Our study shows 88% ASHAs and 97% ASHA facilitators and ANMs were aware of ORS packets and able to explain correctly the steps of ORS solution preparation. Similar results were seen in previous evaluations where knowledge scores of ASHAs regarding ORS were found to be good.[18,19] We further assessed the availability and knowledge of eligible mothers regarding ORS use. Only 53% of mothers were having ORS packets on the day of interview and only 18% knew the correct method of ORS preparation. These findings highlight a major gap where despite the high level of knowledge and adequate availability of ORS packets with ASHAs, the ORS distribution and transfer of knowledge to mothers about ORS preparation were found to be grossly suboptimal. Similar low levels of knowledge and practices of mothers regarding ORS preparation are shown in previous evaluations conducted in similar settings.[20,21] It is recommended that infants were to be exclusively breastfed for the first six months, followed by breastfeeding along with complementary foods for up to two years of age.[22] Our study revealed that more than 80% of health functionaries had adequate knowledge on exclusive breastfeeding and timely initiation of complementary feeding. Our study confirms findings of previous evaluations conducted to assess their knowledge in varied settings of India.[23] However, only 40% of mothers in our study were found to have adequate knowledge about timely initiation of complementary feeding. This proportion is similar to Madhya Pradesh state average (39.5%) found in NFHS 5 for children aged 6–8 months receiving solid or semi-solid food.[13] A significant difference in our findings vis-à-vis NHFS 5 was in the status of child immunization. We found 48% children received age appropriate vaccinations, whereas it was 77% as per NFHS-5.[13] In our study, more than 80% health functionaries had knowledge on timely initiation, correct frequency, and dosage of IFA syrup. These findings were supported by another study conducted regarding correct dosage of IFA syrup, but their finding of only 34.21% ASHAs having knowledge regarding correct frequency of IFA syrup is less than our finding.[24] Only 40% of mothers interviewed reported washing their hands at all critical times during a day. These proportions are lower than the findings of a previous evaluation conducted where 80% of mothers were found to wash their hands after defecation, 45% before cooking, and around 41% before feeding the child.[25] The knowledge of danger signs and timely referral of sick child to higher facilities is extremely essential to increase the probability of survival. Any delay in the recognition of the danger signs of child illnesses at household level leads to further delays at all further levels.[26] We found a grossly suboptimal knowledge of danger signs among ASHAs (44%), ANM (27%) as well as in mothers (34%) which could be a key contributor to infant mortalities in aspirational districts of Madhya Pradesh. CONCLUSION The current study indicates low coverage of HBYC trainings with close of half health functionaries remains untrained against the target set by the state. The field-level implementation of program activities including quality of homes visits conducted by ASHAs is also found to be suboptimal. The program envisages optimal care of every newborn at community level. However, despite satisfactory level of program knowledge among health functionaries, a gap was observed in knowledge and skills transfer to the ultimate beneficiaries. The state is requiring to revisit the existing implementation strategy, identify key bottlenecks, develop contextualized district-level action plans, and conduct periodic reviews to achieve intended outcomes of HBYC program. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. Critical danger signs that require an infant referral include blood in stool; fever for last seven days; convulsions; not able to drink or feed anything; very sleepy or unconscious; red value on mid-upper arm circumference (MUAC) tape ==== Refs REFERENCES 1 Under Five Mortality. UNICEF DATA Available from: https://data.unicef.org/topic/child-survival/under-five-mortality/ [Last accessed on 2021 Jun 05] 2 Khurmi M Mathur N Kaur P Nichale A Child survival status in india child survival status in India 2015 May 0 4 Available from: https://aimdrjournal.com/wp-content/uploads/2021/06/104-107-Brief-C.pdf 3 Pappachan B Choonara I Inequalities in child health in India BMJ Paediatr Open 2017 1 e000054 4 National Family Health Survey 5 Factsheet for India, IIPS 2019-21 Available from: http://rchiips.org/nfhs/NFHS-5_FCTS/India.pdf 5 Onyeneho NG Ozumba BC Subramanian SV Determinants of childhood anemia in India Sci Rep 2019 9 16540 31719548 6 Ministry of Health and Family Welfare, GoI. National health policy 2017 Available from: https://www.nhp.gov.in/nhpfiles/national_health_policy_2017.pdf 7 Neogi SB Sharma J Chauhan M Khanna R Chokshi M Srivastava R Care of newborn in the community and at home J Perinatol 2016 36 S13 7 27924109 8 Guidelines O Home Based Newborn Care 2014 Available from: https://nhsrcindia.org/sites/default/files/2021-03/Revised%20HBNC%20Operational%20Guidelines%202014%20English.pdf [Last accessed on 2022 Sep 01] 9 Guidelines O Home Based Care for Young Child (HBYC) 2018 April Available from: https://www.aspirationaldistricts.in/wp-content/uploads/2019/02/Home-Based-Care-for-Young-Child-Guidelines.pdf [Last accessed on 2022 Sep 01] 10 National Health Mission National health mission, Government of India Operational Guidelines for Improving Health and Nutrition Status in Aspirational Districts 2018 Available from: https://nhm.gov.in/New_Updates_2018/NHM_Components/RMNCHA/ADP/Operational_Guidelines_for_Aspirational_Districts_Print_Ready_File_18icts.th_October.pdf 11 Government of India Aspirational district Available from: https://my.msme.gov.in/MyMsme/Reg/home.aspx 12 Meena R Raj D Saini L Tomar A Khanna M Gaur K Knowledge status of accredited social health activist (ASHA) of Jaipur City IMJ Health 2016 2 18 25 13 National Family Health Survey 5 (2019-21) Factsheet for Madhya Pradesh Available from: http://rchiips.org/nfhs/NFHS-5_FCTS/Madhya_Pradesh.pdf 14 Infant and Young Child Feeding WHO June 2021 Available from: https://www.who.int/news-room/fact-sheets/detail/infant-and-young-child-feeding 15 National Guidelines on Infant and Young Child Feeding, One Day Sensitization Module, National Health Mission, Ministry of Health and Family Welfare, Government of India August 2016 Available from: https://www.nhm.gov.in/MAA/One_Day_Sensitization_Module/One_Day_Sensitization_Module_English_lowres.pdf 16 HBYC Training Status Report of Madhya Pradesh, National Health Mission 2021 17 Assessment of Home Based Care of Young Child 2019. Evaluation Report 2019 18 Choudary M Varia K Kothari N Ghandhi S Makwana NR Parmar D Evaluation of knowledge of ASHA workers regarding various health services under NRHM in Saurashtra Region of Gujarat Natl J Community Med 2015 6 193 7 19 Lamberti LM Fischer Walker CL Taneja S Mazumder S Black RE The association between provider practice and knowledge of ORS and Zinc supplementation for the treatment of childhood Diarrhea in Bihar, Gujarat and Uttar Pradesh, India PLoS One 2015 10 e0130845 26098305 20 Divasha Pasi R Ravi KS Level of knowledge of mothers (18-35 years of age) of under 5 children regarding ORS therapy J Family Med Prim Care 2020 9 4747 50 33209794 21 Singh D Lakhwani S Gaharwar DPS Knowledge, attitude and practice study about use of ORS in diarrhea in mothers with children 02-05 year age group residing in various urban slums of Bhopal City Asian J Clin Pediatr Neonatol 2019 7 31 3 22 The official webpage of WHO Available from: https://www.paho.org/en/topics/breastfeeding-and-complementaryfeeding#:~:text=The%20World%20Health%20Organization%20recommends,two%20years%20old%20or%20beyond 23 Kohli S Knowledge and Counselling Skills of Community Health Workers for Promotion of Optimal Infant and Young Child Feeding (IYCF) Practices: A Review 2017 Available from: https://www.researchgate.net/publication/328517528_Knowledge_and_Counselling_Skills_of_Community_Health_Workers_for_Promotion_of_Optimal_Infant_and_Young_Child_Feeding_IYCF_Practices_A_Review 24 Pal J RoySwagata S Satapathy N 2019. Assessment of knowledge and practices of ASHA work Available from: https://www.researchgate.net/publication/336051439_Assessment_of_knowledge_and_practices_of_ASHA_workers_related_to_maternal-child_health_and_their_performance_affecting_factors_A_mixed_method_study_in_Deganga_block_North_24_parganas_district_West_Ben 25 Begum RU Bhavani K Study of knowledge and practices of hand washing among mothers having children under 5 years of age in urban field practicing area of Kakatiya Medical College, Warangal, Telangana, India Int J Community Med Public Health 2017 3 2035 9 26 Ekwochi U Ndu IK Osuorah CD Amadi OF Okeke IB Obuoha E Knowledge of danger signs in newborns and health seeking practices of mothers and care givers in Enugu state, South-East Nigeria Ital J Pediatr 2015 41 18 25888409

PMC010xxxxxx/PMC10353668.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-436 10.4103/ijcm.ijcm_160_23 Original Article Estimating the Burden of Tuberculosis in India: A Modelling Study Mandal Sandip Rao Raghuram 1 Joshi Rajendra 1 Senior Advisor Data Analytics and Mathematical Modelling, John Snow Institute, New Delhi, India 1 Central TB Division, Ministry of Health and Family Welfare, Govt. of India, New Delhi, India Address for correspondence: Dr. Raghuram Rao, Joint Director, TB, Central TB Division, National TB Elimination Programme, Ministry of Health and Family Welfare, Govt. of India, Jeevan Vihar, Ground Floor, Sansad Marg, New Delhi, India. E-mail: raor@rntcp.org May-Jun 2023 24 3 2023 48 3 436442 14 3 2023 21 3 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Background: Measurements TB incidence and mortality are crucial for monitoring progress towards SDG goals for TB. Until recently, WHO estimated TB burden in India with applied simple, transparent equilibrium models to data from Gujarat, an Indian state where the first state-level prevalence survey was conducted in 2011. However, since then there has been several interventions in India including national TB prevalence survey, infection survey, sub-national survey & certification which gives opportunity for national and sub-national estimates for TB incidence and mortality. Methods: We developed a model is a compartmental, deterministic framework, taking account of TB natural history, as well as India’s healthcare system including health care seeking from public and private sector. To address changes in TB burden owing to COVID disruptions, we followed same model that used by WHO in the global TB Report 2022 with additional impact of delta wave in 2021. Major sources of data included National TB Prevalence survey, trends in caseloads in public and private sector including their contribution and mortality information. Results: We estimated total TB incidence of 2.77 million in the year 2022 as against 2.97 in the year 2015 and corresponding TB mortality of 0.32 and 0.36 million respectively. In terms of rate per 1,00,000 TB incidence in 2022 was 196 as compared to 225 in the year 2015 and mortality was 23 and 27 respectively. TB incidence estimates are similar to what was estimated by WHO, while mortality estimates appear different in our estimates due to different calibration targets depending on in-country published data. Conclusion: Even if TB burden is infeasible to measure directly, a range of data can nonetheless offer indirect evidence for its estimation: mathematical modelling can be a helpful tool for bringing together these diverse sources of evidence, and deriving estimates that are consistent with them all. While the RGI reported mortality is an important source of information, its quality and coverage for medically certified cause of deaths requires improvement in India. Tuberculosis Incidence mortality estimation India ==== Body pmcINTRODUCTION The END TB Strategy and SDG goals for TB call for 80% reduction in TB incidence rates and 90% reduction in TB deaths in 2030, compared with 2015.[1,2] However, there are substantial challenges in monitoring both of these measures. Direct and representative measurement of TB incidence requires an infeasibly large population to be followed up over a year or more, while in high burden countries, vital registration systems generally lack the coverage to provide reliable estimates of mortality. Therefore, incidence and mortality often must be estimated indirectly, using the available data. The need for robust TB burden estimates is especially pressing in India, the country with the world’s largest TB burden, which in 2021 accounted for over a quarter of global TB incidence.[3] Until recently, WHO estimates of TB burden in India applied simple, transparent equilibrium models to data from Gujarat, an Indian state where the first state-level prevalence survey was conducted in 2011.[4] Although Gujarat accounts for around 4% of India’s population, this was the largest survey of its time in India, and thus provided the best available information on TB burden in the country. India subsequently embarked on a national prevalence survey, beginning in 2019.[5] Despite a temporary interruption during the first wave of COVID-19 in the country, the survey went on to cover 3,22,480 individuals in 25 different states, placing it amongst the largest prevalence surveys conducted so far. Findings from this survey have already yielded important insights into the distribution of TB burden in a country as vast as diverse as India. However, they also provide an invaluable opportunity to update estimates of TB incidence in India. In addition, recent years have seen a substantial increase in the availability of other, indirect sources of evidence for TB burden, such as systematic tracking of the sales of anti-TB drugs in the private healthcare sector[6,7] and a steady expansion of district-level initiatives to monitor TB control efforts.[8] In the present work, we brought together all of these data sources in a Bayesian evidence synthesis framework, to inform estimates for TB incidence and mortality in India over the last decade. METHODS Figure 1 shows an illustration of the model structure, with further details shown in the supporting information. Briefly, the model is a compartmental, deterministic framework, taking account of TB natural history, as well as India’s healthcare system. It distinguishes TB care in the public and private sectors, in order to account separately for notifications from both of these sectors. For simplicity the model does not include HIV/TB coinfection or rifampicin resistance, which in 2019 accounted respectively for an estimated 2.3% and 4.2% of estimated TB incidence in the country.[9] Also for simplicity, the model does not incorporate age structure, and does not distinguish between pulmonary and extra pulmonary TB. In this analysis first we calibrated our model with the available data and estimated incidence in 2019. Then, to address changes in TB burden owing to COVID disruptions, we followed the same methodology as that used by WHO in the global TB Report 2022, [10] with one difference: WHO estimations assume a temporary reduction in TB transmission, during periods of lockdowns However in India, mobility data suggests that the severe ‘delta wave’ from May to September 2021 showed stark reductions in mixing, as severe as during the national lockdown from April to July 2020, (see Figure S3, supporting information). Accordingly, we assumed a 50% reduction in TB transmission during the period of the delta wave as well as during the national lockdown, assuming (consistent with the WHO model) that TB transmission was at pre-pandemic levels at other times. Figure 1 Schematic illustration of the model structure. Infectious compartments contributing the force-of-infection are shown in red. For clarity, the diagram omits certain rates incorporated in the model, including: self-cure; exogenous reinfection; and background mortality. Details of parameters and model equations are given in Table S1 of the supplementary document Table S1 List of model parameters, values and sources Parameter Symbol Value Source/Notes Natural history Infection rate (number of annual infections per case) β 4.4 (3.4 – 6.2) Model calibration, with priors U[0, 30] Per-capita annual rate of progression from ‘fast’ latent infection u 0.07 (0.01 – 0.13) Calibration: Menzies (2018)[1] for central value, and U[0.1-20] on multiplying factor Per-capita annual rate of stabilisation from ‘fast’ to ‘slow’ latent status v 0.85 (0.78 – 0.95) Menzies (2018)[1] for central value, and taking uniform priors of +/- 25% Per-capita annual rate of reactivation from ‘slow’ latent infection w 0.0035 (0.0004 – 0.0061) Calibration: Menzies (2018)[1] for central value, and U[0.1-20] on multiplying factor Per-capita annual rate of self-clearance of latent TB c 0.031 (0.022 – 0.035) Emery (2021)[2] for central value, and taking uniform priors of +/- 25% Per-capita annual rate of TB mortality while untreated μ TB 0.07 (0.05 – 0.10) Calibration: corresponding to case fatality rate (CFR) of undetected TB of 29% [20-40], during treatment in public sector 4% [3.7 – 4.4] and during treatment in private sector 1.2% [1 – 1.7] to meet the mortality target in 2015 i.e 28 [20-36] per 100,000 population. Per-capita annual rate of TB mortality while in treatment in public sector μTB(pu) 0.103 Per-capita annual rate of TB mortality while in treatment in private sector μTB(pr) 0.051 Per-capita annual rate of TB self-cure σ 0.17 (0.13 – 0.21) Tiemersma et al., 2011[3] Protection from reinfection amongst those with prior infection h 0.38 (0.13 – 0.73) Andrews (2012)[4], assuming uniform priors of +/-25% Per-capita annual rate of relapse in first two years after treatment completion ρ (lo) 0.083 (0.065 – 0.106) Thomas A et al (2005)[5], Romanowski (2019)[6], Menzies (2009)[7] and Weis (1994)[8], with uniform prior using intervals of ±5% Per-capita annual rate of relapse in first two years after self-cure or incomplete treatment ρ (hi) 0.14 (0.11 – 0.17) Per-capita annual rate of relapse>two years after last TB episode ρ 0.0015 (0.0011 – 0.0018) Most relapse occurs in first two years after recovery: Guerra-Assuncao (2015)[9] Per-capita annual rate of ‘stabilising’ from high to low relapse risk S 0.5 TB services Rate-of-presentation to care, first careseeking visit In 2011 g (2011) 0.30 (0.19 – 0.44) Model calibration, with priors U[0.1, 10] for 2011; and for 2020 taking U[1, 10] as multiplying factor on g (2011). As described in the main text, we allow for increasing rate over time motivated by increasing presumptive examination rates reported by the programme. In 2020 g (2020) 0.55 (0.46 – 0.71) Rate-of-presentation to care, second and subsequent careseeking visits Assuming same in 2011 and 2020 g~ 4.9 (0.8 – 12.4) Model calibration, taking U[1, 40] for multiplying factor on g (2011) Probability that a TB patient visits public provider, per careseeking attempt In 2011 p (2011) 0.50 (0 – 0.98) Model calibration In 2020 p (2020) 0.6 (0.4 – 0.8) Per-capita rate of offering diagnosis d 52 Assumption, corresponding to 1 week Probability of successful TB diagnosis and treatment initiation per careseeking visit Public sector 0.67 (0.34 – 0.89) U[0.3, 0.9], motivated by Subbaraman (2016)[10] Private sector 0.64 (0.33 – 0.88) U[0.3, 0.9], assumption Per-capita annual rate of treatment completion t 2 Corresponds to average treatment duration of 6 months Per-capita annual rate of treatment interruption Public sector ∈ (pu) 0.52 (0.26 – 0.66) Calculated using , for treatment completion rate P, and assuming U[0.75, 0.95] for P Private sector ∈ (pr) 2.2 (0.9 – 2.9) As above, but assuming U[0.4, 0.8] for P Demographics Per-capita annual rate of background mortality μ 1/70 Corresponds to average lifespan of 70 years (World Bank 2021)[11] Programme data shows that the presumptive examination rate by the public sector has been increasing over time. Accordingly, we allowed the rate-of-presentation to care (γ(t)) to increase in a linear way from 2011 to 2020, with starting and ending values to be calibrated. We also estimated the probability p that a patient chooses the public (rather than the private) sector on each careseeking visit, again allowing this parameter to change in a linear way from 2011 to 2020. Sources of data National prevalence survey: Results from India’s prevalence survey suggested a prevalence of 312 [CI 286 - 337] per 100,000 population, when adjusted for age and other factors. In addition to this data, we also calibrated to the data for the proportion of prevalent TB that was on TB treatment (irrespective of bacteriological status) in the survey. A similar approach had been taken in earlier years for WHO incidence estimates based on Gujarat prevalence survey data, but assuming equilibrium conditions for the purpose of analytical tractability.[11] The role of this data is to inform the average duration of untreated TB (the higher this proportion, the shorter the duration). Thus, combined with prevalence, this data informs incidence estimates. In our current analysis, by virtue of using a dynamical transmission model, we were able to incorporate this data without need for equilibrium assumptions. Trends in TB caseload in the private sector: The healthcare data company IQVIA collects comprehensive data on drugs sold through the private market in India, with data available from 2015. Tracking the volume of rifampicin-containing drugs gives a direct measure of the volume of anti-TB treatments being sold through the private sector each year. As described in previous work, it is challenging to translate directly from drug volumes to patient numbers,[12] because two factors necessary for this conversion are unknown: the average duration of TB treatment in the private sector, and the extent of overdiagnosis for TB in this sector. Accordingly, we did not attempt to use drug sales data directly to inform numbers of TB patients, but rather to inform trends. In particular, private sector sales of TB drugs have declined nationally by 44% from 2015 to 2019. After adjusting for private providers increasingly prescribing publicly supplied drugs in recent years (see supporting information), we calibrated the model to this proportionate decline in the number of patients initiating treatment in the private sector, between 2015 and 2019. Trends in TB caseload in the public sector: We calibrated model simulations for treatment initiations in the public sector against data for public notifications in 2011, and 2019. In 2011, patients were notified when initiating treatment and in 2019, when they were diagnosed. We therefore adjusted 2019 notifications downwards by the pre-treatment loss-to-follow up, in order to ensure comparability between these data. We also adjusted 2019 data for notifications that had been inaccurately attributed to the public sector, rather than the private sector. Private sector contribution to notifications: Since 2014, notifications from the private sector have steadily grown over time, as a result of successful efforts by the national programme to engage and coordinate with this sector. We did not aim to match these trends in the models, as they arose directly from programmatic efforts rather than any underlying epidemiological changes. Instead, we calibrated model-simulated treatment initiations in the private sector to notifications from this sector in 2019, adjusting for the proportion of privately treated patients that are notified (see supporting information). Under-reporting: In recent years, India’s TB programme has implemented subnational certification (SNC), a district-level initiative to generate data for monitoring trends in TB burden over time. At the time of writing, this initiative covers more than 500 districts in the country, providing representative data for 21 states. An important part of this initiative is a sustained effort to find TB cases at the community level (see ref [8] for further details). Over the course of a year, all identified TB cases are compared against Nikshay, India’s TB register, to quantify the proportion that had been notified. We used this data as a calibration target in the model, for the proportion of TB patients on treatment that were notified in 2019. Mortality: Since 2015, WHO estimates of TB mortality in India have been based on estimates by the Institute of Health Metrics and Evaluation (IHME), adjusted to align with WHO estimates for all-cause mortality.[11] Data updates for these estimates were last conducted by IHME in 2015.[12] Additionally, RGI data is another valuable source of mortality data in India.[13] However, across the country coverage of RGI data is incomplete. We therefore incorporated IHME mortality estimates but RGI reported moderated mortality rate for India. In this method, we assumed the central value is the average of the upper limit of RGI mortality and the lower limit of IHME mortality (~ 28 per 100,000 populations). And the extreme end of both these estimates (IHME and RGI) inform us the CI of this moderated mortality. (see supporting information). TB infection prevalence: During the prevalence survey, a subset of the population was tested using interferon gamma release assays (IGRA), to estimate the prevalence of TB infection. This prevalence was estimated at around 25%, notably consistent with estimates from a previous statistical study.[14] Although our model parameters for progression to active TB are based on a systematic review of studies from the pre-chemotherapy era, in India undernutrition and other factors are likely to play an important modifying role in these rates. Accordingly, we calibrated the amount by which rates should be increased, in order to match the measured prevalence of TB infection in 2019. Calibration To estimate uncertainty systematically from model inputs to model estimates, we performed calibration using Bayesian Markov Chain Monte Carlo (MCMC). We constructed the posterior density as follows: for each of the calibration targets described above, we constructed beta distributions to capture model proportions, and log-normal distributions to capture population rates, adjusting distribution parameters in order to match the central and uncertainty intervals of each calibration target. We also included wide prior uniform distributions for uncertain model parameters, such as the rates of treatment completion in the private sector.[15] We then took the posterior density to be proportional to the product of all likelihood and prior densities. For all practical purposes we calculated the log-posterior density, therefore taking a sum of the log-probability distributions of each of the individual likelihood components. As an efficient way of sampling from the posterior distribution, we implemented adaptive MCMC,[16] which uses the covariance structure of already-drawn samples to inform the proposal distribution. We first generated 1,000 samples for model parameters using Latin hypercube sampling, then choosing the three parameter sets with the highest posterior density as starting conditions for independent MCMC chains. We ran each chain for 50,000 MCMC iterations. After discarding the burn-in and selecting every 50th sample, we drew 250 samples from the posterior density. For all model outputs, we took central estimates as the 50th percentile. We quantified uncertainty using the 2.5th and 97.5th percentiles, denoting this range as the 95% Bayesian credible interval. We compared results from the three independent chains to ensure that they gave convergent estimates. The model was calibrated to the following data: RESULTS The results of model calibration are shown in Figure 2. This shows the resulting comparisons between model outputs and indicator data. Figure 2 Model calibration with the observed data. Dots show central estimates, while error bars show 95% uncertainty intervals Figure 3 shows model projection for incidence and mortality at national level from 2015 to 2022. Results suggest an incidence rate in 2022 of 196 (95% CrI 171 – 228) per 100,000 population and mortality rate of 23 (95% CrI 16 – 33) per 100,000 population. In terms of absolute numbers, these estimates correspond to an incidence and mortality, respectively, of 2.77 million (95% CrI 2.42 – 3.23) and 0.32 million (95% CrI 0.23 – 0.47). The impact of service disruption due to the COVID-19 pandemic is observed in the model estimates. Annual estimates for incidence and mortality from 2015 to 2022 are summarised in Table 1. Figure 3 Estimated incidence and mortality rate from 2015 to 2022. Solid lines show the central estimate, while the shaded regions show the 95% credible intervals (CrI) of the estimates Table 1 Estimated incidence and mortality rate per 100,000 populations and in total numbers Estimates 2015 2016 2017 2018 2019 2020 2021 2022 lo mid hi lo mid hi lo mid hi lo mid hi lo mid hi lo mid hi lo mid hi lo mid hi Incidence (per 1,00,000) 201 225 254 196 217 246 189 211 238 182 206 233 177 201 228 169 194 226 172 197 227 171 196 228 Incidence (numbers in million) 2.66 2.97 3.36 2.62 2.91 3.29 2.56 2.86 3.23 2.50 2.82 3.19 2.45 2.77 3.16 2.36 2.70 3.15 2.42 2.77 3.20 2.42 2.77 3.23 Mortality (per 1,00,000) 20 27 36 19 26 35 18 25 34 18 24 33 17 23 32 17 23 32 17 24 34 16 23 33 Mortality (numbers in million) 0.27 0.36 0.48 0.26 0.35 0.47 0.25 0.34 0.46 0.24 0.33 0.45 0.23 0.32 0.44 0.23 0.32 0.44 0.24 0.34 0.48 0.23 0.32 0.47 Model Calibration indicators Indicator Value Comments and data source Prevalence per 100,000 population, 2020 312 (286 – 337) National TB prevalence surveyv Of prevalent TB, percent on treatment 12 (9.0 – 16) Population prevalence of LTBI (percent) 25 (21 – 29) Notifications per 100,000 population, 2019 (public sector) 94 (80 – 108) Programmatic data allowing +/- 10% uncertainty. Because the model only addresses new and relapse cases initiating treatment, here we counted only new and relapse notifications, as well as accounting for 15% initial loss to follow up. Private sector treatment initiation in 2019 per 100,000 Greater than 42 Accounting for over-diagnosis in private sector, reducing private notifications by 15%. This target is set as a lower limit, because of uncertainty in the proportion of private providers that are notifying TB. Mortality per 100,000 population, 2015 28 (20 – 36) By adjusting IHME estimates with the RGI data (see supporting information) Proportion of reduction of public sector notification in 2019 relative to 2011 0.3 [0.2 – 0.4] Programmatic data Reduction in treatment initiation in private sector in 2019 relative to 2015 0.24 [0.19 – 0.37] From private-sector drug sales data. Accounting for the increasing proportion of privately managed patients on publicly supplied drugs Notification through active case-finding 2017 2018 2019 2020 2021 Programmatic data 26781 47307 62958 52273 73772 Sensitivity analysis To assess the contribution of each of the model parameter, we performed a sensitivity analysis, using partial rank correlation coefficients. Figure 4 shows the result of that analysis that helps to identify the model inputs that are most influential for model outcomes. The model shows partial rank correlation coefficients of each of the model inputs against the model outcomes - estimated incidence and mortality rates in 2019 [Figure 4a and b]. Additionally, we performed a similar analysis to show how the input parameters correlates with the calibration outcome – public sector notifications [Figure 4c]. Figure Inputs are listed in order of decreasing sensitivity from top to bottom. Figure 4 Sensitivity analysis. Partial rank correlation coefficients of each of the model inputs against incidence, mortality and notification rate in 2019 To compare against previous methods of burden estimation conducted by WHO, we also estimated incidence under a ‘reduced’ model, constructed to simulate equilibrium epidemic conditions: to do so, we assumed no temporal change in the rate of treatment uptake in the public or private sectors, and calibrated only to data from the prevalence survey, as well as notifications in 2019 and estimated TB mortality. This simplified model estimated incidence in 2020 is 194 [95% CrI 172 - 221] per 100,000 populations which is very close to our estimated incidence in 2020 [See Table 1]. DISCUSSION Creating robust estimates of TB burden is an important need for public health, especially in light of India’s ambitious commitment to end TB by 2025. Even if TB incidence is infeasible to measure directly, our work illustrates how cross-sectional data from prevalence surveys - together with other sources of evidence on TB burden - can provide valuable information. Our estimates for TB incidence in 2019 are comparable to earlier estimates by WHO, based on an equilibrium model, although our estimates of TB incidence decline in recent years are somewhat steeper though close to interim estimates published in recent Global TB report 2022. In our model, the key sources of data informing trends over time are private sector drug sales, and public sector notifications - respectively, informing TB caseloads in the private and public sectors. As described above, both sources have limitations that mean they cannot directly be used to inform burden estimates: adjustments are needed, for example to extract treatment initiations from notification data, wherever the latter includes all diagnosed patients. Nonetheless, the Bayesian evidence synthesis approach that we have implemented allows us systematically to incorporate these adjustments in the analysis, together with their associated parametric uncertainties. Such approaches could be used in other settings where similar data is available: not only in other countries, but also for individual states within India. Indeed, extending the work shown here to state-level estimates in India is the subject of ongoing analysis, and will contribute to strategic planning for TB control priorities at the state level. In general, the best approach for any given country will depend on the sources of data available. For example, a recent study used modelling approaches to estimate incidence and mortality in Brazil.[17] Importantly in Brazil, vital registration data provides a comparably robust base of evidence on which not just TB mortality, but also incidence estimates, can be developed. While such comprehensive VR data was not available in India at the time of these estimates, it is steadily developing. Currently available information on vital registration of deaths in India suggests increasing level of reporting, >90% of estimated deaths under vital registration system of Registrar General of India. However, coverage of Medically Certified Cause of Death (MCCD) among the registered deaths is still under 22%. Some of the smaller states have much better coverage such as Goa, which has 100% coverage of MCCD. While all States and Union Territories (UTs) in India are reporting MCCD by 2020, some of the larger states like Bihar & Uttar Pradesh still have poor MCCD coverage of less than 10%. Nonetheless in future, this data will be invaluable in improving our estimates of TB incidence and mortality in India. As with any modelling analysis, our work has some limitations to note. Our estimates are undifferentiated by age, and by drug resistance or HIV status; while the latter two factors account for only a small proportion of TB incidence at the country level, at the subnational level they will nonetheless be more important for some states than others. All three factors are important areas for future work to address. Second, while our analysis has been supported by data on private sector drug sales, we have had to accommodate uncertainty in its implications for TB burden, most importantly in the proportion of private patients receiving publicly supplied drugs. The Bayesian calibration approach allows us systematically to incorporate these and other uncertainties into the estimation. Nonetheless, further data to narrow these uncertainties will be valuable for refining our estimates in future. Indeed, further evidence on the average duration of TB treatment in the private sector, and the extent of TB over-diagnosis in this sector, would be invaluable in translating drug sales volumes directly to patient numbers. Finally, our calibration data for TB mortality come from IHME, which themselves are modelled estimates. While these estimates have been valuable measures for understanding TB burden in India, as mentioned above, it is hoped that they can be bolstered and further refined with increasing availability of vital registration data. In conclusion, even if TB burden is infeasible to measure directly, a range of data can nonetheless offer indirect evidence for its estimation: mathematical modelling can be a helpful tool for bringing together these diverse sources of evidence, and deriving estimates that are consistent with them all. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. Acknowledgements Authors acknowledge contribution of IQVIA India team and National Technical Support Unit (NTSU) at Central TB Division for sharing information on estimated drug sales in India between 2013 to 2022 on annual basis collected as a part of their routine business and to Central TB Division for sharing programmatic information on case finding efforts and TB notification as well as treatment outcomes under National TB Elimination Programme (NTEP). Dr Nim Arinaminpathy, Imperial College, London guided the construction of model especially in early stage. The model was also discussed with WHO at all three levels Country, Regional and HQ-TME team and their inputs in the model were incorporated. Global expert group formed by TME will continue to review this model in future as well. Supplementary Information Re-estimating Tuberculosis Incidence and Mortality in India during 2011-2022: A Modelling Study TABLE OF CONTENTS Model structure Equations and parameter tables Further information on data Mortality rate estimation Linear regression for drug sales declines, and how we adjusted for proportion getting public drugs Linear regression for public notifications - and deflation of 2019 numbers Mobility pattern Posterior distributions for model parameters Results of the MCMC calibration References MODEL STRUCTURE Figure S1 gives a schematic illustration of the overall model structure, with all model parameters listed in Table S1, and governing equations given below. Figure S1 Schematic illustration of the model structure Table S2 Estimation of trend of privately treated patients Average patient months 2015 2016 2017 2018 2019 Actual data 1466928 1245305 1056072 984014 815997 Adjusted patient months- assuming 30% treated with public drugs in 2019 (linearly changes from 2015 to 2019) 1466928 1346276 1242438 1269695 1165710 Adjusted patient months assuming 10% treated with public drugs in 2019 (linearly changes from 2015 to 2019) 1466928 1277236 1111655 1063799 906663 Parameters are defined below, and in Table 1. Infectious compartments contributing the force-of-infection are shown in red. For clarity, the diagram omits certain rates incorporated in the model, including: self-cure; exogenous reinfection; and background mortality. EQUATIONS AND PARAMETER TABLES Governing equations of the model are as follows. All state variables are written as proportions of the population (not as absolute numbers). Uninfected (U): for a birth-rate b; force-of-infection λ; rate of clearance of LTBI c; and background mortality rate μ. Latent, ‘fast’ infection (Lf): for a progression rate u; a ‘stabilisation’ rate (to latent ‘slow’ status) v; and protection from reinfection h, amongst those previously infected. Latent, ‘slow’ infection (Ls): for a reactivation rate w. Active TB (I): for relapse rates ρ(hi), ρ(lo), ρ; careseeking rate γ; TB mortality rate μTB; and self-cure rate σ. Here, p and (1 - p) are, respectively, the proportion-of-presentation to healthcare providers in the public and private sectors. The factor γ(t) allows for an change in the notification over time, from 2011 onwards. The term a(t) represents the time-dependent effect of active case-finding. Presented for diagnosis with provider type s (D(s)): for a rate-of-offering diagnosis δ. Here, (t) is analogous to γ (t), but attached to individuals who remain undiagnosed despite having previously sought care (i.e. compartment E below). On TB treatment with provider type s (T(s)): for a treatment completion rate t; and a treatment interruption rate ϵ(s). Missed diagnosis and temporarily disengaged from careseeking (E): Recovered with low relapse risk, following treatment completion (R(lo)): for a rate of ‘stabilisation’ of relapse risk s. Recovered with high relapse risk, following treatment completion (R(hi)): Long-term, ‘stabilised’ relapse risk (R): Force-of-infection (λ): for a rate-of-transmission α. Table of parameters The notation U [x,y] denotes a uniform probability distribution on the range [x,y]. Otherwise, parameter values in round brackets show 95% percentiles from the respective marginal posterior densities. FURTHER INFORMATION ON DATA Mortality rate estimation For estimation of TB mortality in India, WHO currently relies on IHME estimations since 2015. However, the most recent data informing IHME estimates are from 2015. One important source of current data is vital registration collected through the Registrar General of India (RGI). In brief, this data includes verbal autopsies from trained physicians. The national coverage of this data is currently under 22%, and so it cannot be used as a stand-alone source of mortality estimates. However, it provides a valuable source of evidence to complement IHME estimates. For comparison, the IHME estimate (WHO adjusted) for TB mortality in 2015 is 34 [32 – 36] per 100,000 population [WHO Global TB Report 2022]. RGI data suggests a rate of 22 [20 – 23] per 100,000 population. Because of the incomplete coverage of RGI data, it is instructive also to compare the proportions of overall deaths attributed to TB. For IHME in 2015, this proportion 4.97 (4.57 - 5.38), whereas according to the RGI estimate this number is 3.17 (3.14 - 3.20), based on the Annual Reports on Medica Cause of Certified Deaths (MCCD) published by Registrar General of India (RGI) available at https://crsorgi.gov.in/mccd-reports.html. Overall, therefore, the available evidence suggests that IHME mortality estimates should be adjusted downwards. As a simple but transparent approach, we assumed the central value is the average of the upper limit of RGI mortality and the lower limit of IHME mortality (~ 28 per 100,000 population). For uncertainty intervals, we took the upper limit of IHME estimates and the lower limit of RGI estimates, therefore adopting wide uncertainty. Thus, the mortality rate becomes 28 [20 - 36] per 100,000 population. Estimating declines in privately treated patients As described in the main text, we used private sector drug sales data to inform trends in the number of patients being treated by private providers. In doing so, it was necessary to adjust for increasing uptake of publicly supplied drugs by private providers. Data from public-private-mix implementing agencies in India suggest that 20% of patients being managed by private providers in India were treated using public drugs. Accordingly, Table S1 below shows treatment volumes of all patients managed in the private sector under scenarios where 10% and 30% of patients receive public drugs. We fit these data -series using linear regression (see Figure S2) to estimate the average decline rate of patients in the private sector from 2015 onwards. Overall, this suggests the decline rate in private sector treatment initiation from 2015 to 2019 is 24% [19 - 37]. We used this decline rate as a target for model calibration. Figure S2 Linear fit with the adjusted patient months from 2015 to 2019 assuming 10% and 30% treated with public sector drugs in 2019 Linear regression for public notifications and deflation of 2019 numbers The total TB notification from the public sector notification in 2019 was 125 per 100,000 population. While this data reflects all diagnosed patients, the model only counts new/relapse cases who initiated treatment. We adjusted for both, including 15% initial loss-to-followup, ultimately yielding a rate of 94 [80 – 108] per 100,000 populations. Mobility data[12] Figure S3. Figure S3 Mobility data in India during the pandemic period. The two red arrows show the period of national lockdown, and the delta wave in India, respectively POSTERIOR DISTRIBUTIONS FOR MODEL PARAMETERS Figure S4. Figure S4 Posterior distributions for model parameters RESULTS OF THE MCMC CALIBRATION We sampled from the posterior density using adaptive Bayesian Markov Chain Monte Carlo simulation[13] as described in the Calibration section of the main article. Figure S5 below shows the trace arising from the MCMC calibration. Figure S5 Trace plot arising from MCMC calibration, showing the log-posterior density over 50,000 iterations REFERENCES 1 Menzies NA Wolf E Connors D Progression from latent infection to active disease in dynamic tuberculosis transmission models: a systematic review of the validity of modelling assumptions Lancet Infect. Dis 2018 doi:10.1016/S1473-3099(18)30134-8 2 Emery JC Richards AS Dale KD Self-clearance of Mycobacterium tuberculosis infection: implications for lifetime risk and population at-risk of tuberculosis disease Proc R Soc B Biol Sci 2021 288 20201635 doi:10.1098/rspb.2020.1635 3 Tiemersma EW van der Werf MJ Borgdorff MW Natural History of Tuberculosis: Duration and Fatality of Untreated Pulmonary Tuberculosis in HIV Negative Patients: A Systematic Review PLoS One 2011 6 e17601 doi:10.1371/journal.pone.0017601 21483732 4 Andrews JR Noubary F Walensky RP Risk of progression to active tuberculosis following reinfection with Mycobacterium tuberculosis Clin Infect Dis 2012 54 784 91 doi:10.1093/cid/cir951 22267721 5 Thomas A Gopi PG Santha T Chandrasekaran V Predictors of relapse among pulmonary tuberculosis patients treated in a DOTS programme in South India Int J Tuberc Lung Dis 2005 May 9 5 556 61 PMID: 15875929 15875929 6 Romanowski K Balshaw RF Benedetti A Predicting tuberculosis relapse in patients treated with the standard 6-month regimen: an individual patient data meta-analysis Thorax 2019 74 291 7 doi:10.1136/thoraxjnl-2017-211120 30420407 7 Menzies D Benedetti A Paydar A Effect of duration and intermittency of rifampin on tuberculosis treatment outcomes: A systematic review and meta-analysis PLoS Med 2009 6 doi:10.1371/journal.pmed.1000146 8 Weis SE Slocum PC Blais FX The Effect of Directly Observed Therapy on the Rates of Drug Resistance and Relapse in Tuberculosis N Engl J Med 1994 330 1179 84 doi:10.1056/NEJM199404283301702 8139628 9 Guerra-Assunção JA Houben RMGJ Crampin AC Recurrence due to relapse or reinfection with Mycobacterium tuberculosis: a whole-genome sequencing approach in a large, population-based cohort with a high HIV infection prevalence and active follow-up J Infect Dis 2015 211 1154 63 doi:10.1093/infdis/jiu574 25336729 10 Subbaraman R Nathavitharana RR Satyanarayana S The Tuberculosis Cascade of Care in India's Public Sector: A Systematic Review and Meta-analysis PLOS Med 2016 13 e1002149 doi:10.1371/journal.pmed.1002149 27780217 11 The World Bank. India demographic data https://data.worldbank.org/country/india accessed 14 Nov 2018 12 Our world in Data, CIVID-19: Google Mobility Trends https://ourworldindata.org/covid-google-mobility-trends accessed on 10 Mar 2023 13 Haario H Saksman E Tamminen J An Adaptive Metropolis Algorithm Bernoulli Published Online First 2007 doi:10.2307/3318737 ==== Refs REFERENCES 1 https://www.who.int/teams/global-tuberculosis-programme/the-end-tb-strategy 2 https://documents-dds-ny.un.org/doc/UNDOC/GEN/N17/207/63/PDF/N1720763.pdf?OpenElement 3 Global TB Report 2022 Geneva World Health Organization 2022 License CC BY-NC-SA 3.0 IGO 4 Population based survey for assessing prevalence of pulmonary tuberculosis cases in the state of Gujarat, India 2011-2012, Government of Gujarat, published by State TB Cell in 2013 5 National TB Prevalence Survey in India (2019-2021), Summary Report, Indian Council of Medical Research, Ministry of Health & Family Welfare, Govt of India, Published in March 2022 6 Arinaminpathy N Batra D Maheshwari N Tuberculosis treatment in the private healthcare sector in India:an analysis of recent trends and volumes using drug sales data BMC Infect Dis 19 539 (2019) https://doi.org/10.1186/s12879-019-4169-y 7 Nimalan Arinaminpathy The number of privately treated tuberculosis cases in India:an estimation from drug sales data Lancet Infect Dis 2016 16 1255 60 http://dx.doi.org/10.1016/S1473-3099(16)30259-6 27568356 8 Jeyashree K Thangaraj J Rade K Estimation of tuberculosis incidence at subnational level using three methods to monitor progress towards ending TB in India, 2015–2020 BMJ Open 2022 12 e060197 doi:10.1136/ bmjopen-2021-060197 9 Global tuberculosis report 2020 Geneva World Health Organization 2020 Licence: CC BY-NC-SA 3.0 IG 10 https://www.who.int/publications/m/item/methods-used-by-who-to-estimate-the-global-burden-of-tb-disease-2022 11 Glaziou P Sismanidis C Pretorius C Floyd K Methods used by WHO to estimate the Global burden of TB disease https://arxiv.org/ftp/arxiv/papers/1603/1603.00278.pdf 12 https://vizhub.healthdata.org/gbd-compare/ 13 Report on Medical certification of cause of death 2008, 2020, Registrar General of India https://crsorgi.gov.in/mccd-reports.html 14 Menzies NA Wolf E Connors D Bellerose M Sbarra AN Cohen T Hill AN Yaesoubi R Galer K White PJ Abubakar I Salomon JA Progression from latent infection to active disease in dynamic tuberculosis transmission models: a systematic review of the validity of modelling assumptions. Lancet Infect Dis 2018 Aug 18 (8) e228 e238 doi: 10.1016/S1473-3099(18)30134-8. Epub 2018 Apr 10. Erratum in: Lancet Infect Dis. 2018 Nov;18(11):1177. PMID: 29653698; PMCID: PMC6070419 15 Nimalan Arinaminpathy, et al, Tuberculosis treatment in the private healthcare sector in India: an analysis of recent trends and volumes using drug sales data BMC Infect Dis 2019 19 539 doi: 10.1186/s12879-019-4169-y 31217003 16 Steve Brooks Andrew Gelman Galin Jones and Xiao-Li Meng Handbook of Markov Chain Monte Carlo CRC Press 2011 17 Melanie H Chitwood, A spatial-mechanistic model to estimate subnational tuberculosis burden with routinely collected data: An application in Brazilian municipalities PloS Global Public Health https://doi.org/10.1371/journal.pgph.0000725

PMC010xxxxxx/PMC10353669.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-483 10.4103/ijcm.ijcm_196_22 Original Article Implementation Status of Airborne Infection Control Measures in Primary and Secondary Public Health Facilities, Puducherry: A Mixed-Methods Study Talukdar Rounik Sahu Swaroop Kumar Rajaram Manju 1 Department of Preventive and Social Medicine, JIPMER, Puducherry, India 1 Department of Pulmonary Medicine, JIPMER, Puducherry, India Address for correspondence: Dr. Swaroop Kumar Sahu, 3rd Floor, JIPMER International School of Public Health, Department of Preventive and Social Medicine, Jawaharlal Institute of Postgraduate Medical Education and Research (JIPMER), Puducherry - 605 006, India. E-mail: swaroop.sahu@gmail.com May-Jun 2023 30 5 2023 48 3 483491 26 2 2022 19 4 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Background: Poor ventilation in healthcare settings is a concern for airborne infections, particularly in light of the potential for coronavirus disease 2019 (COVID-19) transmission. This study aimed to assess the implementation status of airborne infection control (AIC) measures in primary and secondary public healthcare facilities (HCFs) and to explore the facilitating factors and barriers in the implementation of AIC measures. Methods: A mixed-methods approach was adopted, which includes a cross-sectional descriptive study using a checklist to collect data on the implementation of AIC measures in 22 primary and two secondary public HCFs in Puducherry, South India, between October 2020 and February 2021. Further, key informant interviews (KIIs) were conducted among medical officers (MOs). The qualitative data were manually analyzed, and transcripts created from handwritten notes and audio recordings were deductively evaluated. Results: Of the twenty-four health facilities visited, 54.2% had infection control (IC) committees. Annual IC training was held for housekeeping staff, MOs, nurses, and laboratory technicians in 23 (95.8%), 21 (87.5%), 20 (83.4%), and 14 (58.4%) facilities, respectively. Respiratory symptomatic patients were counseled on cough etiquettes in 22 (91.6%) facilities. Adequate cross-ventilation was present in outpatient departments in 16 (66.6%) institutions. N95 masks and face shields were provided in 21 (87.5%) facilities. Training through the KAYAKALP program and the presence of a separate sputum collection area were facilitators of IC, while lack of patient adherence and delays in fund release were found as barriers. Conclusion: Overall, the AIC measures were well-implemented, but improvements are needed in infrastructure development for patient segregation in outpatient departments and dedicated AIC training for all healthcare personnel. Airborne infection community health centers mixed-methods study primary healthcare facilities ==== Body pmcINTRODUCTION Airborne infection control (AIC) measures play a pivotal role in curtailing the transmission of infectious diseases in healthcare facilities (HCFs). It has been estimated that airborne infection contributes to 10 to 20% of endemic nosocomial infections in a HCF.[1] HCFs cater to patients with a variety of ailments, and some of them may have various underlying immunocompromised conditions, which predispose them to airborne infection. Infections, such as influenza, tuberculosis, aspergillosis, and measles, are believed to be transmitted via the airborne route.[2] The source of airborne infection can be because of sneezing, coughing, vomiting, or even the breath of an infected individual.[3] Healthcare workers (HCWs) are at a greater risk of acquiring airborne infections as they are continuously exposed to infectious patients in a confined environment.[4] This can significantly raise the risk of airborne disease transmission while attending to a large number of patients, particularly in countries such as India where the burden of hospital-acquired infection is already high. Moreover, this also poses a considerable risk in the form of a decrease among the health workforce available for treatment and management of airborne infections.[5–7] Hospital-acquired infections additionally pose medicolegal concerns due to increased morbidity and death, as well as rising costs of illness for both providers and patients.[7,8] Research suggests that 10–24% of all hospital-acquired infections with no epidemic potential transmits via air, whereas 2–4% of hospital-acquired infections, which spread through the air, holds epidemic potential.[9] AIC measures are crucial in preventing outbreaks of infectious diseases in HCFs. The absence or inadequate implementation of these measures can lead to the spread of airborne infections, which is a significant risk to patients, HCWs, and visitors. Therefore, it is imperative to prioritize measures to prevent cross-contamination and transmission of airborne diseases in HCFs.[10] To reduce the spread of any infectious disease, preventive measures are generally employed at different stages of infection transmission pathways.[11] Airborne transmission preventive measures are mainly categorized as managerial (like setting up of infection control (IC) committee, HCW training regarding AIC measures), administrative (related to measures to detect respiratory infectious patients early, separating them from other patients and expediting their treatment, providing counseling regarding cough etiquette, etc.), environmental (maintaining an adequate cross-ventilation in the facilities), and personal protective control.[12–17] The Government of India, in 2010, developed “Guidelines for Airborne Infection Control in Health Care Settings,” which incorporates all these four types of control strategies.[12] Although the guideline was chiefly intended for curtailing tuberculosis (TB) transmission among immunocompromised patients, it can help in preventing the spread of other airborne infections.[7] Few studies in India have evaluated the operational status of these strategies at the facility level. This study is the first to assess the implementation status of AIC measures in Puducherry’s public primary and secondary HCFs. We have also tried to assess the facilitating factors and barriers in the implementation of AIC measures from the provider’s perspective. METHODS This study was carried out in selected public health facilities in Puducherry from October 2020 to March 2021. There were a total of 27 primary health centers (PHCs) and two community health centers (CHCs) for a population of 9.5 lakh (Census 2011) in the Puducherry District. This study involved all the public HCFs (PHCs and CHCs) within a 20 km radius of Jawaharlal Institute of Postgraduate Medical Education and Research (JIPMER). In the process, we included both the existing CHCs and 22 of a total of 27 PHCs present in Puducherry. The decision to include all facilities within the 20 km radius of JIPMER was made to ensure that the study sample was representative of the population in the selected region and to avoid any potential bias that could arise from selective sampling. The inclusion criteria for the study were based on the proximity to JIPMER and not on any other characteristics of the facilities, which further eliminates the possibility of bias in the selection of the PHCs and CHCs. Institute Scientific Advisory Committee and Institute Ethics Committee approval was obtained before data collection. This study was conducted after ethical clearance from the Ethics Committee of JIPMER, Puducherry (project no. JIP/IEC/2019/521 dated January 24, 2020). Administrative approvals were also obtained from the Directorate of Medical Services Office, Government Health Services, Puducherry, before initiating the study. This was an explanatory sequential mixed-methods study that was carried out in two phases. In the first phase, a cross-sectional descriptive study was conducted with the intent of assessing the implementation status of AIC measures in 24 (22 PHCs and two CHCs) selected public health facilities. The data were collected using pretested structured questionnaires. The questions were framed taking into consideration the observational checklist/reporting formats adapted from the “Guidelines on Airborne Infection Control in Healthcare and Other Settings, April 2010, DGHS, MOHFW.”[12] A pilot study was conducted before data collection, and necessary modifications were made in the data collection pro forma. The site for the pilot study was chosen as one primary health facility outside the sampling frame (i.e., PHC beyond a 20 km radius from JIPMER) of this study. Information on managerial, administrative, and environmental control measures and availability of personal protective equipment were collected from the medical officer (MO) in charge of the public health facility or were gathered by direct observation during visits to the selected health facilities. Under the managerial component, a “dedicated IC committee” is defined as a group of individuals appointed by the HCF to be responsible for overseeing and implementing infection prevention and control measures. The committee’s primary purpose is to develop and monitor policies and procedures to prevent and control the spread of infectious diseases within the HCF. The committee generally consists of representatives from different departments and disciplines, including medical, nursing, housekeeping, and administrative staff.[12,18] Other relevant operational definitions can be found in the “Guidelines on Airborne Infection Control in Healthcare and Other Settings, April 2010, DGHS, and MOHFW.”[12] The methodology used to evaluate ventilation in the selected facilities is outlined in Textbox 1. Adequate ventilation was determined using the 20% rule, which was applied to various areas including examination rooms, waiting halls, and outpatient departments.[19,20] For facilities that housed laboratories as designated microscopy centers (DMCs) for tuberculosis diagnosis, ventilation was evaluated using both the 20% rule and the air change per hour (ACH) method. ACH was measured using the formula specified in Textbox 1, and airspeed was measured using a digital anemometer (HTC Instrument AVM06).[20–22] A minimum ACH of 12 was considered adequately ventilated for DMC laboratories.[19,23–25] Textbox 1 Assessment of the ventilation Ventilation was assessed by the following methods[14,23]:  1. Adequate ventilation was assessed by the “20% rule”: (Surface area of smallest opening/surface area of the room) × 100% ≥ 20%  2. Air change per hour measure was carried out using the following formula:   0.65 × airspeed (m/s) × smallest opening area (m2) × 3600 ––––––––––––––––––––––––––––––––––––––––––––––––––––––––    Room volume (m3) Air speed was measured using HTC Instrument AVM06 Digital Anemometer Data were collected and entered in online Google forms, and data analysis was carried out using Statistical Package for the Social Sciences (SPSS), version 25.0 (IBM Corp., Armonk, NY, USA). In the second phase, a total of 10 key informant interviews (KIIs) (eight from PHCs and all the two CHCs) were carried out. In the process, ten MOs in charge of personnel were interviewed from each of the selected health facilities. To select primary health centers (PHCs) for qualitative interviews, 22 facilities were divided into two groups based on the presence or absence of DMCs, with each group further subdivided based on whether the PHCs were located in an urban or rural setting. From each of the resulting four groups (urban with DMC, urban without DMC, rural with DMC, and rural without DMC), one PHC with a high caseload and one PHC with a low caseload were selected, resulting in a total of eight PHCs that were selected for conducting the interviews. The sampling strategy adapted for choosing the PHCs for KII is also depicted in Figure 1. Following the completion of 10 KIIs, further KIIs were not conducted as data saturation was achieved. The chief MOs in charge of each selected health facility were approached during their convenient time for conducting KIIs. The KIIs were conducted by an investigator who was trained in qualitative research methods. The rationale behind choosing MOs for interviews lies in the fact that MO in charge holds the overall supervisory role in PHCs and CHCs. Before beginning the interview, steps were taken to build up a good rapport with the participants. To maintain privacy, interviews were conducted either in MO’s office or at a place convenient to the MO. Before the interview, written informed consent was collected from all the study participants after they were briefed regarding the study’s purpose. At the end of the interview, the key points were summarized and their concurrence was sought before finalizing the results of KII. Figure 1 Sampling strategy to select the PHCs for conduction of Key informant interviews During the KIIs, notes were taken on paper, and recording was performed based on the preference of the participant. The audio recordings were transcribed on the same day it was conducted. To protect the confidentiality of the data, KII notes and PHC/CHC checklists were deidentified and were given unique numeric codes. For analysis, deductive content analysis was carried out. A categorization matrix was adapted based on the four levels of AIC measures (managerial, administrative, environmental, and personal protective controls) depicted in the literature.[12] All of the transcripts were carefully read, the contents were analyzed, and codes were derived to correspond to the categories that already existed. A subject matter expert reviewed the categories to see how well they represented the codes, themes, and quotations for ensuring viewpoint validity. Further, both phase 1 and phase 2 study data were integrated by explaining qualitative findings in the context of quantitative findings (narrative technique) and visually displaying potential relationships between qualitative and quantitative findings using the joint display.[26] RESULTS The general characteristics of 24 facilities (22 PHCs and two CHCs) assessed are presented in Table 1. Each facility served a median of 150 patients per day in the outpatient department. The objective of this study was to primarily assess the implementation status of AIC measures in public primary and secondary health facilities in Puducherry. In this assessment, PHCs and CHCs have not been separately presented and the results have been analyzed cumulatively considering that there was comparable population coverage between these centers. Though at both levels there are differences in service availability and workforce, population coverage under primary- and secondary-level facilities did not vary significantly. Among the selected 22 primary facilities, the median population coverage was 25176, with an interquartile range between 13037 and 53801. In the two CHCs, the population coverage was 33666 and 8535, respectively. Of 24 facilities, 10 (41.6%) had DMCs for diagnosing tuberculosis. In these DMCs, ventilation was examined using both the 20% rule and ACH estimation, and the results have been presented separately. Table 1 General healthcare facility (HCF) information for primary and secondary healthcare centers of Puducherry, 2020 [n=24 (22 PHCs and two CHCs)] (n=24) Median IQR Number of facilities operating 24/7 (n (%)) 13 (54.2) -- Population catered by facilities 25176 12043-51936 Number of OPD patients attended/day 150 108-200 Number of facilities with designated microscopy center (n (%)) 10 (41.6) -- Burden of airborne infection—in last 3 months (August–October) Number of TB suspects examined by smear microscopy for the facility with DMC 9 4-19 Number of TB patients registered for treatment (August–October) 3 1-5 CHC 1 CHC 2 Inpatient bed (including emergency dept.) —for CHC only 34 21 Average number of patients admitted in a month (average of August–October)—for CHC only ## 18 ##At the time of the visit, it was reported that, in light of COVID-19, CHC had stopped admitting patients temporarily from March 2020, with the exception of emergency cases and patients put on observation Managerial component Quantitative findings Of 10 facilities with DMC, eight (80%) facilities had dedicated IC committee. The presence of an IC committee was even found in five (35%) of the non-DMC facilities. In the previous year (2020), 12 (92%) of the 13 facilities that had an IC committee had conducted committee meetings twice a year as per guidelines given by the Ministry of Health and Family Welfare (MoHFW). In 11 (46%) facilities, IC guidelines were available at the time of the visit. Last year, IC training was attended by housekeeping staff and laboratory technicians in 23 (95%) and 14 (54%) facilities, respectively. Similarly, IC training was attended by MOs and staff nurses in 21 (87%) and 20 (83%) facilities, respectively [Table 2]. Table 2 Implementation status of managerial and administrative control measures of airborne infection control in primary and secondary health centers of Puducherry, 2020 [n=24 (22 PHCs and two CHCs)] n (%) 95% CI Number of facilities having infection control committee 13 (54.2) 33-75 Facilities having infection control committee and conducted committee meeting >2 times a year (n=13) 12/13 (92.3) -- Number of facilities having written infection control guideline 11 (45.8) 26-67 Number of facilities where the AIC plan was present or covered in infection control guideline (n=11) 3/11 (27.3) -- Number of facilities where housekeeping staff has attended any infection control training in the last year 23 (95.8) 79 – 99 Number of facilities where laboratory technician has attended any infection control training in the last year 14 (58.4) 38 – 75 Number of facilities where MO has attended IC training in last one year (2019) 21 (87.5) 67 – 97 Number of facilities where staff nurses have attended IC training in the last one year (2019) 20 (83.4) 62-95 Number of facilities where n(%) 95% CI Respiratory symptomatic patients are counseled regarding cough etiquettes and mask wearing 22 (91.6) 73-99 Separating respiratory symptomatic patients through early screening 20 (83.3) 63-95 Fast-tracking of respiratory symptomatic patients 20 (83.3) 63-95 Separate sputum collection area 19 (79.2) 57-93 Segregation of respiratory symptomatic patients in the separate waiting area 3 (12.5) 12-51 Display of IEC materials/posters of cough etiquette information in registration counter/waiting areas 22 (91.6) 73-99 Patient without mask was provided mask in facility 5 (21.0) 9 – 40 Qualitative findings Figure 2 is a joint display that contains both quantitative and qualitative data on various managerial control measures. Figure 2 Joint display of quantitative and qualitative findings on managerial airborne infection control measures in public primary & secondary healthcare facilities in Puducherry All MOs in charge stated that it has been convenient for them to maintain the facility cleaning policy after the formation of the IC committee. Regular infection control review meetings in facilities with a dedicated IC committee helped them improve their adherence to IC guidelines. It was also found that the recent recruitment of MOs at the subcenter level has helped to decrease overcrowding at PHCs and CHCs. The initiation of the KAYAKALP program by MoHFW has also facilitated the implementation of AIC measures, as the focus of this program is on improving the public health infrastructure and maintenance of cleanliness in public health facilities. This has also served as a motivator for HCWs to maintain strict hygiene standards. It was also felt that multiple webinar sessions and online conferences during the coronavirus disease 2019 (COVID-19) pandemic have contributed to increased knowledge regarding AIC and practices among HCWs. Administrative control measures Quantitative findings In 22 (91%) facilities, respiratory symptomatic patients were counseled on cough etiquette, mask wearing, and other personal protective measures. Early screening and fast-tracking of respiratory symptoms were observed in 20 (83.3%) facilities. Information, Education and Communication (IEC) display about cough etiquette and mask wearing was displayed in the registration and waiting areas of all the facilities. Among the 23 facilities where sputum collection for TB diagnosis was happening, 19 (79.2%) of them had a secluded, well-ventilated, open space. Three (12.5%) facilities identified segregating respiratory symptomatic patients in a separate waiting area as per guidelines. Masks were given to respiratory symptomatic patients who arrived at the hospital without a mask in five (21%) facilities [Table 2]. Qualitative findings Figure 3 is a joint display that contains both quantitative and qualitative data on various administrative and personal protective control measures. During the KIIs, it was found that setting up a screening outpatient department (OPD) at the entrance of each facility helped in the early screening of people having respiratory symptoms. Figure 3 Joint display of quantitative and qualitative findings on administrative & personal protective airborne infection control measures in public primary & secondary healthcare facilities in Puducherry The provision of a dedicated sputum collection area in an open space, outside at facility premises, was reported as another facilitating factor. It assisted in limiting the proximity of suspected tuberculosis patients to other patients at the health center. Steps for early detection of TB were being done in all the PHCs irrespective of whether it was a DMC or not. Sputum samples collected in non-DMC PHCs are sent to DMCs on a triweekly basis. This strategy has also helped in reducing patient crowding in the DMCs. At the time of diagnosis, TB patients were getting counseled for personal protective measures by a dedicated counselor/medical social worker (MSW). Environmental control measures Quantitative findings In 17 (70.8%) facilities, adequate cross-ventilation was present in examination rooms and waiting rooms. In seven (29%) facilities, proper seating arrangements for doctors and patients were maintained while considering directional airflow as per recommendation [Table 3]. Table 3 Implementation status of environmental control measures (ventilation aspect) and availability of personal protective equipment for airborne infection control in primary and secondary health centers of Puducherry, 2020 [n=24 (22 PHCs and two CHCs)] Total (n=24) n (%) 95% CI Number of facilities where unrestricted fixed opening of more than 20% of floor area and opening on opposite sides were found as per recommendation  1. Examination room and waiting room 17 (70.8) 48 – 87  2. Outpatient department (OPD) 16 (66.6) 46-82 Number of facilities where optimal seating arrangement of patient and doctor was maintained 7 (29.2) 12-51 Environmental control in pathology laboratory with the provision of sputum microscopy (DMCs) (n=10), PHCs-8, CHCs-2 Number of DMC laboratories where unrestricted fixed opening of more than 20% of floor area was found as per recommendation 7 (70.0) Number of DMC laboratories with two opposite openings (window/doors) 5 (50.0) Number of DMC laboratories where laboratory placed at the blind end of the building or physically isolated from hospital environments 6 (60.0) Number of DMC laboratories with anterooms 2 (20.0) Air change per hour (ACH) in DMC laboratories Median 28.5 IQR (25-31) Number of facilities where adequate availability of gloves and surgical masks to all cadre of staff was found 24 (100) Number of facilities where adequate availability of protective eyewear/face shield to frontline staff (doctors and nursing staff) was found 21 (87.5) Number of facilities where N95 masks were made available to different health cadres  1. Doctors and all nursing staff 22 (91.6)  3. Laboratory technician 6 (25)  4. Field workers (ASHA and ANM) 21 (87.5) Number of facilities where proper disposal of used masks was followed 24 (100) Ventilation status in DMCs Seven (70%) of the total ten DMC laboratories assessed had unrestricted fixed openings, and five (50%) DMCs had adequate cross-ventilation. In six (60%) facilities, the laboratory was located at the far end and was physically separated from the rest of the hospital. The presence of anteroom was found in two of 10 (20%) healthcare centers with sputum microscopy facilities as per recommended guidelines. ACH was measured in the DMC laboratories of eight PHCs and two CHCs. The median ACH was calculated to be 28.5 with an interquartile range (IQR) of 25 to 31, which was better than the minimum recommended ACH of >12. Personal protective measures Quantitative findings Gloves and surgical masks were adequately available for all the staff in all the facilities visited. Every health staff was getting one surgical mask a day for single shift duty, and for a double shift, two masks per day were made available. In 21 (87.5%) health centers, doctors and nursing staff received protective eyewear or a face shield; in 22 (96%) facilities, doctors and nursing staff received N95 masks. Field workers, such as Accredited Social Health Activist (ASHAs) and health inspectors, were also given N95 masks in 21 of 24 facilities. N95 masks were provided to laboratory technicians in six (25%) facilities [Table 3]. Qualitative findings regarding environmental control and personal protective measures: Facilitating factors from provider’s perspective During the KIIs, the MO in charge of several public HCFs reported that physical separation of the laboratory aided in the prevention of infection transmission within the facility. Statement: “Since our pathology lab and DMC setup are in a separate building, physically separated from the hospital setting, and since there is enough space, proper cross ventilation is possible, I believe this significantly decreases the risk of infection transmission.” --Medical officer (8) (Male) Barriers identified from the provider’s perspective Dedicated AIC Training After the advent of COVID-19, many webinar-based pieces of training were held at the state and national levels addressing AIC strategies, but they had not attended any training program or workshop, which gave dedicated training in AIC. Statement: “After COVID only many trainings were there, in that airborne aspect was touched, but we did not attend any dedicated training for airborne infection control. I feel it would have been beneficial if at least a training program regarding AIC is conducted once a year.” --Medical officer (4) (Male) Human Resource Even though all public health facilities had a sufficient number of doctors according to Indian Public Health Standards (IPHS) guidelines, a shortage of group IV personnel (especially sanitary staff and ward attendants) was identified as a barrier to the implementation of AIC initiatives. Statement: “We have only 4 housekeeping staff presently working here and with that maintaining disinfection practices is very difficult. If all the posts are occupied, it will be easy for us to mobilize manpower to tackle airborne infections in a better way.” --Medical officer (1) (Male) Financing Delay in receiving funding was perceived by MOs as a barrier. It was also felt that the purchase mechanism for public HCFs needed modifications to make it more convenient for the timely procurement of necessities. Statement: “Previously for any emergency requirement we could do local purchases. Now every spending has to be digitalized, so any transaction that happens has to happen through online or cheques only. Any purchase above rupees 5000 has to get approved by the deputy collector. Though it is a good process to limit unnecessary spending, it is a bit inconvenient at times.” --Medical officer (2) (Male) Infrastructure In all the facilities, there was no provision of a separate OPD area for respiratory symptomatic patients. So, segregation of chest symptomatic patients was difficult even if they were being identified early. One of the CHC MOs also pointed out regarding unavailability of a separate block for respiratory infectious diseases Statement: “We don’t have a dedicated block for respiratory infectious diseases, so we had to make isolation wards within the hospital. Since we have space available outside, a separate infectious disease block would have been preferable.” --Medical officer (6) (Male) DISCUSSION AIC measures were assessed under four domains—managerial, administrative, environmental, and personal protective measures. The literature suggests that managerial and administrative controls are regarded as the priority measures as they work on source control to interrupt transmission.[27] The proper implementation of these measures assures early detection of respiratory symptomatic and ensures isolation, thereby fast-tracking their care. Triage and management of symptomatic patients in outpatient clinics are crucial to limit exposure to other patients and HCWs.[27–30] The quantitative analysis revealed that dedicated IC committees were present in 13 (54%) facilities, and this was mainly present in the health centers where DMCs were present (seven of eight facilities with DMC had ICC). These findings fared well when compared to the findings in a study by Raj et al. (2019), They reported that an IC committee was in place at 70% of all secondary and tertiary facilities evaluated.[10] It was observed that although AIC is a part of Revised National Tuberculosis Control Program (RNTCP) training, dedicated AIC training was not imparted to the MOs of the health centers. Dedicated training on AIC for different health cadres will help in further streamlining the quality of implementation of AIC measures. MOs felt that there was a need to review the procurement mechanism for the timely purchase of the consumables and disbursement of funds for AIC. As per the facility-based observation, it was noticed that although the patients with respiratory symptoms were screened early, unfortunately, a dedicated place to segregate them was not available. Several studies have documented the beneficial effect of a separate well-ventilated waiting area for reducing the transmission potential of airborne pathogens.[31,32] Thereby, the focus should be given to necessary infrastructure development for adequate segregation of patients. A dedicated MSW offered counseling on cough etiquette and other hygienic procedures to respiratory symptomatic patients in 91.6% of the facilities. This finding was better compared to the study by Parmar et al. (2010), which reported that similar counseling was provided in 61% of all selected secondary and tertiary facilities assessed in India.[33] The fundamental aim of environmental control measures is to improve ventilation to lower the number of infectious particles. Evidence from several studies suggests that natural ventilation is the most easily adaptable option for maintaining adequate ventilation in resource-constrained settings.[22,34] It was found that 67% and 71% of the facilities had sufficient cross-ventilation in outpatient departments and the patient waiting room, respectively. Moreover, median ventilation of 28.5 ACH (recommended minimum ACH for any airborne isolation room is >12[12]) was achieved only by keeping windows and doors open or unobstructed in all DMC laboratories. These results were similar to those of a study that compared mechanical and natural ventilation in eight hospitals in Lima, Peru.[34] Strength and limitations The strength of this study was that it included nearly all primary HCFs (22/27) and both existing CHCs. A wireless anemometer was used to objectively measure ACH. Standard precautions (hand hygiene, use of personal protective equipment (PPE), etc.) are the minimum prerequisites for any infection prevention.[35] In this study, a standard checklist provided by the Ministry of Health and Welfare’s Central TB Division was utilized to assess the AIC implementation status, which includes some components of standard precaution (respiratory hygiene and cough etiquette). The present study could have been strengthened further had we included a complete assessment of standard precaution implementation. It is possible that some of the status indicators used to measure AIC may have been impacted by the COVID-19 pandemic. To better understand this impact, it would have been helpful to know the different specific timeframes at which these parameters were present or absent. CONCLUSION The study found that the administrative controls in the health facilities were well-implemented. The setting up of an institutional IC committee was found to be one of the facilitators in the implementation of AIC measures. Natural ventilation was identified as an effective method to maintain adequate air exchange in resource-constrained settings. A facility-based regular risk assessment by the facility IC committee, dedicated training for healthcare staff about AIC measures, streamlining fund disbursement methods, etc., were identified as areas for further improvement. Financial support and sponsorship This work was supported by an institutional research grant from the Jawaharlal Institute of Postgraduate Medical Education and Research, Puducherry (JIP/Res/Intramural/Sub com/2020-21). Conflicts of interest There are no conflicts of interest. Abbreviations ASHA - Accredited Social Health Activist ANM - Auxiliary Nursing Midwifery DGHS - Directorate General of Health Services MOHFW - Ministry of Health and Family Welfare Acknowledgement The authors convey their thanks to the Department of Health and Family Welfare, Puducherry, for giving us an opportunity to conduct this study, and to the medical officers and other staff of the facilities who have spent the time to give their insights into this study. ==== Refs REFERENCES 1 Beggs CB The airborne transmission of infection in hospital buildings: Fact or fiction? Indoor and Built Environment 2003 12 9 18 2 Gonzalez-Martin C Airborne infectious microorganisms Encyclopedia of Microbiology 2019 4 52 60 3 Fernstrom A Goldblatt M Aerobiology and its role in the transmission of infectious diseases J Pathog 2013 2013 1 13 4 R AK M K M RR S SK Are healthcare workers safe?Facility assessment of airborne infection control measures in public hospitals of Kerala Int J Community Med Public Health 2020 7 2723 9 5 Hudson B Toop L Mangin D Brunton C Jennings L Fletcher L Pandemic influenza A (H1N1) pdm09: Risk of infection in primary healthcare workers Br J Gen Pract 2013 63 e416 22 23735413 6 Tudor C Van der Walt M Margot B Dorman SE Pan WK Yenokyan G Tuberculosis among health care workers in KwaZulu-Natal, South Africa: A retrospective cohort analysis BMC Public Health 2014 14 891 25174848 7 Kaushal P Sangwan G Rana K Biswal M Kaur M Lakshmi PVM Implementation status of national airborne infection control guidelines in the health care facilities of a North Indian State: A mixed method study Public Health Pract 2021 2 100149 8 Klevens RM Edwards JR Richards CL Horan TC Gaynes RP Pollock DA Estimating health care-associated infections and deaths in U. S. hospitals, 2002 Public Health Rep 2007 122 160 6 9 Eickhoff T Proceedings of the workshop on engineering controls for preventing airborne infections in workers in health care and related facilities 2020 94 106 10 Raj A Ramakrishnan D Thomas CRMT Mavila AD Rajiv M Suseela RPB Assessment of health facilities for airborne infection control practices and adherence to national airborne infection control guidelines: A study from Kerala, Southern India Indian J Community Med 2019 44 S23 31728084 11 Seventer JM van Hochberg NS Principles of infectious diseases: Transmission, diagnosis, prevention, and control Int Encyclopedia Public Health 2017 22 39 12 Directorate General of Health Services M of H &FW Guidelines on Airborne Infection Control in Healthcare and Other Settings Available from: https://tbcindia.gov.in/showfile.php?lid=2858 13 Shrivastava SR Shrivastava PS Ramasamy J Airborne infection control in healthcare settings Infect Ecol Epidemiol 2013 3 21411 14 Eames I Tang JW Li Y Wilson P Airborne transmission of disease in hospitals J R Soc Interface 2009 6 (Suppl 6) S697 702 19828499 15 Zayas G Chiang MC Wong E MacDonald F Lange CF Senthilselvan A Effectiveness of cough etiquette maneuvers in disrupting the chain of transmission of infectious respiratory diseases BMC Public Health 2013 13 811 24010919 16 Malangu N Mngomezulu M Evaluation of tuberculosis infection control measures implemented at primary health care facilities in Kwazulu-Natal province of South Africa BMC Infect Dis 2015 15 117 25887523 17 Centers for Disease Control and Prevention (CDC). World Health Organization (WHO) Tuberculosis Infection-Control In The Era Of Expanding HIV Care And Treatment 1999 Available from: http://apps.who.int/iris/bitstream/handle/10665/66400/WHO_TB_99.269_ADD_eng.pdf;jsessionid=8F12E3A87B25AC02A2B8BA9512986FF6?sequence=2 18 World Health Organization Natural ventilation for infection control in health care settings 2009 Available from: https://apps.who.int/iris/handle/10665/44167 [Last accessed on 2023 Apr 12] 19 Buregyeya E Nuwaha F Verver S Criel B Colebunders R Wanyenze R Implementation of tuberculosis infection control in health facilities in Mukono and Wakiso districts, Uganda BMC Infect Dis 2013 13 360 23915376 20 Brouwer M Katamba A Katabira ET van Leth F An easy tool to assess ventilation in health facilities as part of air-borne transmission prevention: A cross-sectional survey from Uganda BMC Infect Dis 2017 17 1 8 28049444 21 Vaniman Manufacturing Co Air Changes per Hour (ACH &ACPH) Calculator Tool-CFM (Cubic Feet per Minute) 2021 Available from: https://www.vaniman.com/air-changes-per-hour-calculator/ [Last accessed on 2023 Apr 12] 22 Atkinson J Chartier Y Pessoa-Silva CL Jensen P Li Y Seto WH Understanding natural ventilation World Health Organization 2009 Available from: https://www.ncbi.nlm.nih.gov/books/NBK143285/ [Last accessed on 2022 Feb 13] 23 Jensen P lambert L Iademarco MF Ridzon R Guidelines for preventing the transmission of mycobacterium tuberculosis in health-care settings, 2005 MMWR Recomm Rep 2005 54 1 141 24 Centers for Disease Control and Prevention (CDC) Guidelines for Environmental Infection Control in Health-Care Facilities (2003) 2019 Available from: https://www.cdc.gov/infectioncontrol/guidelines/environmental/appendix/air.html [Last accessed on 2023 Apr 14] 25 ANSI, ASHRAE. Ventilation for acceptable indoor air quality 2001 62 Available from: https://www.ashrae.org/ 26 Fetters MD Curry LA Creswell JW Achieving integration in mixed methods designs—Principles and practices Health Serv Res 2013 48 2134 56 24279835 27 Kowalski WJ Hospital Airborne Infection Control Boca Raton, FL CRC Press 2012 28 Collins AS Preventing health care–associated infections Hughes RG Patient Safety and Quality: An Evidence-Based Handbook for Nurses Rockville (MD) Agency for Healthcare Research and Quality (US) 2008 Available from: http://www.ncbi.nlm.nih.gov/books/NBK2683/ [Last accessed on 2022 Aug 01] 29 Kucharski AJ Klepac P Conlan AJK Kissler SM Tang ML Fry H Effectiveness of isolation, testing, contact tracing, and physical distancing on reducing transmission of SARS-CoV-2 in different settings: A mathematical modelling study Lancet Infect Dis 2020 20 1151 60 32559451 30 Islam MS Rahman KM Sun Y Qureshi MO Abdi I Chughtai AA Current knowledge of COVID-19 and infection prevention and control strategies in healthcare settings: A global analysis Infect Control Hosp Epidemiol 2020 41 1196 206 32408911 31 McCreesh N Karat AS Baisley K Diaconu K Bozzani F Govender I Modelling the effect of infection prevention and control measures on rate of Mycobacterium tuberculosis transmission to clinic attendees in primary health clinics in South Africa BMJ Glob Health 2021 6 e007124 32 Yatmo YA Putra N Harahap MMY Saginatari DP Evaluation of spatial layout in health care waiting areas based on simulation of droplet movement trace IJTech Available from: https://ijtech.eng.ui.ac.id/article/view/2106 33 Parmar MM Sachdeva KS Rade K Ghedia M Bansal A Nagaraja SB Airborne infection control in India: Baseline assessment of health facilities Indian J Tuberc 2015 62 211 7 26970461 34 Escombe AR Moore DAJ Gilman RH Pan W Navincopa M Ticona E The infectiousness of tuberculosis patients coinfected with HIV PLoS Med 2008 5 e188 18798687 35 Department of Health. Victoria A. Infection control-standard and transmission-based precautions Available from: http://www.health.vic.gov.au/infectious-diseases/infection-control-standard-andtransmission-based-precautions [Last accessed on 2022 Aug 01]

PMC010xxxxxx/PMC10353670.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-501 10.4103/ijcm.ijcm_685_22 Short Communication Attention to the Bone Health of a Neglected Rural-Tribal Population in India: A Pilot Study Raj Vikash Barik Sitanshu Raj Manish Department of Orthopedics, All India Institute of Medical Sciences, Deoghar, Jharkhand, India Address for correspondence: Dr. Sitanshu Barik, Flat 503, Sri Ram Tower, Bompas Town, Deoghar, Jharkhand - 814 112, India. E-mail: Sitanshubarik@gmail.com May-Jun 2023 30 5 2023 48 3 501504 07 8 2022 20 4 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Objective: The aim of this pilot study was to analyze the need and feasibility of conducting prospective research on the epidemiological factors of bone mineral density (BMD) in the at-risk population in a rural and tribal-dominated area based on a screening test. Methods: This community-based retrospective cross-sectional study was based on data from the medical records from July 2021 to September 2021 at community camps organized in a rural area of Deoghar district of Jharkhand, India, and the at-risk patients who had undergone ultrasound-based BMD measurement were included in this study. Results: The mean age of the patients (N = 216) was 68.2 ± 10.2 years (range 35–73 years) with a mean T-score of -0.83 ± 0.09 (range -2.78–0.3, 95% CI 0.19). 70.3% (n = 152) of the patients were diagnosed as either osteopenic or osteoporotic based on the T-score. 14.8% (n = 32) of the patients had a comorbidity making them susceptible for low BMD. BMD was significantly low in patients above 60 years of age (t - 3.36, P – 0.0005), presence of comorbidity (t – 3.12, P – 0.001), and urban population (t - -1.93, P – 0.02). Conclusion: Although DEXA remains the study of choice, QUS can be used in primary healthcare systems in the developing world for the purposes of screening. Females, elderly, and urban residence have an increased chance of low BMD. This pilot study shows that a large-scale prospective study analyzing various aspects of bone health including dietary and lifestyle practices is the need of the hour. Bone mineral density developing nation epidemiology rural tribal ==== Body pmcINTRODUCTION Low bone mineral density (BMD) in the elderly and susceptible population leads to multiple musculoskeletal complaints like pain, deformity, and fractures. This causes a major burden on the patients, family, and the healthcare systems.[1] Fracture and deformity lead to possible loss of function and entail surgical intervention for its treatment, while pain in bones and joints is present in up to 84% of the elderly population which affects their activities of daily living.[2] These factors all the more become important in the rural areas of a developing country where the acute healthcare needs of the population are barely met, let alone the needs for improvement in quality of health care. The number of fractures in Asian countries, more specifically in southeast Asia, due to osteoporosis is higher than all the European countries combined.[3] Although dual energy x-ray absorptiometry (DEXA) is the gold standard for measuring BMD, it is not widely available in developing countries, in the setting of which quantitative ultrasound (QUS) has been proved to be clinically useful.[4] QUS has been used to assess bone status for almost two decades and has proven to be widely and clinically useful.[5] The current literature lacks any community-based epidemiological study on BMD among the at-risk population in a rural and tribal-dominated area of any developing nation. The aim of this pilot study was to analyze the need and feasibility of conducting prospective research on the epidemiological factors of BMD in the at-risk population based on a low-cost screening test. MATERIAL AND METHODS Study details This study was a retrospective study based on data from the medical records from July 2021 to September 2021 at community camps organized in a rural area of Deoghar district of Jharkhand, India, performed after obtaining institutional ethical clearance (2021-19-IND-02, 24/11/21). Consent was obtained from the patients participating in the study after explaining to them the benefits of taking part in this study. The at-risk patients who had undergone BMD measurement were included in this study. Suspected low BMD due to undiagnosed secondary causes was excluded. Area of study The study was conducted in the Deoghar district of Jharkhand state which is located in the eastern part of India. Tribals like Santhals, Paharias, and Lohras make a sizeable chunk of the population (27.1%).[6] 82.6% (12,33,712/14,92,073) of the population residing in the district belong to rural area, and of these, females constitute 48.2% (5,95,376/12,22,712) of the rural population.[6] Exposure to healthcare facilities is quite low in the district, present in only 15.3% of all villages. Inclusion criteria The method of purposive sampling was used to include subjects in the study based on the inclusion criteria. Any subject fulfilling any one of the criteria mentioned was included in the study. The at-risk population were defined as a) all women 65 years and older and men 70 years and older, b) women younger than 65 years old who have the following history: menopause, history of maternal hip fracture before the age of 50, low body mass (less than 60 kg), and history of amenorrhea more than 1 year before the age of 42, c) history of cigarette smoking, loss of height, thoracic kyphosis, d) individuals at any age with fragility or insufficiency fracture, e) individuals receiving (or expected to receive) steroid therapy, f) individuals with an endocrine disorder known to affect BMD adversely like hyperparathyroidism, hyperthyroidism, and Cushing’s syndrome, and g) individuals with medical conditions that could alter BMD like chronic renal failure, rheumatoid arthritis and other inflammatory arthritis, eating disorders including anorexia nervosa and bulimia, organ transplantation, prolonged immobilization, and conditions associated with secondary osteoporosis, such as gastrointestinal malabsorption or malnutrition, sprue, osteomalacia, vitamin D deficiency, endometriosis, acromegaly, chronic alcoholism or established cirrhosis, and multiple myeloma. BMD calculation The BMD was measured by calcaneal quantitative ultrasound (QUS) using a Pegasus device (Beam Med Ltd., Tel Aviv, Israel), which is designed to measure the SOS (m/s) of ultrasonic waves that travel longitudinally along the bones at a center frequency of 1.25 MHz. The machine uses gel as a coupling agent between the probe and skin. QUS can be measured at either the left or right calcaneus. The device was calibrated each day before data collection using a verification phantom provided by the manufacturer. The QUS T-score was calculated according to the normative data derived from a sex- and age-matched Asian population, provided by the manufacturer. The precision of BMD measurement was denoted by co-efficient of variation which was noted from five consecutive scans in five volunteers at the beginning of each week and was calculated to be 4.7%. Data collection The data collected in this cross-sectional study were age, sex, weight, height, place of residence, musculoskeletal complaints and its duration, and other relevant history like diet, menstrual, drug, and prior treatment history during the camps. The bone density and T-score derived from the QUS were be taken as a measure of BMD. T-score is the number of standard variations of BMD with relation to younger age group. According to WHO definition, normal bone mass—T-score >-1, osteopenia—T-score between -1 and -2.5, and osteoporosis—T-score <-2.5.[4] Statistical analysis Descriptive statistical analysis of the collected data was done. Continuous variables were presented as mean ± SD and categorical variables were presented as absolute numbers and percentages. The Kolmogorov–Smirnov test was applied to check for normality of data. Relevant statistical tests like Student’s t-test, Spearman correlation, and Mantel–Haenszel Chi-square test were used for the appropriate data using SPSS v 26 for Windows v11 (Chicago, USA). Subgroup analysis was done based on age (above or below 60), sex, rural or urban location, and comorbidities (presence or absence). Stratified analysis was done in the rural and urban population group based on age and sex. A P value of less than 0.05 was considered significant. RESULTS A total of 216 patients were analyzed for the study who underwent BMD measurement. The majority of the patients were females (127, 58.7%) and above 60 years of age (184, 85.2%). The mean age of the patients was 68.2 ± 10.2 years (range 35–73 years) with a mean T-score of -0.83 ± 0.09 (range -2.78–0.3, 95% CI 0.19). 70.3% (n = 152) of the patients were diagnosed as either osteopenic or osteoporotic based on the T-score. The most common complaints with which the patients presented were generalized body ache (189/216, 87.5%) and back pain (146/216, 67.5%). 14.8% (n = 32) of the patients had a comorbidity making them susceptible for low BMD (patients receiving steroid therapy = 15, history of amenorrhea before 42 years = 9, previous low velocity fracture = 4, Cushing disease = 2, hyperthyroidism = 2, hyperparathyroidism = 1, chronic renal failure = 1). Of the 15 patients receiving steroid therapy, inflammatory arthritis was the leading cause (n = 12). The most common cause of BMD calculation was the age criteria (females >65 years (107/127, 84.2%) and males >70 years (77/89, 86.5%)). The majority of the screened population was rural (145/216, 67.3%). BMD was significantly low in patients above 60 years of age (t = 3.36, P = 0.0005), presence of comorbidity (t = 3.12, P = 0.001), and urban population (t = -1.93, P = 0.02) [Table 1]. There was no significant difference in BMD between males and females (t = 0.69, P = 0.244). Among the rural population, BMD was significantly lower in patients above 60 years (mean T-score = -2.41 vs -1.34, P = 0.001), while both males and females did not differ significantly in BMD (mean T-score = -1.93 vs -2.12). In analysis of the urban population, females (mean T-score = -2.45 vs -1.24, P = 0.02) and patients above 60 years (mean T-score = -2.41 vs -1.37, P = 0.01) had significantly low BMD. Table 1 Comparison on bone mineral density of the calcaneum between various subgroups based on age, gender, presence of comorbidity, and area of residence Bone mineral density (g/cm2) P >60 years 0.47±0.01 0.0005 <60 years 0.5±0.01 Males 0.51±0.02 0.24 Females 0.49±0.04 With comorbidity 0.5±0.09 0.001 Without comorbidity 0.53±0.09 Urban 0.49±0.03 0.02 Rural 0.53±0.07 DISCUSSION Osteoporosis is a major health problem, with around 250000 hip fractures occurring annually, and of these, 10 to 15% end in a mortality within a year of fracture occurrence.[7] They are also associated with significant morbidity with loss of function and pain being common problems encountered.[8] Early diagnosis and treatment can prevent these problems, and it has been postulated that lack of early and regular screening in the at-risk population is one of the important causes for its problem severity.[9] The study population in the current study belongs to a province which is primarily rural and tribal dominated of a developing country which have a lacuna in the current literature regarding the epidemiology of BMD in the at-risk population. The aim of this pilot study was to analyze the need and feasibility of conducting prospective research on the epidemiological factors of BMD in the at-risk population based on a screening test. 70.3% (n = 152) of the patients with any clinical risk factor were diagnosed as either osteopenic or osteoporotic based on the T-score. Even with such a high percentage, there exists no data regarding the epidemiology of BMD in this geographical area. This pilot study shows that a large-scale prospective study analyzing various aspects of bone health including dietary and lifestyle practices is the need of the hour. Low BMD is a major concern among the elderly and at-risk population, and its screening leads the population to early diagnosis, treatment, and greater awareness about fracture reduction strategies.[10,11] The cheap and effective method used in this pilot study in the form QUS can be used on a wide scale in various healthcare programs, specifically targeting geriatric population which would lead to early detection of osteoporosis and reduction of fracture burden among the population. The majority of the sample population in this study were rural and elderly. Although studies report, a prevalence of secondary osteoporosis in about 30% of perimenopausal women, this study could find secondary causes in only 14.8% of the patients which could be attributed to the recall bias and small sample size.[12] Prior studies from India have noted low BMD (osteopenia or osteoporosis) in upward of 50% in the female population above 40 years of age.[13] The BMD was significantly low in urban population as compared to the rural population although it has been found to be similar in another study.[14] Increased physical activity right since childhood and adolescence due to their involvement in primary farming and livestock rearing activities and increased sunlight exposure in the rural population can be one of the causes of improved BMD. Among the rural population, the BMD of males and females did not differ significantly, whereas in the urban population, the BMD of females was significantly lower as compared to males. The same reasons of increased physical activity and sunlight exposure among the rural females may be one of the causes for the same. QUS of calcaneum, which has been used in this study to calculate BMD, has been shown to correlate well with BMD values of hip and spine by DEXA in various studies.[4,15] Although the diagnosis of osteoporosis by QUS remains controversial, it is more due to the present state of T-scores rather than the technique.[5] Previous studies in the Asian population have shown that T-scores are reliable when the reference population is same as the study population and the current study has derived T-scores based on the data of the Asian population as provided by the manufacturer.[16] Currently, around 20 QUS devices are approved for usage around the globe and Pegasus is one of them which has been used in this study.[5] It has a great potential for widespread usage due to its low-cost, portability, and non-ionizing radiation. Cost-effectiveness studies of QUS have shown that it can be used widely in the primary healthcare population to detect low bone mass in the at-risk population.[17] The retrospective cross-sectional nature of this study is the major limitation of this study along with the small sample size. The natural course of BMD and its rate of change as age progresses needs to be studied in future prospective studies. Also, a large sample size in a prospective study would increase the power as well as internal validity of the study. The normative data for calculation of T-score in the study population are not available, and it was derived from data from other population which are ethnically different, although from the same continent. Further, the interpretation of T-scores in QUS is not available, and the inference is derived from its close correlation with DEXA T-score. Although DEXA is the gold standard for diagnosing low bone mass, its availability and cost of the investigation limited its use in the current study. A correlation study of BMD with physical activity and dietary intake should be conducted to better understand the preventive measures to be implemented. Despite these limitations, this study is one of the few studies dealing with epidemiology of BMD in the at-risk population in a predominant rural and tribal region of a developing country. CONCLUSION This study does an epidemiologic analysis of BMD in at-risk population in a predominant rural and tribal region of a developing country using QUS. Although diagnosis of osteoporosis cannot be done on the basis of T-scores from QUS, they give an indication toward adequacy or inadequacy of BMD. Although DEXA remains the study of choice, QUS can be used in primary healthcare systems in the developing world for the purposes of screening. Such screening would lead to increased awareness about awareness and fracture reduction strategies among the lesser literate population. Females, elderly, and urban residence have an increased chance of low BMD and should be counseled accordingly for the improvement of bone and general health. This pilot study shows that a large-scale prospective study analyzing various aspects of bone health including dietary and lifestyle practices is the need of the hour. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. ==== Refs REFERENCES 1 Al-Saeed O Mohammed A Azizieh F Gupta R Evaluation of bone mineral density in patients with chronic low back pain Asian Spine J 2013 7 104 10 23741547 2 Woolf AD Pfleger B Burden of major musculoskeletal conditions Bull World Health Organ 2003 81 646 56 14710506 3 Ho-Pham LT Nguyen UD Pham HN Nguyen ND Nguyen TV Reference ranges for bone mineral density and prevalence of osteoporosis in Vietnamese men and women BMC Musculoskelet Disord 2011 12 182 21831301 4 Glüer CC Quantitative ultrasound techniques for the assessment of osteoporosis: Expert agreement on current status. The International Quantitative Ultrasound Consensus Group J Bone Min Res 1997 12 1280 8 5 Yang NP Jen I Chuang SY Chen SH Chou P Screening for low bone mass with quantitative ultrasonography in a community without dual-energy X-ray absorptiometry: Population-based survey BMC Musculoskelet Disord 2006 7 24 16526954 6 Census of India Website: Office of the Registrar General & Census Commissioner, India Available from: https://censusindia.gov.in/census.website/data/data-visualizations/PopulationSearch_PCA_Indicators [Last accessed on 2022 Mar 28] 7 Juby AG The use of calcaneal ultrasound evaluation of bone mineral density in cognitively impaired seniors J Am Med Dir Assoc 2004 5 377 81 15530175 8 Chrischilles EA Butler CD Davis CS Wallace RB A model of lifetime osteoporosis impact Arch Intern Med 1991 151 2026 32 1929691 9 Hajcsar EE Hawker G Bogoch ER Investigation and treatment of osteoporosis in patients with fragility fractures CMAJ 2000 163 819 22 11033708 10 Khinda R Valecha S Kumar N Walia JPS Singh K Sethi S Prevalence and predictors of osteoporosis and osteopenia in postmenopausal women of Punjab, India Int J Environ Res Public Health 2022 19 2999 35270692 11 Aggarwal A Pal R Bhadada SK Ram S Garg A Bhansali A Bone mineral density in healthy adult Indian population: The Chandigarh Urban Bone Epidemiological Study (CUBES) Arch Osteoporos 2021 16 17 33479804 12 Mirza F Canalis E Secondary osteoporosis: Pathophysiology and management Eur J Endocrinol 2015 173 R131 51 25971649 13 Khadilkar AV Mandlik RM Epidemiology and treatment of osteoporosis in women: An Indian perspective Int J Womens Health 2015 7 841 50 26527900 14 Uppal M Kaur R Assessment of bone mineral density using calcaneal ultrasound bone densitometer in college-going boys and girls of district Gurdaspur, Punjab Anthropologist 2017 28 166 72 15 Agren M Karellas A Leahey D Marks S Baran D Ultrasound attenuation of the calcaneus: A sensitive and specific discriminator of osteopenia in postmenopausal women Calcif Tissue Int 1991 48 240 4 2059875 16 Ishikawa K Ohta T Radial and metacarpal bone mineral density and calcaneal quantitative ultrasound bone mass in normal Japanese women Calcif Tissue Int 1999 65 112 6 10430641 17 Lippuner K Fuchs G Ruetsche AG Perrelet R Casez JP Neto I How well do radiographic absorptiometry and quantitative ultrasound predict osteoporosis at spine or hip?A cost-effectiveness analysis J Clin Densitom 2000 3 241 9 11090231

PMC010xxxxxx/PMC10353671.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-492 10.4103/ijcm.ijcm_901_22 Original Article Awareness and Hesitancy of COVID-19 and other Vaccines among People Living with HIV/AIDS Attending Antiretroviral Therapy (ART) Center in North India Singh Aman Dev Upal Naina 1 Oberoi Simmi Singh Namrata 2 Garg Archit 3 Kaur Avneet 4 Kaur Ashujot 5 Department of Community Medicine, GMC Patiala, India 1 Department of Distance Education, Punjabi University, Patiala, India 2 Department of Rheumatology, University of Washington, Seattle, USA 3 EMO District Hospital Phase 6 Mohali, Punjab, India 4 MBBS, Intern, Mata Kaushalya Hospital, Patiala, Punjab, India 5 MO, Mata Kaushalya Hospital, Patiala, Punjab, India Address for correspondence: Dr. Simmi Oberoi, Department of Community Medicine, New Academic Building, GMC Patiala - 147 001, Punjab, India. E-mail: oberoisimmi89@gmail.com May-Jun 2023 30 5 2023 48 3 492496 05 11 2022 20 4 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Introduction: Approximately 40 years have passed since we first learned about the human immunodeficiency virus (HIV), but several people living with HIV (PLHIV) in developing countries such as India cannot avail treatments. This makes preventive measures, such as vaccinations, critical in these persons to avoid vaccine preventable diseases (VPDs). However, little is known about the willingness and perceptions of PLHIV regarding these vaccines. Therefore, we explored vaccine awareness and hesitancy, especially during the recent COVID-19 pandemic. Objectives: The primary objective was to determine the uptake of the Covid-19 vaccine and other VPD’s among PLHIV and factors affecting the same in Antiretroviral therapy (ART) centers in a tertiary care hospital in North India. Research Design and Methods: This was a cross-sectional study of HIV patients who attended our Antiretroviral Therapy center (ART). Clinical data were collected using a questionnaire on general profile, disease information, knowledge, attitude, and practice (KAP) regarding vaccinations, and vaccination status for different VPDs. Results/Findings: We enrolled 300 subjects and found that 82% of the patients attending our ART center were aware of vaccinations, most of whom were aware of the polio vaccine (n=91, 30.33%), followed by tuberculosis (n=61, 20.33%), and the majority of them were not aware of vaccines indicated in PLHIV. We also found that the majority (n= 240, 80.23%) of patients had vaccine hesitancy, especially regarding the new COVID-19 vaccine. Conclusion: There is a need to create awareness among people about the benefits and uses of vaccination to achieve the greater goal of reduced morbidity and mortality among PLHIV. There is a need for free vaccination programs for VPDs in PLHIV patients. AIDS COVID-19 hesitancy HIV vaccination ==== Body pmcINTRODUCTION The COVID-19 pandemic has brought unprecedented challenges for public health worldwide, and vaccines have emerged as the most effective tool to control its spread.[1] However, many individuals remain hesitant or unwilling to receive the COVID-19 vaccine, posing unprecedented challenges for public health worldwide. This phenomenon is vaccine hesitancy and has become a significant obstacle in achieving herd immunity and controlling the COVID-19 pandemic, despite the widespread availability of vaccines[2,3] As per World Health Organization, vaccine hesitancy refers to delay in acceptance or refusal of vaccination despite availability of vaccination services. Vaccine hesitancy is complex and context-specific, varying across time, place, and vaccines. It is influenced by factors such as complacency, convenience, and confidence.[4] Vaccine hesitancy can have a significant impact on public health by reducing vaccine coverage and increasing the risk of vaccine-preventable diseases. For example, the recent measles outbreaks in the United States have been linked to vaccine hesitancy, and the COVID-19 pandemic has highlighted the potential consequences of vaccine hesitancy on a global scale. According to a study published in The Lancet, increasing vaccine confidence and addressing vaccine hesitancy are crucial for achieving herd immunity and controlling the spread of infectious diseases.[5] Vaccine hesitancy has been observed in previous vaccination campaigns such as measles and polio, but it has been amplified during the COVID-19 pandemic due to unique circumstances[6,7] People may hesitate to get vaccinated against COVID-19 due to various reasons, such as concerns about the vaccine’s safety and efficacy, distrust of the medical system, and misinformation on social media.[8] One group that has been particularly impacted by vaccine hesitancy is individuals living with HIV or AIDS. Vide the data published by Indian government as of 2020, there were 23.19 lakh PLHIV, indicating the magnitude of this disease. Decreased immunity in AIDS patients can lead to more frequent and severe infectious diseases, as well as increased susceptibility to opportunistic infections. Timely vaccinations can prevent or suppress many opportunistic infections in AIDS patients. Additionally, AIDS patients have higher mortality rates from COVID-19.[9,10] In a similar study conducted in China by Huang et al.,[11] in which it was reported that people living with HIV and AIDS (PLWHA) had comparatively low willingness to receive the COVID-19 vaccination as compared to general population around the world The Centers for Disease Control and Prevention, the National Institutes of Health, and the HIV Medicine Association of the Infectious Diseases Society of America have published guidelines on vaccinations for HIV/AIDS patients, which are widely followed globally, and according to these guidelines, the vaccines recommended for patients with AIDS are pneumococcal, influenza, hepatitis B virus (HBV), diphtheria, tetanus, and pertussis.[12] Some high-risk patients also require additional vaccines, such as meningococcal, hepatitis A vaccine (HAV), and human papillomavirus vaccine (HPV). Live vaccines like MMR, zoster, and varicella can be given to HIV/AIDS patients with preserved CD4 counts (CD4 ≥ 200 cells/mm3). However, Hib has been recently removed from the vaccination guidelines.[13] Therefore, we conducted this study to understand the awareness and vaccine hesitancy for the eligible vaccines among patients with HIV attending our antiretroviral therapy (ART) center. MATERIAL AND METHODS This was a cross-sectional, observational study. The study was conducted at ART center situated at a public tertiary care center in North India. Data were collected from May and June 2021. All PLHIVs above the age of 18 years attending the ART clinic were enrolled. Consecutive sampling was done. Admitted and critically ill participants were excluded. The questionnaire had four main parts: general profile, disease information, knowledge, attitude, and practices regarding vaccination, and vaccination status for different diseases (Supplemental File 1). Data were collected using a pre-structured questionnaire, after explaining in vernacular language to the participant. The questions were close ended with two or more options. The interviewer ensured that the participants understood the questions and chose appropriately to the best of their knowledge and awareness in a manner that they did not feel judged at any point during the interview. The formula used to calculate the sample size, n = 4pq/d2. Assuming expected frequency p = 50% and acceptable margin of error = 6%, total population is 5417 so at 95% confidence interval the sample size comes out to be 255 and assuming 15% non-response rate and rounding off the sample size is n = 300. Data were collected and entered in Microsoft Excel 2016. Data were cleaned, and descriptive and inferential statistics was applied. This study was granted permission by Punjab State AIDS Control Society (PSACS), which is the nodal agency for conducting and publishing any type of study on HIV-positive individuals in Punjab vide diary no-77 dated-7/05/2021 to conduct study (Supplemental File 2) and Ref No./PSACS/SIMU/2022/253 dated-19.04.2022, to publish the findings of this study (Supplemental File 3). RESULTS Patient demographics We enrolled 300 patients during the study period. The majority of the subjects were in age range 30–40 years (27.67%) followed by 50–60 years (18.67%), 40–50 years (18.33%), 20–30 years (18.33%), with a median age of 40 years; however, those with age more than 60 years constituted 13% of sample. Males were 59.33% and females 41.34%, while others were 0.66%. Majority of them were married (n = 274, 91.33%), and almost all 97.33% were adherent to treatment. Most patients who attended the clinic had a low socioeconomic status, with a mean disease duration of five years and three months, and the longest was 23 years. The majority (n = 203, 67.67%) of patients had CD4 counts below 500. A total of twenty individuals had CD4 counts higher than 1000/mm3 [Figure 1]. We found that even with a high illiteracy rate, only 1.67% of the patients did not adhere to antiretroviral therapy (ART) treatment, and most patients 292/300 received regular treatment. Furthermore, 29.3% of the patients reported improvement in their health after antiretroviral therapy (ART) treatment. Figure 1 CD4 count of study participants Participants’ knowledge attitude and practices A total of 249 (83%) patients strongly believed that vaccines are important, while 17% were not in agreement [Table 1]. As we questioned patients in their native language, we found that 45 (15%) patients were of the view that vaccination was harmful to health and can even cause death, 51 (16.6%) patients had a strong belief that they can cause disease against which they were being used, and 0.66% of the patients did not believe that they were safe for use in any condition. Table 1 Distribution of PLHIVs regarding knowledge, attitude, and practices about vaccines Knowledge, Attitude, and Practices regarding Vaccines among People Living with HIV Yes No Do you think vaccination is important? 83% 17% Do you think it is important to complete vaccination schedule? 84.67% 15.33% Do you think vaccination has side effects? 55.4% 44.6% Do you think certain vaccines are hot/cold in nature? 70.33% 29.67% Awareness about diseases against which vaccinations are available 38.33% 61.67% Do you think having HIV will lead to the vaccine becoming ineffective/less effective? 85.4% 14.6% Do you think vaccine will flare HIV? 83.33% 16.67% Are you willing to pay for vaccines that are not available free of cost/in government supply? 41.67% 58.33% Whether COVID vaccination taken? 19.33% 80.67% Did the COVID-vaccinated people have any side effects after vaccination? 62.07% 37.93% Was the COVID vaccine provided free of cost? 100% 0% Any other vaccination taken other than COVID vaccine? 0.01% 99.99% As we further questioned the patients about vaccination, we found that 98.33% preferred government centers for vaccination. The main reason (68.81%) was the availability of vaccines free of cost. Very few participants (less than 2%) stated that private centers were better because of less time consumption and more hygienic practices. Of the 134, 44.6% experienced any adverse effects from vaccination. Most of them reported that fever (n = 128,42.67%) was the most common effect they had followed by body aches (n = 105, 35%) and (n = 2, 0.66%) subjects were of the view that vaccination can cause infertility as a presumed adverse effect. We also recorded data on patient awareness about the type and availability of vaccination [Table 2]. Every seventh patient (n = 44, 14.66%) believed that HIV can make the vaccine ineffective, even if taken as per schedule. Conversely, (n = 50,16.67%) patients stated that vaccines can even cause a flare-up of their HIV infection. Of the 300 patients, very few (n = 58, 19.33%) had taken the COVID-19 vaccine and out of them only (n = 17,5.6%) completed their vaccine dose schedule. The majority of these patients were reluctant to receive the second dose because of adverse effects such as fever (n = 23, 7.6%), pain at the injection site (n = 12, 4%), and fatigue with malaise in a few patients. Among those who were aware of and desired to get vaccinated, we found that 33 (11%) were motivated by their peers for vaccination, 14 (4.6%) were motivated by health workers, and 11 (3.6%) were motivated by various sources such as social media and newspapers. These vaccinated subjects felt comfortable with the vaccination program, and the main reason for this was the efficiency of the vaccine and the free cost. Many were of the opinion that vaccines are indeed important, but opposite results were observed when asked about the diseases against which the vaccinations are available [Table 2]. Similarly, not many were willing to pay for the vaccines that are in government supply free of cost though they have taken the COVID vaccination, and the differences were found to be statistically significant [Table 3]. Table 2 Distribution of patients’ regarding their awareness about vaccines availability Awareness on Availability of Vaccine (n) Percentage (%) Polio (n=91) 30.33 Tuberculosis (n=61) 20.33 COVID-19 (n=58) 19.33 Rabies (n=54) 18.00 Hepatitis B (n=21) 7.00 Hepatitis C (n=20) 6.66 HIV (n=18) 6.00 Pneumonia (n=15) 5.00 Table 3 Distribution of PLHIVs regarding their knowledge about importance of vaccination vs availability and patient’s willingness to pay vis-à-vis uptake Variable Yes No Chi-Square and P Knowledge about importance of vaccination 83 17 42.3684 and <0.00001 Knowledge about vaccine-preventable diseases 38 62 Willingness to pay for vaccines (not in government supply) 42 58 12.4779 and 0.00412 COVID vaccination taken 19 81 COVID-19 Vaccine hesitancy We found that 242 (80.66%) participants were not vaccinated against COVID-19. The reasons behind COVID-19 vaccine hesitancy are shown in [Figure 2]. One hundred and forty-two (47.3%) had fear and anxiety associated with the vaccine and the vaccination process. This fear and anxiety are mainly associated with the myths and adverse effects of the vaccine. Sixty-one subjects had doubts about the outcomes, were also affected by the rumors associated with it, and were concerned about the vaccines not being efficacious. A few, that is, 2% (8) stated that they did not want to sit in queue for it, while 5.3% (16) were of the view that they did not want to add a vaccine among the long list of medications they were already on. Finally, 15 (5%) participants stated that they did not want to visit the hospital again and again in this low immunity state, and they also mentioned a shortage of vaccine they might face even if they visited the hospital. Figure 2 Causes of COVID-19 vaccine hesitancy among participants As per the latest guidelines published,[11] we queried whether patients were aware of the need for vaccinations to their near contacts or other household persons. We observed that none of the patients were aware of it. DISCUSSION Our study sheds light on the vaccination status of patients with HIV/AIDS, as well as the reasons behind underutilization of vaccination in such patients attending an antiretroviral clinic at a tertiary care center in India. About 80% (a large proportion) were not vaccinated against COVID-19. A similar proportion believed that vaccines might flare their disease. Additionally, half of the participants had a fear of adverse effects, which were the main reasons reported for low vaccine uptake. It is now well known that vaccines decrease mortality and morbidity from various preventable diseases, and a recent meta-analysis showed that any type of vaccine can reduce mortality related to a specific disease.[14-18] Hence, it is recommended to vaccinate PLHIV to reduce mortality. We found that 83% of the participants were aware of vaccination, but unfortunately, 61.67% were not aware of vaccinations specifically recommended for AIDS. The majority (74.6%) knew about vaccinations for newborns provided in hospitals free of cost. Similar poor awareness results have been reported in other countries.[19-21] In response, various governments have taken steps, such as observance of HIV vaccine awareness day on May 18.[22] However, there is a need to create further awareness through various social media platforms, tele advertisements, and national policies. Seventeen percent of our patients strongly denied vaccination. Reasons for this were fear, anxiety, and efficacy. A similar study was published in 2013, 2021 indicating that 14.9% of eligible candidates were in denial[23,24] The reasons mentioned in the above studies were similar to the reasons we found in our patients. All 17% of the patients were illiterate, so a lower education level may also be considered a factor behind vaccine hesitancy. We asked patients about their knowledge of the adverse effects of vaccination and found that most patients reported adverse effects similar to a large study on 6,55,590 doses by Menni et al.,[25] but the frequency of reporting varies in comparison with both studies. We did not find any data related to vaccinations causing infertility in human subjects, indicating that there is a misunderstanding of the effects of vaccines in the general population. Such misunderstandings can be improved by patient education and vaccination awareness programs, but unfortunately, these are lacking in our country. We asked patients about their socioeconomic status and stability; most of our enrolled patients (96.5%) were below the poverty line (BPL). This was the main factor behind many other factors for failure of the vaccination program, which is paid in our country, and if we include private sector insurance, they did not provide any vaccine under insurance. The universal immunization program (UIP) of India does not cover patients affected by HIV,[26,27] so we recommend that there must be a program for AIDS immunization programs along with ART treatment to prevent vaccine-preventable diseases (VPDs). Policymakers can also extend the UIP to AIDS patients. More effectively, UIP can also be integrated with the National AIDS Control Program (NACP).[28] This is a unique study that asked patients about vaccines that could be used in AIDS for better treatment and decreased morbidity. The majority of the responses we received were for polio vaccine, indicating that the pulse polio program is effective in communication. Hence, we can strategically follow the steps of such an effective program to aid VPDs in PLHIV. The main strength of this study was its novelty, as very little is known regarding vaccination prevalence and attitudes toward vaccines among PLHIV in India. The information gathered from this study can directly inform policymakers to guide efforts to increase awareness among people regarding the necessity and safety of vaccines.[29] The main limitation of this study is that it was a single-center study, and the data could not be extrapolated to the whole country. CONCLUSION Majority (69%) of the participants were below 50 years of age, married (91.33%), and adhered to the treatment (97.33%). The participants were hesitant to receive vaccines, despite knowing the significance of getting vaccinated and completing its schedule (83% and 84.67%, respectively). However, their knowledge regarding its adverse effects, availability, and its effect in the presence of HIV disease was poor and inaccurate. They had certain concerns that vaccines not being effective in people with HIV (85.4%). Moreover, less than 50% of people are willing to pay for vaccines and almost all felt that they should be provided for free. Thus, inferring that there is a need to alley their fears, misconceptions, and false beliefs and provide the accurate knowledge and motivation for vaccination. RECOMMENDATIONS We urge local and national organizations to come to action, so that there could be a robust vaccination program for PLHIV based on local endemicity of VPDs and media support for its success. We also recommend that there be focus group discussions with PLHIV regarding vaccines that can help reduce their hesitancy. There is an urgent need for vaccination to decrease morbidity and mortality, especially also from the COVID-19 infection, in persons with a suppressed immune system such as PLHIV. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. Acknowledgements The authors would like to thank PSACS for providing valuable support and permission to conduct and publish study. ==== Refs REFERENCES 1 Wouters OJ Shadlen KC Salcher-Konrad M Pollard AJ Larson HJ Teerawattananon Y Challenges in ensuring global access to COVID-19 vaccines: Production, affordability, allocation, and deployment Lancet 2021 397 1023 34 33587887 2 Vaccine hesitancy: Definition, scope and determinants – PubMed Available from: https://pubmed.ncbi.nlm.nih.gov/25896383/ [Last accessed on 2023 Feb 28] 3 COVID-19 Vaccine Hesitancy Worldwide: A Concise Systematic Review of Vaccine Acceptance Rates – PMC Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7920465/ [Last accessed on 2023 Feb 28] 4 (Reticencia a la vacunación: Un desafío creciente para los programas de inmunización Available from: https://www.who.int/news/item/18-08-2015-vaccine-hesitancy-a-growing-challenge-for-immunization-programmes [Last accessed on 2023 Mar 31] 5 Larson HJ Jarrett C Eckersberger E Smith DM Paterson P Understanding vaccine hesitancy around vaccines and vaccination from a global perspective: A systematic review of published literature, 2007–2019 Vaccine 2020 38 7067 74 6 Dubé E Laberge C Guay M Bramadat P Roy R Bettinger JA Vaccine hesitancy Hum Vaccin Immunother 2013 9 1763 73 23584253 7 Schoch-Spana M Brunson EK Long R Ruth A Ravi SJ Trotochaud M The public's role in COVID-19 vaccination: Human-centered recommendations to enhance pandemic vaccine awareness, access, and acceptance in the United States Vaccine 2021 39 6004 12 33160755 8 Skafle I Nordahl-Hansen A Quintana DS Wynn R Gabarron E Misinformation about COVID-19 vaccines on social media: Rapid review J Med Internet Res 2022 24 e37367 35816685 9 Yin Z Rice BD Waight P Miller E George R Brown AE Invasive pneumococcal disease among HIV-positive individuals, 2000-2009 AIDS 2012 26 87 94 22008657 10 Bhaskaran K Rentsch CT MacKenna B Schultze A Mehrkar A Bates CJ HIV infection and COVID-19 death: A population-based cohort analysis of UK primary care data and linked national death registrations within the OpenSAFELY platform Lancet HIV 2021 8 e24 32 33316211 11 Huang X Yu M Fu G Lan G Li L Yang J Willingness to receive COVID-19 vaccination among people living with HIV and AIDS in China: Nationwide cross-sectional online survey JMIR Public Health Surveill 2021 7 e31125 34543223 12 What's New in the Guidelines |NIH 2023 Available from: https://clinicalinfo.hiv.gov/en/guidelines/hiv-clinical-guidelines-adult-and-adolescent-opportunistic-infections/whats-new [Last accessed on 2023 Feb 28] 13 Steinhart R Reingold AL Taylor F Anderson G Wenger JD Invasive Haemophilus influenzae infections in men with HIV infection JAMA 1992 268 3350 2 1453528 14 Tasker SA Treanor JJ Paxton WB Wallace MR Efficacy of influenza vaccination in HIV-infected persons. A randomized, double-blind, placebo-controlled trial Ann Intern Med 1999 131 430 3 10498559 15 Yamanaka H Teruya K Tanaka M Kikuchi Y Takahashi T Kimura S HIV/Influenza Vaccine Study Team. Efficacy and immunologic responses to influenza vaccine in HIV-1-infected patients J Acquir Immune Defic Syndr 2005 39 167 73 15905732 16 Hung CC Chen MY Hsieh SM Hsiao CF Sheng WH Chang SC Clinical experience of the 23-valent capsular polysaccharide pneumococcal vaccination in HIV-1-infected patients receiving highly active antiretroviral therapy: A prospective observational study Vaccine 2004 22 2006 12 15121313 17 Grau I Pallares R Tubau F Schulze MH Llopis F Podzamczer D Epidemiologic changes in bacteremic pneumococcal disease in patients with human immunodeficiency virus in the era of highly active antiretroviral therapy Arch Intern Med 2005 165 1533 40 16009870 18 Huang YZ Kuan CC Vaccination to reduce severe COVID-19 and mortality in COVID-19 patients: A systematic review and meta-analysis Eur Rev Med Pharmacol Sci 2022 26 1770 6 35302230 19 Tedaldi EM Baker RK Moorman AC Wood KC Fuhrer J McCabe RE Hepatitis A and B vaccination practices for ambulatory patients infected with HIV Clin Infect Dis 2004 38 1478 84 15156488 20 Bailey CL Smith V Sands M Hepatitis B vaccine: A seven-year study of adherence to the immunization guidelinesand efficacy in HIV-1-positive adults Int J Infect Dis 2008 12 e77 83 18723381 21 Gallagher KM Juhasz M Harris NS Teshale EH Adult and Adolescent Spectrum of HIV Disease Group Predictorsof influenza vaccination in HIV-infected patientsin the United States, 1990–2002 J Infect Dis 2007 196 339 46 17597447 22 Dieffenbach CW Fauci AS The search for an HIV vaccine, the journey continues J Int AIDS Soc 2020 23 e25506 32418357 23 Dubé E Laberge C Guay M Bramadat P Roy R Bettinger J Vaccine hesitancy: An overview Hum Vaccin Immunother 2013 9 1763 73 23584253 24 Troiano G Nardi A Vaccine hesitancy in the era of COVID-19 Public Health 2021 194 245 51 33965796 25 Menni C May A Polidori L Louca P Wolf J Capdevila J COVID-19 vaccine waning and effectiveness and side-effects of boosters: A prospective community study from the ZOE COVID Study Lancet Infect Dis 2022 22 1002 10 35405090 26 Wodi AP Murthy N Bernstein H McNally V Cineas S Ault K Advisory committee on immunization practices recommended immunization schedule for children and adolescents aged 18 years or younger-United States, 2022 MMWR Morb Mortal Wkly Rep 2022 71 234 7 35176011 27 Kumarasamy N Venkatesh KK Mayer KH Freedberg K Financial burden of health services for people with HIV/AIDS in India Indian J Med Res 2007 126 509 17 18219077 28 Tanwar S Rewari BB Rao CD Seguy N India's HIV programme: Successes and challenges J Virus Erad 2016 2 (Suppl 4) 15 9 28275445 29 Available from: https://ijhpr.biomedcentral.com/articles/10.1186/s13584-022-00527-8

PMC010xxxxxx/PMC10353672.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-377 10.4103/ijcm.ijcm_315_23 Editorial Commentry Effective, Respectful and Affordable Care: A call for Decentralized Maternal Care Mohan Pavitra Ramanathan Srividya Primary Health Care Initiative, A Partnership of Basic Health Care Services and Indian Institute of Management, Udaipur, Rajasthan, India Address for correspondence: Dr. Pavitra Mohan, Basic Health Care Services, 39 Krishna Colony, Bedla Road, Udaipur, Rajasthan, India. E-mail: pavitra@bhs.org.in May-Jun 2023 30 5 2023 48 3 377378 17 5 2023 23 5 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. ==== Body pmcIn 2005, for the first time, the Indian government incentivized institutional deliveries by launching the Janani Suraksha Yojana (JSY), an intervention under the National Rural Health Mission (NRHM) that promised cash assistance to mothers delivering at a medical facility. While the implementation of JSY increased the number of institutional deliveries from 40.8% in 2005–2006 (National Family Health Survey (NFHS) III) to 88.6% in 2019–2021 (NFHS V), there is a recognition that the maternal and newborn health indicators did not improve commensurately. Further, there are huge variations in levels of maternal and newborn mortality across different parts of the country and are higher than in other similar economies. Apart from the absence of a proportionate decrease in maternal and infant mortality ratios in tandem with the overall increase in institutional childbirths, there is also a growing concern about rising cesarean section (C-section) rates. Overall, the C-section rates in India have crossed the World Health Organization’s (WHO) threshold of 15%. On the one hand, there are districts, regions, and states where C-section facilities are not available even when required; on the other hand, there are states and regions where most childbirths occur through C-sections. In the state of Telangana, for example, 60% of the total deliveries in 2021 occurred by a C-section. Such an “over-medicalization” of childbirth is likely to result in increased costs and higher risks of childbirth. Respectful maternity care is another cornerstone of quality care. However, reports of ill-treatment, abuses, and negative experiences associated with institutional birth are common in hospitals across India. A study conducted at public and private maternity facilities in Uttar Pradesh reveals that physical violence was often used while performing fundal pressure. Slapping often occurred when fundal pressure was applied. Verbal abuse included speaking down to the expecting woman, using insulting language, and threatening to perform a C-section if expectant mothers did not stop screaming or weeping.[1] Despite the advantage of readiness to manage obstetric complications, many women are mistreated throughout pregnancy and childbirth in hospital births, including the conduction of vaginal examination without their permission, labor induction, and epidural anesthesia.[2] Decentralized maternal care Primary health care is a comprehensive, equitable, cost-effective, and efficient approach to improving people’s physical and mental health and social well-being. There is a pervasive and growing need for primary health care, especially in low- and middle-income countries (LMICs) like India, where primary health centers (PHCs) are the bedrock of rural health service, often being the first point of contact with the healthcare system. Maternal and infant health services are an essential component of the service package provided by the PHCs. To this end, PHCs could be the ideal setting for low-risk pregnancies, as they are safe, cost-effective, and treat women with the lowest degree of complexity. Research from underdeveloped regions in developed countries shows that decentralization of maternal health care during pregnancy and childbirths, provided by skilled attendants, can indeed lower maternal and infant mortality, is less expensive, and is more responsive and respectful to women.[3] However, only a small fraction of deliveries occur at PHCs in India. There is also a growing discourse in policy circles in India that childbirths are safest if conducted in hospitals by obstetricians as opposed to doctors or nurses/midwives in primary care facilities. This argument is partly based on the evidence from the studies, largely from affluent countries, which evidence a volume–outcome linkage for deliveries, with lower mortality for births occurring at hospitals equipped with superior-level neonatal intensive care units (NICUs).[4] Decentralized maternal care is effective, equitable, and respectful We recently concluded a review of evidence and experience across developed and developing countries to compare the effectiveness, costs, equity, and acceptability of maternal care in primary care settings, delivered by nurses/midwives (decentralized maternal care) versus hospital-based care delivered by doctors and obstetricians.[5] The review revealed that decentralized care delivered by nurses/midwifes improves maternal and newborn outcomes at significantly lower costs. Not surprisingly, being closer to communities, decentralized maternal care also improves access of marginalized communities to maternal care, enhancing equity. We also found strong evidence that decentralized care by nurses/midwives reduces unnecessary interventions and potentially harmful practices. For example, episiotomy rates are much lower among women whose deliveries are conducted by midwives. In this review, we also examined why only a small proportion of all childbirths occur in PHCs in India. We found out that this is largely on account of PHCs not being adequately equipped and PHC teams not being adequately supported and skilled. In the state of Tamil Nadu, where PHCs are well equipped and PHC teams are skilled and supported, a large proportion of births do take place in PHCs and community health centers (CHCs). A call for decentralized maternal care in India Based on the review and on our own experience of delivering maternal–newborn care in primary care settings in rural Rajasthan, we call for an urgent shift to decentralized maternal care in India. This would require three actions: First, it would require strengthening primary care facilities to deliver maternal–newborn care. The example of Tamil Nadu shows that when such investments are made, a large proportion of childbirths occur in PHCs and CHCs, and there is a proportionate decline in maternal–newborn mortality. Second, we need to strengthen the training, skilling, and supportive supervision of nurses (general nurse midwives and auxiliary nurse midwives) and that of the new cadre of midwives. Third, the linkage of primary care facilities with secondary and tertiary hospitals for emergency obstetric care will need to be strengthened so that women who develop complications can be timely and appropriately referred for emergency obstetric care. Acknowledgements This study is inspired by the tireless efforts of all nurses and midwives who provide respectful and skilled care to women in remote and rural areas. The policy brief on which this editorial is based was financially supported by the United Nations Children’s Fund (UNICEF) India Country Office. Dr Evita Fernandez and Dr Kirti Iyengar reviewed the draft manuscript and provided useful inputs. ==== Refs REFERENCES 1 Sharma G Penn-Kekana L Halder K Filippi V An investigation into mistreatment of women during labour and childbirth in maternity care facilities in Uttar Pradesh, India: A mixed methods study Reprod Health 2019 16 7 30674323 2 The baby arrived at a Natural Birth Centre Economic and Political Weekly Available from: https://www.epw.in/journal/2022/22/postscript/baby-arrived-natural-birth-centre.html [Last accessed on 2022 Nov 10] 3 Sandall J Soltani H Gates S Shennan A Devane D Midwife-led continuity models versus other models of care for childbearing women Cochrane Database Syst Rev 2016 4 CD004667 27121907 4 Phibbs CS Bronstein JM Buxton E Phibbs RH The effects of patient volume and level of care at the hospital of birth on neonatal mortality JAMA 1996 276 1054 9 8847767 5 Decentralised Maternal Care in India Available from: https://bhs.org.in/wp-content/uploads/2023/05/Policy-Brief-3UNICEF-PHI.pdf [Last accessed on 2023 May 16]

PMC010xxxxxx/PMC10353673.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-418 10.4103/ijcm.ijcm_777_22 Original Article Adverse Events among Beneficiaries who Received a Dose of Sputnik V Vaccine at a Tertiary Care Hospital in Coastal Karnataka, India Ameer Basma R. Akshaya Kibballi Madhukeshwar Bhargava Madhavi Jamal Jeshela Department of Community Medicine, Yenepoya Medical College, Yenepoya (Deemed to be University), Mangaluru, Karnataka, India Address for correspondence: Dr. Kibballi Madhukeshwar Akshaya, Professor, Department of Community Medicine, Yenepoya Medical College, Yenepoya (Deemed to be University), Mangaluru, Karnataka, India. E-mail: docakshay@gmail.com May-Jun 2023 30 5 2023 48 3 418421 14 9 2022 24 2 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Introduction: Vaccination has played a vital role in containing the COVID-19 pandemic. Sputnik V was the third vaccine approved for emergency use in India. The objectives of the present study were to document the adverse events following Sputnik V vaccination and the factors associated with adverse events. Methodology: This cross-sectional study was conducted during September and October 2021 in a teaching hospital of Karnataka. Ethics approval and CTRI registration were obtained before collecting the data. All persons receiving at least one dose of vaccine were invited to participate and baseline information was collected after written informed consent. They were contacted telephonically to enquire about the adverse events. Data were entered in Microsoft Excel and analyzed using SPSS Version 23 to describe percentages and proportions. Results: The median age of 2532 participants was 31 (IQR 25-39) years and 60.4% were males. Minor adverse events were seen among 29.4% participants. Most common symptoms with first dose were fever, vaccination site tenderness, myalgia and headache, and with second dose were fever, myalgia, headache, and vaccination site tenderness. No severe adverse events were reported in our study. The adverse events were seen more among females (P < 0.05) and with the first dose (P < 0.05). Conclusion: Most common adverse events were similar to symptoms suggested by the vaccine manufacturers with fever being the most common one. A follow-up after a longer lag time may be recommended to enquire whether the vaccinees developed serious adverse events. Adverse events COVID-19 Sputnik V vaccination ==== Body pmcINTRODUCTION Coronavirus which causes COVID-19 infection has a dynamic nature and mutates frequently. Currently, there are 5 variants of concern (VOC) recognized by the World Health Organization (WHO), the latest being the Delta and Omicron variants.[1] In this scenario, the only best hope is vaccination against the disease. India launched its vaccination campaign on January 2021 in a 3-phased manner starting with Covishield. Currently, 12 vaccines are approved for use and 16 vaccines are undergoing clinical trials in India.[2] As of September 14, 2022, according to the COWIN portal (official website of the Government of India for vaccination against COVID-19), 94.56 crore beneficiaries in India and 5.50 crore beneficiaries in Karnataka have received both the doses of COVID-19 vaccine.[3] The Sputnik V vaccine, also known as “Gam-COVID-Vac,” was developed in Russia at The Gamaleya National Research Center of Epidemiology and Microbiology.[4] Sputnik V has an efficacy of 91.6% based on a study conducted among 19,866 volunteers.[5] It is a heterologous recombinant adenovirus using adenovirus 26 (Ad26) and adenovirus 5 (Ad5) as vectors for the expression of the SARS-CoV-2 spike protein in the two consecutive doses given to beneficiaries, respectively. It is given at a dose of 0.5ml each via intramuscular route in the upper third of the deltoid muscle. The phase 3 trial by the parent institute recruited 21977 participants and randomly assigned to vaccine group (n = 16,501) or the placebo group (n = 5476). Of them, 94% of volunteers complained of minor side-effects like flu-like syndromes, injection site reactions, headache, and asthenia.[5] These symptoms were similar to those developed after vaccination with Covishield and Covaxin. Seventy episodes of severe adverse events (SAE) were reported from the same trial but the side effects were decided to be not related to the vaccine by an Independent Data Monitoring Committee. In the Indian version of the clinical trial conducted, it was reported that out of 1500 volunteers recruited, 33.1% reported 1,784 adverse events but there were no SAE.[6] The safety of vaccines and the consecutive side effects have been studied in controlled settings by researchers. Our objectives were to document adverse events following administration of Sputnik V vaccine and examine the factors associated with adverse events among the vaccine beneficiaries from a teaching hospital of Mangalore, Karnataka. METHODOLOGY This was a cross-sectional study of all adult recipients (complete enumeration) of the Sputnik V vaccine from Yenepoya Medical College, Mangalore, situated in coastal Karnataka from September 6, 2021, to November 6, 2021. The vaccine recipients usually waited for 30 minutes post-vaccination to monitor for any immediate side effects as per the national guidelines for vaccination during the study period. During this waiting period, they were explained by postgraduate residents pursuing Community Medicine and trained medical interns, about the side effects of the vaccine as mentioned by the manufacturer in the information leaflet, in their native language (Kannada/Malayalam). Most of these vaccine recipients were from coastal Karnataka and northern part of Kerala. Informed written consent was obtained from those who were willing to be a part of the study. Any effects that the participants had during the post-vaccination observation period were recorded by the doctor in-charge of the vaccination session. Information on socio-demographic data, contact details, and presence of any co-morbidity was collected. The beneficiaries were asked whether they developed any discomfort after receiving the vaccination and were asked to describe it in their own words. Later, these participants were contacted telephonically on the 7th day. A minimum of three telephonic calls were made before declaring the participant as non-respondent. The same protocol was followed when a participant took the second dose also. Any symptom reported by the participant that did not require hospitalization was considered as an adverse event (AE) and any symptom that needed medical intervention and hospitalization was considered as a serious adverse event (SAE). A dedicated phone number was provided to the participants to contact or message for any queries regarding the vaccine and to receive guidance on any side effects the participant may develop. Data were entered in Microsoft Excel and analyzed using SPSS Version 23. Descriptive statistics were reported in the form of the mean (SD) and median (IQR) for continuous variables, depending on the distribution of the data, and categorical variables were presented as frequency (proportions). Exposure variables considered were age, gender, presence of comorbidities, and the dose of vaccine. Statistical significance was considered when the P value was less than 0.05. Ethics approval was obtained from the institutional ethics committee (Protocol number: YEC2/887) of Yenepoya Medical College, Mangalore, on 12 August 2021. CTRI registration (CTRI/2021/09/036228) was obtained on September 3, 2021 before starting the study. RESULTS A total of 3110 vaccine beneficiaries were approached to be part of the study. Of the 3099 participants who consented, 2532 (81.7%) could be contacted telephonically and were finally enrolled. Of the 1553 participants who took the first dose, 1237 (79.7%) responded, and of the 1546 participants who received the second dose, 1295 (83.8%) responded to the telephonic call. The median (IQR) age of participants was 31 (25-39) years. The youngest was 18 years and the oldest participant was 84 years age. Most participants were males (1701/2532, 67.1%). A total of 99 (3.9%) participants reported of co-morbidities; 89 (3.5%) among these had diabetes mellitus, 23 (0.9%) had hypertension, 12 (0.5%) were with other co-morbidities (hypothyroidism, dyslipidemia, bronchial asthma). No adverse events were reported among any of our participants within 30 minutes of receiving both doses of the Sputnik V vaccine. Symptoms suggestive of adverse events were reported by 29.4% (745/2532) participants of which 450 (36.4%) was seen among those who received the first dose and 295 (22.8%) who received the second dose [Table 1] when they were contacted on the 7th day post-vaccination. All of them were considered as mild symptoms. No SAE were reported in our study. Of these who experienced an adverse event, 351 (75.7%) took some form of oral medication for symptomatic relief. Female gender and the first dose of the vaccine were associated with higher adverse events (P < 0.05) [Table 2]. Table 1 Symptoms suggestive of adverse events among participants who received first or second dose of Sputnik V vaccine at a tertiary medical college in coastal Karnataka, India, 2021. (n=2532) Symptom suggestive of adverse event Total beneficiaries (2532) Dose 1 beneficiaries (1237) Dose 2 beneficiaries (1295) Any symptom* 745 (29.4) 450 (36.4) 295 (22.8) Fever 450 (17.8) 282 (22.8) 168 (13) Myalgia 154 (6.1) 86 (7) 68 (5.3) Vaccination site tenderness 147 (5.8) 105 (8.5) 42 (3.2) Headache 124 (4.9) 75 (3.8) 49 (4.9) Asthenia 37 (1.5) 23 (1.9) 14 (1.1) Nasopharyngeal symptoms 39 (1.5) 16 (1.2) 25 (1.9) Arthralgia and pain in extremities 19 (0.8) 13 (1.1) 6 (0.5) Cough 16 (0.6) 5 (0.4) 11 (0.8) Dizziness 17 (0.7) 10 (0.8) 7 (0.5) Increased temperature at vaccination site 14 (0.6) 7 (0.6) 7 (0.5) Nausea and vomiting 13 (0.5) 6 (0.5) 7 (0.6) Decreased appetite 5 (0.2) 1 (0.1) 4 (0.3) Pruritis 2 (0.1) 0 2 (0.2) Figures within brackets indicate percentages. *The sum total of each column is more than the individual symptoms put together as an individual would have experienced more than one symptom Table 2 Comparison of presence of symptoms with sociodemographic and clinical variables among participants who received first or second dose of Sputnik V vaccine at a tertiary medical college in coastal Karnataka, India, 2021. (n=2532) Variable Category Presence of symptoms n (%) χ 2 P Age ≤ 31 years** 400 (30.1) 0.654 0.419 > 31 years 345 (28.7) Gender Male 464 (27.3) 11.487 0.001* Female 281 (33.8) Comorbidities Present 37 (37.4) 3.136 0.077 Absent 708 (29.1) Dose 1st dose 450 (36.4) 56.338 <0.001* 2nd dose 295 (22.8) * Statistically significant, **Median age of participants was 31 years DISCUSSION This is one of the first few studies from India that has looked into adverse effects of Sputnik V vaccine in India after its launch. A large sample size and a good response rate of the participants are strengths of this study. The telephonic interviews were conducted by doctors who were well versed with the symptoms that may appear after taking the vaccine. The study revealed that three out of every ten participants complained of adverse effects following Sputnik V vaccination. This proportion was much more than the studies done from Iran and Argentina, which included only healthcare workers.[7,8] Our study participants reported no SAE, similar to studies conducted in Iran, Italy, and Russia.[5,9,10] A follow-up contact after a longer lag time may be recommended to gain knowledge on whether the vaccine recipients developed any serious adverse events. Fever was the common side effect seen in our study with either of the doses. This was different from other studies done in Argentina, Iran, Italy, or Russia, where the most common event reported was vaccination site pain or tenderness. This was the third most common adverse event in our study.[7-9,11] The population included in our study may not have perceived this as an important side effect or their pain threshold would be higher. The most common symptoms, other than fever were myalgia, vaccination site tenderness or pain, headache, asthenia, and nasopharyngeal symptoms. These were similar to studies from Iran, Italy, and also from Russia during the phase 1/2 trial related to the vaccine, but not in the same order or proportion of participants reporting them.[7-9] In a study documenting SAE following Covishield or Covaxin, the most common symptoms were feeling generally unwell, headache, fever, fatigue, and myalgia.[12] These symptoms were similar to those from our study, although the proportions were different. We did not find any association of the adverse events with the age categories. A study conducted by Jarynowski showed that the symptoms decreased with age and in the study by Pagotto et al.,[7] symptoms were reported more by participants who were 55 years or younger.[8,9,11] Participants who were younger than 40 years reported higher symptoms after COVID-19 vaccination in Iran.[10] A study by Kamal et al.[13] in India, reported that symptoms after Covishield or Covaxin were seen with participants who were more than 50 years of age. Women seemed to be developed more symptoms similar to the findings from the studies by Pagotto et al.,[7] Montalti et al.,[9] and Jarynowski et al.[11] The study by Zare et al.,[10] comparing adverse events after taking either Covishield, Sputnik V, or Covaxin also showed that females reported more symptoms and it was more when they have received Sputnik V vaccination, while a study by Kamal et al.[13] shows that males reported more symptoms after vaccination with Covishield or Covaxin. The participants with comorbidities reported having more side effects in our study which is, similar to studies by Montalti et al.[9] and Jarynowski et al.[11] The symptoms were higher after receiving the first dose, compared to the second dose which is similar to the study from Argentina.[7] This significance may be due to the fact that those who received vaccine for the first-time reported symptoms more. The National Technical Advisory Group on Immunization (NTAGI) in India has recommended that Sputnik Light which contains the Ad26 can be taken as precautionary dose starting from May 2022.[14] Knowing the adverse events will help participants make informed decisions to receive the vaccine. An important limitation of our study is that there might have under-reporting as the participants were contacted telephonically. The non-response rate is near 20% as a result of which some adverse events especially SAE may have been missed. The study takes into account only beneficiaries from a specific population; hence, the results may not be generalizable. CONCLUSION This study revealed that the most common adverse event following Sputnik V administration was fever in general and also with either of the first or second doses. The adverse events were similar to symptoms suggested by the vaccine manufacturers. Adverse events were seen more among females and with the first dose. Consent to participate Participants were explained the side effects of the vaccine as mentioned by the manufacturer in the information leaflet, in their native language. Informed written consent was obtained from those who were willing to be a part of the study. Ethics approval Ethics approval was obtained from the institutional ethics committee (Protocol number: YEC2/887) of Yenepoya Medical College, Mangalore, on 12 August 2021. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. Acknowledgement We thank all the postgraduate students pursuing MD in Community Medicine and the medical interns who helped in the active surveillance of adverse events among the participants via telephonic calls. ==== Refs REFERENCES 1 World Health Organization. Tracking SARS-CoV-2 variants 2022 Available from: https://www.who.int/en/activities/tracking-SARS-CoV-2-variants/ [Last accessed on 2022 May 13] 2 India – COVID19 Vaccine Tracker Available from: https://covid19.trackvaccines.org/country/india/ [Last accessed on 2022 May 13] 3 Ministry of Health and Family Welfare, Government of India. CoWIN Dashboard 2022 Available from: https://dashboard.cowin.gov.in/ [Last accessed on 2022 Sep 14] 4 Press Information Bureau. The National Regulator grants Permission for Restricted Use in Emergency Situations to Sputnik-V vaccine [Internet] Delhi, India 2021 Available from: https://pib.gov.in/PressReleasePage.aspx?PRID=1711342 [Last accessed on 2021 Dec 30] 5 Logunov DY Dolzhikova IV Shcheblyakov DV Tukhvatulin AI Zubkova OV Dzharullaeva AS Safety and efficacy of an rAd26 and rAd5 vector-based heterologous prime-boost COVID-19 vaccine: An interim analysis of a randomised controlled phase 3 trial in Russia Lancet 2021 397 671 81 33545094 6 Dr. Reddy's Laboratories Ltd. Sputnik V vaccine leaflet Telengana, India 2021 [Last accessed on 2021 Dec 30] 7 Pagotto V Ferloni A Soriano MM Díaz M Braguinsky Golde N González MI Active monitoring of early safety of Sputnik V vaccine in Buenos Aires, Argentina Medicina (B Aires) 2021 81 408 14 34137701 8 Babamahmoodi F Saeedi M Alizadeh-Navaei R Hedayatizadeh-Omran A Mousavi SA Ovaise G Side effects and Immunogenicity following administration of the Sputnik V COVID-19 vaccine in health care workers in Iran Sci Rep 2021 11 21464 72 34728696 9 Montalti M Soldà G Di Valerio Z Salussolia A Lenzi J Forcellini M ROCCA observational study: Early results on safety of Sputnik V vaccine (Gam-COVID-Vac) in the Republic of San Marino using active surveillance EClinicalMedicine 2021 38 101027 doi: 10.1016/j.eclinm. 2021.101027 34505029 10 Zare H Rezapour H Mahmoodzadeh S Fereidouni M Prevalence of COVID-19 vaccines (Sputnik V, AZD-1222, and Covaxin) side effects among healthcare workers in Birjand city, Iran Int Immunopharmacol 2021 101 108351 doi: 10.1016/j.intimp. 2021.108351 34801416 11 Jarynowski A Semenov A Kamiński M Belik V Mild adverse events of Sputnik V vaccine in Russia: Social mediacontent analysis of telegram via deep learning J Med internet Res 2021 23 e30529 43 34662291 12 Basavaraja CK Sebastian J Ravi MD John SB Adverse events following COVID-19 vaccination: First 90 days of experience from a tertiary care teaching hospital in South India Ther Adv Vaccines Immunother 2021 9 25151355211055833 doi:10.1177/25151355211055833 34841193 13 Kamal D Thakur V Nath N Malhotra T Gupta A Batlish R Adverse events following ChAdOx1 nCoV-19 Vaccine (COVISHIELD) amongst health care workers: A prospective observational study Med J Armed Forces India 2021 77 S283 8 34334895 14 Bhardwaj S COVID vaccine: Sputnik Light to be given as precaution dose for those vaccinated with Sputnik V ANI 2022 Available from: https://www.aninews.in/news/national/general-news/covid-vaccine-sputnik-light-to-be-given-as-precaution-dose-for-those-vaccinated-with-sputnik-v20220430211453/ [Last accessed on 2022 May 13]

PMC010xxxxxx/PMC10353674.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-422 10.4103/ijcm.ijcm_80_23 Original Article A Validation Study on the Physiological Parameters Recorded by COVIDBEEP - An Indigenous Remote Health Monitoring System Designed for COVID-19 Care in India Taranikanti Madhuri Mudunuru Aswin Kumar 1 Fatima Farheen 1 Maddur Srinivas 2 Bandaru Rajiv Kumar 3 Taranikanti Sai Shriya 4 Guntuka Rohith Kumar 1 Yerra Aruna Kumari 5 Dhanunjaya G 6 Raju A. P. 6 Department of Physiology, All India Institute of Medical Sciences, Bibinagar, Telangana, India 1 Department of Physiology, ESIC Medical College and Hospital, Hyderabad, Telangana, India 2 Director, All India Institute of Medical Sciences, New Delhi, India 3 Department of General Medicine, ESIC Medical College and Hospital, Hyderabad, Telangana, India 4 MBBS Student, Agartala Government Medical College, Agartala, Tripura, India 5 Department of Obstetrics and Gynecology, ESIC Medical College, Hyderabad, Telangana, India 6 Electronics Corporation of India Limited, Hyderabad, Telangana, India Address for correspondence: Dr. Madhuri Taranikanti, Additional Professor, Department of Physiology, All India Institute of Medical Sciences, Bibinagar, Telangana State, India. E-mail: madhuri.tarani@gmail.com May-Jun 2023 30 5 2023 48 3 422429 11 2 2023 27 2 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Background: COVID-19 pandemic has affected mankind globally. After the three waves since March 2020, the threat continues instilling fear in the minds. Vital parameter monitoring through remote health monitoring system (RHMS) becomes critical for effective disease management and manpower safety and confidence. In a low resource setting like India, a comprehensive, wearable, and remotely operable device that is economical was required to be introduced for COVID-19 care. Present study validated the remote health monitoring device named COVIDBEEP with gold standard equipment. Materials and Methods: Six parameters, namely heart rate, SpO2, respiratory rate, temperature, blood pressure, and ECG were acquired in the supine position using the devices. Result: Analysis was performed using Graph Pad Prism. Intraclass correlation coefficients were used to measure concurrent validity. Bland–Altman graphs were plotted to know the agreement for each vital parameter. Confidence limits were set at 95%. All the parameters recorded from the devices showed a significant correlation with an “r” value between 0.5 and 0.9 with P value between 0.001 and 0.0002. Bland–Altman plots showed a minimum bias of 0.033 for heart rate and maximum of 3.5 for systolic blood pressure and respiratory rate. Conclusion: The association between the parameters recorded by the devices strengthened as the time of collection of data increased. Agreement between the two methods in 95% confidence interval was also proven to be significant for the parameters. Therefore, the indigenously developed COVIDBEEP has shown good validity in comparison to standard monitoring device. COVIDBEEP COVID-care device remote health monitoring system IOT validation ==== Body pmcBACKGROUND COVID-19 pandemic has drastically affected mankind globally and after the few waves since March 2020, the threat from the disease continues with fear in the community. Many were infected by the virus, admitted to hospitals in different stages of the disease with various outcomes. Monitoring patients during the pandemic was a highly challenging task with constant exposure leading to high risk of transmission of virus to frontline healthcare workers. To minimize the contact, a remote health monitoring system (RHMS) became critical in a hospital setup which would not only minimize physical contact but also reduced the need for manpower and enable implementation of safety precautions among staff.[1] Problems associated with such pandemics were further complicated in a limited resource country like India. Objective and automated approaches have been proposed for syndromic surveillance in the past using statistical algorithms that compared data to non-outbreak trends.[2] COVID-19 pandemic has shown a high fatality rate due to respiratory failure.[3,4] Continuous monitoring of physiological parameters had become critical for SARS-CoV-2-infected individuals due to the unpredictable outcome of the disease. Parameters like body temperature, respiratory rate, and oxygen saturation had to be monitored continuously. Clinical manifestations in COVID-19 were found to be heterogenous and many patients infected had co-morbidities like diabetes mellitus, cardiovascular disorders, and hypertension and were immune-compromised leading to complications. Monitoring additional parameters indicative of hemodynamic stability like heart rate, non-invasive blood pressure monitoring (NIBP), and ECG became essential.[5] In addition, various drugs like Hydroxychloroquine and Azithromycin initially used for prophylaxis and management of COVID-19 and were known to have effects on cardiovascular balance in the body.[6-8] Therefore, the objective was to develop a complete comprehensive, accurate, wearable, and remotely operable device that would be economical and safe in a low resource setting as India. MATERIALS AND METHODS Conception of the idea to design a Remote Health Monitoring System (RHMS) To address the above issue, researchers from the department of Physiology at ESIC Medical College, Hyderabad (under the Ministry of Labour and Employment) and team of engineers from Electronics Corporation of India Limited (ECIL—A Government of India enterprise) came together to design a novel comprehensive remote monitoring system. It was named COVIDBEEP; Continuous Oxygenation and Vital Information Detection Biomed ECIL ESIC Pod which is Internet of Things (IOT)-based versatile indigenous instrument with a mobile app and web browser by which patients could be monitored remotely. Testing and validation was done in the Department of Physiology of the medical college. The idea was to have its utility beyond the pandemic even for routine clinical management of patients living in remotely accessible areas, in radiotherapy units, high risk maternal, and child groups, along with any other group(s) that required constant surveillance. Technical Description of the device COVIDBEEP is shown in Figure 1 with hardware description: A. Wearable Device: It consists of a small box of 85 X 41 X 75 mm (dimensions) mounted on a strap. The strap serves dual purpose used for securing around the wrist and also as the inflatable blood pressure cuff. The device was equipped to measure six parameters, heart rate, SpO2, respiratory rate, body temperature, blood pressure, and ECG by non-invasive method and transmits information over long range BLE to IOT-based gateway and is designed around a Digital Signal Processor. It also has a built-in GPS and GSM module that was enabled to transmit the parameters directly to cloud/server in the absence of an IOT gateway. Automatic switching facility was created between IOT and GPS and GSM module when one moves toward or away from the range of IOT, respectively. The block diagram of the device is shown in Figure 2. B. IOT Gateway: It was enabled to receive data from a maximum of 20 wearable devices (maximum 20 devices per 1 gateway) to transmit the concatenated data to cloud/private server using GSM. Data were transferred to local LAN or WAN. The Gateway concatenates the data received from different wearable devices (max 20) and pushes to cloud or private server via internet. The internet connectivity to gateway can be accomplished by LAN, WI-FI, or CELLULAR networks. It has provision to use existing GSM network and works with any Indian cellular service provider. These gateways were placed at strategic locations in the hospital wards for good coverage. The data were gathered by Central monitoring server using HTTP web services like REST/JSON/MQTT, etc., and equipped with huge memory for back up during loss of connectivity. The Embedded Linux operating system running on gateway would allow collecting data and push the same to the cloud simultaneously. In case of usage by home-quarantine patients or in remote locations where internet was not available, the wearable device itself could communicate to cloud or server with inbuilt GSM/GPRS network without talking to gateway. C. Database Management System: A Cloud/Server with database engine was used to maintain and manage data for facilitating dashboard and various kinds of reports. D. Mobile App: Patients were allocated to doctors and health care workers using this app to monitor the physiological parameters of their COVID-19 patients. Automatic alert signals are generated when parameters cross the preset safety limits. In addition, an SOS button when used by the patient immediately alerts the concerned health professional and ambulance services [Figure 3] E. Dashboard: A web-based application for dashboard provides a summary of all the patients of a particular quarantine room/hospital/districts/states/country. This could be viewed on Mobile/Tablet/Desktop/Large Display [Figure 3]. F. Bhuvan Maps: The real time location tracking of COVID-19 patients and Red, Green, and Orange Zones of COVID-19 pandemic are shown on maps using Bhuvan on-line portal mapping service with active support from NRSC, Hyderabad, ISRO [Figure 3]. The following are the salient features of the application: Hospital/Home/Cluster-based health monitoring. A real-time Database Management System for pandemics like COVID-19 Location tracking through indigenous Bhuvan on-line mapping Portal of ISRO [Department of Space, Government of India]. Display of geographical distribution of suspected or infected cases like those of COVID-19 pandemic. Real time monitoring of patients. Real time alerts when the monitored patient parameters exceed the pre-set threshold level. Real time display of six physiological parameters including heart rate, SpO2, respiratory rate, body temperature, blood pressure, and ECG. Figure 1 Remote wearable health monitoring system with IOT [Provided by Electronics Corporation of India Limited (ECIL)] Figure 2 Block diagram of wearable device Figure 3 Mobile app display: Parameters, statistics, Bhuvan maps Communication The wearable device communicates the six physiological parameters of the patients to remote distances in the following two configurations: a. The device sends the data to IoT gateway over a long range BLE. The gateway has the communication features to connect to hospital LAN/WAN or dedicated PCs for monitoring, database, etc. The gate way also has GSM/GPRS feature to communicate with cloud/private servers over mobile network communications. b. The device directly sends the data to cloud/server using GSM/GPRS over mobile network communications. No gateway is required in this mode. However, mobile network should be available. Testing and validation methods: The prototype of the device (DEVICE-A) was made and compared with the standard device (DEVICE-B). After obtaining clearance from Institutional Ethics Committee of ESIC Medical College and Hospital, Hyderabad (ESICMC/SNR/IEC-F0179/04/2020), this validation study was performed in the department of physiology. 30 healthy volunteers belonging to both sexes between 20 and 60 years of age were included in the study after taking informed consent from the subjects or their guardians. COVIDBEEP display unit (DEVICE-A) mounted on the BP cuff was attached snugly to the left forearm 5 cm above the wrist crease. The finger sensor also incorporated with the IR sensor to record temperature was clipped to the index finger of the left hand and its wire connected to the display unit. ECG electrodes were attached to the chest/trunk and the lead wire connected to the display unit. Physiological parameters, i.e., heart rate, spO2, respiratory rate, digital temperature, blood pressure, and ECG were acquired in the supine position. The standard ICU monitor (DEVICE-B) of model number and make AQUA-15 manufactured by Shenzhen Unicare Electronic Co., Ltd. routinely used in the Intensive Care Unit (ICU) of the hospital was used for validation purpose. SpO2 sensor, temperature sensor, NIBP, and ECG lead wires of the ICU monitor were simultaneously connected to the right side of the subject. Once connected to both the systems, the subject was made to rest for 5 minutes. The parameters were then periodically noted down at the end of 1, 3, and 5 minutes after the resting period. The data were instantly entered in an MS-Excel sheet. Measurement philosophy/parameters a. Oxygen Saturation Level (SpO2):The finger sensor having Red and Infrared light sources measures the Oxygen saturation (SpO2). By measuring the intensity of Red and Infrared light passing through capillary blood of finger, the SpO2 is measured. The data collected from these reflected lights are digitized and processed in Digital Signal Processor of the device. b. Blood Pressure: Oscillometric method is used for measurement of blood pressure. In this method, a cuff is wrapped around the patient’s lower part of forearm, just above the wrist. The Oscillometric method is based on the principle of occluding the artery of the patient by inflating the tourniquet above systolic pressure; this compresses the artery against the underlying bone and shuts off the flow of blood in the vessel. The pressure is slowly released by deflating the tourniquet in gauged intervals till the pressure falls slightly below the systole. As soon as pressure falls to systolic value, blood breaks through the occlusion and flows turbulently through the artery. This causes the artery walls to vibrate or oscillate and they create a counter pressure against the tourniquet. A piezoelectric pressure transducer continuously monitoring the tourniquet pressure detects these vibrations. The onset of pressure oscillations correlates with the systolic pressure. The diastolic pressure is marked by the sudden change in slope of decreasing amplitude. The pressure is continuously reduced till the tourniquet is fully deflated, the blood flow returns to normal, and oscillations cease. c. Heart Rate: The heart rate is taken from ECG. In the case of regular cardiac rhythm, the heart rate can be determined by calculating the interval between two successive QRS complexes. d. Body Temperature sensor: The body temperature sensor, incorporated into the finger sensor, is based on semiconductor sensing with clinically approved standards: ASTM E1112 and ISO 80601-2-56. e. Respiratory Rate: Impedance plethysmography technique is used to measure respiration with the help of two current injecting electrodes placed across the chest. The voltage across the electrodes and current is measured to calculate changes in trans-thoracic impedance which represents the respiratory rate. In case of non-connectivity of the chest electrodes, the respiratory rate is delivered from the photo-plethysmography signal from the finger sensor. f. Electrocardiogram (ECG): Bio-potential sensing method with four surface mounted electrodes is used for ECG. The electrodes are placed according to the standard protocol. Einthoven triangle concept is used for selection of location for placement of electrodes. Device specifications (DEVICE-A): Parameter Range Accuracy Body temperature (Fahrenheit) 95-110 <0.5% Heart rate (Beats/min) 30-200 <2% Oxygen saturation (SpO2%) 30-100% <2% Respiratory rate (Breaths/min) Using Infra/ECG probe 5-40 <4% Blood pressure Systolic and diastolic pressures <2% ECG 3 leads waveform <2% Statistical Analysis: The data entered into MS-Excel sheet were calculated for Mean and Standard Deviation of each parameter recorded by both the devices. Further analysis was performed using Graph Pad Prism Version 9.2.0 (332) software. To know the concurrent validity, Intraclass correlation coefficients were used. To know the agreement for each vital parameter, Bland–Altman graphs were plotted. Confidence limits were set at 95%. Result: Table 1 shows the distribution of the Mean and Standard Deviation values for the physiological parameters recorded from both the devices. It can be observed that the difference in mean values of respiratory rate recorded by both the devices is narrowed down by the end of 5 minutes. Intra-class correlation coefficient between the 2 devices for all parameters was analyzed Pearson’s correlation coefficient revealed a good association between the parameters recorded by both the devices which only strengthened as the time of collection of data points increased. Agreement between the two methods in 95% confidence interval was also proven to be significant for the parameters. As Fever is one of the important symptoms for COVID-19, recording of temperature was a mandatory requirement. The digital temperatures recorded from both the devices showed a significant correlation with an “r” value of 0.64 and “P” value of 0.0002 [Table 2]. The strength of relationship between the parameters recorded by both instruments was established by Bland–Altman comparison showing no consistent bias [Table 3, Figures 4 and 5]. When data on heart rate and blood pressure were collected from both devices and subjected to statistical analysis, a significant correlation was found with “r” values of 0.9792, 0.9814 and 0.9847 for heart rate, SBP and DBP, respectively. “P” values for all these parameters were found to be <0.0001. Respiratory system was one of the earliest systems to have been involved in the patho-physiology of this disease. Alteration of respiratory rate was prognostic of the course of disease. Monitoring respiratory rate was crucial. COVIDBEEP was designed to extract respiratory rate values from both finger sensor as well as from the ECG electrodes. Analysis of the data on the respiratory rate from both the devices showed a significant correlation between the two with an “r” value of 0.9813 and a “P” value of <0.0001. The data were also subjected to Bland–Altman analysis, a tool to know the degree of agreement between 2 methods within the given confidence interval. The mean difference was found to be 0.14, 0.03, 3.5, 0.2, 3.5, and 1.8 for temperature, heart rate, respiratory rate, SpO2, SBP, and DBP, respectively, with 95% confidence interval. Table 1 Distribution of mean and standard deviation values Time Parameter tested Device Mean SD 1 Minute HR Device-A 74.33 8.46 Device-B 74.10 7.75 SpO2 Device-A 97.10 0.91 Device-B 97.43 1.15 RR Device-A 13.60 3.21 Device-B 18.57 3.21 Temperature Device-A 96.68 1.06 Device-B 98.29 0.34 3 Minutes HR Device-A 74.53 7.20 Device-B 74.73 7.09 SpO2 Device-A 97.03 0.75 Device-B 97.27 0.89 RR Device-A 14.43 3.27 Device-B 18.23 3.29 Temperature Device-A 98.05 0.56 Device-B 98.31 0.43 5 Minutes HR Device-A 73.97 8.00 Device-B 74.00 7.15 SpO2 Device-A 96.77 1.09 Device-B 97.00 1.06 RR Device-A 14.17 3.45 Device-B 17.67 3.28 Temperature Device-A 98.10 0.52 Device-B 98.24 0.43 SBP Device-A 112.97 7.52 Device-B 116.47 7.07 DBP Device-A 71.33 6.65 Device-B 73.20 6.75 Table 2 Intra-class correlation coefficient between the 2 devices for all parameters Parameter Statistic 1 Min 3 Min 5 Min Heart rate R 0.9923 0.9816 0.9792 P <0.0001 <0.0001 <0.0001 SpO2 R 0.5035 0.3346 0.6344 P 0.0046 0.0707 (ns) 0.0002 Respiratory rate R 0.8432 0.9338 0.9813 P <0.0001 <0.0001 <0.0001 Temperature R 0.3764 0.5923 0.6370 P 0.0404 0.0006 0.0002 SBP R - - 0.9814 P - - <0.0001 DBP R - - 0.9847 P - - <0.0001 Table 3 Bland–Altman comparison between the 2 devices Parameter Bland–Altman values Heart rate Bias 0.0333 SD 1.79 Confidence Interval (95%) -3.476, +3.543 SpO2 Bias 0.2333 SD 0.9353 Confidence Interval (95%) -1.6, +2.066 Respiratory rate Bias 3.500 SD 0.6823 Confidence Interval (95%) +2.163, +4.837 Temperature Bias 0.1433 SD 0.4166 Confidence Interval (95%) -0.6732, +0.9599 SBP Bias 3.5 SD 1.5 Confidence Interval (95%) +0.5544, +6.446 DBP Bias 1.867 SD 1.196 Confidence Interval (95%) -0.477, +4.210 Figure 4 Bland–Altman comparison plots for the physiological parameters—(a) temperature (b) heart rate (c) respiratory rate (d) SpO2 recorded using both devices. X-axis in each plot corresponds to the mean values of a particular parameter. Y-axis corresponds to difference in the mean values (bias) to the mean values of a particular parameter Figure 5 Bland–Altman comparison plots for physiological parameters DISCUSSION The need for comprehensive remote health monitoring of patients has been immensely felt during the ongoing COVID-19 pandemic due to the critical requirement of continuous surveillance of patients as well as protection of the healthcare givers. The pathophysiology of the disease and the medication used in the management of COVID-19 patients was expected to affect the heart rate, blood pressure and ECG, and any co-morbidities resulting out of the above needed to be monitored carefully. To make the monitoring system comprehensive, the blood pressure and ECG recording were incorporated into it. COVIDBEEP was designed to record forearm blood pressure and 6 lead ECG. Though the display unit shows single lead ECG, the physician can monitor the six lead ECG through the server. The configuration of the device is shown in Table 4. Therefore, the indigenously developed COVIDBEEP has shown good validity in comparison with standard monitoring device. With judicious usage of COVIDBEEP, the following benefits can be seen. Table 4 Configuration of the device Parameters Specifications Monitoring Interval Configurable between range of 1 sec to 1hour Alerts to gateway Beyond threshold settings Wireless Communication Long Range BLE (~300-400 m) Built in GPS receiver Tracking through Bhuvan map/Google Built in Display Graphics Display GSM/GPRS interface Connect to cloud/server patient tracking Battery Life 12 Hours Operating temperature 0 – 50 deg centigrade Unique ID 10 Alphanumeric string Communication Secure Communication through https:// I. Beneficiaries/Target population: a. COVID patients both symptomatic and asymptomatic. b. Patients in isolation who cannot visit hospital frequently. c. Vulnerable population: Heart Disease, Metabolic Disorders (Diabetes mellitus), Cancers, Immune compromised patients, etc. Utility in Resource Crunch Situations: The country needed doctors, nurses, and technicians, etc., more than ever during the COVID-19 pandemic. To protect the healthcare workers was vital to serve the community. Not to forget that many healthcare workers/professionals themselves could be associated co-morbidities and immune compromised conditions. Long leave due to ill health and fear among the staff lead to manpower crunch. In such a situation, the remote health monitoring system would help the hospitals in overcoming these deficiencies. The device was expected to strongly supplement and support the existing model of healthcare facilities. II. Advantages of the COVIDBEEP: The device has mltiple advantages for COVID-19 and non-COVID-19 care. A. In COVID-19 care: 1. The country needed a strategic plan to attend to the disease at multiple levels of health care like in Primary Health Care, Secondary Health Care, and Tertiary Health Care. For effective utilization of oxygen beds, monitoring played a vital role to decide the further course of management. The following points will elaborate on this issue. a. At Primary Health Care level (L1): When a case is identified with potential risk of COVID-19 disease with symptoms, a preliminary monitoring was essential to know the physiological stability of the patient. However, for the fear of contracting the disease patients were kept at distance from the time symptoms appeared. COVIDBEEP helped in monitoring patients throughout the period from sample collection, RT-PCR testing to consultation and treatment. A decision could be taken based on parameters whether to refer patients to secondary level care (L2 Level). This constituted an effective triaging based on the saturation and other vital information. Triaging of the cases: All COVID-19 suspected and confirmed cases could be triaged using the baseline parameters and appropriately allotted to groups such as 1) Asymptomatic 2) Symptomatic stable; 3) Symptomatic requiring Oxygen bed; 4) Symptomatic requiring HDU bed; and 5) Symptomatic requiring ICU bed b. The feasibility of the device being carried by the patient to home and still be monitored by the doctor/nurse made COVIDBEEP a novel robust gadget. c. At Secondary Health Care level (L2): Suitable Oxygen support along with continuous monitoring of parameters. d. At tertiary Health Care level (L3): All tertiary health care centers may be equipped with a COVID COMMAND ROOM (CCR) to monitor the data of the patients at primary and secondary levels under the tertiary Centre. The CCR of a tertiary hospital can supply the devices to its secondary and primary centers with the device IDs and their locations. The data of the patients who were connected to the devices will be continuously monitored by dedicated personnel in the CCR. This triage is required for suitable allocation of hospital beds, which is especially essential owing to the current condition of scarcity of resources for patient care. In case of admission, the CCR alerts the Oxygenation and vital support units and allocates bed to the patient. If the CCR is already aware that all the beds are occupied in their Centre, the command room can allot the patient to another hospital. e. The technological advantages in the device will help the treating physician/nurse in proper management of a COVID-19 patient: i. Blood Pressure is one of the vital parameters that requires to be monitored in these patients may be due to pre-existing Hypertension or secondary to drug side effects which may produce cardiovascular or hemodynamic instability. COVIDBEEP was incorporated with a forearm cuff which easily worn by the patients themselves and the BP icon touched to start NIBP recording. ii. Heart rate was extracted from PPG signal from finger pulse as well as from the RR interval of the ECG. This ensures, recording of vital data even during connection issues. iii. Respiratory rate was also extracted from both PPG signal from finger pulse and via impedance plethysmography from chest ECG electrodes. iv. This robustness in the data acquisition helped doctor/nurse take timely decisions such as continuing isolation/quarantine or shifting the patient to ward/ICU/HDU or continue to monitor even after discharge. B. In non-COVID-19 care: COVIDBEEP devices are designed with a future utility to function beyond the pandemic in certain conditions: High risk pregnant mothers—to monitor their BP, Oxygenation, heart rate, etc., to know the cardiovascular stability. As they cannot be frequently mobilized, such monitoring at home in a user-friendly manner. General health monitoring of seriously ill patients in a geographically isolated or secluded zones like in flooded areas, earthquake affected zones, dense forests, soldiers, etc., and guide local staff in management of patients. For patients with leukemia (blood cancers), cancer therapy, etc., to monitor the changes which arise due to the drug dosage or disease progression per se. In people who are immuno-compromised where patient is kept in isolation. For patients in radiotherapy units or chemotherapy units who need isolation. For geriatric age groups patients to monitor general well-being, remotely. In children or adults with sleep apnea or other sleep disorders. Unique features of the device make it robust and user-friendly: a. An SOS alarm button for use by patient or attender to call for immediate help. SOS buttons address such grave situations like washroom syncope, hypoxia, sleep apnea, etc. b. GPS tracking of the device and hence the patient who received the device will enable the Hospital personnel to reach out to the patient through dedicated ambulances and Oxygen cylinders. c. The wearable device on the wrist of the patient is robust enough to bear the temperature or physical pressures of the immediate environment. The interference from adjacent electronic devices is also very minimal not affecting the signal acquisition by the device. d. Blood pressure recording is also a unique feature of the device which forms an important factor in the management of a COVID-19 patient. Whether due to side effect of drug or as an associated risk factor, BP needs prompt monitoring to identify any deviations. Unique feature was that the BP cuff could be operated/inflated remotely by the Physician when needed. e. Device is user friendly. f. Relevance of such RHMS device as a mitigation plan to address manpower issues: g. To address the manpower issue, usage of such devices especially in a government run hospital with huge numbers of cases, would help mitigate the deficiency of health care personnel and promise continued monitoring of patients for health outcomes. CONCLUSION COVIDBEEP is an indigenous, low-cost, effective, user-friendly, and comprehensive remote health monitoring device with a robust technology that has tremendous application especially in the present pandemic and beyond. As it is the first of its kind, validation was done against a standard recording device for the physiological parameters like temperature, heart rate, respiratory rate, SpO2, and blood pressure. A significant correlation between the two devices has proved that the device is a valid tool. In future, it can be utilized in ICUs, HDUs, and Isolation wards for other diseases like leukemia, bone marrow transplantation, HIV patient wards, immune disorders, and all of whom require to be strictly monitored in an aseptic environment. Another area where its use will immensely benefit is in maternal and fetal health monitoring during pregnancy. Geriatric patients can be closely observed using this device. It may find an important place in pulmonology for monitoring patients with sleep apnea. COVIDBEEP has tremendous applications especially in the present pandemic and even beyond. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. Acknowledgements We thank Electronics Corporation of India Limited (ECIL), Government of India, for their technical support in designing the device. ==== Refs REFERENCES 1 Mirza MB Hamid G Smart health monitoring systems: An overview of design and modeling J Med Syst 2013 37 1 14 2 Chretien J-P Burkom HS Sedyaningsih ER Larasati RP Lescano AG Mundaca CC Syndromic surveillance: Adapting innovations to developing settings PLoS Med 2008 5 e72 18366250 3 Wang D Hu B Hu C Zhu F Liu X Zhang J Clinical characteristics of 138 hospitalized patients with 2019 novel coronavirus-infected pneumonia in Wuhan, China JAMA 2020 323 1061 9 32031570 4 Tian S Hu W Niu L Liu H Xu H Xiao S-Y Pulmonary pathology of early-phase 2019 novel coronavirus (COVID-19) pneumonia in two patients with lung cancer J Thorac Oncol 2020 15 700 4 32114094 5 Sardu C D’Onofrio N Balestrieri ML Barbieri M Rizzo MR Messina V Outcomes in hyperglycemic patients affected by COVID-19: Can we do more on glycemic control? Diabetes Care 2020 43 1408 15 32430456 6 Giustino G Croft LB Stefanini GG Bragato R Silbiger JJ Vicenzi M Characterization of myocardial injury in patients with COVID-19 J Am Coll Cardiol 2020 76 2043 55 33121710 7 Sandoval Y Januzzi JL Jaffe AS Cardiac troponin for assessment of myocardial injury in COVID-19: JACC review topic of the week J Am Coll Cardiol 2020 76 1244 58 32652195 8 The Novel Coronavirus Pneumonia Emergency Response Epidemiology Team The epidemiological characteristics of an outbreak of 2019 novel coronavirus diseases (COVID-19) —China, 2020 China CDC Wkly 2020 2 113 22 34594836

PMC010xxxxxx/PMC10353675.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-505 10.4103/ijcm.ijcm_1006_22 Letter to Editor Zinc/ORS Co-packaging: A Step Towards Bridging the Gap in Preventable Childhood Diarrhoeal Deaths in India Behera Priyamadhaba Pradhan Somen Kumar Behera Surama Manjari 1 Rao E. Venkata 2 Department of Community Medicine and Family Medicine, All India Institute of Medical Sciences, Bhubaneswar, Odisha, India 1 Department of Community Medicine, IMS and SUM Hospital, Bhubaneswar Odisha, India 2 Department of Community Medicine, IMS and SUM Hospital, Bhubaneswar Odisha, India Address for correspondence: Dr. Somen Kumar Pradhan, Department of Community Medicine and Family Medicine, 3rd Floor, Academic Building, AIIMS, Bhubaneswar - 751 019, Odisha, India. E-mail: somenpradhan@yahoo.com May-Jun 2023 30 5 2023 48 3 505506 25 12 2022 18 4 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. ==== Body pmcSir, Diarrhoeal disease is one of the leading causes of mortality and morbidity among children aged five years and below in the world. It accounted for approximately 9% of all deaths among children aged under five worldwide in 2019.[1] The majority of the burden of morbidity as well as mortality associated with diarrhea has been borne by developing countries worldwide.[2] India is the country with the second highest burden of deaths from diarrhea among those under-five where 7% of all under-five deaths are caused by diarrhea.[3] In 2004, the World Health Organization (WHO) issued a global recommendation to formalize ORS + zinc as the gold standard for the treatment of diarrheal disease. According to this, ORS along with Zinc for 10-14 days can significantly reduce the incidence of diarrhea among under-five children.[4] In 2013, WHO and UNICEF devised the “Global Action Plan for Prevention and Control of Pneumonia and Diarrhoea” to end diarrheal deaths among children by 2025. The “Protect-Prevent-Treat” strategy focused on ORS and Zinc as an essential components in diarrhea treatment.[5] Oral Zinc has been shown to reduce both the duration and frequency of diarrhea among under-5 children.[6] There has been enough evidence on the cost as well as clinical effectiveness of Zinc in controlling acute and bloody diarrhea.[7] In a Lower Middle-Income country (LMIC) like India, where the prevalence of undernutrition is high, the role of Zinc in the prevention of diarrheal death is even more significant.[8] However, the recent National Family Health Survey-5 (NFHS-5) from India reported that only 30.5% of the under-5 children suffering from diarrhea received Zinc as compared to 60.6% who received ORS.[9] Under these circumstances, India is yet to incorporate zinc-ORS co-packaging into its national diarrhea control strategy. Co-packaging can help in scaling up the Zinc uptake along with ORS among under-five suffering from acute diarrhea and reduce the gap between compliance to ORS and Zinc during episodes of diarrhea.[10] This strategy has also been shown to reduce the irrational use of antibiotics during acute diarrhea management helping in tackling the threat of anti-microbial resistance.[11] Co-packaging and rebranding it in the form of a “diarrhoea kit” can help in improving the medical prescription practice among healthcare providers for diarrhea patients.[12] The zinc/ORS co-pack can further facilitate Zinc utilization by frontline health workers during community-based management of diarrhea.[13] This would also help correct the inventory management issues related to Zinc at the primary care level through combined stock management and dispensation along with ORS.[14] Panel 1 Priorities for zinc/ORS co-packaging implementation Health system-based:  1. Upgradation of national drug policy  2. Supply chain management  3. Public-private partnership  4. Community engagement  5. Service provider engagement Community-based:  1. Information, education, and communication  2. Acceptability  3. Affordability  4. Compliance There are certain issues that need to be addressed based on previous experiences before introducing zinc/ORS co-pack into the healthcare delivery system. Strengthening the functional supply chain management should be one of the utmost priorities for the delivery of zinc/ORS co-pack at the grassroots level. The introduction of zinc/ORS co-pack into the health system and market will need upgrading national drug policy under the appropriate regulatory framework. A national-level action plan for the integration of the co-packaged product into national guidelines for diarrhea management and health programs would be the first step in this direction.[15] Utilization of zinc/ORS co-pack would require the sincere engagement of both the providers and community through IEC and capacity building respectively. Partnerships with the private sector and professional bodies would assist in scaling up this strategy effectively. Designating zinc/ORS co-pack as an Over counter (OTC) product would be essential in the implementation of the strategy as this would improve its accessibility for beneficiaries. On the other hand, incorporating flavored Zinc tablets and making the co-packaged product more affordable would improve the acceptability as well as compliance at the consumer level. Long-term interventions like improving WASH practices and Rotavirus vaccination have been instrumental in reducing the burden of preventable diarrhoeal deaths. However, the existing opportunity for further reduction in childhood diarrhea mortality through greater coverage of ORS/zinc co-packaging has received several calls for action worldwide. World Health Organization has already added zinc/ORS co-pack to the Essential Medicine List in 2019. It’s time that other LMIC countries including India introduce the zinc/ORS co-pack under their national health programs as this would go a long way in reducing the burden of diarrhoeal deaths worldwide. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. ==== Refs REFERENCES 1 United Nations International Children's Emergency Fund Diarrhoea 2022 Available from: https://data.unicef.org/topic/child -health/diarrhoeal-disease/ 2 Bulled N Singer M Dillingham R The syndemics of childhood diarrhoea: A biosocial perspective on efforts to combat global inequities in diarrhoea-related morbidity and mortality Glob Public Health 2014 9 841 53 25005132 3 Ugboko HU Nwinyi OC Oranusi SU Oyewale JO Childhood diarrhoeal diseases in developing countries Heliyon 2020 6 e03690 32322707 4 WHO-UNICEF. WHO/UNICEF Joint statement: Clinical management of acute diarrhoea 2004 5 UNICEF. The Integrated Global Action Plan for Pneumonia and Diarrhoea (GAPPD) 2013 6 Bhutta ZA Bird SM Black RE Brown KH Gardner JM Hidayat A Therapeutic effects of oral zinc in acute and persistent diarrhea in children in developing countries: Pooled analysis of randomized controlled trials Am J Clin Nutr 2000 72 1516 22 11101480 7 World Health Organisations Implementing the new recommendations on the clinical management of diarrhoea: Guidelines for policy makers and programme managers Geneva World Health Organization 2006 Available from: http://scholar.google.com/scholar?hl=en&btnG=Search&q=intitle: Implementing+the+New+Recommendations+on+the+Clinical+Management+of+Diarrhoea.+Guidelines+for+Policy+Makers+and+Programme+Managers#0 8 Lazzerini M Wanzira H Oral zinc for treating diarrhoea in children Cochrane Database Syst Rev 2016 12 CD005436 27996088 9 International Institute for Population Sciences (IIPS) and ICF.2021 National Family Health Survey (NFHS-5), India, 2019-21: IIPS Mumbai (2022) 10 Ezezika O Ragunathan A El-Bakri Y Barrett K Barriers and facilitators to implementation of oral rehydration therapy in low- And middle-income countries: A systematic review PLoS One 2021 16 1 23 11 Baqui AH Black RE Arifeen SEl Yunus M Zaman K Begum N Zinc therapy for diarrhoea increased the use of oral rehydration therapy and reduced the use of antibiotics in Bangladeshi children J Heal Popul Nutr 2004 22 440 2 12 Behera P Bhatia V Sahu DP Sahoo DP Kamble R Panda P Adherence of doctors to standard diarrhoeal management guideline during treatment of under-five diarrhoeal episodes: A study from Eastern India Cureus 2021 13 e13433 33763320 13 Habib MA Soofi S Sadiq K Samejo T Hussain M Mirani M A study to evaluate the acceptability, feasibility and impact of packaged interventions (“Diarrhea Pack”) for prevention and treatment of childhood diarrhea in rural Pakistan BMC Public Health 2013 13 922 24090125 14 Lam F Kirchhoffer D Buluma DM Kabunga L Wamala-Mucheri PN Schroder K A retrospective mixed-methods evaluation of a national ORS and zinc scale-up program in Uganda between 2011 and 2016 J Glob Health 2019 9 010504 31217963 15 Child Health Division, Ministry of Health and Family Welfare G of I. IDCF 2014, Intensified Diarrhoea Control Fortnight, Operational Plan for States, UTs and Districts 2014 Available from: https://nhm.gov.in/images/pdf/IDCF/Important_Document/IDCF_Guidelines_for_States.pdf

PMC010xxxxxx/PMC10353676.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-459 10.4103/ijcm.ijcm_285_22 Original Article Access to Maternal and Child Health Services during the COVID-19 Pandemic: An Explorative Qualitative Study in Odisha, India Mishra Bijaya K. Kanungo Srikanta Panda Subhashree Patel Kripalini Swain Swagatika Dwivedy Subhralaxmi Karna Sonam Bhuyan Dinesh Som Meena 1 Marta Brajesh 1 Bhattacharya Debdutta Kshatri Jaya S. Pati Sanghamitra Palo Subrata K. Regional Medical Research Centre, Bhubaneswar, Odisha, India 1 UNICEF, Bhubaneswar, Odisha, India Address for correspondence: Dr. Subrata K. Palo, ICMR-Regional Medical Research Centre, Chandrasekharpur, Bhubaneswar, Odisha - 751023, India. E-mail: drpalsubrat@gmail.com May-Jun 2023 30 5 2023 48 3 459464 02 4 2022 15 4 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Background: Maternal and child health (MCH) care is one of the essential routine healthcare services, which got affected during the coronavirus disease 2019 (COVID-19) pandemic. Modeled projections had anticipated an 8.3%–38.6% rise in maternal mortality from different countries globally. In view of limited studies pertaining to issues related to accessing MCH services in the event of a pandemic, this study was carried out on pregnant and postnatal mothers in Odisha, India. Methods: An explorative qualitative study through 36 in-depth interviews (IDIs) was conducted among 16 (44.4%) antenatal and 20 (55.5%) postnatal mothers in six of thirty districts of Odisha, India, from February to April 2021. The districts and blocks were randomly selected for better representativeness. The IDIs were conducted using a predesigned and pretested guide among mothers who had undergone delivery or availed of antenatal, postnatal, or child health services from October 2020 to April 2021. The IDIs were conducted till data saturation. The data were analyzed using MAXQDA software. Results: The average age of mothers was 27.6 (+/- 2.2) years. Among the participants, 16 (44.4%) were antenatal and 20 (55.6%) were postnatal mothers; 19 (52.8%) were primipara and 17 (47.2%) were multipara. The majority explained that they received enormous support including door-to-door services from the community health workers (CHWs) even during the difficult times of the pandemic. Reduced transportation facility and fear of contracting the infection were reasons behind the unwillingness to visit health facilities and preference for home delivery. Furthermore, the pandemic had physical, mental, social, and financial impacts among pregnant and postnatal women. Conclusion: The unprecedented COVID-19 pandemic has affected access to MCH services by antenatal and postnatal mothers. Health system preparedness and appropriate strategies including better community engagement and participation could avert such challenges in the future. Antenatal COVID-19 healthcare access MCH care Odisha postnatal ==== Body pmcINTRODUCTION The coronavirus disease 2019 (COVID-19) pandemic created several barriers in provisioning and accessing various healthcare services, leading to substantial morbidity, mortality, and loss of economy across the globe.[1-4] The health system got challenged with priority for managing the pandemic, leading to the diversion of resources including healthcare providers from routine health care toward COVID-19-related activities. Routine healthcare services including maternal and child health (MCH) care got affected in most of the regions globally due to the pandemic.[5] Many modeled projections anticipated an 8.3%–38.6% increase in maternal mortality from different countries during the COVID-19 pandemic.[6] The quality and timely maternal healthcare services were beyond the reach of millions of women in the world even before the surge of COVID-19.[7] So, with the pandemic having a foothold, the access and utilization of routine MCH care got significantly compromised. According to a systematic review, 55.5% of expectant mothers missed their antenatal care (ANC) checkup (more among mothers in the 1st and 3rd trimesters), a drop of 38.3% and 59.6% in Tetanus Toxoid (TT) 1st and 2nd doses, respectively, drop in institutional delivery between 9 and 20%, and a drop of 13 and 22% in postnatal visits during the COVID-19 pandemic.[8] Various underlying factors such as mobility restrictions, fear of contracting COVID-19 infection, unavailability of health staff, fear of getting tested for COVID-19 and getting isolated if positive, and preference for private health facilities have been reported for the lapses.[8-10] Such a situation during the pandemic has not only affected the physical health of the MCH care seekers but also affected their mental health and social well-being. The study found that anxiety and obsessive–compulsive symptoms were increased among pregnant women.[11] The disruptions in the provision of MCH services had impacted women’s health undesirably.[12] Till now, COVID-19-related obstetric research has primarily focused on the clinical dimensions, and there is scanty information available on the challenges encountered by pregnant women and postnatal mothers in seeking MCH care during the pandemic. This study intended to explore the challenges faced by pregnant women and postnatal mothers in seeking MCH care during the pandemic. METHODS Study design, setting, and participants This study adopted an explorative qualitative study design, which was conducted in the state of Odisha from February to April 2021. The districts, blocks, and participants for the study were selected and enrolled to achieve maximum sampling variance. From each of the three revenue divisions of the state (northern, central, and southern), two districts were randomly selected, and from each district, two blocks were randomly selected for better sample representativeness. The selected study districts and blocks are detailed in Table 1 and Figure 1. From each of the study blocks, one community health center (CHC), one primary health center (PHC), and one subcenter (SC) were selected to identify and recruit the study participants. Both antenatal and postnatal mothers were purposively selected to participate in the study. Table 1 Details of the study sites and no. of interviews done Revenue division Name of the study district Name of the study block Data collection method—no. of interviews done Northern Sundargarh Bargaon IDI-4 Balisankara IDI-2 Angul Angul IDI-3 Talcher IDI-3 Central Khordha Khordha Sadar IDI-3 Chilika IDI-3 Balasore Remuna IDI-2 Simulia IDI-4 Southern Ganjam Hinjilicut IDI-3 Purusottampur IDI-3 Rayagada Kalyansingpur IDI-3 Kashipur IDI-3 Figure 1 Distribution of the study districts and the study blocks Data collection and analysis For data collection, the participants were interviewed in Odia (the vernacular language of Odisha), using predesigned and pretested in-depth interview (IDI) guides. A total of 36 IDIs were conducted among the study participants, maintaining all necessary prescribed measures to prevent the transmission of COVID-19. With prior informed consent from the participants, the interviews were conducted using a predesigned and pretested guide and were audio recorded. The confidentiality and anonymity of the participants were maintained. The researchers transcribed the audio recordings in the local language and then translated them into English. MAXQDA software was used for analyzing the transcripts and generating the codes. Three researchers reviewed the code tree and came up with the themes and categories in consensus. Categories were organized, and themes and subthemes were generated. The study findings were summarized and narrated accordingly. Ethical consideration Ethical approval for this study was obtained from the Institutional Ethics Committee of Indian Council of Medical Research (ICMR)—Regional Medical Research Centre, Bhubaneswar (ICMR-RMRCB/IHEC-2020/043/28.11.2020), and the State Research and Ethics Committee, Department of Health and Family Welfare, Government of Odisha (13260/MS, Bhubaneswar/18.06.2020). Before all interviews, informed consent from the participants was obtained after explaining the objectives and substance of the study to them. RESULTS Among 36 total study participants, their average age was 27.6 (+/- 2.2) years. While 16 (44.4%) were antenatal and 20 (55.6%) were postnatal mothers, 19 (52.8%) were primipara and 17 (47.2%) were multipara. The qualitative analysis of the data brought out two main themes: 1. access to MCH care services during the pandemic and 2. impact of the pandemic on the life of MCH care recipients [Figure 2]. Figure 2 Impact on MCH care access during the COVID-19 pandemic and possible strategies to address the underlying challenges Theme 1: Access to MCH care services during the pandemic Support from community health workers Participants revealed that they received enormous support from community health workers (CHWs) during the pandemic. Services such as health assessment, medications, and nutritional supplements were provided at their doorstep by CHWs. In case of any problem or issue with seeking MCH care, the participants shared their problems with CHWs. “Yes, I was given iron and calcium supplements, and some common medicines. They gave me one packet of a mixture of cereals and pulses, 12 eggs and some other things per month. The quality was also very good.” (Postnatal care recipient) Most of the participants mentioned that they had received the take-home ration and items such as rice and dal at their doorstep from the CHWs during the pandemic. “Yes, I have availed all the services given by government, like ration card services.” (Antenatal care recipient) A few participants had experienced challenges in accessing MCH services at the community level. They could receive the nutritional supplements neither at Anganwadi centers nor at their doorstep. Access to MCH services at community and facility levels Some MCH care recipients could not avail routine antenatal checkups and nutritional supplements either at the community level or at the facility level. Participants ascribed the lack of information regarding community-based healthcare services such as village health and nutrition day (VHND) as a cause for non-availing the MCH services. “I didn’t know when and where those sessions were held. No one informed me about them.” (Antenatal care recipient) Some pregnant women preferred not to visit health facilities for their routine antenatal checkups and remained at home in the fear of contracting COVID-19 infection at health facilities “No, I did not go to any hospital because I was afraid of the infection. During the lockdown, I remained at home.” (Antenatal care recipient) Restricted mobility and transportation issues According to participants, though they had problems visiting to health facilities because of transportation restrictions, they were supported by the police and system upon showing them the medical records. “We always wore a mask while going out. Hence the police never prevented us from going to the hospital. Oh yes, they would stop us, and check our papers and medical reports though.” (Antenatal care recipient) However, as public transportation was not operational during the lockdown, many had to hire private vehicles causing additional financial burden. “The auto rickshaw demanded extra fare, which we were unable to afford. Hence, we didn’t go for check-up.” (Postnatal care recipient) COVID-19-related barriers Participants had to wait long for their turn at health facilities. “Even though prior appointment was taken, we had to wait in queue for long. It took us 1-2 hours.” (Postnatal care recipient) Preference for home delivery Some pregnant mothers did not visit health facilities for their delivery because of the fear of contracting COVID-19 infection, inaccessibility to health facilities, and unavailability of transportation services. They preferred to have their delivery at home. “Everyone was in fear. Everyone advised me not to go to the hospital. By God’s grace, my delivery was done smoothly at home.” (Postnatal care recipient) Theme 2: Impact of the pandemic on the life of MCH care recipients Increased physical strain Most of the participants expressed an increased household workload as all family members were staying at home all the time. Further, in fear of COVID-19 infection, the cleaning activities increased a lot. “Yes, I had to cook for everyone. There would be a lot of dishes to wash too. My mother-in-law and sister-in-law also helped me though.” (Postnatal care recipient) Fear of the infection MCH care recipients were stressed in fear of contracting COVID-19 infection during pregnancy and childbirth. The fear was mostly regarding the safety of their newborn at hospitals because of the higher risk of infection. “Many people were there in the hospital and anyone of them might be carrying the infection. I was bothered what if my child contracted the infection during our stay in the hospital.” (Postnatal care recipient) Social life The COVID-19 pandemic had a negative impact on the socialization and social relationships of the participants. “Yes, my family members always go out and mingle with others. I had to be very cautious with them. I was a little afraid of mingling with them.” (Antenatal care recipient) Financial burden During the pandemic, families were hit hard by the economic crisis due to the dual burden of the inability to go for work and increased expenses. Some respondents incurred additional out-of-pocket expenditure while seeking MCH services, due to expenses such as hiring of transportation, cost of medicines, and diagnostic tests needed during pregnancy. “At that time, we could not go to work and hence did not earn money. And the pregnancy related expense was an additional burden for me.”(Antenatal care recipient) DISCUSSION The recent COVID-19 pandemic has impacted health care including access to MCH care in many ways. Several Low- and Middle-Income Countries (LMIC) experienced restricted access to Reproductive, Maternal, New-born, Child, and Adolescent Health (RMNCH) services during this pandemic.[6] The antenatal and postnatal mothers were affected physically, mentally, and financially. Some of them could not avail the antenatal care and nutritional services mostly due to a lack of proper prior information such as where to visit and when to visit. Dissemination of prior information through proper planning among the care seekers could have averted this. Most of the participants were in anxiety, stress, and fear about how to undergo delivery, their own health, and that of their babies and family members. Other studies have also similarly found that stress, fear, and anxiety were common among antenatal mothers during situations of a health emergency.[13-15] In the event of a pandemic, such mental health issues among MCH care seekers need to be addressed as an integral part of antenatal and postnatal care. Factors such as limited transportation services, movement restrictions, and fear of contracting the infection played a critical role in the decision of undergoing home delivery and avoiding antenatal and postnatal checkups. According to a research study, 69.3% of pregnant women failed to avail of antenatal service and 24.2% received inadequate antenatal services during the COVID-19 pandemic.[16] Similar shortfalls with respect to antenatal and postnatal care were reported from almost all corners of the world during the pandemic. Preparedness with proper planning for adequate measures by the system during a pandemic would establish trust and confidence among care seekers to visit and avail of the desired services including routine care. Another study in India by Kumar et al.[3] reported that the nutritional and immunization services were reduced during this pandemic. Similar to this, we also observed the nutritional services to be somehow compromised. However, in contrast to their finding, we found that the immunization services were ensured by the health workers and the immunization coverage was maintained throughout the pandemic. This achievement was possible because of the system’s priority for immunization program and alternative strategies adopted at the district and local levels to ensure immunization coverage. The CHWs visited the home of antenatal and postnatal mothers to provide necessary MCH services such as weight measurement, counseling services, provision of medications, and nutritional supplements. Similar to this, other studies have also stressed introducing home-based care for high-risk pregnancies,[17,18] forming social media groups among faculty members, midwives, and pregnant women to provide proper guidance,[19] and ensuring tele-consultations for MCH care seekers.[20] Moreover, our study participants suggested actions such as provision of uninterrupted quality health care, adequate transportation facilities, and additional nutritional food that would help them better during an emergency situation. Implication for policy and practices It is of utmost importance to plan and develop a context-specific evidence-based policy for ensuring MCH care services during any future pandemic or health emergency. While mitigating an emergency event like the COVID-19 pandemic has a paramount priority, at the same time the importance of sustaining routine health services such as MCH care cannot be compromised. Strategies communicating with care seekers, building their trust and faith on the health system, promoting infection prevention measures, ensuring transportation services, having clear pandemic management guidelines, and leveraging technology support for tele-consultation are some of the possible solutions. Strengths and limitations The scope of the study is limited to MCH services based on the qualitative findings. The study findings are restricted to the event of the COVID-19 pandemic and hence need to be interpreted accordingly. As the study participants were from different places and mostly from rural areas, the results in urban settings may vary. Because the wide geographic variability in selecting and recruiting the study participants was done, this study provides a deeper insight into the challenges encountered among antenatal and postnatal mothers. The findings would be more or less similar in other similar settings. This will provide an opportunity for developing appropriate strategies to ensure quality MCH care services even in the event of any similar future health emergencies. CONCLUSION In the event of the COVID-19 pandemic, factors such as mobility restriction, limited transportation facility, fear of contracting infection, and lack of information, posed a challenge for accessing MCH care services among the seekers. This resulted in avoiding a visit to health facilities for ANC, PNC, and even for delivery in some cases. The MCH care seekers not only encountered physical problem but also faced mental stress, anxiety, and fear. During a pandemic or health emergency, while mitigation strategies for the pandemic are of utmost importance, the system also needs to be prepared for ensuring the provision of routine healthcare services such as MCH care. While our findings attempted to find out the root causes for the MCH care-related lapses, more research is needed to develop appropriate strategies to ensure quality MCH care during any future pandemic or health emergency. Financial support and sponsorship Source(s) of support: This work was supported by the United Nations Children’s Fund (UNICEF), Bhubaneswar, Odisha, India (grant number: BHU/100/2020/31). The funding agency has no role either in the collection, analysis, or interpretation of data. Conflicts of interest There are no conflicts of interest. Acknowledgement We would like to acknowledge and show our gratitude to all the state and district health officials for their cooperation in carrying out the project at the study sites smoothly. The authors also thank the study participants for providing support and time. ==== Refs REFERENCES 1 Singh K Kaushik A Johnson L Jaganathan S Jarhyan P Deepa M Patient experiencesand perceptions of chronic disease care during the COVID-19 pandemic in India: A qualitative study BMJ Open 2021 11 e048926 2 Kumar C Sodhi C CP AJ Reproductive, maternal and child health services in the wake of COVID-19: Insights from India J Glob Health Sci 2020 2 e28 3 Kumar SU Kumar DT Christopher BP Doss C The rise and impact of COVID-19 in India Front Med (Lausanne) 2020 7 250 32574338 4 Osanan GC Vidarte MF Ludmir J Do not forget our pregnant women during the COVID-19 pandemic Women Health 2020 60 959 62 32880229 5 Aggarwal R Sharma AK Guleria K Antenatal care during the pandemic in India:the problem and the solutions Int J Pregnancy Child Birth 2021 7 15 7 6 Balogun M Banke-Thomas A Sekoni A Boateng GO Yesufu V Wright O Challenges in access and satisfaction with reproductive, maternal, newborn and child health services in Nigeria during the COVID-19 pandemic: A cross-sectional survey PLoS One 2021 16 e0251382 33961682 7 Stein D Ward K Cantelmo C Estimating the potential impact of COVID-19 on mothers and newborns in low-and middle-income countries Health Policy Plus 2020 8 e863 4 8 Palo SK Dubey S Negi S Sahay MR Patel K Swain S Effective interventions to ensure MCH (Maternal and Child Health) services during pandemic related health emergencies (Zika, Ebola, and COVID-19): A systematic review PLoS One 2022 17 e0268106 35536838 9 UNICEF. Maternal and Newborn Health and COVID-19 UNICEF Data 2020 10 Tadesse E Antenatal care service utilization of pregnant women attending antenatal care in public hospitals during the COVID-19 pandemic period Int J Womens Health 2020 12 1181 8 33335430 11 Yassa M Yassa A Yirmibeş C Birol P Ünlü UG Tekin AB Anxiety levels and obsessive compulsion symptoms of pregnant women during the COVID-19 pandemic Turk J Obstet Gynecol 2020 17 155 60 33072418 12 Esegbona-Adeigbe S Impact of COVID-19 on antenatal care provision Eur J Midwifery 2020 4 16 33537618 13 Freitas-Jesus JV Rodrigues L Surita FG The experience of women infected by the COVID-19 during pregnancy in Brazil: A qualitative study protocol Reprod Health 2020 17 1 7 31915022 14 Dodgson JE Tarrant M Chee YO Watkins A New mothers'experiences of social disruption and isolation during the severe acute respiratory syndrome outbreak in Hong Kong Nurs Health Sci 2010 12 198 204 20602692 15 Lohm D Flowers P Stephenson N Waller E Davis MD Biography, pandemic time, and risk: Pregnant women reflecting on their experiences of the 2009 influenza pandemic Health (London) 2014 18 493 508 24481774 16 Patabendige M Gamage MM Jayawardane A The potential impact of COVID-19 pandemic on the antenatal care as perceived by non-COVID-19 pregnant women: Women's experience research brief J. Patient Exp 2021 8 2374373521998820 34179402 17 Nguyen NH Nguyen AQ Duong PX Van Nguyen T Using emerging telehealth technology as a future model in vietnam during the COVID-19 pandemic: Practical experience from Phutho General Hospital JMIR Form Res 2021 5 e27968 34078590 18 Royal College of Obstetricians & Gynaecologists. Coronavirus (COVID-19) infection in pregnancy JMIR Form Res Version 2020 4 19 Mortazavi F Ghardashi F The lived experiences of pregnant women during COVID-19 pandemic: A descriptive phenomenological study BMC Pregnancy Childbirth 2021 21 193 33685398 20 Gupta A Yadav S Seduchidambaram M Singh N Pradhan PK Pradhan M Triage of antenatal care through telehealth during COVID-19 pandemic in a tertiary care centre of North India J Family Med Prim Care 2022 11 1055 8 35495822

PMC010xxxxxx/PMC10353677.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-407 10.4103/ijcm.ijcm_305_22 Original Article An Evaluation of Malaria Surveillance System in Punjab, India, 2020 Sharma Sahil Goel Kapil 1 Kaushal Kanica 2 Grover Gagandeep S. 3 Dikid Tanzin 4 Singh Gurinder B. 5 National Centre for Disease Control (NCDC), Delhi, India 1 Department of Community Medicine and School of Public Health, Post Graduate Institute of Medical Education and Research (PGIMER), Chandigarh, India 2 South Asia Field Epidemiology and Technology Network, NCDC, New Delhi, India 3 Integrated Disease Surveillance Programme, Directorate Health Services, Department of Health and Family Welfare, Punjab, India 4 Epidemiology Division, National Centre for Disease Control (NCDC), Delhi, India 5 Directorate Health Services, Department of Health and Family Welfare, Punjab, India Address for correspondence: Dr. Kapil Goel, Department of Community Medicine and School of Public Health, Post Graduate Institute of Medical Education and Research (PGIMER), Chandigarh - 160 012, India. E-mail: drkapil123@gmail.com May-Jun 2023 30 5 2023 48 3 407412 07 4 2022 09 12 2022 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Background: India accounted for 6% of global burden of malaria with 95% population residing in malaria endemic areas. However, Punjab is in the malaria elimination phase with annual parasite incidence (API) <1/1000 population. Objectives: We evaluated malaria surveillance system in Punjab using CDC’s updated guidelines for evaluating public health surveillance systems to provide recommendations for strengthening the existing system and to overcome the challenges in the path of malaria free Punjab. Methods: We chose two districts of Punjab, Amritsar (lowest API) and Mansa (highest API), interviewed stakeholders, and performed a retrospective desk review. We evaluated the overall usefulness of the system and assessed seven attributes at state, district, health facility, and village level during July–August 2020. Results: In Punjab, there was progressive decline in the malaria cases from 2,955 cases in 2009 to 1,140 in 2019 and no malaria deaths since 2011. Regarding various attributes, overall score for flexibility was good (85.9%); average for simplicity (77%), acceptability (74%), data quality (74%), and timeliness (70%); and poor for representativeness (59%) and stability (57%). Conclusions: Malaria surveillance system was useful in analyzing the trends of morbidity and mortality and for generating data to drive policy decisions. To improve stability, representativeness, and acceptability, surveillance staff should not be engaged in supplemental work, and reports from private sector must be ensured. Supportive supervision and regular trainings should be carried out regarding reporting formats, guidelines, and timely epidemiological investigations to improve timeliness, data quality, and simplicity. Evaluation India malaria Punjab surveillance ==== Body pmcINTRODUCTION Malaria is a potentially life-threatening vector borne parasitic infection that affected approximately 228 million population globally in 2018. The majority (93%) of malaria cases were from the African region; India accounted for 6% of global burden of malaria.[1,2] Most of the cases were reported from eastern and central parts of India with 95% of the population residing in malaria endemic areas.[3,4] Malaria not only contributes to morbidity and mortality but also imposes a social and economic burden. Each case of malaria had a direct out-of-pocket expenditure of at least US$ 2.67 and additional minimum indirect cost of US$ 10.85 due to loss of productivity of 3–4 days.[5] There has been a progressive decline in malaria transmission in India from an annual parasite incidence (API) of 2.1/1000 in 2001 to 0.25/1000 in 2019.[3] In Punjab also there has been 82% reduction in total malaria cases between 2010 and 2015.[6] Punjab is in the malaria elimination phase with an API of <1 case/1000 population in all of the 22 districts for last five years.[2] The National Vector Borne Disease Control Programme (NVBDCP) is an umbrella program under the National Health Mission for the prevention and control of six vector borne diseases, including malaria. In alignment with the 2016 National Malaria Elimination Campaign, the Government of Punjab launched the Punjab Malaria Elimination Campaign (PMEC) 2017–2021 with a vision to achieve zero case of indigenous malaria in the state of Punjab by 2021. Punjab malaria elimination campaign has objectives to interrupt the malaria transmission from areas where cases are still being reported to identify the foci of infection and efforts to eliminate it with integrated vector management and to prevent reintroduction of malaria transmission in areas where interruption has been achieved.[6] Epidemiological surveillance, both active and passive, is a core intervention under NVBDCP to eliminate malaria from Punjab. The surveillance reports are maintained on the malaria format registers (MF), and the information in the system flows from village to national level [Figure 1]. Figure 1 Flow of data in the malaria surveillance system, Punjab, July–August 2020 The results of this evaluation will provide recommendations to the Department of Health and Family Welfare, Punjab on overcoming challenges on the path of a malaria free Punjab. MATERIALS AND METHODS Usefulness is defined as the ability of a surveillance system to meet its objectives and to support disease-control programs and policy decisions.[7] We assessed the overall usefulness of the malaria surveillance system and other attributes by using the CDC’s updated guidelines for evaluating public health surveillance systems[7] during July to August 2020. To evaluate usefulness, we also analyzed the trend of malaria cases (both Plasmodium vivax and P. falciparum) and malaria deaths from 2009 to 2019. We chose seven attributes; simplicity, flexibility, acceptability, data quality, timeliness, stability, representativeness, and evaluation were conducted at district, health facility, and village level. Two of Punjab’s 22 districts were selected according to the API, one district with lowest API (Amritsar) and another with highest API (Mansa) in 2019. We randomly selected two community health centers (CHC) from the rural areas of each district, one primary health center (PHC) from each CHC, one subcenter from each PHC, and one village from that subcenter. We interviewed key stakeholders using semi-structured questionnaire and performed a retrospective desk review. In urban areas, health personnel under the Urban Malaria Scheme at district level were selected for interviews. The data were collected and compiled using Microsoft Excel and presented in frequencies and proportions [Table 1]. The overall attribute score was calculated by taking average of the indicator scores of both districts. Likert scale was used to evaluate the performance of each attribute. An attribute with a score >90% was ranked “excellent,” a score between 80% and 89% was ranked “good,” a score between 60% and 79% was ranked “average,”, a score between 40% and 59% was ranked “poor,” and a score <40% was ranked “very poor.” Table 1 Description of attributes, indicators, data source, and collection method with analysis for the evaluation of malaria surveillance system in Punjab, India, July—August 2020 Attributes and Indicators Data collection Analysis (Proportions) Simplicity  Simplicity in malaria case definition and reporting formats Interviews of #DE, *MO, †AMO, ‡MPHS (§M), ||MPHW (M), MPHW (¶F), **LT, ‡‡ASHA, ††IC Respondents who were able to explain the case definition and reporting formats in native language  Knowledge about malaria case definition “ Respondents aware of case definition for malaria  Knowledge about treatment and follow-up Interviews of DE, MO, AMO, MPHS (M), MPHW (M), MPHW (F), LT Respondents aware of malaria treatment and follow-up Flexibility  Flexibility in the surveillance period and reporting formats Interviews of DE, AMO, MPHS (M), MPHW (M), MPHW (F), ASHA, LT, IC Respondents who agreed on changing the surveillance period and reporting formats Data Quality  Completeness of filled reporting formats Reporting formats filled by MPHS (M), MPHW (M), LT, IC for May-June 2020 Completely filled reporting formats filled by respondents  Completeness of reports at state ¶¶HQ Monthly reports received at state HQ for May–June 2020 Completely filled reporting formats received at state HQ Acceptability  Satisfaction with malaria surveillance Interviews of DE, MO, AMO, MPHS (M), MPHW (M), LT, MPHW (F), ASHA, IC Respondents satisfied with malaria surveillance  Willingness to test fever cases Review of OPD registers for fever cases in June 2020 Fever cases tested for malaria by blood slide  Willingness for online entry of malaria cases Observe online portal entries of malaria cases in 2019 Malaria cases in 2019 entered online on website portal Timeliness  Reporting, investigation and treatment of malaria cases Record review of case investigation forms (CIFs), July-Dec 2019 Cases reported within 24 hours of slide examination, investigated with 48 hours and treated within 24 hours  Reporting at state HQ Reports received at state HQ via emails within first five days for May-June 2020 Reports received at state HQ on time Stability  Existence of feedback mechanism Interviews of DE, MO, AMO, MPHS (M), MPHW (M), LT, IC, MPHW (F), ASHA Respondents who received feedback in last 6 months  Training for malaria surveillance “ Respondents trained in malaria surveillance in the last 1 year  Engagement in other work Interviews of AMO, MPHS (M), MPHW (M), LT, IC Respondents not engaged in other work  Loss of surveillance date Key informant interview of DE Health facilities with stable computerized system Representativeness  Passive surveillance Record review of monthly reports received at district HQ for June 2020 Govt. and private health facilities shared malaria reports  Active surveillance “ Villages covered under active surveillance *MO=Medical officers. #DE=District epidemiologist. †AMO=Assistant malaria officers. ‡MPHS=Multipurpose health supervisors. §M=Male. ||MPHW=Multipurpose health workers. ¶F=Female. **LT=Lab technicians. ††IC=Insect collectors. ‡‡ASHA=Accredited social health activists Ethical considerations Necessary permissions from state government and NCDC, Delhi were obtained. Interviews were conducted after taking written informed consent, and information was kept confidential. RESULTS We conducted 42 interviews (22 from Amritsar District and 20 from Mansa District) at two district HQs, 12 health facilities, and four villages. Key informants included two district epidemiologists, two medical officers (MO), five assistant malaria officers (AMO), 12 multipurpose health supervisors (male) [MPHS (M)], six multipurpose health workers (male) [MPHW (M)], four multipurpose health workers (female) [MPHWs (F)], five lab technicians (LT), two insect collectors (IC), and four accredited social health activists (ASHA). Usefulness Malaria surveillance system was useful in analyzing the trend of malaria cases over time, and trend of cases can also be compared with the same time period of previous years. There was progressive reduction in the malaria cases since 2010 (3,476) till 2019 (624) and then cases again rise in 2019 (1,140) due to increased surveillance among the migratory population. There were three malaria deaths in 2011, since then were no malaria deaths in Punjab [Figure 2]. Figure 2 Trend of malaria cases and deaths in Punjab, India, 2009–2019 The data generated by the malaria surveillance system were utilized by the World Health Office Country Office of India and NVBDCP, India to prepare the strategic action plan for malaria elimination in Punjab for 2018–2020 in collaboration with NVBDCP, Government of Punjab.[2] Simplicity The malaria case definition was easily understood by 100% (42) respondents, and around 90% respondents had knowledge of malaria case definition. About 74% respondents were aware of treatment of malaria, and 61% had correct responses regarding the number of follow-up slides prepared for each malaria case. Only 60% respondents were able to explain the complete information required on the malaria reporting formats. Overall mean score for simplicity was 77%. Flexibility During the transmission season in 2019, the surveillance reporting period was changed from monthly to weekly reporting with minimal cost and effort. Reporting formats were provided by the NVBDCP HQ in New Delhi; however, formats were modified as per the needs with due permission from the Government of India. Out of 40 respondents, 92% agreed on editing the information required on malaria reporting formats and 80% agreed on changing the surveillance period from monthly to weekly. Overall mean score for flexibility was around 86%. Data quality The completeness of the malaria formats received from state HQ from May to June 2020 was 100%. But below the district level, 48% malaria reporting formats (56% in Mansa and 40% in Amritsar) were completely filled by the respondents. Overall mean score for data quality was 74%. Acceptability Of 42 respondents, 55% were satisfied with the surveillance system. In June 2020, at the health facility level, 69% fever cases (86% in Mansa and 52% in Amritsar) were tested for malaria parasite. All the malaria cases in Mansa and Amritsar Districts were entered in an online portal in 2019. Overall mean score for acceptability was 74%. Timeliness There were 240 malaria cases (237 in Mansa and 3 in Amritsar) from July to December 2019. However, CIFs were available for 226 cases (223 in Mansa and three in Amritsar). Complete information was available for 212 cases for treatment, 218 cases for epidemiological investigation, and 225 cases for reporting from the CIFs. The majority (98%) of malaria cases were reported within 24 hours of blood slide examination; 85% (69% in Mansa and all three in Amritsar) cases received anti-malarial drugs within 24 hours; and 21.5% (10% in Mansa and one of three in Amritsar) cases were investigated within 48 hours of reporting. At state HQs, 75% of monthly reports were received on time from May to June 2020. Overall mean score for timeliness was 70%. Stability Of 42 respondents, 32 (76%) received feedback from their supervisors in last six months, and 17 (41%) were trained in malaria surveillance in 2019. Only 3 of 32 malaria surveillance staff were completely engaged in surveillance; the remaining 29 were also involved in other jobs. There were two computer systems (one NVBDCP and one personal) in Mansa and one IDSP computer system in Amritsar. All the computer systems were stable as there was no loss of data due to a system crash in 2019. Overall mean score for stability was 57%. Representativeness Of the 1100 villages (241 in Mansa and 859 in Amritsar), all of them were covered under the active surveillance. Regarding passive surveillance, 100% (31) government health facilities were doing malaria reporting at district level, but there was zero reporting from private facilities in both districts. Age distribution showed that patients diagnosed with malaria ranged from 1.5 years to 90 years old. Overall mean score for representativeness was 59%. DISCUSSION Of the seven attributes, flexibility was good; timeliness, acceptability, data quality, and simplicity were average; and stability and representativeness were poor. Surveillance system was useful in analyzing the morbidity and mortality pattern of malaria cases over time and was used to support policy decisions. System was simple in terms of structure and data flow from village to state HQs, but it was average in terms of ease of operations. Many MPHSs and MPHWs were not able to understand the reporting formats and the number of follow-up slides to be prepared for each malaria case. However, the system was flexible. It could accommodate changes in reporting formats as well as surveillance period as per the needs during transmission season. This finding was consistent with malaria surveillance evaluations conducted in the Bhutan and Nigeria.[8,9] Data received at the state HQs were of excellent quality in terms of completeness of reports. But at district level and below, almost half of reporting formats examined were incomplete, especially in Amritsar. A study in Chipinge district of Zimbabwe showed that completeness of reports was 100% at district level, but gaps were still there in completeness at health facility level.[10] Overall, the surveillance system had an average level of data quality. Acceptability of the surveillance system at the health facility level was poor. When comparing the out-patient department (OPD) data with surveillance data, 31% (14% in Mansa and 49% in Amritsar) fever cases were not investigated for malaria at health facility level. However, reports received at state/district HQs showed that all fever cases were investigated for malaria, raising the issue of authenticity of these reports. Almost half of respondents wanted to change the surveillance system because of the shortage of manpower; many health personnel were involved in other health programs like immunization and administrative work like death-birth registration, etc., In comparison with Amritsar District, Mansa experienced considerable shortage of manpower, especially lab technicians. However, willingness to enter malaria cases on the web portal was excellent. Overall, acceptability was average. This is in contrast to a study conducted in Kano State, Nigeria that showed all respondents were willing to continue with same surveillance system and 84.6% were fully involved in malaria surveillance system, depicting the high level of acceptability.[11] The surveillance system was excellent at reporting malaria cases within 24 hours of blood slide examination; however, the epidemiological investigation of malaria cases in terms of active case search in community, entomological investigation, etc., was very poor, especially in Mansa. Timeliness in sharing monthly reports at state HQs was average, especially in Amritsar. Overall, timeliness was average. Stability was poor as majority of respondents were engaged in other activities in addition to malaria surveillance. These malaria surveillance staff were simultaneously engaged in COVID-19 duties, death-birth registrations, and other health programs activities. About 60% respondents had not had refresher training in the last year and one-fourth had not received feedback from their supervisors in last six months. Regarding logistics, there were shortages of MF registers, rapid diagnostic tests (RDTs), and anti-malarial drugs at health facility level. Amritsar District did not have dedicated computer system by NVBDCP for generating and sharing the reports. Similarly, according to the study in Kano state in Nigeria, 68% of respondents indicated stock-out of the malaria commodity as one of the major challenges, along with irregular supply of RDTs, other data tools, guidelines, etc.[11] For active surveillance, all the villages were covered, but passive surveillance was poor because no private health facility was reporting for malaria surveillance. Government health facilities shared monthly reports even in in the absence of malaria cases. Private health sector only provided information when confirmed malaria case was detected, otherwise there was no regular (or even nil) reporting. Population sub-group analysis showed malaria cases reported from a wide range of ages. Overall, representativeness was poor due to lack of reporting from the private health sector. The study had few limitations. Since the study was conducted in two of the 22 districts of Punjab, the results of the surveillance evaluation cannot be generalized to the whole state; however, the study findings will definitely help the decision-makers to understand the challenges ahead of malaria free Punjab. Due to COVID-19 pandemic, the district hospital in Amritsar was not visited as it was converted into COVID-19 hospital. However, some extra efforts were carried out to collect the data telephonically for the surveillance evaluation. Due to lack of literature on malaria surveillance in India, the results of the study were compared with the studies conducted in other Asian and African countries. CONCLUSION Malaria surveillance system under NVBDCP was useful in estimating the morbidity and mortality and allowed trend analysis of malaria cases over time. Overall, it contributed to the prevention and control of malaria as there have been no malaria deaths in Punjab since 2011. However, the system’s attribute performance was only average. First and foremost, stability and representativeness need to be strengthened. Malaria surveillance staff should not be engaged in supplemental work; even if they are involved in other activities due to shortage of health staff, malaria surveillance should not be hampered. Regular trainings on reporting formats and updated guidelines should be prioritized. Government must ensure regular reporting including nil/zero reporting from private health facilities. Regular and uninterrupted supply of MF registers, RDTs, and anti-malarial drugs are needed. To increase willingness to prepare malaria blood slides, we recommend supportive supervision to ensure that every fever case gets investigated for malaria at health facility and village level. For each malaria case, timely epidemiological investigation should be carried out within 48 hours. Financial support and sponsorship This public health activity was conducted by India Epidemic Intelligence Service (EIS) program of the National Centre for Disease Control (NCDC). The NCDC receives funding support for the India EIS Program through a cooperative agreement with the U.S. Centers for Disease Control and Prevention, Center for Global Health, Division of Global Health Protection. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. This present study received no funding from anyone. Conflicts of interest There are no conflicts of interest. Acknowledgements We acknowledge the Malaria consultant (World Health Organization) and health staff under the National Vector Borne Disease Control Program (NVBDCP) in Mansa and Amritsar districts for their cooperation and participation in the data collection process for this surveillance evaluation. We thank Dorothy L. Southern for providing scientific writing advice and critically reviewing the manuscript. ==== Refs REFERENCES 1 World malaria report 2019. World Health Organization Available from: https://www.who.int/news-room/feature-stories/detail/world-malariareport-2019 [Last accessed on 2020 Feb 26] 2 Micro Strategic Action Plan for Malaria Elimination in the State of Punjab, India. National Vector Borne Disease Control Programme Available from: http://origin.searo.who.int/india/publications/strategic-action-planfor-malaria-lowress-reference-file.pdf [Last accessed on 2020 Sep 07] 3 National Vector Borne Disease Control Programme, Ministry of Health and Family Welfare, Government of India. Magnitude of the problem Available from: https://nvbdcp.gov.in/inde× 4.php?lang=1 and level=0 and linkid=420 and lid=3699 [Last accessed on 2020 Feb 26] 4 World Malaria Day 2019. National Health Portal of India Available from: https://www.nhp.gov.in/world-malaria-day-2019_pg [Last accessed on 2020 Aug 21] 5 National Framework for malaria elimination in India, 2016–2030. National Vector Borne Disease Control Programme, Ministry of Health and Family Welfare, Government of India Available from: https://nvbdcp.gov.in/WriteReadData/l892s/National-framework-for-malariaelimination-in-India-2016%E2%80%932030.pdf [Last accessed 2020 Aug 22] 6 Punjab Malaria Elimination Campaign, 2017-2021. National Vector Borne Disease Control Programme, Ministry of Health and Family Welfare, Government of Punjab Available from: https://nvbdcp.punjab.gov.in/Download/Punjab%20Malaria%20Elimination%20Campaign%202017-2021.pdf [Last accessed on 2020 Sep 06] 7 Updated Guidelines for Evaluating Public Health Surveillance Systems: Recommendations from the guidelines working group. Centres for Disease Control and Prevention. MMRW 2001 Available from: https://www.cdc.gov/mmwr/preview/mmwrhtml/rr5013a1.htm [Last accessed on 2020 Feb 26] 8 West N Gyeltshen S Dukpa S Khoshnood K Tashi S Durante A An evaluation of the national malaria surveillance system of Bhutan, 2006–2012 as it approaches the goal of malaria elimination Front Public Health 2016 4 167 27595095 9 Joseph A Evaluation of malaria surveillance system in Ebonyi state, Nigeria, 2014 Ann Med Health Sci Res 2017 7 4 10 Kureya T Chadambuka E Mhlanga M Ndaimani A Makoni P An evaluation of the malaria surveillance system of Chipinge District, Manicaland Province Int J Health Sci 2017 11 14 11 Visa TI Ajumobi O Bamgboye E Ajayi I Nguku P Evaluation of malaria surveillance system in Kano State, Nigeria, 2013–2016 Infect Dis Poverty 2020 9 15 32036790

PMC010xxxxxx/PMC10353678.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-497 10.4103/ijcm.ijcm_663_22 Short Communication QTc Interval of Healthcare Workers from India: Baseline and Effect of Hydroxychloroquine Prophylaxis during the COVID-19 Pandemic Gutte Shreyas Gurjar Mohan Sanjeev Om Prakash 1 Bhadauria Dharmendra 2 Kapoor Aditya 3 Mishra Prabhaker 4 Azim Afzal Poddar Banani Department of Critical Care Medicine, Sanjay Gandhi Post Graduate Institute of Medical Sciences (SGPGIMS), Raebareli Road, Lucknow, India 1 Department of Emergency Medicine, Sanjay Gandhi Post Graduate Institute of Medical Sciences (SGPGIMS), Raebareli Road, Lucknow, India 2 Department of Nephrology, Sanjay Gandhi Post Graduate Institute of Medical Sciences (SGPGIMS), Raebareli Road, Lucknow, India 3 Department of Cardiology, Sanjay Gandhi Post Graduate Institute of Medical Sciences (SGPGIMS), Raebareli Road, Lucknow, India 4 Department of Biostatistics, Sanjay Gandhi Post Graduate Institute of Medical Sciences (SGPGIMS), Raebareli Road, Lucknow, India Address for correspondence: Prof. Mohan Gurjar, Department of Critical Care Medicine, Sanjay Gandhi Post Graduate Institute of Medical Sciences (SGPGIMS), Raebareli Road, Lucknow - 226 014, Uttar Pradesh, India. E-mail: m.gurjar@rediffmail.com May-Jun 2023 30 5 2023 48 3 497500 02 8 2022 19 4 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Background: The aim of this study was to access the incidence of prolonged QTc interval and changes, if any, among Indian healthcare workers (HCWs) taking hydroxychloroquine (HCQ) prophylaxis while managing coronavirus disease 2019 (COVID-19) cases. Methods: At the beginning of the COVID-19 pandemic, as per the Indian Council of Medical Research (ICMR) policy, HCWs were advised to take HCQ as prophylaxis after getting an electrocardiogram (ECG) while being posted to look after COVID-19 patients. A follow-up ECG was repeated for those who took HCQ. The normal upper limit for QTc interval of 460 milliseconds (ms) for females and 450 ms for males was considered. Results: A baseline ECG was analyzed for 250 HCWs with a median age of 35 (30–43) years. The median QTc was 410 (395–421) ms with the prevalence of prolonged QTc of 1.8% in females and 0% in males. A follow-up ECG after HCQ intake for 43 HCWs was further analyzed. They had a median age of 35 (31–39) years and took an average dose of HCQ of 2372 ± 839 mg. Pre- and post-HCQ chemoprophylaxis QTc interval (ms) was as follows: 408 (386–419) and 405 (387–417), with P = 0.434, respectively. Conclusion: Among Indian HCWs, the prevalence of prolonged QTc is 1.8% and 0% in females and males, respectively. HCQ intake as chemoprophylaxis for COVID-19 did not affect their QTc interval. Electrocardiogram healthcare workers hydroxychloroquine QTc SARS-CoV-2 infection ==== Body pmcINTRODUCTION During the recent coronavirus disease 2019 (COVID-19) pandemic, before the availability of vaccine in the year 2020 itself, the National Task Force by the Indian Council of Medical Research (ICMR) recommended the use of hydroxychloroquine (HCQ) as chemoprophylaxis for asymptomatic healthcare workers (HCWs) involved in the care of suspected or confirmed cases of COVID-19 and asymptomatic household contacts of laboratory-confirmed cases.[1] Their recommendation was based on in vitro studies, which found that HCQ is effective against COVID-19.[2,3] In view of ICMR recommendations, the Indian Heart Rhythm Society (IHRS) proposed a scientific statement for the use of HCQ and strongly discouraged its use for the general public without medical supervision and prescription.[4] As there is a scarcity of known prevalence of QT interval prolongation in nonhospitalized population in general, and none in specific population such as HCWs from India, we conducted this study to assess the baseline-corrected QT (QTc) interval, the incidence of prolonged QTc interval, and changes, if any, among HCWs taking HCQ prophylaxis while managing COVID-19-confirmed or COVID-19-suspected cases. METHODS This was a prospective observational study conducted at a tertiary care university hospital in North India, after approval from the Institutional Ethics Committee (IEC code 2020-124-IP-EXP-18) and registered with the Clinical Trial Registry of India (CTRI/2020/05/025089). Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines have been followed for the study [Supplement file 1]. Study participants’ inclusion During the study period (May–September 2020) at the beginning of COVID-19, our hospital followed ICMR recommendations to advise all HCWs to take HCQ as chemoprophylaxis during their posting to manage COVID-19-confirmed or COVID-19-suspected patients. As a part of good clinical practice, HCWs were also suggested to get an electrocardiogram (ECG) done before starting HCQ intake. Investigators routinely received the ECG of HCWs for review and advice for safety to take HCQ. These HCWs were further followed if they took HCQ as chemoprophylaxis, to get their follow-up ECG. Data collection For all included participants, we collected information on age and gender. We collected more variables for whom we had follow-up ECG after HCQ intake, which includes the presence of comorbidities, use of concurrent medication, dose of HCQ, and adverse effect of HCQ, if any. QTc interval measurement A 12-lead ECG was used to measure the QT interval (and standard correction for heart rate (HR) by Bazett’s formula) to calculate the corrected QT for HR (QTc). Data acquisition and ECGs were read by a single independent observer and confirmed by a certified cardiologist led to a reduction in bias. The normal upper limit for QTc interval is considered 460 ms for females, 450 ms for males, and delta QTc (absolute increase in QTc post-HCQ) by 60 ms.[5] Sample size For the calculation of sample size to determine the effect of HCQ on the QTc interval, the effect size (Cohen’s d) range between 0.2 and 0.49 was considered a small effect size when estimated between two paired means.[6] Assuming a small change (effect size = 0.49) to be detected in the QTc interval between pre- and post-observations, at a two-sided 95% confidence interval and 90% power of the study, the estimated sample size for paired observations came out to be 46. The sample size was estimated using software G Power version 3.1.9.2 (Düsseldorf University, Germany). Statistical analysis Descriptive statistics of the continuous variables were presented as mean ± standard deviation/median (interquartile range (IQR)), whereas categorical data were presented in frequency (%). A paired-samples t-test was used to calculate the significance level as appropriate. A P value <0.05 was considered statistically significant. Statistical Package for Social Sciences version 23 (SPSS 23, IBM, Chicago, USA) software was used for data analysis. RESULTS Of the 294 ECGs collected over 6 months, 250 were analyzed (44 ECGs excluded due to artifacts and fainting of ink). Baseline ECG was analyzed for 250 HCWs with a median age of 35 (30–43) years. The median QTc was 410 (395–421) with the prevalence of prolonged QTc in females (n = 107) and males (n = 143) being 1.8% and 0%, respectively [Supplementary Figure 1]. In follow-up, there were 43 HCWs who took HCQ and had a second ECG. They were included for further analysis to determine the effect of HCQ on QTc interval, with a median age of 35 (31–39) years and body mass index (BMI) of 24 (23–27.5), and 19% were female. None of the participants were taking any known medication that causes prolongation of QTc. Five had comorbidities (diabetes mellitus with hypertension: 1 (2.3%), asthma: 1 (2.3%), hypothyroidism: 1 (2.3%), and arthritis: 2 (4.7%)), with median QTc of 421 (412.5–427.5) [Table 1]. The average dose of HCQ taken by HCW was 2372 ± 839 mg (1200–4000 mg). Table 1 Demographics, HCQ dose, and ECG variables among HCW (n=43) follow-up Demographic, HCQ dose, and ECG variables Mwedian (IQR)/n (%) Demographic  Age, median in years 35 (31,39)  Female sex (n (%)) 8 (18.6%)  BMI 24 (23.0, 27.5) Comorbidities (n (%))  Diabetes mellitus/hypertension 1 (2.3%)  Asthma 1 (2.3%)  Hypothyroidism 1 (2.3%)  Arthritis 2 (4.7%) HCQ dose  Average dose of HCQ (mean (in mg)) 2372±839 ECG variables at baseline  Heart rate (beats/min) 76 (67–83)  QT interval (ms) 360 (340–390)  QTc interval (ms) 408 (386–419) Changes (Δ) in ECG variables in follow-up  Δ HR (beats/min) 2 (-6 to 7)  Δ QT interval (ms) -6.0 (-16 to 10)  ΔQTc interval (ms) -2.0 (-20 to 7) Adverse effects of HCQ (n (%))  Headache 1 (2.3%)  Anxiety 1 (2.3%)  Palpitation 1 (2.3%)  Loose motion 3 (6.9%)  Gastritis 2 (4.7%) Pre- and post-HCQ chemoprophylaxis HR per minute was as follows: 76 (67–83) and 75 (68–84), with P = 0.520; QT interval (ms): 360 (340–390) and 358 (340–382), with P = 0.388; and QTc interval (ms): 408 (386–419) and 405 (387–417), with P = 0.434, respectively [Figure 1 and Supplementary Figure 2]. The linear coefficient of correlation between dose of HCQ and delta QTc is very weakly positive (R2 = 0.007) [Supplementary Figure 3]. Post-HCQ median delta (D) changes were as follows: HR of 2 (IQR: -6 to 7) per minute; QT interval of -6 (IQR: -16 to 10) ms; and QTc interval of -2 (IQR: -20 to 7) ms [Table 1]. Figure 1 Pre- and post-intakes of total HCQ doses and changes in ECG variables Adverse effects related to HCQ were nonserious and occurred in 18.6% (n = 43) of the HCWs (headache: 1 (2.3%), anxiety: 1 (2.3%), palpitation: 1 (2.3%), loose motion: 3 (6.9%), and gastritis: 2 (4.7%)) [Table 1]. DISCUSSION In our study, the prevalence of prolonged QTc interval is 1.8% and 0% in females and males, respectively, which is very low in comparison with other nonhospitalized studied populations in various countries, viz. Chinese population (70.6% and 29.3%),[7] Uganda population (14% and 9%),[8] and Italian population (10% and 5%).[9] We reviewed the demographic and clinical characteristics of populations in these studies and possibilities of other variables, which might influence the prevalence of prolonged QTc interval like the definition of prolonged QTc interval, age, comorbidities, and use of concurrent medications. First, varied prevalence might be due to differences in the definition of prolonged QTc (for males it is 440 ms in the Italian and Chinese population, while 450 ms for the Uganda population; also, for females it is 440 ms in the Italian and Chinese population, while 460 ms for Uganda population). Second, with aging, there is an increase in QTc interval seen by the median age of the Chinese population of 54 years versus our study of 35 years. Third, simultaneously with a growing population, there is a rise in comorbidities, as it is clearly demarcated in a Chinese population that hypertension, dyslipidemia, and obesity are associated with an increase in the prevalence of prolonged QTc while our study has not studied comorbidities in all HCWs. Lastly, as seen with the Chinese population, the prevalence of prolonged QTc is associated with 59.9% population using medications that cause prolonged QTc interval, which is not the case with ours. An Indian study in 366 healthy adult males showed a mean QTc of 372 ms, but the prevalence of prolonged QTc has not been calculated.[10] In our cross-sectional study among 43 HCWs, the median age of the population was 35 years. The baseline characteristics of HCWs revealed that all were healthy with comorbidities only in five (11%) HCWs. By week 4, nearly 32.5% of HCWs had stopped HCQ, and we followed up participants with ECG and found a cumulative dose of 2372 mg (mean). All HCWs who took HCQ were aware of the side effects and investigated with ECG to rule out prolonged QT, QTc, and HR and reported no significant difference in the changes in QTc, change in QT, and HR. Our findings are also supported by Jha et al.[11] who reported 33 HCW’s follow-up and found to have a median QTc of 389 ms and a cumulative dose of HCQ of 2000 mg. However, there is no meta-analysis that evaluated the HCQ effect in a healthy population, but studies were done in COVID-19 patients or other diseases such as systemic lupus erythematosus (SLE) and rheumatoid arthritis (RA). A meta-analysis of 28 studies on HCQ use in severe acute respiratory syndrome (SARS)-COVID-19 patients did not show a significant incidence of prolonged QTc.[12] Another meta-analysis of 13 studies on HCQ or chloroquine use with or without azithromycin in SARS-COVID-19 patients produces a significant risk of QTc prolongation, but the data were heterogeneous.[13] On further evaluation, our study showed that the median delta QTc of -2 (-20 to 7) ms is not significant (an absolute value of 60 ms is considered significant). A study of HCQ on 152 COVID-19-positive Turkish outpatients showed similar results on delta QTc of -7 (-10.5 to 23.5) ms.[14] In our study, none of the study participants had a delta QTc variation of more than 1%, which is similar to a study on 219 COVID-19 patients, which showed that the vast majority (75%) of patients had a variation of less than 5%.[15] In our study, delta QTc and variation of delta QTc were not significantly changed by HCQ, possibly because of relatively young and absence of comorbid population, and also, none were taking concurrent medication causing prolonged QTc. Limitation and strength of the study The limitation of our study is that being a single center and involving a small number of participants, the results cannot be generalized to a larger population due to the possibility of regional or ethnic predisposition. Also, we did not ascertain abstinence from caffeine, tobacco, and alcohol at least 24 hours before the ECG was recorded, which might have affected QTc interval. However, the strength is that our study is the first among HCWs so far to know the prevalence of prolonged QTc interval. CONCLUSION Among Indian HCWs, the prevalence of prolonged QTc is 1.8% and 0% in females and males, respectively. The short-term use of HCQ by HCWs as chemoprophylaxis for COVID-19 did not affect their QTc interval. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. Supplementary Figure 1 Baseline QTc interval in milliseconds among 250 HCW’s Supplementary Figure 2 Heat map corresponding to pre-exposure and post-exposure of HCQ on QTc conditions for each case. The transverse axis of the heat map represents the leads of the pre-QTc and post-QTc, and the longitudinal axis represents the cases. Each lattice color represents different QTc, with blue representing QTc =320, red representing QTc =460, and blue-to-red lattice color changes representing the QTc = 440 to 340. Supplementary Figure 3 Correlation between intakes of total HCQ doses (in milligram) with ΔQTc SUPPLEMENTARY FILE 1 STROBE Statement—checklist of items that should be included in reports of observational studies Item No. Recommendation Page No. Relevant text from manuscript Title and abstract 1 (a) Indicate the study’s design with a commonly used term in the title or the abstract 01 (b) Provide in the abstract an informative and balanced summary of what was done and what was found 01-02 Introduction Background/rationale 2 Explain the scientific background and rationale for the investigation being reported 03 Objectives 3 State specific objectives, including any prespecified hypotheses 04 Methods Study design 4 Present key elements of study design early in the paper 04 Setting 5 Describe the setting, locations, and relevant dates, including periods of recruitment, exposure, follow-up, and data collection 04-05 Participants 6 (a) Cohort study—Give the eligibility criteria, and the sources and methods of selection of participants. Describe methods of follow-up Case-control study—Give the eligibility criteria, and the sources and methods of case ascertainment and control selection. Give the rationale for the choice of cases and controls Cross-sectional study—Give the eligibility criteria, and the sources and methods of selection of participants 04-05 (,b) Cohort study—For matched studies, give matching criteria and number of exposed and unexposed Case-control study—For matched studies, give matching criteria and the number of controls per case 04-05 Variables 7 Clearly define all outcomes, exposures, predictors, potential confounders, and effect modifiers. Give diagnostic criteria, if applicable 04 Data sources/ measurement 8* For each variable of interest, give sources of data and details of methods of assessment (measurement). Describe comparability of assessment methods if there is more than one group 04 Bias 9 Describe any efforts to address potential sources of bias 05 Study size 10 Explain how the study size was arrived at 05 Quantitative variables 11 Explain how quantitative variables were handled in the analyses. If applicable, describe which groupings were chosen and why 05 Statistical methods 12 (a) Describe all statistical methods, including those used to control for confounding 05 (b) Describe any methods used to examine subgroups and interactions - (c) Explain how missing data were addressed - (d) Cohort study—If applicable, explain how loss to follow-up was addressed Case-control study—If applicable, explain how matching of cases and controls was addressed Cross-sectional study—If applicable, describe analytical methods taking account of sampling strategy - (e) Describe any sensitivity analyses - Results Participants 13* (a) Report numbers of individuals at each stage of study—eg numbers potentially eligible, examined for eligibility, confirmed eligible, included in the study, completing follow-up, and analysed 05-06 (b) Give reasons for non-participation at each stage - (c) Consider use of a flow diagram - Descriptive data 14* (a) Give characteristics of study participants (eg demographic, clinical, social) and information on exposures and potential confounders 05-06 (b) Indicate number of participants with missing data for each variable of interest 05-06 (c)Cohort study—Summarise follow-up time (eg, average and total amount) Outcome data 15* Cohort study—Report numbers of outcome events or summary measures over time Case-control study—Report numbers in each exposure category, or summary measures of exposure Cross-sectional study—Report numbers of outcome events or summary measures 05-06 Main results 16 (a) Give unadjusted estimates and, if applicable, confounder-adjusted estimates and their precision (eg, 95% confidence interval). Make clear which confounders were adjusted for and why they were included (b) Report category boundaries when continuous variables were categorized 05-06 (c) If relevant, consider translating estimates of relative risk into absolute risk for a meaningful time period - Other analyses 17 Report other analyses done—eg analyses of subgroups and interactions, and sensitivity analyses Discussion Key results 18 Summarise key results with reference to study objectives 6-7 Limitations 19 Discuss limitations of the study, taking into account sources of potential bias or imprecision. Discuss both direction and magnitude of any potential bias 7,8,9 Interpretation 20 Give a cautious overall interpretation of results considering objectives, limitations, multiplicity of analyses, results from similar studies, and other relevant evidence 9 Generalisability 21 Discuss the generalisability (external validity) of the study results 9 Other information Funding 22 Give the source of funding and the role of the funders for the present study and, if applicable, for the original study on which the present article is based NIL *Give information separately for cases and controls in case-control studies and, if applicable, for exposed and unexposed groups in cohort and cross-sectional studies. Note: An Explanation and Elaboration article discusses each checklist item and gives methodological background and published examples of transparent reporting. The STROBE checklist is best used in conjunction with this article (freely available on the Web sites of PLoS Medicine at http://www.plosmedicine.org/, Annals of Internal Medicine at http://www.annals.org/, and Epidemiology at http://www.epidem.com/). Information on the STROBE Initiative is available at www.strobe- statement.org. ==== Refs REFERENCES 1 Indian Council of Medical Research: Recommendation for empiric use of hydroxychloroquine for prophylaxis of SARS-CoV-2 infection Ministry of Health and Family Welfare, Government of India 2020 Available from: https://www.mohfw.gov.in/pdf/AdvisoryontheuseofHydroxychloroquinasprophy laxisforSARSCoV2infection. pdf [Last accessed on 2020 Mar 23] 2 Vincent MJ Bergeron E Benjannet S Erickson BR Rollin PE Ksiazek TG Chloroquine is a potent inhibitor of SARS coronavirus infection and spread Virol J 2005 2 69 doi:10.1186/1743-422X-2-69 16115318 3 Wang M Cao R Zhang L Yang X Liu J Xu M Remdesivir and chloroquine effectively inhibit the recently emerged novel coronavirus (2019-nCoV) in vitro Cell Res 2020 30 269 71 32020029 4 Kapoor A Pandurangi U Arora V Gupta A Jaswal A Nabar A Cardiovascular risks of hydroxychloroquine in treatment and prophylaxis of COVID-19 patients: A scientific statement from the Indian Heart Rhythm Society Indian Pacing Electrophysiol J 2020 20 117 20 32278018 5 Rautaharju PM Surawicz B Gettes LS Bailey JJ Childers R Deal BJ AHA/ACCF/HRS recommendations for the standardization and interpretation of the electrocardiogram: Part IV: The ST segment, T and U waves, and the QT interval: A scientific statement from the American Heart Association Electrocardiography and Arrhythmias Committee, Council on Clinical Cardiology;the American College of Cardiology Foundation;and the Heart Rhythm Society. Endorsed by the International Society for Computerized Electrocardiology J Am Coll Cardiol. 2009 53 982 91 19281931 6 Kelley K Preacher KJ On effect size Psychol Methods 2012 17 137 52 22545595 7 Ma Q Li Z Guo X Guo L Yu S Yang H Prevalence and risk factors of prolonged corrected QT interval in general Chinese population BMC Cardiovasc Disord 2019 19 276 31783793 8 Magodoro IM Albano AJ Muthalaly R Koplan B North CM Vořechovská D Population prevalence and correlates of prolonged QT interval: Cross-sectional, population-based study from rural Uganda Glob Heart 2019 14 17 25 e4 30584028 9 Leotta G Maule S Rabbia F Del Colle S Tredici M Canadè A Relationship between QT interval and cardiovascular risk factors in healthy young subjects J Hum Hypertens 2005 19 623 7 15905890 10 Roy P Naidu M Raju Y Kumar TR Rani PU Kiran PU Evaluation of QT interval in healthy adult males Indian J Pharmacol 2006 38 135 6 11 Jha S Batra N Siddiqui S Yadav A Misra A Loomba M HCQ prophylaxis in COVID-19 did not show any QTc prolongation in Healthcare workers Indian Heart J 2021 73 74 76 33714413 12 Oscanoa TJ Vidal X Kanters JK Romero-Ortuno R Frequency of long QT in patients with SARS-CoV-2 infection treated with hydroxychloroquine: A meta-analysis Int J Antimicrob Agents 2020 56 106212 doi:10.1016/j.ijantimicag.2020.106212 33164789 13 Agstam S Yadav A Kumar-M P Gupta A Hydroxychloroquine and QTc prolongation in patients with COVID-19: A systematic review and meta-analysis Indian Pacing Electrophysiol J 2021 21 36 43 33075484 14 Sogut O Can MM Guven R Kaplan O Ergenc H Umit TB Safety and efficacy of hydroxychloroquine in 152 outpatients with confirmed COVID-19: A pilot observational study Am J Emerg Med 2021 40 41 6 33348222 15 Jiménez-Jáimez J Macías-Ruiz R Bermúdez-Jiménez F Rubini-Costa R Ramírez-Taboada J Flores PIG Absence of relevant QT interval prolongation in not critically ill COVID-19 patients Sci Rep 2020 10 21417 doi:10.1038/s41598-020-78360-9 33293554

PMC010xxxxxx/PMC10353679.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-430 10.4103/ijcm.ijcm_394_22 Original Article Confirmatory Factor Analysis (CFA) and Psychometric Validation of Healthy Lifestyle and Personal Control Questionnaire (HLPCQ) in India Goyal Himani Aleem Sheema Department of Psychology, Jamia Millia Islamia, New Delhi, India Address for correspondence: Ms. Himani Goyal, Department of Psychology, Jamia Millia Islamia, Jamia Nagar, New Delhi - 110 025, India. E-mail: goyalnhimani@gmail.com May-Jun 2023 30 5 2023 48 3 430435 10 5 2022 15 3 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Background: The silent epidemic of chronic illness has become a significant public health challenge worldwide. The prevention and management of these deadliest health conditions primarily require empowering the individual to make healthy choices every day, e.g., healthy eating, physical exercise, etc., The first step in designing an intervention for this comprises measuring the health empowerment-related factor. However, severe scarcity of practical tools is noted. Objective: The present study aimed at the evaluation of psychometric properties of a healthy lifestyle and personal control questionnaire (HLPCQ)’s English Version in the Indian population. Method: For this, a cross-sectional study was conducted on 618 people enlisted from the Northern state of India with the help of convenience sampling strategy. Data is collected by sharing the google form of HLPCQ through various online platforms. Results: The value of Cronbach alpha and MacDonald’s Omega, was >.70, suggesting HLPCQ has good reliability in the Indian population. While confirmatory factor analysis result shows that the final model with 24 items has a good fit to the data (RMSEA = 0.04, CFI = 0.97, TLI = 0.96, SRMR = 0.03) which implies that HLPCQ has acceptable structural and cultural validity. While the value of average variance extracted (AVE) and composite reliability (CR) for each factor were found to be more than .50 and .70 respectively, indicating the presence of convergent and discriminant validity for HLPCQ. Conclusion: These findings indicate that the HLPCQ has sound psychometric properties and can be used in the Indian population. Confirmatory factor analysis healthy lifestyle personal control psychometric properties ==== Body pmcINTRODUCTION Healthy lifestyle choices like regular physical activity, healthy eating, etc., have a crucial role in averting chronic or lifestyle illnesses like heart diseases, strokes, diabetes, etc., by up to 80%.[1,2] Together, these diseases constitute the leading cause of death, killing an estimated 41 million people per year, accounting for seven out of ten death globally.[3] India is not different, according to its annual health report 2020–2021, the economic burden due to chronic illnesses has surpassed that of infectious diseases such as TB, HIV, etc., and comprises 60% of all deaths.[4] Based on another study, researchers estimate that between 2012 and 2030, the economic burden of these illnesses will reach $ 6.2 trillion, nearly nine times as much as the previous 19 years total health expenditure. These diseases are primarily attributable to unhealthy lifestyle choices, such as poor diet, lack of exercise, etc.,[1-4] Therefore, the prevention, treatment, and management of these deadliest illnesses rely heavily on lifestyle modification and making healthy choices in everyday life.[1-3] It implies that substantial management of emerging health issues cannot be executed merely via biomedical interventions and requires the integration of health promotion programs into public health policies.[1,3,5] The interventions under the health promotion Programs were supposed to emphasize lifestyle modification by empowering individuals to have personal control over their life circumstances and make healthy choices in everyday life.[1,5] However, these programs focus haphazardly on healthy choices and disregard personal control training, thereby failing to produce the desired results in lifestyle modification and chronic disease management.[1,5] Although, personal control is found to have a fundamental role in making and shaping daily life choices.[6] It is evident from various studies that making healthy choices even in a supportive and conducive environment is difficult for individuals if they lack control over their circumstances.[7,8] Additionally, the current stressful environment has made decision-making and self-control practices more challenging.[8] Work requirements, family development, and technological progress have exposed individuals to a variety of stressors and to alleviate this stress, they are compelled to favor the choices of immediate physical rewards and indulgence in unhealthy lifestyles.[9] Thus, it becomes imperative to design health promotion interventions that encourage both personal control training and making healthy choices in everyday life. The initial step in this direction entails measuring both lifestyle choices and personal control simultaneously. According to a review of the relevant literature, all available instruments assess either personal control or lifestyle choices. None of them measures these two factors simultaneously. For instance, the multidimensional health locus of control scale,[10] self-esteem scale,[11] etc., measure individual control. In contrast, the healthy lifestyle scale[12] the healthy lifestyle screening instrument,[13] etc., measure healthy lifestyle decisions. To date, only one instrument, the healthy lifestyle and personal control questionnaire (HLPCQ) has been developed to assess the twin determinants of health promotion, namely healthy lifestyle choices and personal control.[1] Initially, this instrument was developed in the Greek language. But, due to the ubiquitous need for health promotion interventions for the management of chronic illnesses, this tool was validated very soon in other cultures also, e.g. Persian,[14] English,[15] and Polish.[16] The validation of the tool in desired cultures should be preferred over developing a new one as it reduces costs and time.[17] Furthermore, Indian health care researchers also have an urgent need for effective health promotion interventions to control the unprecedented rise in chronic illnesses.[4] Consequently, HLPCQ’s psychometric properties are required to be evaluated in Indian culture also. To ensure the quality of results, an instrument must have sound psychometric properties, which have been estimated by evaluating its reliability and validity.[18] Provided this, the present research aims to test the following hypothesis. Hypothesis 1: HLPCQ and its all dimensions would have significant construct validity (i.e., structural, cultural, convergent, and discriminant validity) in the Indian population. Hypothesis 2: HLPCQ and its five dimensions have significant reliability, (i.e., internal consistency) in the Indian population. MATERIALS AND METHODS Ethics The present study was carried out following ethical standards of seeking informed consent from each participant. Setting and participants In this cross-sectional research, participants are enlisted from North India, mainly Delhi-NCR, Dehradun, Lucknow, etc., using a convenience sampling strategy. The age of the participants varies from 20-60 years, with a mean (M) =32.34 years and standard deviation (SD) =9.52. The other socio-demographic characteristics of the participants are listed in Table 1. Table 1 Socio-demographic characteristics of the sample Variable Category Number of respondents in each category Percentage (%) of total respondents (n=618) Gender Male 348 56.3 Female 266 43.0 Other 4 0.6 Education Postgraduate 482 78.0 Undergraduate 110 17.8 Intermediate 18 2.9 High School 8 1.3 Employment Status Professional 395 63.9 Semi-professional 107 17.3 Skilled worker 58 9.4 Semi-skilled worker 5 0.8 Other 28 4.5 Unemployed 25 4 Procedure The data is collected by sharing the google form of the questionnaire to 700 individuals through academic groups available on various social networking platforms like What’s App, Facebook, electronic mail, etc. Out of this, 618 respondents matched the criteria for final analysis. Measures The scale used in this study assesses healthy lifestyle choices and personal control simultaneously with the help of 26 positively stated sentences.[1] The items are divided into five different dimensions - dietary health choices (DHC), dietary harm avoidance (DHA), daily routine (DR), organized physical activity (OPA), and social and mental balance (SMB) with 7, 4, 2, 8 and 5 items, respectively.[1] Responses to each item are recorded on a 4-point Likert Scale (1 = never, 2 = sometimes, 3 = often, 4 = always.). Reliability, i.e. Cronbach alpha coefficient of each subscale of the questionnaire’s original version, i.e. DHC, DHA, DR, OPA, and SMB, is found to be 0.75, 0.65, 0.81, 0.78, and 0.63 respectively.[1] Total scores and the score of each dimension of this questionnaire were calculated by summing the scores. Higher scores indicated a healthier lifestyle or higher empowerment to have personal control and make healthy daily life choices.[1] Statistical analysis First, the data were screened to check outliers, normality, and multicollinearity with the help of SPSS-21. After this, reliability assessment is done by using Cronbach’s alpha and Macdonald’s Omega. Afterward, confirmatory factor analysis for the specification and testing of best model fit, and calculation of Average Variance Extracted (AVE) and Composite Reliability (CR) is done to assess HLPCQ’s construct validity by providing evidence of structural, cultural, convergent and discriminant validity in the Indian population by using AMOS 22. RESULTS Initially, the data was screened to test the hypotheses of outliers via using leverage indices for each participant; multicollinearity via tolerance and variance inflation factor (VIF), and normality via kurtosis value.[19] Sample adequacy was assessed using the Kaiser-Meyer-Olkein measure.[19] The results of all these tests verify the suitability of data to run further appropriate statistical analyses to estimate measurement properties like confirmatory factor analysis (CFA). Validity estimation The validity ensures the accuracy of an instrument, and it has been estimated in the present research through construct validity by testing its evidence i.e. structural, and cultural validity and convergent and discriminant validity.[18] For this CFA with maximum likelihood method is used as researchers have enough information on the dimensionality of HLPCQ, thus CFA with maximum likelihood method is used to test all evidence of construct validity[18,20] Structural and cultural validity According to the previous literature, the measurement model of this scale has 26 observed variables, which are further divided into five latent variables, i.e. DHC, DHA, DR, OPA, and SMB with 7, 4, 8, 2, and 5 observed (indicator) variable respectively.[1,14-16] To assess the consistency of hypothesized factorial structure in the Indian population, structural and cultural validity is estimated by evaluating standardized factor loading and fit indices.[18,20] The result of the initial parameter estimation given in Figure 1 shows that two items (DHC 6 and DHC 7) have low standardized factor loading compared to an acceptable cut-off value >.60.[20] Both the items were removed sequentially, and the parameter estimation is rerun for the measurement model with 24 items. The results presented in Figure 2 show that each of the 24 items has factor loading above the accepted value of .60.[21] Figure 1 Measurement model of HLPCQ with 26 items Figure 2 Measurement model of HLPCQ with 24 items Afterwards, model fit for both models, one with 26 items and another with 24 items, were compared using multiple fit indicators i.e., absolute fit indices assessed via Chi-square statistics, SRMR (Standardized Root Mean Square Residual), and GFI (Goodness of Fit Index) (GFI); adjusted for parsimony indices assessed with RMSEA (Root Mean Square Error of Approximation); relative fit indices which was assessed using CFI (Comparative Fit Index) and TLI (Tucker Lewis Index).[19,21] Adequate model fit is defined using an established threshold value for each fit indicator given in Table 2. Table 2 Goodness of fit index and cut-off values Model 1 5 Factor Model(With 26 variables) Model 2 5 Factor Model (With 24 variables) Good fit value Acceptable fit Value Chi square 1.9 1.8 0< Chi square/df <2 2< Chi square/df <3 SRMR 0.04 0.03 0< SRMR <0.05 0.05 SRMR <0.08 GFI 0.93 0.94 0.95< GFI <1.00 0.90< GFI <0.95 RMSEA 0.05 0.04 0.00<RMSEA <0.05 0.05< RMSEA <0.08 CFI 0.96 0.97 0.95< CFI <1.00 0.90< CFI 0.95 TLI 0.95 0.96 0.95< TLI <1.00 0.90< TLI <1.00 Reference: Kline (2015) Standardized Root Mean Square Residual (SRMR), Goodness of Fit Index (GFI); Root Mean Square Error of Approximation (RMSEA); Comparative Fit Index (CFI) and Tucker Lewis Index (TLI) The results in Table 2 suggest that both the models have good to acceptable model fit values. But after removing two items with very low factor loadings, the five-factor model with 24 items shows an equally better model fit value as compared to other one.[19,21] This indicates that in the Indian population HLPCQ with 24 items with five factors has acceptable structural and cultural validity. Convergent and discriminant validity Convergent validity ensures that latent constructs meant for measuring the underlying factor are related as they are expected to be related theoretically.[22] While discriminant validity ensures that latent constructs should differ empirically at the same time.[22] Both types of validity are assessed by using the Fornell-Larcker criterion i.e., by estimating AVE (average variance extracted) and CR (composite reliability) calculated by using factor loading of each item.[23] To observe the convergent validity the threshold value of AVE and CR should be ≥0.50 and ≥0.70 respectively.[23,24] Referring to the results in Table 3, the AVE and CR values of each latent factor are ≥0.50 and ≥0.70 which implies that HLPCQ has good convergent validity.[25] Table 3 Results for Cronbach alpha, Omega coefficient, Composite reliability (CR); Average variance extracted (AVE), square root of the AVE (in bold), and correlations between constructs (off-diagonal) Factors Omega Cronbach alpha CR AVE DHC DHA DR OPA SMB DHC 0.88 0.88 0.87 0.59 0.77 DHA 0.81 0.80 0.81 0.51 0.58 0.71 DR 0.89 0.89 0.89 0.52 0.29 0.25 0.72 OPA 0.73 0.73 0.72 0.57 0.09 0.07 0.14 0.75 SMB 0.86 0.86 0.85 0.55 0.26 0.28 0.27 0.06 0.74 DHC- Dietary harm control; DHA- Dietary harm avoidance; DR- Daily routine; OPA-Organised physical activity; SMB- Social and mental balance According to Fornell- Larcker criterion discriminant validity is assessed by comparing the square root of each AVE given in the diagonal with the correlation coefficients (off-diagonal) for each construct.[22,23] The results in Table 3 show that the square root value of AVE for each latent factor is high compared to its correlation value with other factors. It suggests that the latent construct of the proposed measurement model has significant discriminant validity.[25] Results suggest that hypothesis 2 is also accepted as all four evidence of construct validity demonstrate that HLPCQ has good validity. Reliability estimation The scale’s reliability is assessed through internal consistency by using Cronbach’s alpha (α)[26] and Omega coefficient (w total)[21] as each has its limitation and strength. Internal consistency signifies that items of a scale are interrelated and can reflect the underlying construct.[27] It has been the most used measure to assess reliability.[27] The results given in Table 3 show that the value of alpha and omega for each latent factor is ≥0.70, suggesting that hypothesis 1 is verified, HLPCQ has good internal consistency i.e., reliability in the Indian population.[25,27] DISCUSSION A standardized and valid research instrument is the first requirement to get psychometrically relevant results.[18,28] But construction and validation of a new tool each time involves a complex process and consumes much time and resources.[28] Thus, the current research is conducted to assess model fit and psychometric validation of the already available tool HLPCQ in the Indian population for designing and evaluating health promotion programs. This scale is reported to have adequate empirical and theoretical evidence, thus carried directly to CFA without running EFA.[29] CFA represents a theory-driven technique that tests how the proposed factor structure can be replicated in another sample.[30] The HLPCQ is a novel and robust tool developed initially in Greek and later adapted and validated in Persian, Polish, and U.S cultures.[14-16] But these studies have overlooked the detailed estimation of psychometric properties. For instance, only HLPCQ’s Persian version opted for CFA to estimate the validity.[14] However, in the present study detailed analysis of psychometric properties was carried out to ensure its usability in Indian culture. For this, concerning hypothesis 1 the reliability of HLPCQ in Indian culture is calculated using the Cronbach alpha and Omega coefficient.[24,26] The previous studies have intensely articulated the limitation of Cronbach alpha and suggested using other better options such as the Omega coefficient.[31] From the reliability analysis results given in Table 3, it is estimated that the value of Cronbach alpha and Omega coefficient for all the subscales DHC, DHA, DR, OPE, and SMB is.88, .80, .89, .73, .86 and .88, .81, .89, .73, .86 respectively. It suggests that all the subscales have good reliability in the Indian population compared to its other versions such as Greek, Persian, and Polish, in which values range from .60 to .85.[14-16] Afterwards, to evaluate the second hypothesis i.e., to estimate all the four evidences of construct validity i.e., structural and cultural validity and convergent and discriminant validity, CFA was performed using two five-factor models: one with 26 items [Figure 1] and the second with 24 items [Figure 2]. Results of parameter estimation from Figure 1 show that two items under Dietary Health Control (DHC) factor, DHC 6 and DHC 7, have low factor loadings of .31 and .47 compared to the acceptable value of .60 or higher.[21] It suggests that these two items are incongruent with Indian culture. For instance, cooking is seen as a female job in India and has been supported intensely by various religious and social factors. Thus, item DHC 6 (Do you like cooking) might have biased responses.[32] While concerning item DHC 7 (Do you eat products with whole grains), a recent report of ICMR has found that Indians are essentially not familiar with the taste, texture, appearance, and nutritional value of whole grains.[33] It hints that both the items are not consistent with Indian culture and thus have been removed subsequently. Furthermore, the results of fit indices from Table 2 show that the value of fit indicators for both models lies in the good to an acceptable range. But the model with 24 items has slightly better fit (Chi-square = 1.8; SRMR =0.04; GFI =0.95; RMSEA =.03; CFI =.97; TLI =.96 in comparison to model with 26 items (Chi-square = 1.9; SRMR =.05; GFI =.93; RMSEA =.04; CFI =.95; TLI =.94.[19,25] It suggests that item removal does not have any effect on model fit. Also, shortening the questionnaire was reported to have a positive effect in increasing the responses.[34] Thus, the HLPCQ with 24 items is accepted to be used in Indian culture. The values of model fit indicators for the model with 24 items agree with values in HLPCQ’s Persian version. However, the HLPCQ’s Persian version has not investigated the parameter estimation, which is imperative to know the strength of items.[14,17] Afterward, the convergent validity results from Table 3 showed that the AVE and CR values for all five subscales exceeded the threshold value of >0.50 and >.70 which suggests that HLPCQ has good convergent validity in the Indian population. To establish the discriminant validity, the square root value of AVE for each latent variable should have a higher value than the correlation value with any other latent variable.[23] The results given in Table 3 showed that the square roots of AVE for all the five latent variables were ..77, .71, .72, .75, and .74 which were higher than the value of inter-construct correlation. It suggests that the subscales of HLPCQ have also fulfilled discriminant validity criteria.[22,23] However, in most validation studies of HLPCQ, a detailed discussion on construct validity i.e., on the suitability of its structure in respective culture is not given. Its English version validated on the nursing population in U.S culture and reported good convergent and discriminant validity.[15] The original version of HLPCQ found criterion validity, which cannot be considered a substitute for construct validity.[1,17] In the present research, a detailed analysis of HLPCQ’s measurement properties is conducted to assess the scope of its applicability in Indian culture. The findings suggest that HLPCQ has good reliability and construct validity i.e., structural, cultural, convergent, and discriminant validity in the Indian population. Thus, public health researchers and clinicians can consider using this tool in the Indian population to assess the individual’s health empowerment via measuring their healthy choices and ability to have personal control in daily life. It will help to conduct epidemiological studies, design interventions for lifestyle modification under the health promotion program and to evaluate the future outcome of these interventions. This study covers approximately all parameters required to establish the sound psychometric properties of an instrument.[18] But in the future, behavioral medicine researchers should consider assessing the psychometric properties of a questionnaire consistent with globally accepted COSMIN guidelines to ensure the better quality of the measurement tool. Acronym list: Healthy lifestyle choices and personal control questionnaire (HLPCQ), Dietary Harm control (DHC), Dietary harm avoidance (DHA), Organised physical activity (OPA), Social and mental balance (SMB), Daily routine (DR). Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. ==== Refs REFERENCES 1 Darviri C Alexopoulos EC Artemiadis AK Tigani X Kraniotou C Darvyri P The Healthy Lifestyle and Personal Control Questionnaire (HLPCQ): A novel tool for assessing self-empowerment through a constellation of daily activities BMC Public Health 2014 14 995 25253039 2 Dean E Söderlund A What is the role of lifestyle behavior change associated with non-communicable disease risk in managing musculoskeletal health conditions with special reference to chronic pain? BMC Musculoskelet Disord 2015 16 1 7 25637090 3 World Health Organization. 2020 Available from: https://www.who.int/news-room/fact-sheets/detail/the-top-10-causes-of-death [Last accessed 2021 Nov 03] 4 Department of health and family welfare, ministry of health and family welfare. Annual report 2020-2021 New Delhi, India Government of India 2021 Available from: https://main.mohfw.gov.in/sites/default/files/Annual%20Report%202020-21%20English.pdf 5 Bloom DE Cafiero ET Jané-Llopis E Abrahams-Gessel S Bloom LR Fathima S The global economic burden of non-communicable diseases: A report by the World Economic Forum and the Harvard School of Public Health Geneva, Switzerland In World Economic Forum 2011 2 Available at: https://www3.weforum.org/docs/WEF_Harvard_HE_GlobalEconomicBurdenNonCommunicableDiseases_2011.pdf 6 Kumar S Preetha GS Health promotion: An effective tool for global health Indian J Community Med 2012 37 5 12 22529532 7 Sela A Berger J Kim J How self-control shapes the meaning of choice J Consum Res 2017 44 724 37 8 Koelen MA Lindström B Making healthy choices easy choices: The role of empowerment Eur J Clin Nutr 2005 59 S10 6 16052175 9 Starcke K Brand M Decision making under stress: A selective review Neurosci Biobehav Rev 2012 36 1228 48 22342781 10 Wallston KA Strudler Wallston B DeVellis R Development of the multidimensional health locus of control (MHLC) scales Health Educ Monogr 1978 6 160 70 689890 11 Rosenberg M Rosenberg self-esteem scale (RSE). Acceptance and commitment therapy Measures Package 1965 61 52 12 Dong W Xiao-Hui X Xian-Bo W The healthy lifestyle scale for university students: Development and psychometric testing Aust J Prim Health 2012 18 339 45 22950906 13 Kim CH Kang KA The validity and reliability of the Healthy Lifestyle Screening Tool Phys Ther Rehabil Sci 2019 8 99 110 14 Zahra D Ahmadipour H Persian version of healthy lifestyle and personal control questionnaire (HLPCQ): A confirmatory factor analysis J Prev Med Care 2018 2 15 9 15 Grace-Farfaglia PM Pickett-Bernard DL Gorman AW Dehpahlavan J Keep calm and lead by example: Healthy lifestyles of dietitians and satisfaction with life J Prev Med 2018 3 6 16 Czapla M Juárez-Vela R Rozensztrauch A Karniej P Uchmanowicz I Santolalla-Arnedo I Psychometric properties and cultural adaptation of the polish version of the healthy lifestyle and personal control questionnaire (HLPCQ) Int J Environ Res Public Health 2021 18 9190 34501778 17 Arafat SY Chowdhury HR Qusar MM Hafez MA Cross-cultural adaptation and psychometric validation of research instruments: A methodological review J Behav Health 2016 5 129 36 18 Souza AC Alexandre NM Guirardello ED Psychometric properties in instruments evaluation of reliability and validity Epidemiol Serv Saude 2017 26 649 59 28977189 19 Kline RB Principles and practice of structural equation modeling New York The Guilford Press 2015 20 Mokkink LB Terwee CB Patrick DL Alonso J Stratford PW Knol DL COSMIN Checklist Manual Amsterdam University Medical Center 2012 21 Kilic AF Doğan N Comparison of confirmatory factor analysis estimation methods on mixed-format data Int J Assess Tools Educ 2021 8 21 37 22 Ab Hamid MR Sami W Sidek MM Discriminant validity assessment: Use of Fornell and Larcker criterion versus HTMT criterion J Phys: Conf Ser 2017 890 012163 23 Fornell C Larcker DF Evaluating structural equation models with unobservable variables and measurement error J Mark Res 1981 18 39 50 24 McDonald RP Test theory: A unified treatment London Lawrence Erlbaum 1999 25 Hair JF Jr Advanced issues in partial least squares structural equation modeling Inc: SAGE publications 2017 26 Cronbach LJ Coefficient alpha and the internal structure of tests Psychometrika 1951 16 297 334 27 Bonett DG Wright TA Cronbach's alpha reliability: Interval estimation, hypothesis testing, and sample size planning J Organ Behav 2015 36 3 15 28 Sousa VD Rojjanasrirat W Translation, adaptation and validation of instruments or scales for use in cross-cultural health care research: A clear and user-friendly guideline J Eval Clin Pract 2011 17 268 74 20874835 29 Hurley AE Scandura TA Schriesheim CA Brannick MT Seers A Vandenberg RJ Exploratory and confirmatory factor analysis: Guidelines, issues, and alternatives J Organ Behav 1997 18 667 83 30 Jackson DL Gillaspy JA Jr Purc-Stephenson R Reporting practices in confirmatory factor analysis: An overview and some recommendations Psychol Methods 2009 14 6 23 19271845 31 Deng L Chan W Testing the difference between reliability coefficients alpha and omega Educ Psychol Meas 2017 77 185 203 29795909 32 Singh P Pattanaik F Unfolding unpaid domestic work in India: Women's constraints, choices, and career Palgrave Commun 2020 6 1 3 33 ICMR-National Institute of Nutrition Recommended Dietary Allowances and Estimated Average Requirements Nutrient Requirements for Indians-2020: A Report of the Expert Group Indian Council of Medical Research National Institute of Nutrition 2020 34 Singer E Ye C The use and effects of incentives in surveys Ann Am Acad Polit Soc Sci 2013 645 112 41

PMC010xxxxxx/PMC10353680.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-443 10.4103/ijcm.ijcm_721_22 Original Article Development and Validation of Clinical Schedule for Primary Care Psychiatric Nursing (CSP-N) for Primary Care Nurses Paul James Govindan Radhakrishnan Manjunatha Narayana 1 Kumar Channaveerachari Naveen 1 Math Suresh Bada 1 Department of Nursing, NIMHANS, Bengaluru, Karnataka, India 1 Psychiatry, NIMHANS, Bengaluru, Karnataka, India Address for correspondence: Dr. Radhakrishnan Govindan, Department of Nursing, NIMHANS, Bengaluru - 560 029, Karnataka, India. E-mail: dr.rk76@hotmail.com May-Jun 2023 30 5 2023 48 3 443452 21 8 2022 06 4 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Introduction: As per the World Health Organization’s mental health report for 2022, nearly a billion people have mental health issues, and 82% of them are in low and middle-income countries where mental health services are largely absent. For the successful integration of mental health into primary health care, proper training and education of primary care professionals are mandatory. Primary care nurses are in an excellent position to screen, identify, dual collaboration for treatment planning/referral, and follow-up of persons with mental illness (PMI), but they often lack the confidence and competence to tackle mental health problems. The study aimed to develop and validate the clinical schedule for primary care psychiatric nursing (CSP-N). Materials and Methods: It is conducted in two phases: the development and validation phases. An extensive literature search has been conducted, and the ten themes derived from the two-focused group discussions and three-direct one-to-one interviews and input from mental health experts were used to design the CSP-N. The CSP-N was checked for content validity by a panel of 17 experts using the item-level content validity index (I-CVI) and the scale-level content validity index (S-CVI). Results: The draft version 1 of the CSP-N showed high content validity for individual items (I-CVI range: 0.82 to 1.00) and high overall content validity (S-CVI = 0.95), and suggestions from the experts were incorporated. The CSP-N was developed in four modules. The single-measure two-way mixed absolute agreement ICC value was calculated (for 32 subjects) for the reliability test, and the ICC value was 0.97 with a 95% CI (0.94, 0.99). Conclusions: Using an iterative approach, the development and validation of the CSP-N demonstrated high I-CVI and S-CVI for screening and identification, dual collaboration for the treatment plan, referral, and follow-up of a person with mental illness by the nurses in the community. Clinical Mental illness Primary care nursing Psychiatric nursing Schedules Screening Validation ==== Body pmcINTRODUCTION As per the National Mental Health Survey of 2016, the prevalence of mental illness in India is 10.6%, and the treatment gap ranges from 60% for common mental disorders to 90% for substance use disorders.[1-3] The mental disorders include common mental disorders (CMDs), substance use disorders (SUDs), and severe mental disorders (SMDs) such as schizophrenia and mania.[4-6] The burden of mental disorders accounts for 25.3% to 33.5% of all disease burden, and the treatment gap for severe mental disorders is 76% to 85% in low- and middle-income countries, respectively, due to the scarcity of specialist human resources and the large inequities in resource allocation.[3,7] In lower and middle-income countries, different cadres of primary care workers, like doctors, nurses, and other paraprofessionals without any specialization in mental health, are effective in providing essential mental health care in the community.[3,8,9] The National Mental Health Survey 2016 recommended that all the states of India focus on human resource management by building the skills and knowledge of doctors, nurses, ASHAs, and others and hand-holding them in identifying and managing mental health issues at the primary care level.[2,8] The primary care nurses are the people who reach the patients who are not attending the PHC, and training them to identify the patients with mental health issues at the community level and refer them for treatment can bring a significant change. However, the nurses working in PHCs’ have to address all 23 programs, and they find it very difficult to give more focus to psychiatry based on the currently available standard MSE and history collection formats.[3,7,10] In dual clinical collaboration (DCC), the nurse decides what kind of care their patients require and then collaborates with the health team accordingly, which makes health services easily accessible and untroubling to patients, helps to understand their individual needs, and strengthens the overall healthcare system.[11] The nurses’ scope in the DMHP can double up as counsellors and community health workers as social workers to help access social welfare benefits and tackle other families’ psychological issues.[12] The CSP-N is an integrated module that helps primary care nurses do a rapid screening and identify, refer, and follow-up with a person with mental illness (PMI). CSP-N version-1 consists of a screening and identification of mental illness tool (Appendix-1) using culturally appropriate twenty-three screening questions, diagnostic guidelines for the provisional impression for psychiatric disorders adapted for use in primary care settings, and nursing management guidelines for CMD, SMD, IDD, and SUD, including handling side effects of medication and the red flags for the referral. This study aimed to develop and validate the CSP-N for the primary care nurses to screen, identify, and provide dual collaboration of care for the treatment, referral, and follow-up of the person with mental health problems. MATERIALS AND METHODS The CSP-N was developed and validated in four stages: 1) an extensive literature search; 2) focused group discussion and in-depth interviews with primary care nurses and experts; 3) development of the CSP-N Module; and 4) stabilizing the face and content validation, as explained in Figure 1. Figure 1 Stages of CSP-N development 1. Literature Search: An extensive literature search was done using various search MeSH terms such as screening, identification of mental illness, primary care nurse’s role, and management of the mentally ill from several databases such as PubMed, PsycINFO, Science Direct, and CINAHL for studies related to the role of the primary care nurse in screening and identification, referral, and management of a PMI. The researcher has referred to 182 studies and, in those, reviewed 57 various studies related to primary care nurses’ role in mental health issues. 2. FGDs and in-depth interviews with nurses and experts: After the literature review, a draft of the CSP-N was prepared in line with the clinical schedule for primary care psychiatry (CSP) for the medical officers and discussed individually with several mental health experts from the multidisciplinary team. Two FGDs were done with DMHP nurses (12 psychiatric nurses), primary care nurses (six field nurses) from Jigani PHC, and three in-depth direct interviews for 30 minutes each with the primary care nurses from Gottigere PHC [Table 1]. The sample size for the number of FGDs and in-depth interviews was decided upon data saturation during the qualitative phase. Table 1 Socio-demographic data of the participants for the FGD and in-depth interview (n=21) Variable Mean (S.D.) Age in years 32.95 (5.05) Education in years 15.67 (2.06) Experience in psychiatry (in years) 0.81 (0.87) Variables Category Frequency (%) Gender Male 3 (14.3) Female 18 (85.7) Educational Qualification ANM 7 (33.3) GNM 11 (52.4) BSc Nursing 3 (14.3) Cadre DMHP Community Nurse 6 (28.6) DMHP Psychiatric Nurse 6 (28.6) Primary Healthcare Officer 6 (28.6) Nursing Officer 3 (14.3) *ANM: auxiliary nurse midwife, GNM: general nursing and midwifery and DMHP: district mental health program The theme analysis was done using ATLAS-ti software, and the major themes raised from the FGD and direct interviews were used to develop the CSP-N. The major themes identified are explained in Table 2. Table 2 Major themes, subthemes, and verbatim of respondents during the FGDs and interviews (n=21) Themes Subtheme Verbatim 1) Understanding the need for mental health care in primary health care. Unwillingness to seek treatment in mental hospitals. Early identification can reduce the treatment gap. Stigma and misconceptions lead to delay in availing treatment. Lack of specialists in the remote areas. “Even if we ask them to take the patient to NIMHANS, many are not willing or agree to take the patient…. They are worried that if it is known to someone in the village, it will spoil their life” (D.I-3, N.O-6). “Our people are more interested in going to “Devasthana and Masjid” than to the hospital.” (C.N-4, C.N- 5). 2) Nurses’ experience in screening, referral, and follow-up of PMI. Lack of confidence in identifying a PMI. Lack of motivation to do the screening and extra work. Able to identify serious mental illness. “… very busy with our regular work related to the RCH programme and not looking much at the mental health aspects …”(C.P-3, C.P-5, D.I-2). “….focus only on patients who are…very violent or depressed… not taking food properly…. We are seeing more common problems like alcoholism and wandering behaviour.”(DI-3, CN-5, CP-4). 3) Confidence and educational preparedness to manage mental health problems. Fewer opportunities exist during diploma in nursing. Lack of knowledge on mental illness and its management. “… just GNM qualified, and … very little theoretical knowledge about mental health disorders… can identify the patients who have some serious problems: violent behaviour, very odd Behaviour, wandering, talking and smiling to oneself or some hallucinations…. we do not know much about other illnesses, and not get much time to focus specifically on mental health aspects….”(NO-3). 4) Management of PMI. Medication adherence and compliance with treatment. Handling the side effects of medication. Handling expressed emotions. Handling psychiatric emergencies “……they will stop the medication once the symptoms are reduced and … family members also will not give much focus to give the medication correctly.”(CP-3, CP-6, DI-2, NO-4, NO-6). “….It is easy to handle the patients, but it is very difficult to convince the parents…They create more issues.”(CP-3, NO-5). “We can just give some counselling to the family members and encourage them to go to the hospital and give the medicine properly…”(NO-6). 5) Barriers in routine screening. Frustration due to the workload. Lack of knowledge on mental illness and its management. Personal, organizational, and job-related barriers. Time constraints lead to the neglect of mental health aspects. “As a part of the MCH program… so much work to do…. mainly focusing on the MCH and RCH programs…”(NO-1, NO-3). “submit various reports to the government on a weekly and monthly basis… and otherwise we will get scolded….”(NO-1, NO-4, DI-3). “All the programmes end up with the primary care nurses ….”(NO-3, DI-2). “….focus on mental health only on Tuesdays as a part of the Manochaithanya programme. not getting enough time to focus much on the mental health ….”(NO-3, NO-6). 6) Role of nurse in the prevention of relapse. Regular follow-up and home visits. Family education on handling PMI. The importance of drug compliance. Teach the family about the early signs of relapse. “….follow up will be done by the ASHA workers and . monitor for any worsening of symptoms…. If not improved, then ask them to revisit the PHC for consultation…” (NO-3). “…educating the family members about the illness … Give the medication properly and do not stop without advise….”.(CP-4, CN-2, NO-6). “….Once the illness started, only the family members would understand that the patient was not taking medicine. ….” (CP-4, CP-6, NO-5) 7) Referral services. Confirm the referral through ASHA. Refer to the DMHP psychiatrist if there is no improvement. “We are not specifically looking for any mental health issues… link with ASHA workers…. If there are any cases in their locality… do a home visit to. evaluate the main complaints… minor complaints will be given medicines… if unable to solve the problem, refer to the PHC … Do a follow-up with the ASHA workers .We rarely get mental health problems”…. (DI-1, NO-4). 8) Empowerment of nurses. Express the need for guidance. Showing interest in training. Enhance the confidence level in handling PMI. “Unable to focus … due to our lack of knowledge…”(NO-5). “If we get some training, it should be very useful for us…. less knowledge on what to do if we get any cases.”(CN-1, DI-2) “….Got some basic training on myths and mis- concepts about mental illness from Dr. Adarsh, DMHP Psychiatrist”(NO-1, NO-4, DI-3). 9) Support services. Family and financial support. Supply of free essential drugs in PHC. “. they do not have money to get even the bus ticket to go to NIMHANS . Sometimes we also feel so helpless and do not know what to do….”(DI-2). “Most of the time, the psychiatric medicines are unavailable in the PHCs” (NO-1, NO-2, DI-3). 10) Components to be included in CSP-N. Move from general to specific. Special focus on patient and family education and counseling. “.No time to read big manuals. need simple tips on how to handle them in the community…. no time to do all physical examinations and other things” (CP-3). “It is easy to find severe mental illness….”(CP-3, NO-5). “…We will say given counselling and family education for the sake of the name. But in reality, nothing has been done….” (CP-3, CN-2, DI-1, DI-2). *PMI: person with mental illness, FGD: focus group discussion, RCH: reproductive and child health, MCH: maternal and child health, PHC: primary health center, DMHP: district mental health program, and GNM: general nursing and midwifery course Development of CSP-N: Based on the review of the literature and the inputs from the FGD and in-depth interviews and discussions with the experts on the role of primary care nurses in screening, identification, referral, and follow-up of a person with mental health issues, the CSP-N developed under four modules explained in Figure 2. Figure 2 Modules of CSP-N Validation of CSP-N: The researcher approached 28 mental health experts from three universities, three mental health institutions, two medical colleges, and one community health center across India. Out of 28, only 17 mental health professionals from the multidisciplinary team sent their comments, and the remaining 11 did not respond [Table 3]. Table 3 Socio-demographic data of the experts who validated the CSP-N (n=17) Variable Experts for content validation Mean (S.D.) Age in years 39.53 (11.62) Education in years 24.41 (2.51) Experience in psychiatry (in years) 10.41 (9.45) Variables Category Frequency (%) Gender Male 8 (47.1) Female 9 (52.9) Occupation Psychiatrist 6 (35.3) Psychiatric nurses 7 (41.2) Primary care nurses 3 (17.6) Medical Officer 1 (5.9) The mental health expert evaluated each item of the CSP-N using a structured questionnaire to determine whether the contents described were applicable. For face validation, the mental health professional gave their opinion on appropriateness and relevancy on a three-point scale (1 = completely meets the criteria, 2 = partially meets the criteria, and 3 = does not meet the criteria) to calculate the I-CVI. The S-CVI was calculated using the average of the I-CVI of each item of the CSP-N.[13,14] After incorporating the suggestions from the experts, CSP-N Module Version 1 was developed. The item-wise content validity index of the CSP-N Version-1 was developed and explained in Tables 4 and 5. Table- 4 Expert opinion on face validity of CSP-N Modules (n=17) S.I No Evaluation criteria 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Expert agreement Item CVI 1. Socio demographic profile 1 1 1 1 1 1 1 2 1 1 1 2 1 1 1 1 1 15 0.88 2. Clinical Schedule for Primary care psychiatric Nursing (CSP-N) Module-  1. Relevant of the module to the study 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17 1  2. Content Organisation. 1 2 1 1 1 1 1 1 1 1 1 1 1 2 1 1 1 15 0.88  3. Clarity of the Items used. 1 1 1 1 1 1 1 2 1 1 1 1 1 1 1 1 1 16 0.94  4. Any Other Suggestions. 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17 1  2.1 Part-1 Case Record Form:  1. Relevant of the content to the study 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17 1  2. Content Organisation. 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17 1  3. Clarity of the Items used. 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 16 0.94  4. Any Other Suggestions. 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17 1  2.2.1 Part-II. Management Guidelines:  1. Diagnostic Guideline  1. Relevant of the content to the study 1 1 1 1 1 1 1 1 1 1 1 2 1 1 1 1 1 16 0.94  2. Content Organisation. 1 1 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1 15 0.88  3. Clarity of the Items used. 1 1 1 2 1 1 1 1 1 1 1 2 1 1 1 1 1 15 0.88  4. Any Other Suggestions. 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17 1 2.2.2  2. Nursing Management Guidelines:  1. Relevant of the content to the study 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17 1  2. Content Organisation. 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17 1  3. Clarity of the Items used. 1 1 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1 16 0.94  4. Any Other Suggestions. 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17 1 2.2.3  3. Non Pharmacological Management:  1. Relevant of the content to the study 1 2 1 1 2 1 1 1 1 1 1 1 1 2 1 1 1 14 0.82  2. Content Organisation. 1 1 1 1 2 1 1 1 2 1 1 1 1 1 1 1 1 15 0.88  3. Clarity of the Items used. 1 2 1 1 1 1 1 1 1 1 1 1 1 1 2 1 1 15 0.88  4. Any Other Suggestions. 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 1 1 16 0.94 2.3 Part-III. Follow Up Guidelines:  1. Relevant of the content to the study 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17 1  2. Content Organisation. 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17 1  3. Clarity of the Items used. 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17 1  4. Any Other Suggestions. 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17 1 2.4 Part-IV. Overview of Mental illness  1. Relevant of the content to the study 1 1 1 1 1 1 1 1 1 1 1 2 1 1 1 1 1 16 0.94  2. Content Organisation. 1 1 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1 15 0.88  3. Clarity of the Items used. 1 1 1 2 1 1 1 1 1 1 1 2 1 1 1 1 1 15 0.88  4. Any Other Suggestions. 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17 1 *I-CVI: Item level content validity index, Completely meet the criteria=1, Partially meet the criteria=2, and Does not meet the criteria=3 Table 5 I-CVI score and experts’ comments on each session of the CSP-N (n=17) Topic I-CVI Remarks by Validators Module-1: Case record form  a) Screening and identification of mental illness tool (SIM Tool): 20 questions very specific to the CMD (panic disorder, generalized anxiety disorder, somatization and depressive disorder along with suicidal risk), SMD (schizophrenia and mania), SUD (alcohol and tobacco addiction), and IDD. 0.96 a) To add more questions to the socio-demographic profile: 1. Have you ever identified and referred any PMI? b) EPS to be explained in full form as “extrapyramidal side effects.” c) To simplify the content.  b) Provisional impression: Make a provisional impression and refer to the medical officer. 0.98 a) Case vignettes can be used to discuss arriving at a provisional impression. Module-2: Management Guideline  a) Diagnostic guidelines: It briefly explains the signs and symptoms of each disorder in a very simplified manner and helps to form a provisional impression. 0.92 a) Detailed sessions need to be planned in the intervention session, and more focus should be given to the diagnostic guidelines with suitable examples.  b) Nursing Management Guidelines:   i. Nursing care on medication adherence and handling side effects: Common psychotropic medications’ dosage, side effects, and nursing management of side effects with red flags.   ii. Non-pharmacological guidelines: Counseling and family education are the keys to preventing relapse and ensuring medication compliance, which varies with each disorder. 0.98 0.88 a) Simplify the dosage and action and give more focus on the nurse’s role in monitoring and handling side effects. b) To add a “red flag” for referrals. c) To emphasize non-pharmacological management as it will be more beneficial to the nurse. Module-3: Follow-Up Guidelines  Assessment and frequency of follow-up. 1 Module- 4: Overview of Mental illness 0.92 Average I-CVI of the CSP-N (S-CVI) 0.95 *I-CVI: item-level content validity index, S-CVI: scale-level content validity index, CMD: common mental disorder, SMD: severe mental disorder, SUD: substance use disorder, and IDD: intellectual developmental disorder The face validity of the scale was calculated by sending it to experts, and for each item, the item-level content validity index (I-CVI) was computed as the number of experts who gave it a rating of 1 or 2, divided by the number of experts, and the scale-level CVI (S-CVI) was calculated by taking an average of the I-CVI. The S-CVI for the CSP-N was 0.95, very close to 1; it indicated the average proportion of items judged relevant across the 17 experts was 0.95, which is very good.[13,14] The final manuscripts were sent to the experts for evaluation after including all their comments and suggestions. Reliability test The test–retest reliability of the CSP-N was measured through the single-measure two-way mixed absolute agreement ICC for thirty-nine subjects recruited from the PHC. The test was administered through online google forms, and the participants filled out the Google form in the researcher’s presence. Thirty-two subjects completed the retest after 15-day intervals through online Google forms. Seven subjects did not complete the retest, so they were excluded from the process. The single-measure two-way mixed absolute agreement ICC value was calculated (for 32 subjects) for the reliability test, and the ICC value was 0.97 with 95% CI (0.94, 0.99). As this ICC value is very close to 1, it indicates that the test–retest reliability of the developed module is very good. RESULTS The present study aimed at developing the CSP-N, which helps and guides the primary care nurse to screen, identify, refer, and follow-up with the person with mental health issues, especially with a particular focus on CMD, SMD, IDD, and SUD. Network analysis was done using Atlas-ti version 8, and the themes and categories derived from the codes and quotations from all the transcribed documents grouped by code manager and network analysis helped to identify the common patterns and major themes between the codes explained in Figures 3 and 4. Figure 3 Barriers for nurses in screening and identification of mental illness Figure 4 Role of nurse in screening and identification of mental illness The ten major themes identified [Tabel 2] from the FGDs and direct interviews were used to frame the domains of the CSP-N and underwent various rounds of revision from various expert groups, including the nurses who work in primary care settings. The I-CVI calculations for the relevancy of each item are explained in Table 4. Most experts agreed on the content prepared for the CSP-N regarding the topic’s relevance, organization, and clarity of the content included in the schedule. All the expert suggestions, including the red flags for referral, were incorporated and revised. The item-level content validity index (I-CVI) of the CSP-N ranged from 0.82 to 1.00. The average scale-level content validity index (S-CVI) of the CSP-N is 0.95, which shows high content validity. It takes approximately 20–30 minutes to screen a person for any mental health issues using the CSP-N, and the comprehensive assessment yields specific targets or goals of counseling that can be tailor-made to suit the person with SMD, CMD, IDD, or SUD. The major strength lies in the scientific method used to develop the CSP-N through the iterative process and the simplified diagnostic criteria for the provisional impression. The CSP-N is a comprehensive schedule that helps the nurses screen the person for any mental illness and guides the nurses on where to refer, how often to do the follow-up and home visit, and what assessments must be done during the follow-up. DISCUSSION The nurses need to be trained and empowered with basic knowledge using simple guidelines to overcome the negative attitude, which may worsen the recovery of PMI due to stigma and discrimination.[15] It is one of the first studies of its kind to develop a simple module that helps the nurses in the primary care setting to screen, identify, refer, and do the follow-up of a person with mental health issues. The CSP-N demonstrated high content validity of 0.95 to screen, identify, refer, and follow-up on PMI. Unlike the previously designed questionnaires to screen for a particular condition like depression, the CSP-N is designed to rapidly screen for highly prevalent mental health disorders like CMD, SMD, IDD, and SUD in a primary care setting.[5,16] During the focus group discussion, the end users emphasized the importance of a simple and comprehensive module to screen for commonly seen mental health problems within the limited time available in their busy schedules, even though assessment tools were available in the resource book for the DMHP nurses.[17] Most of the screening tools, like the Patient Health Questionnaire (PHQ-9) for depression and the Screening Tool for Autism, are specifically focused on one illness, and a common screening tool to screen mental illness as a whole is the need for the hour during the busy schedule of primary care nurses. The CSP-N is tailor-made for primary care nurses based on the needs and suggestions of the nurses and further refined by the experts on the multidisciplinary team. It was adopted from the clinical schedule for primary care psychiatry (CSP) for doctors, with a high sensitivity of 91% and a fairly high specificity to detect mental illness.[4] The major strength of CSP-N was the qualitative strategy adopted for the schedule’s development that took into consideration the needs and requirements of the end users (nurses) as well as feedback on how to overcome the barriers while screening, identifying, and following up with people with mental illness in the community. As with any preliminary module, its design had some limitations. The study’s limitations include 1) the potential lack of generalizability, 2) the risk of using a self-reported measure, and 3) the length of the module. In therapeutic settings, CSP-N should be viewed as a knowledge booster rather than a competence enhancer. Therefore, rather than focusing on the skill-improving effects of a clinical training program, the results of this paper should be viewed as translating the CSP-Ns’ knowledge-enhancing effect into a clinical situation. Although the CSP-N was designed for the primary care nurses who work in the PHC and sub-centers, their generalizability to other nurses in a different setting is unknown and must be tested. There is a risk of recall bias or inflated answers in the self-reported measures due to the high workload among the nurses. The CSP-N Module also takes about 25 to 30 minutes to complete. The next step will be the validation and field testing of the CSP-N with concurrent validity and inter-rater reliability with a larger sample to improve the clinical skill-based training of primary care nurses in primary care settings. CSP-N refinement would also be necessary to improve its usefulness, effectiveness, and acceptability by primary care nurses working in real-life community settings. The module also should be prepared in various local languages for extensive utilization by primary care nurses. CONCLUSION The CSP-N is designed to use the ( Appendix-1) by primary care nurses in screening and identifying people with mental health issues in primary care, thereby preventing the delay in the PMI reaching mental health professionals. Empowering and equipping primary care nurses through CSP-N can become a powerful strategy to bridge the treatment gap present in the mental health area in developing countries like India. Ethical clearances This research project received ethical clearance from the NIMHANS IEC, Ref No. NIMH/DO/IEC (BEH.Sc. DIV)/2018 Dated: 17/12/2018 with approval from the TAC, Govt. of Karnataka, Ref No. D.D./Mental Health/50/2019-20. Permission was obtained from the DHO and MO of each PHC. The ICF was obtained from all the mental health experts and nurses, and confidentiality was maintained. Ethics Approval Ref No NIMH/DO/IEC (BEH.Sc. DIV)/2018 Dated: 17/12/2018 and approval from the TAC, Govt of Karnataka, Ref No. DD/Mental Health/50/2019-20. Permission was obtained from the DHO and MO of each PHC. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. Acknowledgements The author acknowledges the support and cooperation from the nursing staff at the primary care centers in Ramanagara, DMHP nurses of the state of Karnataka, India. APPENDIX-1: SCREENING AND IDENTIFICATION OF MENTAL ILLNESS TOOL (SIM TOOL) ==== Refs REFERENCES 1 Math SB Chandrashekar CR Bhugra D Psychiatric epidemiology in India Indian J Med Res 2007 126 183 92 18037711 2 Pradeep BS Gururaj G Varghese M Benegal V Rao GN Sukumar GM National Mental Health Survey of India, 2016-Rationale, design and methods PLoS One 2018 13 e0205096 30359382 3 Kohn R Saxena S Levav I Saraceno B The treatment gap in mental health care Bull World Health Organ 2004 82 858 66 15640922 4 Kulkarni K Adarsha AM Parthasarathy R Philip M Shashidhara HN Vinay B Concurrent validity and interrater reliability of the “clinical schedules for primary care psychiatry.” J Neurosci Rural Pract 2019 10 483 8 31595121 5 Shia N The role of community nurses in the management of depression Nurse Prescr 2009 7 548 54 6 Fleury MJ Imboua A Aubé D Farand L Lambert Y General practitioners'management of mental disorders: A rewarding practice with considerable obstacles BMC Fam Pract 2012 13 19 22423592 7 mhGAP Intervention Guide for Mental, Neurological and Substance Use Disorders in Non-Specialised Health Settings: Mental Health Gap Action Programme (mhGAP): Version 2.0 Geneva World Health Organization 2016 8 Saxena S Thornicroft G Knapp M Whiteford H Resources for mental health: Scarcity, inequity, and inefficiency Lancet Lond Engl 2007 370 878 89 9 Kakuma R Minas H Van GN Dalpoz MR Desiraju K Morris JE Human resources for mental health care: Current situation and strategies for action Lancet Lond Engl 2011 378 1654 63 10 World Mental Health Report: Transforming Mental Health for All World Health Organization Geneva 2022 14 109 11 Ohri U Nirisha PL Poreddi V Manjunatha N Kumar CN Math SB Dual clinical collaborator: A pragmatic role of nurses from developing countries Investig Educ Enferm 2022 40 e01 12 Patel V Cohen A Mental health services in primary care in developing countries World Psychiatry 2003 e6842202 13 Polit DF Beck CT Owen SV Is the CVI an acceptable indicator of content validity?Appraisal and recommendations Res Nurs Health 2007 30 459 67 17654487 14 Polit DF Beck CT The content validity index: Are you sure you know what's being reported?Critique and recommendations Res Nurs Health 2006 29 489 97 16977646 15 Reilly S Planner C Hann M Reeves D Nazareth I Lester H The role of primary care in service provision for people with severe mental Illness in the United Kingdom PLoS One 2012 7 e36468 22615769 16 Kennedy CW Polivka BJ Chaudary R Public health nurses'role in the care of adults with mental disabilities Psychiatr Serv 1997 48 514 7 9090736 17 Gandhi S Nattala P Radhakrishnan G Jothimani G Resource Book for Karnataka District Mental Health Programme- Psychiatric Nurses Bengaluru NIMHANS Publication No-147 2018 51 2

PMC010xxxxxx/PMC10353681.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-478 10.4103/ijcm.ijcm_837_22 Original Article Parents’ Caring Approach for Their Children Suffering from Pneumonia—A Study among Bhil Tribes of Maharashtra Kulkarni Prashant S. Kurane Anjali D. Department of Anthropology, Savitribai Phule Pune University, Pune, Maharashtra, India Address for correspondence: Prashant S. Kulkarni, Department of Anthropology, Savitribai Phule Pune University, Pune, Maharashtra, India. E-mail: prashant1505@gmail.com May-Jun 2023 30 5 2023 48 3 478482 08 10 2022 19 4 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Context: In India, pneumonia deaths in the past decade show a decreasing trend in the child mortality rate from 74.6 to 45.4. However, NFHS-5 records an increase in prevalence to 2.8% from 2.7% in NFHS-4. The childhood pneumonia control strategies focus on strengthening the health system, skill-building health workers, counseling, and creating awareness about promoting healthy behaviors regarding the management of sick children. Aims: The study attempts to understand “caregivers” care-seeking behavior and the management of childhood pneumonia. Setting and Design: The study was conducted in the Akkalkuwa block of Nandurbar district, Maharashtra, India. We used episodic interviews, asking caregivers to recollect specific events linked to the need for treatment. Methods and Material: A total of 11 in-depth interviews of mothers were conducted whose under-five children had pneumonia in the past year. These interviews used vignettes from real pneumonia cases to discuss community priorities for health care and actions taken to improve child’s health. In addition, the qualitative data from the in-depth interviews were thematically analyzed. Result: Cough, breathlessness, and disturbance in the routine schedule of the child were the major symptoms to identify pneumonia (vavlya) among children. Branding on the stomach, oil massage, and jadi-buti were commonly observed phenomena to seek help. Low priority, the influence of traditional healers and herbal medicines, and the inaccessibility of quality healthcare services were the main factors that led to the child’s treatment. Conclusion: Culturally appropriate activities are to be imparted on recognition of symptoms and appropriate care seeking, and community health workers need capacity building. Care seeking Caregivers India Tribal Under-five pneumonia ==== Body pmcINTRODUCTION Pneumonia deaths in the past decade show decreasing trend in child mortality rate from 74.6 to 45.4. However, NFHS-5 records 2.8% of the prevalence of ARI increase from 2.7% reported in NFHS-4. Under-five mortality in India remains a challenge in achieving Sustainable Development Goals.[1,2] Most of these childhood deaths are preventable by simple interventions, including maintaining hygiene and early and appropriate treatment of common childhood illnesses. Maharashtra is a middle-priority state considering the morbidity and mortality of under-five pneumonia.[3] Pneumonia among tribal children adds to their risk and vulnerability. Extra efforts have been made by making tribal-specific policies and programs.[4-6] Increasing community participation in health is a key priority area of the National Health Mission in India, and a part of UNICEF recommended Integrated Management of Neonatal and Childhood Illnesses (IMNCI) to reduce long delays in care seeking for children. Further, IMNCI focuses on community-based care and increasing community health workers’ skills to have early referrals.[7,8] Caregivers are the primary care providers to their children, so their knowledge becomes vital in preventive efforts. Barriers to effective care for children with pneumonia have been documented as contextual.[9-11] The current study aims at explaining knowledge, perceived causes, and help-seeking experiences of the Bhil tribal community toward childhood pneumonia. METHODS We conducted in-depth interviews (IDIs) of caregivers between March and October 2021 using a semi-structured interview schedule to understand perceptions and experiences of childhood pneumonia. These IDIs were conducted by PK with assistance from AV and were pilot tested before conducting the main interviews. The interviewers completed a master’s degree in social sciences with experience in conducting qualitative interviews. Setting This study was conducted in Akkalkuwa block of Nandurbar district of Maharashtra State in India. Nandurbar has a population of 16.48 lakhs with a 0–6-year population of 240,222.[12] Akkalkuwa block has 13 PHCs. Public health care is the main source of seeking health and private facilities at the block level. It is a tribal-dominant and inaccessible block of the district with larger percentages of the population living below the poverty and the lowest Human Development Index.[13-15] Study design The present study used a descriptive phenomenological design following and applying Husserl’s philosophical underpinnings that the “lifeworld” is understood as what individuals experience pre-reflectively, without resorting to interpretations.[16] Husserl’s approach has been adopted to explore caregivers’ experiences concerning episodes of pneumonia among their under-five children. We conducted one-time episodic interviews with caregivers of under-five children who were known to have had pneumonia in the previous 12 months. The approach led to discussion asking about specific experiences linked to a meaningful life event (episode of pneumonia), differentiating into various stages of care seeking and referral, perceived understanding of the illness, decision-making process, and treatment and care-seeking behavior. Recruitment Respondents having under-five children from the 35 villages of sampled health facilities were recruited from Akkalkuwa block using systematic random sampling. Data collection We conducted eleven in-depth interviews (IDIs) with caregivers out of 107 mixed survey participants selected using systematic random sampling. Respondents who reported an episode of under-five children in the last year from the date of the interview were selected for a further in-depth interview. Face-to-face interviews using a semi-structured interview schedule lasted for 35–40 minutes. No participant declined to be interviewed. The caregiver interviews were conducted with mothers of under-five children. Data collection was completed by a qualitative researcher assisted by a local assistant who was well-versed in tribal dialect. None of the data collectors was related to the study participants. The respondents were informed about the purpose of the study, risks and benefits of participating in the study, and written informed consent was sought. The in-depth interviews were conducted at locations convenient for the caregivers, mostly in their houses. Data Management and Analysis: Open-ended data from semi-structured interviews were recorded as notes in the interview schedule and later expanded. The interview schedule/notes were translated into English from the tribal dialect/local language. After several readings of the textual data, a thematic coding system was prepared by PK to enable a detailed description of the phenomenon under study. The themes included the identification of childhood pneumonia, help-seeking patterns, the decision-making process to seek care for pneumonia, experiences with providers, preventing pneumonia recurrences, etc. Interviews were coded manually. All the codes/narration were checked for consistent interpretation with local context by PK. The coded segments were used to describe the phenomena of childhood pneumonia, detailing its meanings, experiences, and help-seeking behaviors. Data were analyzed using the framework method incorporating relevant new data into the final description of the essence of the experiences of the phenomenon. Conceptual framework The social representation theory was used to map the complexity of the knowledge system presented by participants. The theory explains the development and operation mechanism of everyday knowledge and has been applied to various health knowledge in different countries. The theory entails that all knowledge is socially constructed, and social representations are the interplay between social structure and individuals. In other words, it is a group of people representing a common thought process and thought systems with common sense. Ethical issues addressed Written informed consent was obtained from all the respondents maintaining confidentiality and anonymity. RESULTS Profile of Caregivers A total of 11 out of 107 surveyed respondents reported an incidence of pneumonia in the past year. Seven respondents were between 18 and 25 years, and four were between 26 and 35 years. Seven respondents had ever attended school. Three of them have studied primary schooling, three respondents have studied secondary schooling, and one has studied junior college. Out of eleven children, six were male and five were female. Identifying the illness as pneumonia The respondents applied different criteria to identify the illness as pneumonia in addition to cough/cold and breathlessness, such as crying continuously (n = 6), stomach in-drawing/extended/pain (n = 6), not sleeping/eating/drinking milk well (n = 7), and vomiting/low voice during the talk (n = 1). Among 11 caregivers whose children suffered from pneumonia in the last year, most of them could not state the danger signs of pneumonia.1 The IMNCI strategy expects strengthened family and community care practices and their involvement in health care. Help-seeking pattern Ten out of eleven approached traditional healers and applied herbs mixed with goat’s excreta on a child’s chest, three of them approached lady oil masseurs [Figure 1], and two of them approached an herbalist at Kukarmunda (a village in Gujarat State) who suggested to consume herb to be consumed with water as a first help-seeking measure. Figure 1 Oil massage by traditional healer Six out of eleven had also approached private healthcare facilities, whereas three of them went to a government health facility as an alternate treatment source, and two reported that they only approached herbalists/traditional healers. [Table 1] Table 1 Help-seeking behavior of caregivers Case Help-Seeking Behavior Case-1 He was given chataka (branding on chest) thinking he will be cured, but he got puss there so took him to the clinic. Case-2 She was ill, so we first went to herbalist and worshipped God. Her health was improved. Now we are going to offer cocks to him. Case-3 She [child] was crying very much, so a neighbor asked to go there (Udaipur, a neighboring village). She used a thread of cloth to burn , then my baby got cured, stopped crying, and she is well now. Case-4 We went to PHC, given three days of medicines, but they were not useful; we went to private at Khapar, but no improvement. Then I went to Bhagat, and he put a burnt on his chest and gave jadi-buti for three days, and then he got cured. Case-5 We went to Bhagat, and our child got cured. Case-6 I went on the first day to the doctor after applying herbs, and it showed some effect, so I went on the second and third days. Still going there; now he started playing, so I feel it worked. Case-7 Taken medicines for two days but did not show effect, so we went there. Case-8 Tied pala (herbs) and went to the lady who gives oil massage (Kali Bai), he was still crying, so we took him to the doctor. Case-9 The child was crying, he was not eating, so we went to Kukarmunda; he gave us herbal medicine, then tied pest of herb and goat excreta. Case-10 She cried a lot, had a cold cough, low sound, and pain. Finally, I tied pala at home; she was cured, so I did not have to go anywhere. Case-11 We were told by neighbors that a woman works for pneumonia in the village, and many children went to her, so we went there, and it worked for her. “I went to the government hospital first with my one and half years old son, but he (her son) did not recover, so I approached lady for giving him oil massage” (Caregiver-1) All the community people residing in the hilly and plain areas show a similar line of thinking about causes of getting pneumonia in children, that is, causes of food ingestion. The causes associated with the belief are that (lactating) women eat something “wrong,” meaning which causes gases in the stomach (vegetables such as brinjal, cowpea (chawali), buffalo milk, cluster beans, etc.). Therefore, pneumonia is perceived as stomach-related illness rather than distress in breathing. It could be due to the fast movement of the stomach being given extra attention than shortness of breath. It was also observed that the causes were also associated with “spoiled breast milk”: (dudh fato/ubalya). Decision to seek help Continuous crying or not playing and not drinking milk were the reasons that the caregivers sought outside help after consultation with family or community members. Eight respondents mentioned that there were some medicines given to cure the illness. Three of them stated that some syrup was given, and one mentioned that cough syrup and injections were given. The rest five of them did not know the name of the medicines. Two of them mentioned that only the “herbs” pest (Areni paala and goat’s excreta) was tied, and the child was branded on the chest with a heated rod. [Figure 2] Figure 2 Branding by hot rod on stomach to cure pneumonia Out of nine respondents, three stated that the medicine was given for four days, five of them said that medicines were given for 5–10 days, and one of them mentioned that she gave the child medicines for 15 days. Except for one, others followed the advice of giving medicines for the said days. One respondent gave medicine for one day and was prescribed medicines for 15 days. She mentioned that the illness seemed not curing, so she discontinued the dose of medicines. All the respondents gave credit to the recent health providers whom they visited because post-visit to him/her, their child started playing or started “normal” life. JOURNEY OF A CHILD WHO SUFFERED FROM PNEUMONIA My son had pneumonia when he was just 21 days old. I delivered at Somaval PHC. He was a low-birth-weight baby weighing 2 kg 200 grams. I was not asked specific things [like prohibiting particular food] to do to increase his weight. They only asked to keep the baby warm. So, I kept him in ‘’’jholi’ and used to (bujhlo) breastfeed him frequently (bujhlo means breast). I (the mother) have a habit of chewing Vimal gutka, but it did not affect breastfeeding. (The baby) had sleeplessness, shortness of breath (madhalya madhe shwas ghet hota), stomach became tough (watdi). First help seeking- First, I went to PHC (government) immediately- I did not feel a cure; they (PHC staff) did not get him admitted and gave syrup and asked for nebulization somewhere else. Then, we went to (place) to the Private Doctor; he also did not look properly, so we went to Akkalkuwa (place) to another private doctor, Dr XXXX (private provider). PVT dr. advised the baby to sleep on their stomach. Why the decision was taken to shift to private: no change in the ‘ ‘child’s health situation; we roamed various places for around 3-4 days. Finally, the private dr asked to go to xxxx dr. Doctor asked not to give a bath to the baby. I came to know that it was pneumonia from Dr XXXX. The child was admitted for three days there. His breathing was unstable. If I had been late by 2-3 days, the child would have been on oxygen. I was there for three days. I also applied the excreta of the goat simultaneously so that it at least showed some effect on his health. Total expenses: Spent a total of around Rs. 6000/-. We could not afford to pay this much. At that time, my mother helped me financially. [Caregiver 3] Experience with the provider Each respondent shared the reason for shifting of health provider as “child does not feel ‘cured’.” The explanation given for discontinuing drugs/herbal medicines were; because of no effect of either of the medicines and two treatment measures [herbal & allopath] were used so that one would work [Table 2]. Table 2 Experience with the health provider No effect after giving medicine, so we went to Kali Bai (lady) for massage and felt better Other illnesses get cured, but he [private doctor] might not have experience in treating vavlya (pneumonia). He [child] was well, improved now. The doctor [private] gave an injection. I went to the doctor, given medicines and injection, and Kargir (herbalist) at Kukarmunda (place) gave jadi-buti, I also went to my natal place, he was given branding/chatka, and now he is playing (˜ cured). Preventing pneumonia recurrence Five out of eleven respondents said that it is difficult to prevent getting pneumonia because, according to them, suffering from pneumonia is not in our hands [destiny]. Two of them suggested avoiding eating cold food or being in a cold environment. Three of them prefer to restrict eating food items which can lead to pneumonia. DISCUSSION The present study documented the low awareness regarding the danger signs of childhood pneumonia. Cough, breathlessness, and disturbance in the routine schedule of the child (continuous crying, unable to drink milk/water, and vomiting) were the major symptoms reported to identify pneumonia (vavlya) among children. A study by Amuka et al. in rural and urban slums in Kenya also reports inadequate knowledge regarding pneumonia causes, risk factors, and prevention. However, another study assessing the knowledge of mothers of Mithi Tharparkar desert in Pakistan regarding ARI indicates high knowledge regarding ARI symptoms and preventive measures.[17] The shopping of providers could be because of no guidance regarding what pathways to follow for treating childhood pneumonia. The current study showed preferred traditional medicines as an initial preferred measure of treatment. The main factors that led to the child’s treatment were the influence of traditional healers, belief in herbal medicines, low educational level, and less availability of quality healthcare services. Branding on the stomach by Bhagat (Shaman), oil massage by a lady massager, and jadi-buti given by an herbalist were commonly observed phenomena to seek help. Traditional knowledge of herbs continued belief in herbalists, and delayed approach to Allopath medicines pushes them to use traditional medicines. Various studies have identified a gap in services utilization, provider practices, and family practices in seeking care.[7,18,19] Continued belief in traditional healers and herbal medicines was prevalent in the study area. In the present study, most caregivers’ first help-seeking contacts were traditional healers. The study villages had shown less belief in the public health sector could be because of inaccessibility and prior negative experiences. The perceived cause was correlated with food especially taken during the lactation period. A study conducted by Kulkarni et al.[20] in tribal areas of Jharkhand on tuberculosis denotes local liquor as the perceived cause of getting TB. Durkheim’s structural-functionalism theory focuses on “social solidarity” and “social facts.” The theory argues that functional differentiation is a prerequisite of society because it strengthens the sense of social solidarity and binds people together, each depending upon the functional services of all to maintain social order. Similarly, in the present study, a pattern of social solidarity of the Bhil tribal group of Akkalkuwa was observed in deciding on a child’s health. The influence of community members when making decisions regarding seeking healthcare providers was observed. Durkheim further argues that ceremonies and ritual practices are of key concern. Through rituals, the members of society are brought together, strengthening their bond. Along similar lines, the pattern of thinking regarding defining the meaning of illness (pneumonia aka Vavlya), correlating the causes with most emphasizing/indicating signs and symptoms, continued belief in herbal medicines or branding on the stomach with a rod, or going for oil massage by a traditional massager could be the outcome of following ritual practices for curing pneumonia.[21] A study assessing caregivers’ knowledge of under-five children regarding pneumonia care in Uganda reflects use of herbs as an inexpensive remedy.[22] The social institution has shown an important role in inculcating values among the individuals of that society. According to Lévi-Strauss’s theories of structuralism, universal patterns in cultural systems are products of the invariant structure of the human mind., that is, our mind takes our varied and potentially chaotic experiences and attempts to logically structure them along binary configurations, utilizing and incorporating the dominant images and symbols that we observe in our world.[23] In this study, the Bhil community group tried to seek help from traditional, private, and, at times, public sector facilities. However, the strong belief in causes being related to food leads them to follow the advice of traditional healers, following food taboos impacting in a child’s cure.[24] CONCLUSION This study has provided insights into the tribal community’s perceptions and care-seeking experiences for childhood pneumonia. The findings suggest the need for community-based approaches to improving caregiver knowledge and care seeking for under-five pneumonia. Messaging should be in the Bhili language and include knowledge of symptoms, risk factors, and community responsibilities in healthcare service delivery and utilization. Findings suggest the need for government interventions that can reduce the potential impacts of care seeking on household finances. It also indicates that the healthcare providers need to be sensitized regarding understanding the community psyche and plan the healthcare delivery accordingly. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. Acknowledgement The authors would like to thank Mr Ashok Vasave (AV) for assisting in facilitating and conducting interviewers of caregivers in the tribal dialect (Bhili). A child aged 2–59 months with cough and/or difficulty in breathing for less than two weeks of duration plus fast breathing and/or chest in drawing ==== Refs REFERENCES 1 NFHS-4 Factsheet India Available from: http://rchiips.org/nfhs/pdf/nfhs4/india.pdf [Last accessed on 2022 May 03] 2 RCH, “NFHS-5 Nandurbar.pdf.” Available from: http://rchiips.org/nfhs/http://rchiips.org/nfhs/NFHS-5_FCTS/MH/Nandurbar.pdf [Last accessed on 2022 Apr 15] 3 Gothankar J Doke P Dhumale G Pore P Lalwani S Quraishi S Reported incidence and risk factors of childhood pneumonia in India: A community-based cross-sectional study BMC Public Health 2018 18 1111 30200933 4 World Bank Improving Health Services for Tribal Populations Available from: https://www.worldbank.org/en/news/feature/2012/02/28/improving-health-services-for-tribal-populations [Last accessed on 2022 Oct 06] 5 8945026795Draft Tribal Development Plan (RTI ACT, 2005).pdf 6 Kulkarni P Kurane A Review of under-five children's health policies and programmes in India with reference to tribal population Man India 2019 99 231 40 7 Mathew JL Patwari AK Gupta P Shah D Gera T Gogia S Acute respiratory infection and pneumonia in India: A systematic review of literature for advocacy and action: UNICEF-PHFI series on newborn and child health, India Indian Pediatr 2011 48 191 218 21478555 8 Pneumonia in Children Statistics-UNICEF DATA Available from: https://data.unicef.org/topic/child-health/pneumonia/ [Last accessed on 2022 Jul 12] 9 Bakare AA Graham H Agwai IC Shittu F King C Colbourn T Community and caregivers'perceptions of pneumonia and care-seeking experiences in Nigeria: A qualitative study Pediatr Pulmonol 2020 55 (Suppl 1) S104 12 31985894 10 Pajuelo MJ Anticona Huaynate C Correa M Mayta Malpartida H Ramal Asayag C Seminario JR Delays in seeking and receiving health care services for pneumonia in children under five in the Peruvian Amazon: A mixed-methods study on caregivers'perceptions BMC Health Serv Res 2018 18 149 29490643 11 Rahman SA Kielmann T McPake B Normand C Healthcare-seeking behaviour among the tribal people of Bangladesh: Can the current health system really meet their needs J Health Popul Nutr 2012 30 353 65 12 District Nandurbar |Government of Maharashtra |India Available from: https://nandurbar.gov.in/ [Last accessed on 2022 Nov 19] 13 Maharastra Human Development Index. MPSC Notes, Dec 13, 2017 Available from: https://maharashtra.pscnotes.com/education/maharastra-human-development-index/ [Last accessed on 2022 Nov 19] 14 Jayakumar T Can Maharashtra be India's California?† 2020 15 Gavit SD A geographical study of tribal families below poverty line in Nandurbar District (State of Maharashtra) J Interdiscip Cycle Res 2022 XII 2397 401 16 Dowling M From Husserl to van Manen. A review of different phenomenological approaches Int J Nurs Stud 2007 44 131 42 16412442 17 Amuka DO Onguru D Ayodo G “Knowledge, Perceptions and Practices of Caregivers on Pneumonia among Children aged below 5 years in Migori County Referral Hospital, Kenya,” Int. J. Health Sci. Res 2020 10 40 57 18 Guindon GE Lavis JN Becerra-Posada F Malek-Afzali H Shi G Yesudian CA Bridging the gaps between research, policy and practice in low- and middle-income countries: A survey of health care providers Can Med Assoc J 2010 182 E362 72 20439448 19 Cowling K Dandona R Dandona L Social determinants of health in India: Progress and inequities across states Int J Equity Health 2014 13 88 25294304 20 Kulkarni P Kudale A Arasu K Lab M Darby W Rangan S Tuberculosis knowledge and awareness in tribal-dominant districts of Jharkhand, India: Implications for ACSM Public Health Action 2014 4 189 94 26400809 21 Gangwar S Structural functionalism: Definition, theories and criticism Sociology Group: Sociology and Other Social Sciences Blog, May 25, 2021 Available from: https://www.sociologygroup.com/structural-functionalism-meaning-theories/ [Last accessed on 2022 Apr 14] 22 Tuhebwe D Tumushabe E Leontsini E Wanyenze RK Pneumonia among children under five in Uganda: Symptom recognition and actions taken by caretakers Afr Health Sci 2014 14 993 1000 25834512 23 Structuralism |anthropology |Britannica Available from: https://www.britannica.com/science/structuralism-anthropology [Last accessed on 2022 Apr 14] 24 Bhasin V Medical anthropology: A review Stud Ethno-Med 2007 1 1 20

PMC010xxxxxx/PMC10353682.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-390 10.4103/ijcm.ijcm_577_22 Review Article Systemic Review and Meta-Analysis to Evaluate Therapeutic Effectiveness of Interferon Beta-1b in Hospitalized COVID-19 Patients Nayudu Greeshma Sai Sree Benny Mamkoottathil Thomas Grace Adil Khan Maria Basutkar Roopa Satyanarayan Department of Pharmacy Practice, JSS College of Pharmacy, JSS Academy of Higher Education and Research, Ooty, The Nilgiris, Tamil Nadu, India Address for correspondence: Dr. Roopa Satyanarayan Basutkar, Department of Pharmacy Practice, JSS College of Pharmacy, JSS Academy of Higher Education and Research, Ooty, The Nilgiris, Tamil Nadu - 643 001, India. E-mail: roopasatyanarayan@gmail.com May-Jun 2023 30 5 2023 48 3 390400 04 7 2022 01 3 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. The COVID-19 pandemic has caused havoc in the health sector. Inflammatory cytokines play an important role in the disease condition. Existing evidence has provided certain insights into the repurposing of the drugs. This meta-analysis and systematic review aimed to explore the efficacy of the administration of interferon beta-1b (IFN β-1b) and standard care versus only standard care as the therapeutic agent for managing COVID-19 patients who are severely ill. The search was conducted in the following databases: Cochrane Central Register of Controlled Trials (CENTRAL), PubMed, Scopus, and Google Scholar, which were published during the period January 1, 2020, to February 16, 2023. All the included three studies were independently assessed for eligibility. The modified data extraction form of Cochrane were used. The quality of the three included studies was assessed using the Cochrane risk of bias tool. GradePro software was used to summarize the quality grading of the primary outcome measures. The time taken for clinical improvement was (MD: -3.28 days; 95% CI: -5.65, -0.91; P value = 0.007) when treated with IFN β-1b. The duration of hospital stays (MD: -2.43 days; 95% CI: -4.45, -0.30; P value = 0.03), and need for intensive care unit (ICU) admission (RR: 0.71; 95% CI: 0.52, 0.97; P value = 0.03) was statistically significant. Interferon beta-1b is proven to reduce the duration of hospital stay, and the improved clinical status may become a cornerstone of COVID-19 treatment. Clinical improvement COVID-19 Hospitalization Hydroxychloroquine ICU admission Interferon Interferon beta-1b Lopinavir/ritonavir Mortality Ribavirin Safety ==== Body pmcINTRODUCTION The COVID-19 pandemic caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has wreaked havoc on humanity. The World Health Organization (WHO) declared the COVID-19 outbreak a pandemic on March 11, 2020.[1,2] According to current epidemiological data, the median incubation period is 6 days, and transmission can happen even before symptoms appear. Furthermore, asymptomatic cases, which account for a significant portion of infections, are likely to contribute to virus circulation.[3] All the positive COVID-19 cases are treated with supportive treatment, symptomatic treatment, oxygen therapy, respiratory and circulation support, and therapeutics. Warm baths and antipyretic medications such as ibuprofen and acetaminophen were used to treat severe cases of high fever as supportive treatment. Patients who are having trouble breathing should be given non-invasive mechanical ventilation (NIV).[4] Most management strategies seek to enhance viral clearance and inhibit the cytokine storm by reducing the need for long hospital stays, mechanical ventilation, and COVID-19-related mortality. Several options, ranging from prophylactic vaccines to targeted antiviral drugs, are considered for this purpose. Anti-inflammatory drugs such as hydroxychloroquine (HCQ), dexamethasone, tocilizumab, and chloroquine (CQ) have been recommended to minimize the release/production of pro-inflammatory cytokines to reduce the cytokine storm caused by SARS-CoV-2.[5] Remdesivir has in vitro activity against SARS-CoV-2. Other antivirals such as arbidol, oseltamivir, favipiravir, interferon beta-1a, darunavir, and cobicistat are under trials. Controversial studies are there for glucocorticoids because there was no improvement in the rate of radiographic recovery; hence, they are not recommended in mild cases. Early, low-dose, and short-term (1–2 mg/kg/d for 5–7 days) corticosteroids were linked to a faster improvement of clinical manifestations and absorption of focal lung lesions in severe COVID-19 cases[6]. Severe acute respiratory syndrome (SARS) and the Middle East respiratory syndrome (MERS) viral loads peak 7–10 days after symptom onset; however, COVID-19 viral loads peak at the time of presentation, similarly to influenza.[7] Inflammatory cytokines are the first line of defense against viruses. Interferons (IFNs) are divided into three families, each of which has several subfamilies. IFN-II is the only isoform with a single isoform 33: IFN-I [α, β, ω, ε, κ]; IFN-II (γ); IFNs-III and IFN λ (λ1, λ2, λ3, λ4). IFNs are naturally occurring anti-inflammatory proteins that bind to receptors on the surface of different cells and activate the Janus kinase (JAK)/signal transducer and activator of transcription (STAT) signaling pathway, which results in transcription of IFN-stimulated genes (ISGs) like the pro-inflammatory chemokine C-X-C Motif Chemokine Ligand 10 (CXCL10) and antiviral enzyme Ribonuclease L (RNase L).[8] Type I and III IFNs are genetically distinct, have different receptors, elicit similar pathogen detection sensors, and activate antiviral, antiproliferative, and immunomodulatory gene expression programs. IFN-l helps to reduce harmful inflammatory responses and limits viral spread from the upper respiratory tract to the lungs by contrasting viral replication in epithelial cells at the entry point. Additionally, by stimulating adaptive immunity, it protects the mucous barrier. Finally, it protects the barrier’s integrity by reducing inflammation and its harmful effects caused by neutrophil activation. Type I IFNs (IFN-α, IFN-β, IFN-ε, IFN-κ, and IFN-ω) bind to the transcription factor type I IFN receptor (IFNAR) in a paracrine and autocrine manner in humans. Type III IFNs bind to the type III IFN receptor, which is expressed preferentially on epithelial and certain myeloid cells. Type I and III IFNs both cause ISG expression to be induced and depleted, but type I IFN signaling causes ISG expression to be induced and depleted more quickly. IFN-α and IFN-β, primarily recombinant and pegylated, have been explored to treat various disorders, including multiple sclerosis and viral hepatitis.[9,10] In contrast to antiviral activity, IFN-β is much more effective than IFN α-2 at activating the antiproliferative program, a finding that was also confirmed using the IFN α-2 variant. As a result, a link is discovered between the administration of IFN β-1a and the improvement in the clinical course of COVID-19 disease. This finding was significant because it suggests that IFN could be a viable therapeutic option in severe COVID-19 cases. Patients admitted to hospitals with a high viral load suggest that a combination of antiviral drugs is more effective than single-drug treatments.[11] In all clinical samples, aerosol inhalation of IFN-k plus trefoil factor 2 (TFF2) in combination with standard care is proven safe and superior to standard care alone in reducing the time to viral ribonucleic acid (RNA) negative conversion.[12,13] Thus, an increasing number of researchers are focusing on the IFN treatment in COVID-19. Interferon beta-1b, ribavirin, and lopinavir/ritonavir were safer and more effective in reducing virus shedding, lessening symptoms, and allowing patients with mild to moderate COVID-19 to be discharged than lopinavir/ritonavir alone.[14,15] However, there is limited evidence to know about the impact of the administration of IFN β-1b and standard care on the prognosis of COVID-19 in severely ill patients. This meta-analysis and systematic review currently aimed to explore the efficacy of the administration of IFN β-1b and standard care role versus only standard care as the therapeutic agent for managing COVID-19 patients who are severely ill. METHODS Types of studies Randomized controlled trials (RCTs), open label. Eligibility criteria Inclusion criteria: RCTs of patients aged ≥18 years of both genders. SARS-CoV-2 patients confirmed the positive result of nasopharyngeal swabs using Real-Time Polymerase Chain Reaction (RT-PCR) with or without co-morbidities. Exclusion criteria: We excluded articles with a single arm and self-comparison studies and papers with mild and moderate infection of COVID-19 subjects. Papers comparing the other types of interferon along with standard care versus standard care alone. Types of interventions Intervention arm: IFN β-1b along with standard care that includes antiviral (lopinavir/ritonavir, ribavirin) therapy, corticosteroids, HCQ, and antibiotics. Control arm: The standard care of antiviral therapy (lopinavir/ritonavir, ribavirin), corticosteroids, HCQ, and antibiotics. Types of outcome measures Primary outcomes Time for clinical improvement from admission to discharge with IFN β-1b: Mortality at the end of the study Need for ICU admission Duration of hospitalization Hospital admission requiring the invasive and non-invasive mechanical ventilation. Secondary outcomes Safety outcome of the study: Electronic search The search was performed for relevant studies which were published from January 1, 2020, to February 16, 2023, on the following databases: Cochrane Central Register of Controlled Trials (CENTRAL), PubMed, Scopus, and Google Scholar. We searched the Clinical Trials Registry of India for ongoing studies. Manual searching was also conducted. The mesh terms “COVID-19, SARS-CoV-2, coronavirus, beta interferon, and interferon beta” were used. The search strategy is mentioned in Appendix 1. The retrieved article was imported into Zotero and converted into “ris” format for importing pooled studies into Rayyan. Appendix 1 Search methods for identification of the studies Population COVID-19 Intervention Interferon beta Comparison Control Outcome Efficacy and Safety (((COVID-19[Title]) OR (SARS-CoV-2 [Title]) OR (Coronavirus))) (((Interferon beta [Title]) OR (beta Interferon [Title])) OR (Interferon-beta [Title])) Lopinavir/ritonavir, ribavirin therapy, corticosteroids, hydroxychloroquine Efficacy [All Fields] COVID-19 Beta interferon Clinical effectiveness COVID-19 virus disease Interferon, beta Effectiveness clinical Coronavirus disease 2019 Fiblaferon Clinical improvement 2019 novel coronavirus disease Interferon beta Improvement clinal Coronavirus disease-19 Interferon, fibroblast Hospital stays Hospital admission COVID-19 virus infection Fibroblast interferon Duration 2019-nCoV disease Interferon, beta-1 Intensive care unit COVID-19 pandemic Beta interferon Mechanical ventilation SARS-CoV-2 infection Beta-1 interferon Mortality 2019-nCoV infection Beta-1 interferon Death 2019 novel coronavirus infection Interferon beta-1 Treatment effectiveness Interferon beta-1 Effectiveness, treatment Data collection and analysis Selection of population in the included studies of this systematic review Inclusion criteria: Patients ≥18 years of age from both genders. Patients with or without co-morbidities like diabetes mellitus, hypertension, coronary artery disease, and malignancy were included. Patient with the confirmed positive result of nasopharyngeal swabs using RT-PCR for COVID-19. Patients on the treatment of IFN β-1b and standard therapy, which includes antiviral, HCQ, and supportive therapy. Exclusion criteria: Patients with known neuropsychiatric disorders, thyroid disorders, and cardiovascular diseases. Patient on other types of different IFN therapy like IFN β-1a, IFN α-2b, etc. Consumption of potentially interfering medications with lopinavir/ritonavir + HCQ, IFN β-1b, or having a history of alcoholism, or any illicit drug addiction within the past five years. Pregnant patients and lactating women. Data extraction and management All records were assessed by the reviewers (GSS, BMB, GT, and MAK) to confirm eligibility in Rayyan. The modified data extraction from Cochrane CENTRAL was used. The following details were extracted: trial ID, general information, methods, participants, interventions, and outcomes. For each intervention and comparison group, a number of participants randomized into each group, description and duration of the treatment, and timing and medium of delivery were abstracted. Similarly, for each outcome, data like relevancy of outcome, the time points, and unit of measurement were reported. Mean and standard deviations were extracted for continuous outcome variables, whereas in some studies, the outcome measures were mentioned in the median and interquartile range from which the mean and standard deviation were derived using Hozo’s method.[16] The retrieved data were reviewed by the RSB. Assessment of risk bias and summary findings of the included studies The reviewers (GSS, RBS, BMB, GT, and MAK) separately examined the following domains using the risk of bias tool: Random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias), and other biases are all involved in determining whether studies have a registration number. With correct judgment and explanation, the above-mentioned biases were labeled as “low,” “high,” or “unclear” in a table. The data were used to create a graph and figure depicting the risk of bias. Any differences were sorted out by discussing with reviewer RSB. The overall quality grading of the outcomes, such as the size of the effect of the treatments employed in the studies and the accessible important information on the outcomes, was also examined by extracting the appropriate data into the summary of results table using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) technique. Registration The protocol of this systematic review and meta-analysis was registered prospectively with PROSPERO [Registration number: CRD42021284428]. Statistical analysis All the analysis was performed with the Review Manager version 5.4 (Cochrane Collaboration, Copenhagen, Denmark), which was used to produce forest plots with pooled estimates by importing the appropriate data. As the outcome measures were continuous, the overall effect size was calculated using mean and standard deviation values. The pooled estimates were calculated using the inverse variance approach, and both random and fixed effect models were utilized. Each outcome›s estimated mean difference (MD) and 95% confidence interval (CI) were graphically and statistically represented, along with the weight assigned by each study and the risk ratio was calculated for a dichotomous variable to know the ratio of the probability of the event to occur in the intervention group to that of the control group. Cochrane›s Q (P values) and Higgins I2 statistics, as well as ocular examination, were used to evaluate heterogeneity. The intertrial variability was represented in percentage with a P value. We used a funnel plot to assess the publication bias. RESULTS Search results Our search from various databases yielded a total of 66 studies, out of which 36 were duplicates. This resulted in getting 30 records, of these 26 studies were excluded due to reasons: The drug used as an intervention was different, and the population included was different such as patients with MERS and articles published in foreign languages. A full-text review was performed for four articles, out of which one report was removed as the route of administration of the drug was different. A set of three studies were included in the qualitative analyses. The complete search is illustrated in Figure 1. Figure 1 Flow diagram of the included studies in the review Characteristics of studies Characteristics of the included studies are provided in Table 1. The included studies were all open-label randomized controlled trials with an intervention group receiving the IFN β-1b plus standard care and the control group receiving only standard care. The IFN β-1b was given subcutaneously in all three studies. The drugs included in the standard care were mainly antiviral, that is, lopinavir and ritonavir. One of the three studies has not administered HCQ to the subjects. The included participants are adults, who had a positive RT-PCR result at the time of admission, and all the patients were having severe COVID-19 and were hospitalized. The primary outcome measure for the RCTs was to monitor the time for clinical improvement. This was measured using scales such as NEW2 and WHO 7 points scale with different parameters to measure improvement status with the therapy. Table 1 Characteristics table of included studies Author, Year Study country Study sample Study design Enrolled period Population Age Gender Interventions Co-morbidities Outcomes Alavi Darazam I et al. 2021[7] Iran 60 Investigator initiated, three-armed, parallel group, individually randomized, open labeled, randomized controlled trial. - Adult hospitalized severe COVID-19 Median and IQR 69 (55–82) Male: 51.7% and female: 48.3% Intervention group: IFNb1b (Ziferon) (subcutaneous injections of 0.25 mg (8,000,000 IU) on days 1, 3, 6) + Hydroxychloroquine + Lopinavir/Ritonavir (Kaletra) Control group: Hydroxychloroquine (single dose of 400 mg on day 1, orally + Lopinavir/Ritonavir (Kaletra) (400 mg/100 mg twice a day for 10 days) Diabetes mellitus, hypertension, coronary artery disease, chronic kidney disease, and malignancy Time for clinical improvement, duration of hospital stays, ICU admission, mechanical ventilation, and mortality. Rahmani H et al. 2020[15] Iran 80 Open-label RCT April 20, 2020 to May 20, 2020 Adult hospitalized severe COVID -19 Median and IQR 60 (50–71) Male: 59.09% female: 40.91% Intervention group: IFN b-1b 250 mcg subcutaneously every other day for two consecutive weeks, lopinavir/ritonavir (400/100 mg BD) or atazanavir/ritonavir (300/100 mg daily) plus hydroxychloroquine (400 mg BD in first day and then 200 mg BD) for 7–10 days. Control group: Lopinavir/ritonavir (400/100 mg BD) or atazanavir/ritonavir (300/100 mg daily) plus hydroxychloroquine (400 mg BD in first day and then 200 mg BD) for 7–10 days. Hypertension, diabetes mellitus, ischemic heart disease, asthma, COPD, malignancy, and transplantation Time for clinical improvement, duration of hospital stays, ICU admission, mechanical ventilation. and mortality. Hung IFN et al. 2020[14] China 127 Phase-2 multi-center open-label RCT February 10, 2020 to March 20, 2020 Hospitalized COVID-19 patients Median and IQR 52 (32–62) Male: 46% and female: 40% Intervention group: lopinavir 400 mg and ritonavir 100 mg, ribavirin 400 mg every 12 h, and subcutaneous injection of one to three doses of interferon beta-1b 1 mL (8 million international units [IU]) Control group: lopinavir/ritonavir (lopinavir 400 mg and ritonavir 100 mg) every 12 h for 14 days. Diabetes, hypertension, coronary artery disease, cardiovascular diseases hyperlipidemia, thyroid diseases, obstructive sleep apnea, and malignancy Time for clinical improvement, duration of hospital stay, and mortality. The participants included in the included RCTs had co-morbidities, mainly hypertension, type-2 diabetes mellitus, and ischemic heart disease. The majority of the selected studies have finished their trial for a shorter duration, and the long-term effects of the drug are not known. Risk of bias assessment The risk for random sequence generation was low in all three studies. The risk for blinding was found to be high, as the included studies were an open label. Selective reporting of primary outcomes measures was found to be low in three studies. The trial registration number was confirmed for two of the included trials, and it remains unclear for one study. The complete risk assessment of all articles is depicted as a graph [Figure 2 and Table 2]. Figure 2 Risk of bias assessment of the included study Table 2 Risk of Bias Table Alavi darazam I et al. (2020)[7] Methods Participants Intervention Outcome Investigator initiated, three-armed, parallel group, individually randomized, open labeled, randomized controlled trial Male, non-lactating, and non-pregnant female patients with at least 18 years of age who had confirmed COVID-19, defined as a positive test of Reverse Transcriptase Polymerase-Chain Reaction (RT-PCR). Intervention group: IFNb1b (Ziferon) (subcutaneous injections of 0.25 mg (8,000,000 IU) on days 1, 3, 6) + Hydroxychloroquine+Lopinavir/Ritonavir (Kaletra). Control group: Hydroxychloroquine (single dose of 400 mg on day 1, orally + Lopinavir/Ritonavir (Kaletra) (400 mg/100 mg twice a day for 10 days). TTCI, defined as the time from enrollment to discharge from the hospital or a decline of two steps on the seven-step ordinal scale. Secondary outcomes included mortality from the date of randomization until day 21. Safety outcome measures. Bias Author’s judgment Support for judgment Random sequence generation (selection bias) Low risk Randomization sequence generation was generated using package “randomized R” in R software version 3.6.1 Allocation concealment (selection bias) Low risk Unstratified randomization was performed in a 1:1:1 ratio utilizing a block balance randomization method. The permuted block [ three or six patients per block] and placed in individual sealed and opaque envelopes for allocation concealment by an outside statistician. Blinding of participants and personnel (performance bias) High risk Open label: The patient and the investigator are aware of the intervention given. Blinding of outcome assessment (detection bias) Low risk Outcome assessor was blinded to study arms Incomplete outcome data (attrition bias) Low risk All the participants who have undergone randomization their data were included for analysis Selective reporting (reporting bias) Low risk All outcomes’ measures were analyzed and reported Other biases Low risk No Hung IFN et al.(2021)[14] Methods Participants Intervention Outcome Phase 2, multi-center, Open-label and randomized controlled trial. Inclusion criteria: Age at least 18 years, a national early warning score 2 (NEWS2) of at least 1, and symptom duration of 14 days or less upon recruitment. Intervention group: lopinavir 400 mg and ritonavir 100 mg, ribavirin 400 mg every 12 h, and subcutaneous injection of one to three doses of interferon beta-1b 1 mL (8 million international units [IU]). Control group: lopinavir/ritonavir (lopinavir 400 mg and ritonavir 100 mg) every 12 h for 14 days. The primary endpoint was time to achieve a negative RT-PCR result for SARS-CoV-2 in a nasopharyngeal swab sample. Secondary clinical endpoints were time to resolution of symptoms defined as a NEWS2 of 0 maintained for 24 h; daily NEWS2 and sequential organ failure assessment (SOFA) score; length of hospital stay; and 30-day mortality Bias Author’s judgment Support for judgment Random sequence generation (selection bias) Low risk Patients were randomly assigned to the groups. Each serial number was computer generated. Allocation concealment (selection bias) Low risk Simple randomization with no stratification was used, and patients were assigned to a serial number by the study coordinator. Blinding of participants and personnel (performance bias) High risk Open label: The patient and the investigator are aware of the intervention given. Blinding of outcome assessment (detection bias) Low risk The blinding of outcome assessor was not mentioned. Incomplete outcome data (attrition bias) Low risk All the participants who have undergone randomization their data were included for analysis. Selective reporting (reporting bias) Low risk All outcomes’ measures were analyzed and reported one patient stopped on day 7 because of adverse events Other biases Low risk No Rahmani H et al. (2020)[15] Methods Participants Open labeled randomized controlled trial Adult patients (≥18 years old) with positive PCR and clinical symptoms/signs of pneumonia (including dyspnea, cough, and fever), peripheral oxygen saturation (SPO2) ≤ 93% in ambient air or arterial oxygen partial pressure to fractional inspired oxygen (PaO2/FiO2) < 300 or SPO2/FiO2<315 and lung involvement in chest imaging. Intervention Outcome Intervention group: IFN b-1b 250 mcg subcutaneously every other day for two consecutive weeks, lopinavir/ritonavir (400/100 mg BD), or atazanavir/ritonavir (300/100 mg daily) plus hydroxychloroquine (400 mg BD in first day and then 200 mg BD) for 7–10 days. Control group: lopinavir/ritonavir (400/100 mg BD) or atazanavir/ritonavir (300/100 mg daily) plus hydroxychloroquine (400 mg BD in first day and then 200 mg BD) for 7–10 days. Clinical status of the patients was assessed by the six-category ordinal scale at days 0, 7, 14, and 28 of the randomizations. Need for supplemental oxygen therapy and also invasive or non-invasive respiratory supports were evaluated regularly. Bias Author’s judgment Support for judgment Random sequence generation (selection bias) Low risk Participants were randomly assigned to the groups. A biostatistician who was not involved in this study did the randomization. Allocation concealment (selection bias) Low risk The method randomization was the permuted block randomization; six patients per block Blinding of participants and personnel (performance bias) High risk Open label: The patient and the investigator are aware of the intervention given. Blinding of outcome assessment (detection bias) Low risk The blinding of outcome assessor was not mentioned. Incomplete outcome data (attrition bias) Low risk All the participants randomized were analyzed in the study group Selective reporting (reporting bias) Low risk All the mentioned primary and secondary outcome were assessed Other biases Low risk No Clinical outcome measures Time of clinical improvement Analysis was performed for all three studies (n = 167) with similar interventional drugs, that is, IFN β-1b along with standard care which includes the antivirals like lopinavir, ritonavir, ribavirin, and the symptomatic treatment to that of only standard care provided to the control group. The COVID-19 patient who was severely ill and treated with IFN β-1b showed rapid clinical response, that is, negative RT-PCR or improvement of any two parameters in the seven-point scale or six-point scale when compared to the only standard care group (MD: -3.28 days; 95% CI: -5.65, -0.91; P value = 0.007). The intertrial variability among the included studies was found to be significantly high (χ2 = 62.56; I2 = 97%; p=<0.001) [Figure 3]. Figure 3 Forest plot showing the effect of interferon beta-1b on the time of clinical improvement versus standard care in COVID-19-infected patient Duration of hospital stay The analysis was conducted to know the duration of hospital stays, that is, the number of days the COVID-19-infected patients were admitted to the hospital. IFN β-1b significantly reduced the duration of the hospital stay when compared to standard care (MD: -2.43 days; 95% CI: -4.45, -0.30; P value = 0.03). On accounting for the degree of intertrial variance, it was found to be high with a statistically significant P value < 0.001 (χ2 = 83.62; I2 = 98%) [Figure 4]. Figure 4 Forest plot showing the effect of interferon beta-1b on the duration of hospital stay versus standard care in COVID-19-infected patient ICU admissions Two included studies have measured the number of patients admitted to ICU, who had experienced severe COVID-19 infection. The results of the analysis stated that the number of admissions to ICU was statistically less when treated with IFN β-1b (RR: 0.71; 95% CI: 0.52, 0.97; P value = 0.03). Low heterogeneity was observed among the trials, but the results are statistically insignificant P value of 0.41 (χ2 = 0.67; I2 = 0%) [Figure 5]. Figure 5 Forest plot showing the effect of interferon beta-1b on the need of ICU admission versus standard care in COVID-19-infected patients Hospital admission with invasive mechanical ventilation The studies have monitored the number of patients needing mechanical ventilation during the hospital stay as most patients included in studies have oxygen saturation (SpO2) <95%. The analysis of the two studies which were included depicts that there is no statistically significant difference between both the groups. The risk for invasive mechanical ventilation was not reduced when treated with IFN β-1b along with standard care (RR: 0.82; 95% CI: 0.39, 1.74; P value = 0.60). The intertrial variability was found to be significantly low (χ2 = 0.57; I2 = 0%; P = 0.45) for the included studies [Figure 6]. Figure 6 Forest plot showing the effect of interferon beta-1b on hospital admission with invasive mechanical ventilation versus standard in COVID-19-infected patients Hospital admission with Non-invasive mechanical ventilation As mentioned above, the studies have also measured the need for NIV between the groups. The analysis from the included studies revealed that there was no statistically significant difference between the groups (RR: 1.27; 95% CI: 0.51, 3.20; P value = 0.61). Low heterogeneity was observed among the trials but with a statistically insignificant P value of 0.57 (χ2 = 0.33; I2 = 0%) [Figure 7]. Figure 7 Forest plot showing the effect of interferon beta-1b on hospital admission with non-invasive mechanical ventilation versus standard care in COVID-19-infected patients Mortality All the included studies have noted several mortalities that occurred in both groups during the study period. The analysis results stated that there is a significant difference between the groups (RR: 0.41; 95% CI: 0.15, 1.14; P value = 0.09). Low heterogeneity was observed between the trials, and the results were statistically insignificant (χ2 = 0.30; I2 = 0% P value = 0.59) [Figure 8]. Figure 8 Forest plot showing the effect of interferon beta-1b on mortality rate versus standard care in COVID-19-infected patients Safety outcome The safety outcomes were analyzed in this meta-analysis to assess if the common adverse drug reactions (ADRs) reported in the included studies are significantly different between both the groups. The ADRs reported in the studies are nausea and vomiting (RR: 1.03; 95% CI: 0.65, 1.62; P value = 0.90), diarrhea (RR: 0.95; 95% CI: 0.63, 1.44; P value = 0.81), injection site reaction (RR: 1.50; 95% CI: 0.26, 8.75; P value = 0.65), flu-like syndrome (RR: 1.13; 95% CI: 0.72, 1.80; P value = 0.59), increased aminotransferase (RR: 0.73; 95% CI: 0.49, 1.08; P value = 0.64), hyperalbuminemia (RR: 0.71; 95% CI: 0.20, 2.53; P value = 0.11), acute respiratory distress syndrome (RR: 0.63; 95% CI: 0.33, 1.19; P value = 0.15), acute kidney injury (RR: 0.70; 95% CI: 0.29, 1.68; P value = 0.42), shock (RR: 0.40; 95% CI: 0.08, 1.98; P value = 0.26), and increased creatinine (RR: 0.70; 95% CI: 0.29, 1.69; P value = 0.43). Our results showed that there was no statistically significant difference found with the severity of ADRs experienced between both the groups. The intertrial variability was found to be insignificantly low for most of the parameters except for flu-like syndrome [Figure S1]. Assessment of Reporting bias We investigated the publication bias by using the funnel plot for all the primary outcome measures for included studies in the analysis. We found that the funnel plot was symmetrical for all the primary outcome measures [Figures S2-S7]. The outcome of GRADE approach assessment Quality grading for the clinical outcomes was performed using the GRADE approach which is depicted in Table 3. The certainty of the extracted evidence was categorized into very low to high grades for six outcomes: time of clinical improvement, duration of hospital stay, mortality, hospital admission with invasive mechanical ventilation, hospital admission with NIV, and number of ICU admission. It was found that for all the six outcomes, the certainty of the evidence was found to be moderate. Thus, further, there is a need to have a substantial number of RCTs to be conducted to achieve the certainty of evidence to be strong. Table 3 Summary of finding Comparison of interferon beta-1b along with standard care with standard care for severe COVID-19 patients Patient or Population: Severe COVID-19 patients Intervention: Interferon beta-1b with standard care Comparison: Standard care Outcomes Anticipated absolute effects* (95% CI) No. of participants (Studies) Certainty of the evidence (Grade) Risk with Standard Care Risk with Interferon Group Time of clinical improvement The mean time of clinical improvement was 8.67 Days MD 3.01 lower (4.97 lower to 1.05 lower) 167 (two RCTs) ⨁⨁⨁◯ MODERATEa Duration of hospital stay The mean duration of hospital stay was 8.57 Days MD 2.43 lower (4.55 lower to 0.3 lower) 233 (three RCTs) ⨁⨁◯◯ LOWa,b Mortality 160 per 1,000 72 per 1,000 (28 to 178) 233 (three RCTs) ⨁⨁⨁◯ MODERATEa Hospital admission with invasive mechanical ventilation 208 per 1,000 170 per 1,000 (81 to 361) 106 (two RCTs) ⨁⨁⨁◯ MODERATEa Hospital admission with non-invasive mechanical ventilation 68 per 1,000 86 per 1,000 (34 to 216) 193 (two RCTs) ⨁⨁⨁◯ MODERATEa ICU admission 717 per 1,000 509 per 1,000 (373 to 695) 106 (two RCTs) ⨁⨁⨁◯ MODERATEa *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; MD: Mean difference. GRADE working group grades of evidence. High certainty: We are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect. Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect. aProper blinding of the investigators and participants was not followed in most of the studies along with incomplete reporting of the outcomes and selection bias. Some of the trials were not registered. bFindings of the study were inconsistent in case of duration of hospital stay. DISCUSSION Since December 2019, the world has been shrouded by the COVID-19 newborn idiopathic coronavirus and an increasing number of infected patients. Since then, antiviral and immunomodulatory drugs have been tested in clinical trials to see if they can prevent the coronavirus from spreading. To date, no certified antiviral drugs with proven efficacy for COVID-19 treatment have been identified. Following that, it appears critical to collect and summarize a variety of evidence to achieve an effective COVID-19 treatment.[7] Type I IFNs are antiviral proteins that aim and prevent the replication and progression of a variety of viral pathogens while also promoting an immune response to clear virus infection. Hepatitis B and C, autoimmune disorders, SARS and MERS, and certain cancers are all treated with type I IFNs. This prompted the start of COVID-19 treatment with IFN. According to the National Health Commission of the People’s Republic of China, the drug of choice for COVID-19 treatment is a combination of IFN and an antiviral.[14] A recent study carried out in Iran proposed the need for IFN therapy in severe COVID-19. IFNs cause the expression of multiple genes in the host cells to shift into an antiviral state, preventing virus propagation and secondary replication. Although there are three major forms of IFNs, type I, II, and III, type I IFN plays a major role in the antiviral response and affects the adaptive immune responses. Given the striking similarities between COVID-19 and the SARS and MERS in terms of changes in total neutrophil and lymphocyte counts in patients, it is widely assumed that SARS-CoV-2 may also inhibit type I IFNs in the early stages of COVID-19 disease.[7] Similarly, we discovered a significant relationship between the administration of IFN β-1a and the amelioration of the clinical course of COVID-19 disease. This finding was particularly significant as it suggests that IFN could be a powerful therapeutic option in severe COVID-19 cases. Because the development of a newer antiviral takes years before it is approved for clinical use, specific high active antivirals are required for any novel emerging infectious disease. As a result, in the event of a pandemic, the most feasible approach is to test existing broad-spectrum antiviral drugs that have previously been used to treat other viral infections.[14] Antiviral agents should be started as soon as possible after the onset of symptoms to control viral replication and prevent tissue viral invasion. Antiviral efficacy decreased significantly after the cytokine release phase was established in COVID-19.[15] As of our knowledge, no meta-analysis states that the standard treatment when combined with IFNs increases the treatment efficacy and reduces hospital stay. Thus, we think it is a necessity to carry out this meta-analysis. In the current study, we found that IFN with antivirals (lopinavir/ritonavir, ribavirin) and HCQ effectively suppresses SARS-CoV-2 by reducing hospitalization time, viral-load clearance, and time for clinical improvement when compared to the standard care (lopinavir/ritonavir, ribavirin, and HCQ). The result of our meta-analysis depicts that the patients administered with IFN along with standard care had a reduced need for ICU admission compared to the standard care group. The IFN β-1b could not prevent invasive mechanical ventilation as there is no significant difference between the groups. The number of patients that needed NIV during the hospital stay was similar in both groups. Although two of the three included studies have reported mortality in both groups, our analysis results have statistically shown no difference in the number of deaths between the groups. The systematic review meta-analysis conducted by Kumar S et al.[17] included the studies which used interferon β-1b and β-1a in the intervention arm, whereas in this review we have included the studies conducted with interferon β-1b alone. The review conducted by Kumar S et al.[17] concluded that there was a significant reduction with respect to clinical improvement, and all other outcome parameters did not show any significant improvement between the intervention and the control arm. In this review, the patients who received interferon β-1b in the intervention arm have shown significant improvement in reducing the number of hospital stays, increase in clinical response, and decrease in ICU admission. Our meta-analysis also includes safety concerns. The adverse outcomes reported by each study were included in the analysis, and it was observed that there is no statistically significant difference seen between the intervention and the control group. As for limitations, all the included studies were open label. Hence, the subjects and investigators were aware of the treatment provided which could lead to performance bias. Though the route of administration was the same, the dose and dosing interval of IFN β-1b varied among the included studies. A substantial group of participants recruited in the studies had co-morbid conditions such as diabetes, hypertension, cardiovascular disease, and malignancy. This may increase the risk of interaction with other concomitant medications which might have an antagonistic or synergistic action with IFN β-1b or standard care treatment provided to COVID-19 patients. The concomitant medications might have an impact on the estimate of the IFN β-1b effect on COVID-19. Future RCTs must be performed to overcome the limitations mentioned in this review. CONCLUSION Our results demonstrate that administration of interferon beta-1b along with the standard care, that is, antiviral such as lopinavir/ritonavir, ribavirin, and HCQ had reduced the need for ICU admission and had better efficacy in shortening the duration of virus shedding, reducing the cytokine response, and reducing the symptoms of COVID-19. This study also gives insight to clinicians and to have informed them about the efficacy and safety of interferon beta-1b and a reasonable therapeutic option that can be used in severely ill COVID-19 patients for a better prognosis. Authors contributions Greeshma Sai Sree Nayudu (GSS), Binit Mamkoottathil Benny (BMB), Grace Thomas (GT), Maria Adil Khan (MAK), and Roopa Satyanarayan Basutkar (RSB) were involved in the conception of the study from the inception of the protocol development and throughout. GSS, BMB, and GT conducted the literature search, and the relevant information was also extracted. Risk assessment was performed by MAK and GSS. Any discrepancies were sorted, discussed, and resolved by RSB. GSS had entered the data into Review Manager 5.3 and contributed to the analysis. All the tables and figures were prepared by GSS, GT, and BMB. The first and subsequent drafts of the manuscript were prepared by GSS, BMB, GT, and MAK. RSB reviewed and corrected all the drafts of the manuscript. The final version of the manuscript was approved by all authors. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. Figure S1 Forest plot showing the adverse events of interferon beta-1b versus standard care in COVID-19-infected patients Figure S2 Funnel plot showing the effect of interferon beta-1b on the time of clinical improvement versus standard care in COVID-19-infected patients Figure S3 Funnel plot showing the effect of interferon beta-1b on the duration of hospital stay versus standard care in COVID-19-infected patients Figure S4 Funnel plot showing the effect of interferon beta-1b on the need for ICU admission versus standard care in COVID-19-infected patients Figure S5 Funnel plot showing the effect of interferon beta-1b on hospital admission with invasive mechanical ventilation versus standard care in COVID-19-infected patients Figure S6 Funnel plot showing the effect of interferon beta-1b on hospital admission with non-invasive mechanical ventilation versus standard care in COVID-19-infected patients Figure S7 Funnel plot showing the effect of interferon beta-1b on mortality rate versus standard care in COVID-19-infected patients Acknowledgement The authors thank JSS College of Pharmacy, JSS Academy of Higher Education and Research, Rocklands, Ooty, The Nilgiris, Tamil Nadu, India, for the support, technical assistance, and resources provided throughout the study. ==== Refs REFERENCES 1 Chuan A Liu Y Shao Y Yang CZ Xu J Yang B Hydroxychloroquine and Interferons for the prophylaxis and early treatment of COVID-19-current clinical advances J. clin. cell. immunol 2020 11 596 doi:10.35248/2155-9899.20.11:596 2 Basutkar RS Sagadevan S Sri Hari O Sirajudeen MJ Ramalingam G Gobinath P A study on the assessment of impact of COVID-19 pandemic on depression: An observational study among the pregnant women J Obstet Gynecol India 2021 71 28 35 34483514 3 Galluccio F Ergonenc T Garcia Martos A Allam AE Pérez-Herrero M Aguilar R Treatment algorithm for COVID-19: A multidisciplinary point of view Clin Rheumatol 2020 39 2077 84 32472459 4 Haji Abdolvahab M Moradi-kalbolandi S Zarei M Bose D Majidzadeh-A K Farahmand L Potential role of interferons in treating COVID-19 patients Int Immunopharmacol 2021 90 107171 doi:10.1016/j.intimp.2020.107171 33221168 5 Bakadia BM He F Souho T Lamboni L Ullah MW Boni BO Prevention and treatment of COVID-19: Focus on interferons, chloroquine/hydroxychloroquine, azithromycin, and vaccine Biomed Pharmacother 2021 133 111008 33227708 6 Peng F Tu L Yang Y Hu P Wang R Hu Q Management and treatment of COVID-19: The Chinese experience Can J Cardiol 2020 36 915 30 32439306 7 Alavi Darazam I Shokouhi S Pourhoseingholi MA Naghibi Irvani SS Mokhtari M Shabani M Role of interferon therapy in severe COVID-19: The COVIFERON randomized controlled trial Sci Rep 2021 11 8059 33850184 8 Gallo CG Fiorino S Posabella G Antonacci D Tropeano A Pausini E COVID-19: Role of the Interferons Preprints 2020 2020080018 9 Park A Iwasaki A Type I and Type III interferons –induction, signaling, evasion, and application to combat COVID-19 Cell Host Microbe 2020 27 870 78 32464097 10 Palermo E di Carlo D Sgarbanti M Hiscott J Type i interferons in COVID-19 pathogenesis Biology 2021 10 829 46 34571706 11 Shalhoub S Interferon beta-1b for COVID-19 Lancet 2020 395 1670 1 32401712 12 Fu W Liu Y Liu L Hu H Cheng X Liu P An open-label, randomized trial of the combination of IFN-k plus TFF2 with standard care in the treatment of patients with moderate COVID-19 EClinicalMedicine 2020 27 100547 32984784 13 Jean SS Lee PI Hsueh PR Treatment options for COVID-19: The reality and challenges J Microbiol Immunol Infect 2020 53 436 43 32307245 14 Hung IFN Lung KC Tso EYK Liu R Chung TW Chu MY Triple combination of interferon beta-1b, lopinavir–ritonavir, and ribavirin in the treatment of patients admitted to hospital with COVID-19: An open-label, randomised, phase 2 trial Lancet 2020 395 1695 704 32401715 15 Rahmani H Davoudi-Monfared E Nourian A Khalili H Hajizadeh N Jalalabadi NZ Interferon β-1b in treatment of severe COVID-19: A randomized clinical trial Int Immunopharmacol 2020 88 106903 doi:10.1016/j.intimp.2020.106903 32862111 16 Hozo SP Djulbegovic B Hozo I Estimating the mean and variance from the median, range, and the size of a sample BMC Med Res Methodol 2005 5 1 10 15636638 17 Kumar S Saurabh MK Narasimha VL Maharshi V Efficacy of Interferon-β in moderate-to-severe hospitalised cases of COVID-19: A systematic review and meta-analysis Clin Drug Investig 2021 41 1037 46

PMC010xxxxxx/PMC10353683.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-401 10.4103/ijcm.ijcm_282_22 Original Article Perspective Toward Complementary & Alternative Medicines in the Prevention of COVID-19 Infection Agrawal Apurva Sharma Ashish 1 Mathur Medha 2 Sharma Anita 3 Modi Gaurav 4 Patel Tarang 5 Department of Pharmacology, RNT Medical College, Rajasthan University of Health Science, Udaipur, Rajasthan, India 1 Department of Biochemistry, Geetanjali Medical College, Geetanjali University, Udaipur, Rajasthan, India 2 Department of Community Medicine, Geetanjali Medical College, Geetanjali University, Udaipur, Rajasthan, India 3 Department of Biochemistry, Himalaya Institute of Medical Science, Dehradun, Uttarakhand, India 4 Department of Biochemistry, GMERS Medical College, Gandhinagar, Gujarat, India 5 Department of Pathology, All India Institute of Medical Science Rajkot, Gujarat, India Address for correspondence: Dr. Ashish Sharma, Department of Biochemistry, Geetanjali Medical College, Udaipur, Rajasthan, India. E-mail: ashishapurva@gmail.com May-Jun 2023 30 5 2023 48 3 401406 01 4 2022 20 4 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Background: Across the globe, people are seeking integrative and holistic measures to prevent coronavirus (COVID-19) infection in the form of complementary and alternative medicines (CAM) with or without conventional medicines. This study was done to know the extent of CAM use for COVID-19 prophylaxis and to know beliefs and attitudes of people related to CAM use in India. Methodology: A pretested and prevalidated questionnaire was circulated on social media. Participants, who completed the online form and gave voluntary consent, were included. The questionnaire included demographic details and questions related to CAM use, preferences with reasons, preparations used, perceived role of CAM in prevention, immunity boosting and side effects, sources of information, etc. Results: Out of 514 responses, 495 were analyzed. 47.07% of respondents were males and 52.93% were females. 66.9% were using CAM for COVID-19 prophylaxis. The association between age, gender, and profession with CAM use was statistically significant (P < 0.05). 41.1% reported CAM use in the past. 36.6% of CAM users were taking “Kadha” and 33% were using ayurvedic medicines. Other frequently used CAM preparations were chyavanprash, giloy, tulsi, ginger, pepper, cloves, honey, sudarshanghanvati, arsenic-30, lemon juice, cinnamon, steam inhalation, ashwagandha, swasarivati, coronil, and warm saline water gargles. 46.9% of the CAM users were on self-medication and 52.3% preferred CAM over allopathy. Conclusion: Complementary and alternative medicine utilization for COVID-19 prophylaxis is widespread and self-medication is prevalent. As no specific cure is available in conventional systems, people believe in traditional medicines more than conventional, yet confusion exists. There is a need of increasing awareness regarding side effects, drug–drug interactions, and self-medication. Complementary and alternative medicine COVID-19 perspective toward CAM ==== Body pmcINTRODUCTION The corona pandemic is unique in several aspects and has challenged the healthcare system all over the world. The conventional system is playing its role as a frontline system for the treatment of infection, yet no optimum treatment is available. Interim results of the solidarity trial have reported that all four treatments evaluated (remdesivir, hydroxychloroquine, lopinavir/ritonavir, and interferon) had very less or no effect on overall mortality and course of disease.[1] Many pluralistic knowledge systems available globally are also trying to intervene through their knowledge base.[2] Ministry of AYUSH; Government of India has recommended self-care guidelines for preventive health measures and enhancing immunity with special reference to respiratory health.[3] Across the globe, people are frightened by the present pandemic situation and are seeking measures for prophylaxis in the form of complementary and alternative medicine (CAM). One such report published by the National Center for Complementary and Integrative Health reveals that people are using alternative systems to prevent corona infection, though data has not been provided.[4] Though few studies have been published reporting the use of Chinese traditional medicine in the treatment of coronavirus (COVID-19) infection[5] and others have been published focusing on possible interventions by Indian traditional and pluralistic medicine systems,[2,6] information on the use of CAM and traditional medicines by the community for COVID-19 prophylaxis is limited. Thus, this study was planned to know the extent of the use of CAM for the prevention of COVID-19 infection as well as to know the beliefs and attitudes of people in India related to CAM keeping COVID-19 infection in focus. METHODOLOGY Study design Cross-sectional Study Study setting and population A study was conducted as an online survey through a web platform. This mode of the survey was chosen due to feasibility, ease to conduct, and cost-effectiveness, and it aids in the collection of diverse information efficiently irrespective of geographical boundaries posed due to the current pandemic condition. An online questionnaire was circulated through social media among general people of India who have access to the internet via “Google forms.” Wide dissemination of online forms was assured by sharing them on multiple social networking sites. A total of 514 responses were obtained. Inclusion and exclusion criteria Participants with social media access, who completed the online form and gave voluntary consent to enroll in the study, were included while those who were less than 18 years of age and whose forms were found incomplete or invalid were excluded from the study. Study duration The study was completed in eight weeks of duration from November to December 2020. Questionnaire designing, piloting, and planning were done in two weeks followed by four weeks for data collection through an online survey. Two weeks were dedicated to data analysis and report writing. Data collection A pretested and prevalidated questionnaire was generated through Google Forms and its Uniform Resource Locator was circulated through social media and data was collected through a snowballing technique where the link was forwarded from one participant to the other to a large geographical area. Initially, the questionnaire was sent to the social networking sites frequented by general people of India and to the acquaintances of authors, who contributed as an initial sample. The questionnaire was sent with a note to forward it to friends, family members, and other acquaintances, and they contributed as the next level of sample. In this way, the questionnaire was forwarded from acquaintances of authors and people visiting social networking sites to their acquaintances and so on. Possibility of bias could not be completely ruled out as it was initially also distributed to the acquaintances of authors and then forwarded to their acquaintances. A survey was conducted from November 15th, 2020 to December 15th, 2020. After that, the link was disabled and no more responses were recorded. Study tool The questionnaire was a mix of closed, semiclosed, and open-ended questions, with the majority belonging to closed ended. Six questions were related to demographic details including age, gender, education, nationality, residential area, and profession of the participants. Fifteen questions were related to CAM use for COVID-19 prophylaxis and beliefs and attitudes of participants toward CAM. The details related to CAM were recorded like the preferences of CAM, the preparations used for ailments other than COVID-19 infection, perceived role in prevention or immunity-boosting, and the side effects experienced with CAM and conventional medication therapies. Only two questions were open-ended where participants were asked to name specific CAM preparation they have used for COVID-19 prevention and to name specific CAM medicine they believe could prevent COVID-19 infection. In questions related to “CAM therapy & conventional medicines they have been using for COVID-19 infection, reasons for using and preferring CAM and who advised for CAM”, semiclosed type of questions were framed to provide the participant an opportunity to write his/her answer if it does not belong to any of the given choices and belonged to “other” category. The respondents were allowed to select multiple responses for the usage of various CAM therapies, allopathic medications, and reasons for using CAM therapy, and the data was analyzed accordingly. The Institutional Human Research Ethics Committee approved this study (Date: 15.09.2020, Ref: GU/HREC/2020/819). Approval was obtained before the commencement of the study. Informed consent was obtained from study participants and anonymity was maintained at all stages of the study. Statistical analysis Data was collected, compiled, and entered in Microsoft Excel software and analyzed using SPSS Version 24 (SPSS Inc, Chicago IL, USA). All the categorical variables were presented as frequencies and percentages, and test of significance like Chi-square test was used to assess the level of significance of variables and P value < 0.05 was considered statistically significant. RESULTS A total of 514 responses were received out of which, 495 responses were analyzed after eliminating incomplete/invalid responses and responses from respondents below 18 years of age. There were 233 (47.07%) male and 262 (52.93%) female respondents in the survey. As the responses were collected through a web platform, 3 (0.61%) of the responses were obtained from overseas too but the majority of respondents were Indian 492 (99.39%) citizens. The responses were received from all parts of the country and no specific cluster could be recognized, rather the distribution was dispersed pan India [Figure 1]. Demographic details of respondents suggested that 387 (78.2%) were residents of urban areas while 108 (21.8%) were rural respondents. The majority of study subjects (238; 48.1%) were in the 18–30 years of age range, followed by 30–45 years of age (154; 31.1%), and more than 45 years of age (103; 20.8%). The education of study subjects indicated that 223 (45.1%) respondents were post-graduates, 191 (38.5%) were graduates, and 81 (16.4%) were below senior secondary level. The occupations of study participants had vast distribution like homemakers (38; 7.7%), business (46; 9.3%), other professionals like engineers, lawyers, and other professionally educated people except health care workers (147; 29.7%), health care workers (117; 23.6%), and students (147; 29.7%). The health care workers were analyzed as a separate entity because they have prior knowledge regarding drugs and CAM. Figure 1 Geographical distribution of participants It was found that 331 (66.9%) participants were using medicine other than conventional medicine (allopathy) or any commercial or homemade herbal medicine or therapy for the prevention of COVID-19 infection in contrast to 164 (33.1%), who never used such preparations. Two hundred and four (41.1%) respondents reported usage of CAM in the past for ailments other than COVID-19. Table 1 shows the distribution of study subjects as per their demographic profile and CAM usage. The association between age, gender, and profession with CAM use was found to be statistically significant (P < 0.05). The age group of 31–45 years was found to use CAM therapy more frequently as compared to other age groups (P < 0.05), gender-wise females were using CAM therapy more frequently as compared to males (P < 0.05), and health care workers were found to be less involved in CAM therapy usage (P < 0.05) as compared to other professions in the study. There was no significant difference in CAM therapy based on the area of residence and educational status of the participants. Table 1 Demographic profile of study participants Variables CAM use P # Present Absent Age (years)  18–30 143 (28.89%) 95 (19.19%) 0.007*  31–45 115 (23.23%) 39 (7.88%)  >45 73 (14.75%) 30 (6.06%) Gender  Male 145 (29.29%) 88 (17.78%) 0.039*  Female 186 (37.58%) 76 (15.35%) Area of residence  Urban 256 (51.72%) 131 (26.46%) 0.520  Rural 75 (15.15%) 33 (6.67%) Profession  Business 32 (6.46%) 14 (2.83%) 0.046*  Professional** 106 (21.41%) 41 (8.3%)  Homemaker 31 (6.26%) 7 (1.41%)  Health care worker 71 (14.34%) 46 (9.3%)  Student 91 (18.38%) 56 (11.31%) Education  School pass out 55 (11.11%) 26 (5.25%) 0.647  Graduate 123 (24.85%) 68 (13.74%)  Post-graduates 153 (30.91%) 70 (14.14%) *P-value is statistically significant #P value is calculated by Chi-square test, **Professionals include engineers, lawyers, and other professionally educated people except health care workers The common CAM preparations used by study participants as shown in Table 2 are vitamins and mineral-rich food products (43.2%), kadha (a polyherbal phytochemical-rich decoction) (36.6%), Ayurveda medicines (33.7%), herbal tea (30.4%), homemade concoctions (29.7%), yoga (25.1%), herbal preparations (20.8%), milk with turmeric (18.3%), homeopathic (15.8%), Unani (0.3%), and reiki-therapy (0.3%). Unani is an Arabic traditional medicine system that is practiced in Asian Muslim culture. Reiki-therapy is an energy-healing therapy that promotes relaxation. Other frequently used CAM preparations reported by respondents are chyavanprash (an Ayurveda health supplement made up of a concentrated blend of nutrient-rich herbs and minerals, the key ingredient is Indian gooseberry), giloy (Tinospora cordifolia) tablets and juice, tulsi (Ocimum tenuiflorum), ginger (Zingiber officinale), pepper (Piper nigrum), cloves (Syzygium aromaticum), honey, sudarshan ghanvati (a multi-ingredient Ayurveda medicine containing Chitrak (Plumbago zeylanica), Ashwagandha (Withania somnifera), turmeric, Mustaka (Nut Grass), Triphala (a polyherb containing Emblica officinalis, Terminalia bellerica, and Terminalia chebula), arsenic 30 tablets (Arsenicum album 30C is a homeopathic medicine), lemon juice, cinnamon (Cinnamomum verum), steam inhalation, swasari vati (a calcium-rich herbal mixture indicated in respiratory disorders. The main ingredients are salt cresse, gall plant, liquorice plant (Glycyrrhiza glabra), dried ginger, black pepper and Indian long pepper (Piper longam), coronil (a tri-herbal formulation containing extracts from Withania somnifera, Tinospora cordifolia, and Ocimum sanctum), and warm saline water gargles. Table 2 The specific Complementary and Alternative Medicine therapy/preparation used by respondents (n=331) Complementary and Alternative Medicine Frequency (percentage)* Food rich in vitamins and minerals 131 (43.2%) Kadha 111 (36.6%) Ayurvedic medicines 102 (33.7%) Herbal tea 92 (30.4%) Homemade concoctions and other methods 90 (29.7%) Yoga 76 (25.1%) Herbal preparations 63 (20.8%) Milk with turmeric 55 (18.3%) Homeopathic 48 (15.8%) Unani 01 (0.3%) Rachytherapy 01 (0.3%) *Respondents selected multiple responses for this item In response to the use of allopathic medication for the prevention of corona infection, 227 (45.8%) respondents denied its use. The most commonly consumed medication was multivitamins (including Vitamin C) and zinc [Table 3]. Table 3 Allopathic medicines used by respondents for prevention of COVID-19 infection Medicine Frequency (percentage)* Hydroxychloroquine 44 (8.9%) Azithromycin 58 (11.7%) Ivermectin 28 (5.6%) Multivitamins and zinc 259 (52.3%) Others 41 (8.3%) *Respondents selected multiple responses for this item The study participants were asked about the reasons for using CAM, 13.7% (68) were not sure about any specific reason while 34.34% (170) used CAM to improve their immunity, 17.2% (85) wanted a preventive action while 34.74% (172) desired both immunity and preventive action from CAM. Out of 331 responders who reported CAM intake, 46.9% (155) reported self-administration, 26.5% (88) reported intake on advice of relatives/friends, and 18.6% (62) reported intake after consultation with CAM practitioners. Knowledge about CAM therapy was also gained from other sources like social media, television, or literature by 8% (26) of responders. Perception of study participants regarding the improvement of immunity revealed that 28.9% (143) believed that CAM preparations can improve immunity, 4.8% (24) thought that conventional therapy is useful in immunity enhancement, 54.9% (272) opted for mixed therapy, while 11.4% (56) were not sure about this. 10.7% (53) respondents believed that there is some CAM therapy/preparation available for the treatment of corona infection while 48.5% (240) were not of this opinion, whereas 40.8% (202) respondents were not sure of it. 52.3% (259) respondents said that they would prefer CAM therapy over allopathy for the prevention or treatment of corona infection, and the reasons for their preference are shown in Table 4. Table 4 Reasons for preferring Complementary and Alternative Medicine therapy Reason Frequency (percentage)* Fewer side effects 181 (36.56%) More effective 165 (33.33%) Belief on therapy 132 (26.67%) Ease of taking 102 (20.6%) Cheaper 98 (19.8%) Popularity of therapy 68 (13.74%) Cultural reasons 45 (9.09%) Influence of media 37 (7.47%) To explore new therapy 33 (6.67%) Previous experience 26 (5.25%) Miscellaneous reasons 64 (12.93%) *Respondents selected multiple responses for this item The participants perceived CAM therapy as free of side effects (19.03%), while 6.73% were not sure about it. 31.3% believed that, if CAM and conventional medication are taken together, they won’t alter each other’s actions while 18.2% believed that combined therapy can lead to interaction [Table 5]. Table 5 Perception of study participants regarding Complementary and Alternative Medicine therapy/preparations Perception Frequency (percentage) Side effects of CAM  Yes 67 (13.5%)  No 223 (45.1%)  May be 205 (41.4%) Interaction of CAM and conventional therapy  Yes 90 (18.2%)  No 155 (31.3%)  May be 250 (50.5%) DISCUSSION The COVID-19 pandemic is unique in several aspects. There is confusion among scientists, healthcare professionals, and also among people regarding prevention as well as its treatment. Conventional modern medicine is at the forefront in managing COVID-19 cases, especially in critical care circumstances. Yet, recent solidarity trial has reported no significant effect of all five medicines which were earlier considered effective.[1] No proven effective medicines are thus available for prophylaxis as well as treatment. Vaccines are now available, but a recent study has reported that immune function in vaccinated individuals after eight months of the last dose, is lower than that in unvaccinated individuals.[7] During such difficult times when the conventional health system is unsure about the treatment strategies, people start looking for other alternatives and self-medication. A recent survey of Ebola outbreak survivors revealed that over 71% of survivors were self-medicating and about 50% of survivors reported the use of traditional and complementary medicines.[8] The use of traditional Chinese medicines for COVID-19 prophylaxis and treatment has been reported,[5] but knowledge regarding the use of traditional Indian medicines for COVID-19 prophylaxis is limited. In the present study, an effort was made to find out the extent of utilization of CAM and traditional Indian medicines by general people in India as well as to find out the reasons behind their inclination toward CAM during the COVID-19 pandemic. More than 65% of respondents were using medicines or therapies other than conventional medicines for the prevention of COVID-19 infection. As India has a long and rich history of Ayurvedic and other traditional medicines, people have strong faith in them. A recent telephonic survey done among asymptomatic patients at an isolation center has revealed that about one-fourth were using traditional therapies even though residing in the isolation center and more than half among them were consuming Ayurveda “Kadha.”[9] In the present study, CAM use was most common in the 31-45 years age group with a statistically significant difference. CAM use was reported more in females than males, and more among homemakers as compared to professionals. There was no significant difference in CAM use depending on the area of residence and level of education. Similar results have been reported by a study on CAM utilization in urban and regional Australia, though the mean age in their study was 53 years.[10] Among CAM users, more than 35% were using “Kadha,” which is an Ayurvedic polyherbal phytochemical-rich decoction for oral use, commonly used to control various respiratory disorders, while 33% were using Ayurvedic medicines. Respondents also reported the use of herbal tea, homemade concoctions, milk with turmeric, homeopathic medicines, giloy (T. Cordifolia), tulsi (Ocimum tenuiflorum), ginger, pepper, cloves, honey, lemon juice, etc., Charan et al. have also reported the use of these herbal and household medicines by corona-positive patients.[9] 45.8% respondents denied of using any modern medicine (allopathic) for the prevention of COVID-19 infection. The most commonly used conventional medicine was multivitamin capsules and almost half of respondents accepted using them which is contradictory to 45% denying using modern medicines. It may be a possibility that multivitamins are considered nutritional supplements by people rather than medicine. The majority of responders were using CAM as self-administered and only 18.6% consulted CAM practitioners. 36.6% of responders believe that traditional medicines are having lesser side effects, which might be a factor for self-administration. This is an alarming fact as it may result in drug–drug interactions and an increased risk of adverse effects in people with chronic or major organ diseases. Self-administration of Ayurvedic and other traditional medicines has been reported by other studies also.[11-14] Eboreime EA et al.[15] have also expressed concerns regarding the safety of traditional and complementary medicines rampantly self-administered by people for the prevention of COVID-19 infection. Clinicians should be aware of the frequent utilization of CAM and herbal remedies by general people and should enquire about the same to prevent possible drug interactions. As many of these are age-old Ayurveda medicines, literature might be available on their indication and possible adverse effects. There is a need for collaboration between modern medicine and Ayurveda medicine systems as well as research organizations to find out evidence related to CAM and traditional drugs, also to enlist possible drug interactions and adverse effects. Though about 35% of responders were using CAM for its immunity improvement and preventive action, only 10.7% believed that there is any CAM therapy/preparation available for its prevention. This shows that people are in desperate need of some preventive medicine or therapy, but they are in confusion or disbelief as no studies are available that prove the efficacy of any preventive drug. More than half of the participants replied that they would prefer CAM over allopathic medicines for prevention or treatment of corona infection, most common reason was a belief that CAM causes fewer side effects, followed by belief in efficacy. A current systematic review and meta-analysis have concluded that adding Chinese herbal medicines to the standard care of COVID-19 patients improves their signs and symptoms,[16] but no such study has been done related to the efficacy of Ayurvedic medicines in India. Another recent systematic review and meta-analysis of randomized controlled trials have concluded a significant effect of the combined use of herbal and western medicine on signs, symptoms, and laboratory parameters of corona-infected patients.[17] Though people have faith in traditional and Ayurvedic medicines, steps must be taken in directions to provide evidence for traditional medicinal approaches. A pragmatic plan for the implementation of the Ayurvedic system in the prevention and treatment of COVID-19 infection has been proposed by Rastogi S et al.,[18] yet more steps are needed in this direction to ensure safe and informed CAM utilization by people. Limitations This study was done through online survey forms circulated through social platforms; the possibility of selection bias was a limitation. Also, its reach to rural areas and people not on social media was limited. The present study thus provides a rough picture of CAM utilization for corona prevention, and its findings could not be applied to the general public. Therefore, more such studies are recommended especially in a community set up to establish the extent and pattern of CAM among people during such a pandemic and factors governing such use. CONCLUSION Complementary and alternative medicine utilization for the prevention of corona infection is widespread and self-administration is prevalent. As no specific cure is available for corona infection in the conventional system, people believe in traditional medicines more than standard modern medicines, yet confusion exists. People believe that CAM is safe and increase immunity, but studies providing evidence of their efficacy and safety are limited. There is a need for clinicians to be vigilant about CAM use and self-medication, as well as to spread awareness among general people regarding possible side effects and drug–drug interactions between traditional and modern medicines, and that these medicines should be taken after proper consultation from CAM consultants. Further, studies are needed to know the extent of various parallel medicine systems used by people and predictors of such usage. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. ==== Refs REFERENCES 1 Consortium WST Pan H Peto R Henao-Restrepo AM Preziosi MP Sathiyamoorthy V Repurposed antiviral drugs for Covid-19-Interim WHO solidarity trial results N Engl J Med 2020 384 497 511 33264556 2 Lyu M Fan G Xiao G Wang T Xu D Gao J Traditional Chinese medicine in COVID-19 Acta Pharm Sin B 2021 11 3337 63 34567957 3 Ministry of Ayush, Ayurveda's immunity-boosting measures for self-care during COVID 19 crisis Available from: https://www.mohfw.gov.in/pdf/ImmunityBoostingAYUSHAdvisory.pdf [Last accessed on 2020 Jun 20] 4 National Institute of health, Covid - 19 and alternative treatments Available from: https://www.nccih.nih.gov/health/in-the-news-coronavirus-and-alternative-treatments-what-you-need-to-know [Last accessed on 2020 Aug 20] 5 Yang Y Islam MS Wang J Li Y Chen X Traditional Chinese medicine in the treatment of patients infected with 2019-new coronavirus (SARS-CoV- 2): A review and perspective Int J Biol Sci 2020 16 1708 17 32226288 6 Ren JL Zhang AH Wang XJ Traditional Chinese medicine for COVID-19 treatment Pharmacol Res 2020 155 104743 32145402 7 Yamamoto K Adverse effects of COVID-19 vaccines and measures to prevent them Virol J 2022 19 100 35659687 8 Dror AA Eisenbach N Taiber S Morozov NG Mizrachi M Zigron A Vaccine hesitancy: The next challenge in the fight against COVID-19 Eur J Epidemiol 2020 35 775 9 32785815 9 James PB Wardle J Steel A Adams J Pattern of health care utilization and traditional and complementary medicine use among Ebola survivors in Sierra Leone PLoS One 2019 14 e0223068 31560708 10 Von Conrady DM Bonney A Patterns of complementary and alternative medicine use and health literacy in general practice patients in urban and regional Australia Aust Fam Physician 2017 46 316 20 28472578 11 Charan J Bharadwaj P Dutta S Kaur R Bist SK Detha MD Use of complementary and alternative medicine (CAM) and home remedies by COVID-19 patients: A telephonic survey Indian J Clin Biochem 2021 36 108 11 33162692 12 Sharma A Agrawal A Complementary and alternative medicine (CAM) use among patients presenting in out-patient department at tertiary care teaching hospital in Southern Rajasthan, India-A questionnaire based study Altern Integ Med 2015 4 187 13 Kristoffersen AE Stub T Broderstad AR Hansen AH Use of traditional and complementary medicine among Norwegian cancer patients in the seventh survey of the Tromsøstudy BMC Complement Altern Med 2019 19 341 31783842 14 Azizah N Halimah E Puspitasari IM Hasanah AN Simultaneous use of herbal medicines and antihypertensive drugs among hypertensive patients in the community: A review J Multidiscip Healthc 2021 14 259 70 33568913 15 Eboreime EA Iwu CJ Banke-Thomas A Any and every cure for COVID-19: An imminent epidemic of alternative remedies amidst the pandemic? Pan Afr Med J 2020 35 108 16 Fan AY Gu S Alemi SF Chinese herbal medicine for COVID- 19: Current evidence with systematic review and meta-analysis J Integr Med 2020 18 385 94 32792254 17 Ang L Song E Lee HW Lee MS Herbal medicine for the treatment of coronavirus disease 2019 (COVID- 19): A systematic review and meta-analysis of randomized controlled trials J Clin Med 2020 9 1583 32456123 18 Rastogi S Pandey DN Singh RH COVID-19 pandemic: A pragmatic plan for Ayurveda intervention J Ayurveda Integr Med 2022 13 100312 32382220

PMC010xxxxxx/PMC10353684.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-465 10.4103/ijcm.ijcm_520_22 Original Article Effect of Self-Directed Learning Module and Assessment on Learning of National Health Programme by Medical Undergraduates – A Mixed Methods Evaluation Rajalakshmi M Ganapathy Kalaiselvan Department of Community Medicine, Sri Manakula Vinayagar Medical College and Hospital, Puducherry, India Address for correspondence: Dr. Kalaiselvan Ganapathy, Professor and Head, Department of Community Medicine, Sri Manakula Vinayagar Medical College and Hospital, Puducherry, India. E-mail: kalaiselvanmd@gmail.com May-Jun 2023 30 5 2023 48 3 465470 21 6 2022 17 4 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Background: Competency-based medical education (CBME) curriculum in India has introduced many new concepts like a foundation course, early clinical exposure, and self-directed learning (SDL). Sometimes SDL simply means self-study. Self-directed learning as defined by Knowles is a process in which individuals take the initiative with or without the help of others in diagnosing their learning needs, setting their own learning goals, identifying appropriate learning resources, and selecting appropriate learning strategies. SDL is seen as a prerequisite for life-long learners, especially medical graduates. We found poor uptake of SDL sessions in terms of learning and attendance by students. To develop and assess the effect of the SDL module in Community Medicine for Phase -3 MBBS students. Materials and Methods: The study design was a program development and evaluation design. The program development consists of free listing and Nominal Group Technique (NGT). The evaluation design consists of a formative assessment, an end-of-module assessment, and feedback from undergraduate students, postgraduates, and faculties. Data collection procedure: SDL module was developed, agreed and implemented among undergraduates of Phase – 3 MBBS students. Results: Free listing was conducted among undergraduate students who had completed the phase 3 MBBS examination and Nominal Group Technique was conducted among the faculties (n = 7) and Postgraduates of the Department of Community Medicine (n = 2) to explore the appropriate topics for SDL in Community Medicine. The topic with the highest ranking and which was finalized for preparation of the SDL module was “National Health Programme”. Three fourth 118 (75%) of the students scored ≥50% at the end of the module assessment. Manual content analysis for the feedback was categorized into three themes such as facilitating factors, challenges, and solutions. Conclusions: Effective implementation and assessment of SDL sessions are one of the new concepts in the CBME curriculum. Community medicine feedback module nominal group technique perception ==== Body pmcINTRODUCTION Competency-based medical education (CBME) curriculum in India has introduced many new concepts like a foundation course, early clinical exposure, and self-directed learning. Sometimes SDL simply means self-study. Self-directed learning as defined by Knowles is a process in which individuals take the initiative with or without the help of others in diagnosing their learning needs, setting their own learning goals, identifying appropriate learning resources, and selecting appropriate learning strategies.[1] Although there are several definitions and interpretations, the essence of SDL remains in its words, i.e., self (learner-oriented), directed (facilitated and monitored), and learning (applicable to lifelong learning).[2] Some of the examples currently being used to cultivate skills of self-directed learning and reflection are problem-based learning, small group learning, self, and peer evaluation, self-study materials, library works, projects, and computer-assisted learning. Now we could see a movement from pedagogy to andragogy in this transformational learning model of SDL in medical education.[3] SDL adds variety to teaching-learning methods and provides an option for curriculum makers to choose this method in alignment with some learning objectives. The conduct of SDL is quite variable at different places.[2,4,5] In several instances, it is confused with self-learning or just asking students to read from books but remaining unobserved. Students and teachers have shown apprehension about the freedom of learning in countries where teacher-oriented learning has been there for a long time. SDL is an active learning approach with the teacher acting as a facilitator of learning. A medical graduate, being a lifelong learner, should instill the habit of SDL. SDL has been receiving increasing attention since the implementation of competency-based medical education (CBME) by the Medical Council of India (MCI).[4,5] Even though dedicated time has been allotted to SDL in the CBME curriculum in each specialty, implementation of SDL is challenging and has become mandatory. Hence in the present study, we developed, implemented, and assessed module-based SDL sessions in Community Medicine for the current batch of students. The challenges faced in implementing the SDL module were also explored by qualitative technique. METHODS The study was carried out among medical undergraduates of Phase - 3, part 1 MBBS, postgraduates, and faculties of the Community Medicine Department in a private medical college located at Puducherry Union Territory. The college admits 150 undergraduate medical students per academic year and is affiliated with Pondicherry University. National Health Programme (NHP) is a part of the medical undergraduate’s curriculum and only the must-know components mentioned in the syllabus are taught during lectures. It was a program development and evaluation design. The program development consists of qualitative techniques like free listing and Nominal Group Technique (NGT). The evaluation design consists of a formative assessment, an end-of-module assessment, and feedback from undergraduate students, postgraduates, and faculties. The module was delivered to 158 Phase - 3, part 1 students of the academic year 2018, over a period of 2 months from November 2021 to December 2021. The steps for the conduct of the SDL session are as follows: Step 1: Selection of topic and development of module Step 2: Actual conduct of the session Step 3: End-of-module assessment Step 4: Feedback Step 1: Selection of topic and Development of module: Free listing was conducted among undergraduate students who had completed the phase 3 MBBS examination to explore the difficult topics for SDL in Community Medicine. [Table 1] Table 1 Perceived as difficult topics by students Item Frequency (%) Average Rank Salience Health programmes in India 100 1 1 Communication for health education 100 2.2 0.82 Health planning and management & Health care of the community 70 3 0.484 Medicine and social sciences 70 4 0.376 Preventive Medicine in OBS, Peds, and geriatrics 70 5.14 0.276 Communicable diseases 60 5.33 0.194 Environment and health 50 6 0.143 Concept of health and disease 30 5 0.129 International Health 30 6 0.086 Health information and basic medical statistics 30 7 0.043 Health planning and management 20 3.5 0.129 Epidemiology 20 5 0.086 Health care of the community 10 5 0.043 Demography and family planning 10 5 0.043 A Nominal Group Technique (NGT) was conducted among the faculties (n = 7) and Postgraduates of the Department of Community Medicine (n = 3) to explore the appropriate topics for SDL in Community Medicine. The technique was conducted by a trained Principal investigator in a place and time convenient for the participants using a semi-structured interview guide with a broad open-ended question. The question in the Nominal Group Technique was “List the appropriate topics for SDL in Community Medicine for Phase -3 MBBS students”. Firstly, every participant in the study was asked to give their suggested list of topics for the SDL session. Secondly, all the participants were asked to proceed to rank the topics according to priority as 1st, 2nd, 3rd, 4th, and so on. Thirdly participants were encouraged to share and discuss the reasons for their choices. It helped to identify common ground and plurality of ideas and approaches by each participant. Fourthly, the rank for each topic received was totalled, and the topic with the highest (i.e., most difficult) total ranking was selected as the final decision for the development of the module. The topic with the highest ranking and which was finalized for preparation of the SDL module was “National Health Programme”. Then participants were again asked to rank all the National Health Programmes according to priority. Finally, among all the National Health Programmes, the top four National Health Programmes with the highest total ranking were selected for the preparation of the module. The top four National Health Programmes were National AIDS Control Program (NACP), National Tuberculosis Elimination Program (NTEP), the Reproductive and Child Health (RCH) program, and National Leprosy Eradication Program (NLEP) were included in the module. All the interviews were audio recorded and the transcripts were prepared verbatim in English [Table 2]. The module was drafted by the first author by following the competencies given by NMC. The draft module was shared with the faculties of community medicine for review and was approved by the curriculum committee. The module consists of subtopics under each National Health Programme with inbuilt self-assessments like Multiple choice questions, short answer questions, fill-in-the-blanks, and case-based or problem-based questions. Table 2 Consensus score by Nominal Group Technique Topics Score by each respondent Total 1 2 3 4 Health programs in India 5 4 - 3 12 Environment & health - - 4 5 9 MDG to SDG 4 - - 4 8 Surface infections 3 5 - 8 Preventive obstetrics, pediatrics 2 - 3 - 5 Health planning - 5 - - 5 Sociology - 1 1 2 4 Health care of the community 3 - - - 3 Concept of health and disease 1 - - 1 2 Rickettsial infection - 2 - - 2 Demography - - 2 - 2 Step 2: Actual conduct of the session: First contact session: Orientation on the process of SDL like division of students into small batches, fixing of learning goals and the milestone by the students, sharing of resources during the intersession period, implementation of the self-directed module, and assessment at the end of each day of the SDL session was briefed to the students. The role of the facilitator was to help students find the resources, and the fixing of venue and timetable adjustments was also briefed. A Whatsapp group for coordination with the students was formed. Intersession period: During the intersession period documents and websites related to National Health Program (NACO, NTEP, NHM, NPCDCS) were shared through the Whatsapp group and SMVMCH Learning Management System to engage them in learning. Second contact session: Before the start of the second contact session, an interactive workshop was held for the facilitators (n = 10) using faculty guide on the implementation of the module and assessment. Through the second contact session, module-based SDL sessions were implemented in Phase - 3, part 1 MBBS students. Students were divided into five small batches. Each batch contains 30 students who were moderated by a faculty and postgraduate. The number of hours allotted for each NHP was six hours, total there were four NHPs and the total time allotted for all the NHPs was 24 hours. The content of each NHP in the SDL module includes important subtopics under each NHP followed by assessment in the form of multiple-choice questions, short answer questions, fill-in-the-blanks, and case-based or problem-based questions. Following the implementation of the module, debriefing was also done by discussing answers to the assessment questions asked at the end of each NHP, and the modules were also marked by the facilitators with the areas to be improved and handed over to the students individually after the end of the module assessment. Step 3: Feedback: Feedback was collected from all the students and facilitators about the implementation of the SDL module. The online feedback was also obtained from the students who appeared in the final Pondicherry University summative examination. Step 4: End-of-module assessment: Students learning was assessed by, Written examination consisting of short answer questions and was evaluated with answer key by the principal investigator. Submission of all the completed modules. Ethical issues: The present study was cleared by the Research Committee and the Institutional Ethics Committee (Human Studies) (Ref no: IEC No- EC/91/2021). Permission was also obtained from the Head of the Institution for implementing module-based SDL sessions. Students’ marks were not displayed on the noticeboard and were communicated individually to students. Marks were stored separately in HOD’s computer. Data analysis: The following analysis was done in the study. The free listing data was entered and analyzed using the Visual Anthropac 1.0 software package and the salience value was calculated. Manual content analysis was done by the first author for feedback obtained from students, postgraduates, and faculties regarding the SDL session. For written assessment frequency was calculated and the Marks were categorized into less than 50%, 50 – 75%, and >75 percentage. The average of marks was also expressed in mean ± SD. RESULTS Program development Out of 158 students, 86 (54.4%) were females and 72 (45.6%) were males. As shown in Table 1, an Exhaustive list of responses that were obtained during the free listing activity was fed into Visual Anthropac software, and 14 salient items were obtained with a Smith salient score. The topic with the highest Smith salient score was National Health Programmes in India. The Nominal Group Technique was conducted among facilitators to obtain consensus for the selection of topics for the development of the SDL module. The topic which was obtained the highest consensus was National Health Programme in India. [Table 2] Program evaluation End of module assessment At the end of all four modules, there was an end-of-module assessment for 50 marks. Out of 50 marks, 30 marks were given to written assessment consisting of structured short answer questions and 20 marks (five marks for each module) for the assignment submission i.e., submission of four completed modules. The average mark at the end of the module assessment was 64 ± 19 (standard deviation). Out of 158 students 25.4%, 41%, and 33.6% of students scored marks <50%, 50-75%, and >75% respectively. [Table 3] Table 3 End of module assessment scores of all modules of SDL Gender of students Mark category n (%) < 50% 50-75% >75% Female 22 (55) 37 (57) 27 (51) Male 18 (45) 28 (43) 26 (49) Total 40 65 53 Feedback from students, postgraduates, and faculties In Table 4, content analysis of students, postgraduates, and faculties feedback was categorized into three themes, the facilitating factors, challenges, and solutions. The categories which were emerged under each theme were the SDL session, session frequency, module development, and assessment. The students felt that the module stressed difficult topics in the curriculum, the simple and easily understandable module, and discussion with peers during activities and assessment was the facilitating factors regarding the SDL session and facilitators felt that students learned new terminologies in NHP. Fewer case scenarios and less space for writing in the module were the few challenges in the module. This was the Kirkpatrick model of level 1, which assesses the immediate reactions of the stakeholders. Table 4 Feedback from students, postgraduates, and faculties regarding the SDL session Students Postgraduates and Faculties Facilitating factors • Stressed on difficult topics for UG students • Time allotment for each topic was sufficient • Student-centered learning • Discussion with peers during activities • Avoids monotony of regular lecture classes • Continuous sessions on SDL • Module was simple and clear, easy to understand, simple language, well organized, easy to revise before exams • Module has problem-based questions in the assessment • Need a similar type of module for communicable diseases • Daily tests can be conducted • Students learned new terminologies in NHP • Both learning and writing practice was given • Marking of module and feedback by the facilitators Challenges • Only a few NHPs were included in the module. • There was less space for writing in the module and also contains fewer case scenarios • Students lost enthusiasm because of continuous SDL sessions Solutions • All topics in NHP can be included • Need more space to write after each question • Instruction page at the beginning of the module • Discuss how to present each question in the examination • SDL sessions can be scheduled once or twice a week. • Consensus can be developed for the selection of questions in the module • Questions in the module can be simplified. • Binding of the module can be done • Applied type of questions should be included more • Credits in the form of bonus marks for successful submission of the completed module to motivate the students • Post-test at the end of each day can be included. Feedback on the performance of questions on NHP in the University Exam (Kirkpatrick level 4) Feedback was also collected from the students after the completion of the university theory and practical examination regarding the SDL module on the National Health Programme. Although the program was implemented on 158 students, feedback after the University examination could be obtained only from 50 students. The module helped to recollect relevant points and many abbreviations in NHP to perform better in university theory and practical examination was the feedback received from the students. This was the Kirkpatrick model of level 4, which analyzes the final results. A male student had given feedback that. I was able to write two NHPs such as NPCDCS and RMNCH+A well only because of the SDL module, which helped me in last-minute revision and remembering the sub-topics under each program. [Table 5] Table 5 Feedback on the performance of questions on NHP in the University Exam (Kirkpatrick level 4) (n=50) • Module helped to recollect relevant points to perform better in university theory and practical examination. (18) • Module helped in last minute revision of NHP and remember the subtopics in each programme in exam. (16) • Two NHP such as NPCDCS and RMNCH + A were directly from the SDL module. (13) • With the help of the Module on NHP we were not new to many abbreviations in NHP in examination. (11) • Module helped to realize the importance of NHP at the level of UG. (10) • Without SDL module it would have not been possible to write about NHP in paper 2 Community Medicine theory examination. (8) DISCUSSION We developed, implemented, and evaluated module-based SDL on NHP. The current module-based SDL teaching demonstrated significant knowledge gains in National Health Programme among medical undergraduates. This was very well evident from the results of the end-of-module assessment, 118 (75%) students scored more than 50 percent. Further as informed by the students they could recollect and answer appropriately the questions related to NHP in the recently conducted summative examination by Pondicherry University. The facilitators felt that the module was simple, well-organized, and easy for the students to understand. Further, the problem-based questions in the module exercise were easy to understand and avoided the monotony of the lecture class. According to NMC, the number of hours allotted for SDL in Community Medicine in second and third-year MBBS was 20 and 5 hours respectively and it has been made compulsory in the curriculum. Similar SDL sessions were happening in the Department of Community Medicine in the Medical College of Delhi and CMC Vellore well before the new NMC curriculum.[6,7] Patra S et al.[6] in Delhi found that students were satisfied and motivated to study the allotted topic further and they also felt that facilitators could have been more active in imparting knowledge and skills. Previous studies showed that the SDL willingness between batches of students was declining, hence the current curriculum should promote SDL by increasing teaching-learning activities. Factors such as curriculum, assessments, and culture do impact SDL readiness.[8] Teaching students regarding SDL usually takes place in the experiential or co-curricular setting, the skills necessary for SDL should be introduced and developed in the didactic portion of the curriculum, which allows students to develop scaffolding. Flipped classrooms have the potential to move students toward self-directed learning and it is one of the strategies to develop self-directed learners.[9] A study showed that e-learning or blended learning requires SDL and may benefit students to know the goals of learning that may impact their engagement. In our study, we developed a module to facilitate SDL.[7] Kohan et al.[10] stated that higher levels of self-direction are essential for successful online learning in higher education institutes. The factors such as information overload, mind wandering, role ambiguity, inadequate coping skills, heavy workload, and inadequate writing skills were the barriers to self-directed learning. However, the study also identified facilitating factors, challenges, and solutions regarding SDL sessions. Some of the facilitating factors were a simple and clear module, which is easy to understand, simple language, well organized, easy to revise before exams and problem-based questions in the assessment. They also suggested the need for a similar type of module for communicable diseases. In the present study, the students felt that SDL sessions were effective which helped them to answer the questions on National Health Programme in the University examination. Facilitators felt that students learned new terminologies in NHP, they were given both learning and writing practice, and marking of modules and feedback by the facilitators was the facilitating factors. They also suggested developing consensus for the selection of questions in the module, simplifying questions in the module, binding the module, and including more applied types of questions. A study done in Delhi also reported positive feedback that sixty-seven percent of students were satisfied and 66% also reported as motivated to study the allotted topic further.[6] The gap between learners’ cognitive development and scientific reasoning must be bridged as a way forward toward a more accurate and integrated understanding of self-directed learning.[11] Our educational project helped students to find the answers to the learning objectives decided by them by thinking, searching, and group discussion. We have used a qualitative design and involved the students and faculties in finalizing the topic for SDL. The problem-solving activities planned during SDL sessions made learners utilize available resources, read, discuss, and come up with solutions, which they might not have done otherwise following lectures or small group teaching. Assessing SDL, which was also included in the module, which usually not done in the didactic teaching-learning process. Each group of students with allotted facilitators identified their objectives, resources, and teaching-learning activities, which might have created experiences that were not uniform for all the students. However, each student was a unique learner with their learning preferences. The SDL sessions can be further improved based on feedback from students, postgraduates, and faculties. Our study found that students enjoyed and were satisfied with the SDL sessions and the assessment methods. Factors such as simple and easily understandable modules, discussion with peers during activities, and assessment were the facilitating factors regarding SDL sessions. As recommended by the students, postgraduates, and faculties scheduling SDL sessions once or twice a week and a few changes in the module suggested were the prioritized action points to improve the SDL session further. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. Acknowledgment This study was done as a part of the Advanced Course in Medical Education (ACME) in Sri Ramachandra Institute of Higher Education and Research (SRIHER) Porur, Chennai. We would like to thank the faculty Dr. Dilara K and other participants of the ACME XI batch for their valuable input and support. We also thank the management of our college for providing permission to conduct this educational research. ==== Refs REFERENCES 1 Towle A Cottrell D Self-directed learning Arch Dis Child 1996 74 357 9 8669942 2 Badyal DK Lata H Sharma M Jain A Triple Cs of self-directed learning: Concept, conduct and curriculum placement CHRISMED J Health Res 2020 7 235 9 3 MacDougall M An overview of preparing medical students for self-directed learning in statistics: What should we expect of tomorrow's doctors? MSOR Connections 2011 11 18 22 4 Bhandari B Chopra D Singh K Self-directed learning:assessment of students abilities and their perspective Adv Physiol Educ 2020 44 383 6 32628525 5 Chaudhuri A Paul S Goswami A A comparative study to evaluate the role of interactive lecture classes and self-directed learning sessions among first MBBS students in the department of physiology during the implementation of competency-based medical education J Evid Based Med Healthc 2020 7 2714 8 6 Patra S Khan AM Upadhyay MK Sharma R Rajoura OP Bhasin SK Module to facilitate self-directed learning among medical undergraduates: Development and implementation J Educ Health Promot 2020 9 231 33209923 7 Sun W Hong JC Dong Y Huang Y Fu Q Self-directed learning predicts online learning engagement in higher education mediated by perceived value of knowing learning goals Asia-Pacific Edu Res 2022 32 307 16 8 Premkumar K Vinod E Sathishkumar S Pulimood AB Umaefulam V Prasanna Samuel P Self-directed learning readiness of Indian medical students: A mixed method study BMC Med Educ 2018 18 134 29884155 9 Robinson JD Persky AM Developing self-directed learners Am J Pharm Educ 2020 84 292 6 10 Kohan N Soltani Arabshahi K Mojtahedzadeh R Abbaszadeh A Rakhshani T Emami A Self-directed learning barriers in a virtual environment: A qualitative study J Adv Med Educ Prof 2017 5 116 23 28761885 11 Lapidow E Walker CM Rethinking the “gap”: Self-directed learning in cognitive development and scientific reasoning Wiley Interdiscip Rev Cogn Sci 2022 13 e1580 doi:10.1002/wcs. 1580 34619809

PMC010xxxxxx/PMC10353685.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-453 10.4103/ijcm.ijcm_165_22 Original Article Workplace Breastfeeding Support and Breastfeeding Practices among Healthcare Professionals Ranjitha R Maroof Khan Amir Rajoura Om Prakash Shah Dheeraj 1 Department of Community Medicine, University College of Medical Sciences and Guru Teg Bahadur Hospital, Delhi, India 1 Department of Pediatrics, University College of Medical Sciences and Guru Teg Bahadur Hospital, Delhi, India Address for correspondence: Dr. Khan Amir Maroof, Department of Community Medicine, University College of Medical Sciences and Guru Teg Bahadur Hospital, 414, 4th Floor, College Block, Dilshad Garden, Delhi - 110 095, India. E-mail: khanamirmaroof@yahoo.com May-Jun 2023 30 5 2023 48 3 453458 17 2 2022 13 4 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Introduction: Working mothers face striking challenges in breastfeeding. It is important to focus on them to further improve breastfeeding rates. Aim and Objectives: To assess the workplace breastfeeding support and breastfeeding practices of healthcare professionals. Methodology: We conducted a cross-sectional study among two hospitals in East Delhi. All mothers having at least one child aged six months to five years and currently employed as healthcare personnel were included. For a sample size of 100, population proportionate to size sampling was done among two hospitals. The participants were randomly selected from a list of healthcare personnel. Employee perception of breastfeeding support questionnaire (EPBS-Q) was used to assess the workplace breastfeeding support. Chi-square test was used to compare proportions, logistic regression, and survival analysis to find the association between workplace breastfeeding support and IYCF parameters. Results: The proportion of mother who perceived poor workplace breastfeeding support was 37%. The mean (SD) score obtained was 103.48 (8.93). The early initiation of breastfeeding within one hour was practiced by 54%, exclusive breastfeeding for at least six months by 60%, and timely initiation of complementary feeding for six to eight months by 64% mothers. Workplace breastfeeding support was significantly associated with exclusively breastfeeding for at least six months. Conclusions: More than one-third of mothers perceived poor workplace breastfeeding, and it was associated with exclusive breastfeeding. Breastfeeding maternity benefits workplace support ==== Body pmcINTRODUCTION World breastfeeding trends initiative assessment report 2018 stated “Arrested Development” in India’s policies and programs on infant and young child feeding (IYCF); that is, country has failed to make progress in key indicators in the years between 2015 and 2018.[1] The National Family Health Survey (NFHS-5) report for 2019–21 also confirms that only minimal change from NFHS-4 (2015-16) for IYCF parameters like early initiation of breastfeeding (41.8% from 41.6%) and timely initiation of complementary feeding (45.9% from 42.7%).[2] With more women joining the workplace, we should focus on working mothers to improve breastfeeding rates. Studies show that employment status affects the status of child feeding practices.[3,4] Comparative studies in India between working and non-working mothers substantiate that exclusive breastfeeding rates were lower among working mothers.[5] The workplace breastfeeding support includes organizational support, co-worker’s support, and time and infrastructure support. Various studies report an association between workplace breastfeeding support and breastfeeding practices.[6-9] The existing national programs related to child feeding are focused on mothers in the community and do not specifically focus on breastfeeding mothers in their respective workplaces. Therefore, we aimed to assess the mothers’ perceptions of workplace breastfeeding support and breastfeeding practices among healthcare professionals. METHODOLOGY We conducted a cross-sectional study among two purposively selected teaching hospitals of East Delhi, in 2019–21. Approval was obtained from the Institutional Ethics Committee–Human Research (IEC-HR) prior to the commencement of the study. All mothers having at least one child aged six months to five years and are currently employed as healthcare personnel such as faculty, resident, medical officers, lab technician, nurses, or medical social worker, and employed in the same hospital at the time of delivery were included in the study. Those mothers who were on leave during the whole period of study were excluded. To the best of our knowledge, there was no similar study published in India. Therefore, we did a pilot study on 20 mothers fulfilling the selection criteria and found that 60% of mothers perceived poor workplace breastfeeding support. For an estimated proportion of 60%, an allowable error of 10%, and for 95% confidence interval, the sample size obtained was 96, rounded to 100. The population proportionate to size sampling was done among the two hospitals, that is, 67 from hospital A and 33 from hospital B to obtain sample size. The participants were randomly selected from a list of healthcare personnel of the respective hospitals. A pretested, pre-validated, semi-structured schedule was used to collect data. It had four sections, viz. sociodemographic details, employee perception of breastfeeding support questionnaire (EPBS-Q), IYCF practices, details pertaining to maternity-related leave. Employee perception of breastfeeding support questionnaire (EPBS-Q) was used to assess the workplace breastfeeding support.[10] EPBS-Q is a validated questionnaire with a reliability of 0.68–0.89.[11] To make it adaptive to the local context, few words have been changed in EPBS-Q like “company” was replaced with “hospital”; “manager” was replaced with “head of department/in charge”; the word “pump” in relation to breast milk was replaced with “express.” The face and content validity were assessed by two experts associated with research and policymaking related to breastfeeding in India. The total score ranged from 40 to 160. For the purpose of the study, the midpoint of the range of scores was taken as cutoff for categorizing as good (total scores ≥100) and poor (<100) support. The questionnaire has four sections, and the number of questions varies in each section; therefore, the scores vary in each section. To be able to make a valid comparison between the scores of the sections, we computed standardized mean scores for each section considering a base of 10 questions, that is, a score of 40. Statistical analysis: Data was entered in Microsoft Excel and cleaned. The cleaned data were imported into SPSS 20.0 software. Categorical variables such as breastfeeding status, maternity leave, workplace breastfeeding support, etc., are presented in the form of proportions, and continuous variables such as age, income, breastfeeding duration, scores of EPBS-Q, etc., as means and standard deviations or median (IQR). Chi-square test was used to compare proportions. Binomial logistic regression was done for finding out the independent predictors of mothers’ perception of workplace breastfeeding support. Variables with P value less than 0.25 by the bivariate analysis were included in multivariate analysis. Survival analysis was done to find the association between the duration of breastfeeding and workplace breastfeeding support. Log-rank test was used to statistically compare the survival curves. All statistical tests were two-tailed. P - value less than 0.05 was considered statistically significant. RESULTS Majority (73%) of the participants were nurses, 16% were doctors, and the rest 11% were other supporting staff like lab technicians, medical social workers. Most of the mothers (80%) were graduates or higher degree holders, and the median family income per month was INR 100000 [Table 1]. Table 1 Sociodemographic details of healthcare professionals who were mothers of children 6–60 m from two teaching hospitals of East Delhi (n=100) Variables Percentages Age of the mothers (in years)  Median (IQR) 32 (30,35) Educational status of mothers  Professional 16%  Postgraduate 12%  Graduate 52%  Diploma 20% Designation of mothers  Doctors 16%  Nurse 73%  Other supporting staffs 11% Distance from the residence to the workplace (km)  Median (IQR) 5 (2,13.75) Type of family  Nuclear 60%  Joint 40% Number of family members  Median (IQR) 4 (4,5) Total family income per month (INR)  Median (IQR) 100,000 (100,000, 200,000) Sex of the child  Male 54%  Female 46% Age of the child (in months)  Median (IQR) 27 (13,42.75) IQR-Interquartile range More than one-third of the participants (37%) were found to have poor workplace breastfeeding support and 63% of them perceived good workplace breastfeeding support. The total possible score range for workplace support from EPBS-Q was 40 to 160, and the observed range was from 76 to 130. The standardized mean score for managerial support, organization support, time and physical space, and co-worker support were 27.30, 26.46, 22.45, and 18.13, respectively. [Table 2] Among these various domains, distribution of responses which contributed to the poorer score was shown in Figure 1. Table 2 Employee perception of breastfeeding support (EPBS) score among mothers of children 6–60 m employed in two selected teaching hospitals of East Delhi (n=100) Domains Minimum score Maximum score Mean score (SD) Mean score (standardized) Organizational support 22 38 29.14 (3.08) 26.46 Managerial Support 23 45 32.56 (3.17) 27.30 Co-worker support 11 23 16.88 (1.97) 18.13 Time and physical space 11 34 24.70 (4.31) 22.45 Total EPBS Score 76 130 103.48 (8.93) 94.34 SD- Standard Deviation Figure 1 Distribution of responses for questions under various domains of employee perception of breastfeeding support (EPBS) questionnaire among study participants (n = 100) The early initiation of breastfeeding within one hour was practiced by 54% of the mothers. Exclusive breastfeeding for at least six months was done by 60% of mothers. Timely introduction of complementary feeding for six to eight months was done 64% by mothers. Maternity leave was availed by 92% of the participants. The next common leave was Child Care Leave (CCL), 50 out of 100 participants. The median duration of maternity leave availed by the mothers was 6 months ranging from 2 months to 6 months. Only around one out of ten mothers were aware of the availability of lactation rooms in the workplace, that is, hospitals. Table 3 shows the association of workplace breastfeeding support and IYCF practices and certain sociodemographic features. The proportion of mothers practicing exclusive breastfeeding for at least six months was higher (70%) among mothers who perceived good workplace breastfeeding support as compared to those who perceived workplace breastfeeding support as poor. Table 3 Association of workplace breastfeeding support with infant and young child feeding practices, certain sociodemographic, and work-related factors among mothers of children 6–60 m employed in two selected hospitals of East Delhi (n=100) Workplace breastfeeding support OR (95% CI) P Good support (n=63) Poor support (n=37) Exclusive breastfeeding (months)  ≥6 44 (73%) 16 (27%) 3.03 (1.30–7.06) 0.009*  <6 19 (47%) 21 (53%) Initiation of complementary feeding (months)  6 to 8 42 (66%) 22 (34%) 1.36 (0.58–3.15) 0.46  <6 and >8 21 (58%) 15 (42%) Age (years)  ≤32 36 (67%) 18 (33%) 1.40 (0.34-4.76) 0.41  >32 27 (59%) 19 (41%) Designation of mothers  Doctors 9 (56%) 7 (44%) 0.71 (0.24–2.11) 0.54  Nurses & others 54 (64%) 30 (36%) Distance between residence and workplace (Km)  ≤5 37 (67%) 18 (33%) 1.50 (0.66–3.44) 0.32  >5 26 (58%) 19 (42%) Night shift performed  Yes 59 (64%) 32 (36%) 1.25 (0.36–4.26) 0.75#  No 7 (58%) 5 (42%) Working hours per week (hrs.)  <48 16 (64%) 9 (36%) 1.05 (0.41-2.71) 0.90#  ≥48 47 (62%) 28 (48%) Awareness on availability of lactation room  Yes 8 (88%) 1 (12%) 5.23 (0.62–43.66) 0.14#  No 55 (60%) 36 (40%) *Statistically significant #Fisher’s exact test Multivariate binomial logistic regression for analysis of factors for workplace breastfeeding support revealed that odds of perceiving poor workplace breastfeeding support was 3.8 times (CI = 1.61 – 9.35) higher among those who did not exclusively breastfeed for six months compared to those who exclusively breastfeed for at least six months (P = 0.002). Figure 2 shows that mothers who perceived good workplace breastfeeding support were breastfeeding longer than the mothers who perceived poor workplace breastfeeding support though it was not statistically significant (P = 0.13). Figure 2 Survival curves depicting the probability of continuation of breastfeeding with respect to the perceived workplace breastfeeding support among mothers of children 6–60 m employed in the teaching hospitals of East Delhi DISCUSSION The proportion of mothers with poor workplace breastfeeding support was 37%. The total observed score ranged from 76 to 130, and the mean (SD) score from our study was 103.48 (8.93). Scott et al.,[9] in 2019, done a study in North Carolina among employees of a large integrated healthcare system using EPBS-Q, revealed that the total observed score ranged from 63 to 158. Another study by Waite et al.,[12] done in Seattle children’s hospital among female employees who had a child born within the last five years, in 2015 revealed that observed scores range from 89 to 156, and the mean (SD) total workplace support score was 124.5 (14.99). The maximum observed score was around 150 in the above-quoted studies, whereas in our study, the maximum was only 130. The observed scores were low in our study. The difference in the lactation program, maternity benefits, working hours, etc., among the countries might be the reason for this difference. Workplace characteristics like working hours, distance between workplace and residence were associated with workplace breastfeeding support in other studies which shows significant difference between full-time workers and part-time workers[13], and also among multinational companies providing job adjustments like relocating mothers’ workplace near to their home were found to have higher workplace breastfeeding support than national companies.[14] In our study, the median (IQR) of working hours was 48 (46, 48). The range of working hours among our study participants was very less, as majority of our study participants were nurses, who have uniform working schedule, and also, no such job adjustments were given in our study hospitals. As there was a lack of comparative group in our study, there was no significant association among mothers who perceived good and poor workplace breastfeeding support in terms of their age, occupation (doctors/nurses and supporting staff), distance between residence and workplace, working hours per week or night shift performed. In India, there were no policies that reduce the working hours per week during post-partum period. Workplace breastfeeding support is multidimensional and, in our study, co-workers support, time, and physical space support were lacking majorly. About 77% of the mothers disagreed that their co-workers helped them continue breastfeeding. More than one-third mothers felt that they could not able to adjust their break schedules in order to breastfeed or express breast milk. A study which compared workplace breastfeeding support among faculty and administrative staffs found significant difference to access to breastfeeding breaks.[15] In our present study, the survival analysis revealed that the median duration of breastfeeding among mothers who perceived good workplace breastfeeding support was higher when compared to those who had poor support, yet it was not statistically significant. Multiple other studies using different scales for workplace breastfeeding support have revealed an association between support and duration of overall breastfeeding[6,16] and also return to work is the reason for discontinuation of breastfeeding.[17] Sattari et al’s study showed each unit increase in organizational support increased breastfeeding duration by 1.3 months.[18] The decreased access to lactation room among our study participants might be the reason for that. A study done in Minnesota found that women with access to workplace accommodations to support breastfeeding were 1.5 times as likely to continue breastfeeding compared with women without access.[19] In our study, only one in ten mothers reported the availability of lactation rooms. Both hospitals in our study had lactation rooms, and these lactation rooms were available for both the general public who were admitted in hospitals and also to employees. Weber et al. (2011)[6] study about the female employees’ perceptions of organizational support for breastfeeding at work found that only a few (19%) had access to a room specially designated for breastfeeding. Non-availability of lactation rooms might be one of the reasons for mothers to think cessation of breastfeeding before returning to work. Women working in other informal and non-hospital settings might have a lesser probability of access to the lactation room. A comparative study was done among hospital and non-hospital employees and found there was a significant association between them for workplace breastfeeding support.[6] Our study population was from a homogenous group, that is, hospital setting, and this might be the reason that our study didn’t find any association. Even if lactation room was available, the additional supportive features like provision of nursing breaks, creche facility needs to be reiterated. The Maternity Benefits Act requires employers to provide fully paid nursing breaks until a child reaches the age of fifteen months and creche facility.[20] The current operational positions of these laws require assessment. More than half (60%) of the mothers practiced EBF for six months. This was similar to the NFHS-5 report for India and Delhi and also to other studies done in Delhi.[2,21-23] There was an association between exclusive breastfeeding for six months and workplace breastfeeding support in our study which was supported by studies done using the same and also different questionnaires.[6,9] Although less sample size was a limitation, our study contributes to identifying the domains which were perceived as poor workplace breastfeeding support and requires improvement at the local level. There is a possibility of recall bias for certain IYCF indicators and maternity leave details, as the mothers who were having at least one child of fewer than 5 years were included, and also the validation study of the EPBS-Q also done for mothers with child of less than 2 years. The internal reliability of the scale for our study was found to be high (Cronbach’s alpha = 0.84). The EPBS-Q score cutoff considered by us for good and poor breastfeeding support was arbitrary and could have implications in the analysis. However, to avoid error on either side, we considered the cutoff as the midpoint of minimum and maximum possible scores. Another limitation was a lack of a qualitative component in our study which could have given more insight into the reasons for the responses given by the mothers. Also, we did not capture the employer’s perspectives in this study. CONCLUSION We found that more than one-third of mothers of children 6–60 m perceived poor workplace breastfeeding support. Exclusive breastfeeding till six months of age was associated with perceived workplace breastfeeding support in our study. More studies are needed from India on the impact of workplace support and breastfeeding. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. Acknowledgements Dr Praveen Kumar Malik (Additional Medical Superintendent, GTB Hospital), Dr Rajni Khedwal (Medical Superintendent, Swami Dayanand Hospital) & Dr Surender Singh Bisht (Department of Pediatrics, Swami Dayanand Hospital). ==== Refs REFERENCES 1 World Breastfeeding Trends Initiative. ‘Arrested Development’ 5th Report of Assessment of India's Policy and Programmes on Infant and Young Child Feeding New Delhi 2018 1 96 2 International Institute for Population Sciences. National Family Health Survey (NFHS-5) 2019-2021 India 2021 Available from: http://rchiips.org/nfhs/factsheet_NFHS-5.shtml [Last accessed on 2023 May 01] 3 Amin R Said Z Sutan R Shah S Darus A Shamsuddin K Work relate determinants of breastfeeding discontinuation among employed mothers in Malaysia Int Breastfeed J 2011 6 4 9 21342506 4 Cooklin AR Donath SM Amir LH Maternal employment and breastfeeding: Results from the longitudinal study of Australian children Acta Paediatr 2008 97 620 3 18394107 5 Amin R Said Z Sutan R Shah S Darus A Shamsuddin K Work relate determinants of breastfeeding discontinuation among employed mothers in Malaysia Int Breastfeed J 2011 6 4 9 21342506 6 Weber D Janson A Nolan M Wen LM Rissel C Female employees'perceptions of organisational support for breastfeeding at work: Findings from an Australian health service workplace Int Breastfeed J 2011 6 1 7 21281517 7 Tsai SY Employee perception of breastfeeding-friendly support and benefits of breastfeeding as a predictor of intention to use breast pumping breaks after returning to work among employed mothers Breastfeed Med 2014 9 16 23 24304034 8 Bai Y Wunderlich SM Lactation accommodation in the workplace and duration of exclusive breastfeeding J Midwifery Womens Health 2013 58 690 6 24325729 9 Scott VC Taylor YJ Basquin C Venkatsubramanian K Impact of key workplace breastfeeding support on job satisfaction, breastfeeding duration and exclusive breastfeeding duration among health care employees Breastfeed Med 2019 14 416 23 30994382 10 Greene SW Olson BH Development of an instrument designed to measure employees'perceptions of workplace breastfeeding support Breastfeed Med 2008 3 151 7 18778209 11 Greene SW Wolfie ED Olson BH Assessing the validity of meaure of an instruments designated to measure employees perception of workplace breastfeeding support Breastfeed Med 2008 3 159 63 18778210 12 Waite M W Christakis D Relationship of maternal perceptions of workplace breastfeeding support and job satisfaction Breastfeed Med 2015 10 222 31 25831141 13 Omer-Salim A Suri S Dadhich JP Faridi MMA Olsson P 'Negotiating the tensions of having to attach and detach concurrently': A qualitative study on combining breastfeeding and employment in public education and health sectors in New Delhi, India Midwifery 2015 31 473 81 25660847 14 Soomro JA Shaikh ZN Saheer TB Employers'perspective of workplace breastfeeding support in Karachi, Pakistan: A cross-sectional study Int Breastfeed J 2016 11 223 9 15 Abdulwadud OA Snow ME Intervention in the workplace to support breastfeeding for women in employment (re-view) Cochrane Database Syst Rev 2012 10 1 13 16 Lutter CK Morrow AL Protection, Promotion, and Support and Global Trends in Breastfeeding Adv Nutr 2013 4 213 9 23493537 17 Bar-Yum NB Workplace lactation support, part I: Return to work breastfeeding assessment tool J Hum Lact 1998 14 249 54 10205439 18 Sattari M Serwint J Neal D Chen S Levine DM Work-place predictors of duration of breastfeeding among female physicians J Pediatr 2013 163 1612 7 24011764 19 Kozhimannil KB Jou J Gjerdingen DK McGovern PM Access to workplace accommodations to support breastfeeding after passage of the Affordable Care Act Womens Health Issues 2015 26 6 13 20 Ministry of labour and employment, Government of India. Maternity Benefit (amendment) Act, 2017 New Delhi 2017 Available from: https://labour.gov.in/sites/default/files/Maternity%20Benefit%20Amendment%20Act%2C2017%20.pdf [Last accessed on 2022 Jan 28] 21 Harne P Batra P Faridi MMA Dewan P Optimal infant and young child feeding practices among working women: A challenge Breastfeed Med 2013 8 511 2 23905756 22 Khan AM Kayina P Agrawal P Gupta A Kannan AT A study on infant and young child feeding practices among mothers attending an urban health center in East Delhi Indian J Public Health 2012 56 301 4 23354143 23 Boralingiah P Polineni V Kulkarni P Manjunath R Study of breastfeeding practices among working women attending a tertiary care hospital, Mysore, Karnataka, India Int J Community Med Public Heal 2016 3 1178 82

PMC010xxxxxx/PMC10353686.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-413 10.4103/ijcm.ijcm_839_22 Original Article Impact of Structured Training Program about Cadaver Organ Donation and Transplantation on Knowledge and Perception of Nursing Students at Public and Private Nursing Teaching Institute of Northern India - An Interventional Study Singh Sukhbir Kaur Kiran 1 Saini Ravinder S. 2 Singh Sunita 1 Aggarwal H. K. 3 Chandra Hem 4 Research Scholar, H. N. Bahuguna Uttarakhand Medical Education University, Dehradun and Sr. Professor, Department of Hospital Administration, Pandit Bhagwat Dayal Sharma, Post Graduate Institute of Medical Sciences, Rohtak, Haryana, India 1 College of Nursing, Pandit Bhagwat Dayal Sharma, University of Health Sciences, Rohtak, Haryana, India 2 Department of Hospital Administration, Himalyan Institute of Medical Sciences, Swami Rama Himalayan University, Dehradun, Uttarakhand, India 3 Department of General Medicine, Pandit Bhagwat Dayal Sharma, Post Graduate Institute of Medical Sciences, Rohtak, Haryana, India 4 Vice-Chancellor, H. N. Bahuguna Uttarakhand Medical Education University, Administrative Block, Govt. Doon Medical College Campus, Dehradun, Uttarakhand, India Address for correspondence: Dr. Sukhbir Singh, Department of Hospital Administration, Pt. B. D. Sharma Postgraduate Institute of Medical Sciences, Rohtak - 124 001, Haryana, India. E-mail: drbrar1980@gmail.com May-Jun 2023 30 5 2023 48 3 413417 09 10 2022 15 2 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Background: “Donation gap” refers to the shortage of organ donors worldwide. The medical/nursing students and various healthcare workers have poor awareness and attitude toward organ donation. Objective: We conducted this study to assess the current level of knowledge and perception regarding cadaver organ donation and transplantation among nursing students and to evaluate the impact of structured training interventions on their baseline knowledge and perception level. Methods: It was a single-group pre-post interventional study done by nursing students of one government and one private nursing college.A pre-tested questionnaire was used as a study tool. Statistical Analysis: Various statistical tests like one-way repeated measure ANOVA, Mauchly’s test of sphericity, and Greenhouse–Geisser correction were used. Pairwise comparisons used Bonferroni corrections. Results: The pre-test group had the lowest mean knowledge (50.2346, SD = 15.35188), and immediately after training group had the highest (57.3900, SD = 14.34626). After one month, knowledge decreased but was still higher than pretraining (mean = 52.3607, SD = 13.28141). Conclusions: The positive attitude of nursing students may augment cadaver organ donation and transplantation in the future. The study has also highlighted the further training needs of the participants. Cadaver organ donation healthcare workers (HCWs) knowledge nurse organ transplantation perception structured training program World Health Organization (WHO) ==== Body pmcINTRODUCTION “Donation gap” refers to the shortage of organ donors worldwide.[1] According to WHO,[2,3] developed countries have an organ donation rate of 20–30 per million (i.e. 70%–80%) compared to 0.26 per million in India.[4] People’s misconceptions about cadaveric organ donation also contribute to low organ donation in our country. Lack of legal and procedural knowledge also causes poor organ donation.[5-7] As the first contact with the donor family, HCWs can persuade them to donate. Living organ donations are the main source, but trade has continued despite national and international laws.[8-10] Donating cadaveric body organs is the only way to increase organ donations. Few studies[11,12] reported that people have poor organ donation knowledge and attitudes. Similarly, medical/nursing students and various healthcare workers have poor awareness and attitude toward organ donation, and physicians’ positive attitude may inspire the public and can improve donor rates.[13-15] This study assessed nursing students’ knowledge and perception of cadaver organ donation and transplantation in a government and private institution. It evaluated the impact of structured training interventions on their baseline knowledge and perception. This study received Institute Ethics Committee approval. MATERIALS AND METHODS Study design It was a pre-post single-group intervention study. This study was approved from Institute Ethics Committee vide No. BREC/21/021, dated 16.04.2021. Participants and sample size This study was conducted in the year 2021 among the nursing students of one government nursing college and one private nursing college in Haryana state. All nursing students in the study institutions (n = 481) were recruited [Table 1]. Non-attending and non-consenting participants were excluded from the study. Table 1 Distribution of participants as per their sociodemographic profile Parameter Count Institution  Govt. College of Nursing 275 (81%)  Pvt. College of Nursing 66 (19%) Age  <=20 145 (43%)  21-25 165 (48%)  26-30 26 (8%)  31-35 4 (1%)  >35 1 (0.30%) Gender  Female 335 (98%)  Male 6 (2%) Place of Residence  Rural 176 (52%)  Urban 165 (48%) Type of Family  Joint 86 (25%)  Nuclear 255 (75%) Educational Qualification  BSc 1st 87 (26%)  BSc 2nd 79 (23%)  BSc 3rd 23 (7%)  BSc 4th 61 (18%)  MSc 1st 35 (10%)  MSc 2nd 14 (4%)  Post Basic 1st 30 (9%)  Post Basic 2nd 12 (4%) Educational Status of Father  <10th 15 (4%)  10th 92 (27%)  12th 93 (27%)  Graduate 92 (27%)  Postgraduate 19 (6%)  Diploma/others 30 (9%) Educational Status of Mother  <10th 47 (14%)  10th 110 (32%)  12th 74 (22%)  Graduate 44 (13%)  Postgraduate 15 (4%)  Diploma/others 51 (15%) Study tool After a thorough literature review, a structured questionnaire was created to assess cadaver organ donation and transplantation knowledge and perception. The survey had three parts. The first part included the participant’s sociodemographic details like age, gender, education level, residence, etc., The second part tested cadaver organ donation and transplantation knowledge. The third part used a five-point Likert scale to measure respondents’ perceptions of the study topic. Ten experts pilot-tested the questionnaire’s content, applicability, comprehension, etc., Twenty knowledge- and fifteen perception-based questions covered different aspects of cadaver organ donation and transplantation, including the associated legal and ethical issues. Tool administration Before the training session, the study tool was administered digitally to evaluate nursing students’ baseline knowledge and perceptions of the study topic. The training program included IEC material and audio-visual classroom lectures. The researcher and other faculty trained participants used the Zoom platform. The same study tool was digitally administered immediately and one month after training to evaluate its impact. Before each study, participants gave digital informed consent. Statistical analysis All knowledge-assessment questions were scored. Correct answers were worth five points. Incorrect answers were not penalized. The Likert scale for assessing participants’ perceptions scored 05 to 01 for a strongly agree to disagree response. There were three groups based on three different periods, i.e. pre-training, immediately after training, and post-one month after training. Three-time periods’ knowledge and perception scores were calculated. Overall mean and question-by-question pre-test, post-test, and one month after post-test differences were calculated. We used one-way repeated measure ANOVA to compare knowledge and perception scores across periods. To use one-way repeated measure ANOVA, we checked its assumption as sphericity or equality of variance between each pair of three time periods. Mauchly’s test of sphericity tests sphericity formally. Mauchly’s W = 0.999 and P = 0.854 for knowledge and W = 0.974009 and P = 0.011519 for perception. Mauchly’s sphericity test showed that data related to perception assessment had violated sphericity (P < 0.05). The data related to mean knowledge score had not violated the sphericity; hence, one-way repeated measure ANOVA was used in mean knowledge scoring. However, the data related to mean perception score violated sphericity; hence, one-way repeated measure ANOVA with Greenhouse–Geisser correction was used in mean perception scoring. Pairwise comparisons used Bonferroni corrections. RESULTS There were a total of 341 participants in the study. Table 1 shows the respondents’ age, sex, education, residence, institution, and family type, and 48% of participants were 21–30 years old, 98% were females, 74% were pursuing BSc, 81% were from government institutions, 52% were from rural areas, and 75% were from nuclear families. In most questions, the correct percentage increases immediately after training and decreases after one month of training. Pre-training, immediately after, and post-one-month training groups were created. All three time periods had knowledge and perception scores. The pre-test group had the lowest mean knowledge (50.2346, SD = 15.35188), and the immediately after training group had the highest (57.3900, SD = 14.34626). Perception decreased after one month but was higher than the pre-training level (mean = 52.3607, SD = 13.28141). Mean knowledge scores differed significantly (f = 32.891, P = 0.0003). Similarly, the mean perception score was lowest in the pre-test group (mean = 52.1701, SD = 4.17204), and immediately after training group had the highest mean perception (54.8211, SD = 7.14103). Perception decreased after one month but was higher than the pre-training level (mean = 53.0968, SD = 6.85132). The mean perception score differed significantly (f = 17.134812, P = 0.00002) [Table 2]. Table 2 Mean knowledge and perception score in different time groups Descriptive Statistics Parameter Mean Sth. Deviation n F, P Mean Knowledge (Pre-training) 50.2346 15.35188 341 32.891, 0.0003 Mean Knowledge (Immediately after Training) 57.3900 14.34626 341 Mean Knowledge (One-month post-Training) 52.3607 13.28141 341 Mean Perception (Pre-training) 52.1701 7.33728 341 17.134812, 0.00002 Mean Perception (Immediately after Training) 54.8211 7.14103 341 Mean Perception (One-month post-Training) 53.0968 6.85132 341 The pairwise comparison of mean knowledge was made using Bonferroni corrections. There was a significant difference in mean knowledge score when the pre-test group was compared with the immediate after training (mean difference = -7.1554, SE = 0.91416, P = .00001), post-one-month training group (mean difference = -2.1261, SE = 0.91177, P = .06088). The immediate after the training group was compared with the pre-test (mean difference = 7.1554, SE = .91416, P = .00001) and post-one month after the training group (mean difference = 5.0293, SE =.89224, P = .00002). The post-one month after the training group was compared with the pre-test (mean difference = 2.1261, SE =.91177, P = .06088) and immediately after the training group (mean difference = -5.0293, SE =.89224, P = .00002) [Table 3]. Table 3 Pairwise comparison of mean difference in knowledge and perception score Pairwise Comparisons Measure: Knowledge (I) Time (J) Time Mean difference (I-J) Std. error P Pre-training Immediate after training -7.1554 0.91416 0.00001 One-month post-Training -2.1261 0.91177 0.06088 Immediate after Training Pre-training 7.1554 0.91416 0.00001 One-month post-Training 5.0293 0.89224 0.00002 One-month post-Training Pre-training 2.1261 0.91177 0.06088 Immediate after training -5.0293 0.89224 0.00002 Measure: Perception Pre-training Immediate after training -2.6510 0.46859 0.00002 One-month post-Training -0.9267 0.48551 0.17143 Immediate after training Pre-training 2.6510 0.46859 0.00002 One-month post-Training 1.7243 0.42250 0.00017 One-month post-Training Pre-training 0.9267 0.48551 0.17143 Immediate after training -1.7243 0.42250 0.00017 Adjustment for multiple comparisons: Bonferroni Similarly, on pairwise comparison using Bonferroni corrections, a significant difference in mean perception score was observed when the pre-test group was compared with the immediate after training (mean difference = -2.6510, SE = 0.46859, P = .00002) and post-one-month training group (mean difference = -0.9267, SE = 0.48551, P = .17143), and the immediate after the training group was compared with pre-test (mean difference = 2.6510, SE =.46859, P = .00002), post-one month after training group (mean difference = 1.7243, SE =.42250, P = .00017), and post-one month after training group was compared with the pre-test (mean difference = 0.9267, SE =.48551, P = .17143), immediately after training group (mean difference = -1.7243, SE =.42250, P = .00017) [Table 3]. DISCUSSION All nursing professionals must have adequate knowledge about cadaver organ donation and transplantation, including the associated legal and ethical issues. However, not many studies related to nursing students are available on this subject in the indexed literature. Therefore, this study was conducted among nursing students in one government and one private nursing college in Haryana to map their knowledge and perceptions about cadaver organ donation and transplantation and to see the impact of the structured training program on it. We discovered that 91% of respondents were aware of organ donation from deceased individuals. This finding is comparable to the Goa study,[16] in which 91.5% of participants reported familiarity with cadaver organ donation, but slightly lower than those of the Delhi[17] and Bangalore[18] studies (96% and 99%, respectively). In the present study, 26% of participants indicated that television was the most common source of information about organ donation from deceased donors, followed by scientific journals (24%), newspapers (16%), radio (1%), and other sources (36%). This finding is supported by a cross-sectional study[19] of medical students in which television was reported as the most common source of information about the study topic. However, these findings differ from those of the Goa study,[16] where newspapers (44.8%), followed by television (40.5%), were reported as the most common source of information about cadaveric organ donation. In the current study, only nursing students were included, whereas consultants, resident doctors, and nurses were included in the Goa study. Consequently, the difference in findings can be attributed to the difference in samples. In the current study, 41% of respondents stated that cadaver organ donation is unpopular due to religious reasons, followed by socio-cultural factors (31%), expensive treatment (10%), inadequate hospital infrastructure (7%), and legal issues (5%). The conclusion is supported by a study conducted in Turkey,[20] which found a statistically significant correlation between religious attitude and the sub-dimension of apprehension of medical neglect on the Organ Donation Attitude Scale. In contrast, 3.6% of participants in the Goa study[16] cited their “religious beliefs” as their unwillingness to donate. Differences in the results may be attributable to differences in sample size and regional variation, and 82% of respondents knew that cadaver liver, kidney, pancreas, heart, lung, and intestines can be donated. In comparison, 57% knew that cadaver cornea, bone, skin, heart valve, blood vessels, nerves, and tendons could also be donated. In contrast, participants in the Goa study[16] revealed that the cornea was the most frequently donated organ (89.3%), followed by the kidney (80%), heart (68.3%), and lungs (25.3%). In another study[7] conducted in South India, kidney donation awareness was found to be the highest (94%), followed by heart (82%), liver (78%), cornea (59%), and lungs (55%) donation awareness. The Goa study and the South India study did not assess the awareness of cadaveric tissue donation. We found that the structured training program positively impacts participants’ knowledge and perception of participants. The overall score was highest after training, followed by one month after training, and was lowest before training. This finding conforms with the results of a few other studies,[21,22] wherein the researchers reported the positive impact of the training program on the study topic. Although our study has shown an overall improvement in the knowledge and perception score among the participants due to training intervention, it has also thrown light on the domains where the participants performed well and areas where their performance was moderate. The improvement after training was moderate on the topics like the category of patients fit for cadaver organ donation, coordination of deceased organ donation activities in hospitals, the organization responsible for promoting cadaver organ donation in the state, legislation covering the cadaver organ donation and transplantation, clinical findings for declaring a patient as brain stem dead, and correct organ and tissue donation form for pledging the organ as per Act. Goa’s study[16] reported that 94% of the participants knew that brain-dead people could donate organs. This finding is higher than our study (52%). Similarly, in our study, it was reported that 62% of participants were not aware of the legislation covering cadaver organ donation and transplantation. However, the knowledge of this aspect among our study participants was better than the finding of another study,[23] where it was reported that 94% of the participants were unaware of the organ donation law. Hence, the current study has also revealed the future training needs of the participants. CONCLUSIONS This study has shown the positive impact of relevant training on the knowledge and perception of nursing students about cadaver organ donation and transplantation. The positive attitude of nursing students may augment cadaver organ donation and transplantation in the future. The study has also highlighted the further training needs of the participants. The importance of frequent training is also concluded in the study. Strength of the study The study was multicentered, involving both government and private nursing institutions. The impact of training intervention was followed over a period of time. However, in the related studies available in indexed literature, only the immediate effect of training was assessed among the participants. This study is also novel for the Haryana state as no such study from this area is available in the indexed literature. Limitation of the study The study was conducted only among nursing students. No other category of HCW was enrolled in the study. Hence, the results of the study cannot be interpolated on other healthcare workers. Recommendations Regular training of nursing students on cadaver organ donation and transplantation, focusing on areas of moderate performance. Plan more multicenter studies with nursing institutions. Other categories of HCWs must be trained. Further research may determine if classroom training increases cadaver organ donation. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. ==== Refs REFERENCES 1 Hamed H Awad ME Youssef KN Fouda B Nakeeb AE Wahab MA Knowledge and attitudes about organ donation among medical students in Egypt: A questionnaire J Transplant Technol Res 2016 6 155 doi:10.4172/2161- 0991.1000155 2 Sugumar JP Padhyegurjar MS Padhyegurjar SB An interventional study on Knowledge and attitude regarding organ donation among medical students Int J Med Sci Public Health 2017 6 402 8 3 Meghana SG Subramanian M Atmakuri SA Tarun S Bera P Nelson J A study on Knowledge, attitude and practice regarding organ donation and transplantation among final year health science students in Bengaluru, Karnataka, India Int J Community Med Public Health 2018 5 1529 34 4 Kishore Y Jothula Sreeharshika D Study to assess Knowledge, attitude and practice regarding organ donation among interns of a medical college in Telangana, India. Int J Community Med Public Health 2018 5 1339 45 5 Hamed H Elhosseny Awad M Knowledge and attitudes about organ donation among medical students in Egypt: A questionnaire J Transplantation Tech Res 2015 6 12 4 6 Kocaay A Celik S Eker T Oksuz N Akyol C Tuzuner A Brain death and organ donation: Knowledge, awareness, and attitudes of medical, law, divinity, nursing, and communication students Transplant Proc 2015 47 1244 8 26093691 7 Alex P Kiran KG Baisil S Badiger S Knowledge and attitude regarding organ donation and transplantation among medical students of a medical college in South India Int J Community Med Public Health 2017 4 3449 54 8 Rudge C Matesanz R Delmonico FL Chapman J International practices of organ donation Br J Anaesth 2012 108 i48 55 22194431 9 Transplantation of Human Organs and Tissues Rules, 2014. MOHFW, The Gazette of India: Extraordinary Available from: https://main.mohfw.gov.in/acts-rules-and-standards-health-sector/acts/transplantationhuman-organs-acts-and-rules [Last accessed on 2022 Sep 28] 10 Highlights of National Organ and Tissue Transplant Programme and Operational Guidelines for its implementation. NOTP Cell, DGHS, MHFW, GOI Available from: https://notto.gov.in/WriteReadData/Portal/News/95_1_guidelines.pdf [Last accessed on 2022 Sep 28] 11 Mithra P Ravindra P Unnikrishnan B Rekha T Kanchan T Kumar N Perceptions and attitudes towards organ donation among people seeking healthcare in tertiary care centers in coastal South India Indian J Palliat Care 2013 19 83 7 24049347 12 Annadurai K Mani K Ramasamy J A study on Knowledge, attitude and practices about organ donation among college students in Chennai, Tamil Nadu -2012 Prog Health Sci 2013 3 59 65 13 Sucharitha ST Siriki R Dugyala RR Mullai P Roshini K Organ donation: Awareness, attitudes, and beliefs among undergraduate medical students in South India Natl J Res Community Med 2013 2 79 148 14 Marqués-Lespier JM Ortiz-Vega NM Sánchez MC Sánchez MC Soto-Avilés OE Torres EA Knowledge of and attitudes toward organ donation: A survey of medical students in Puerto Rico P R Health Sci J 2013 32 187 93 24397216 15 Ali N Qureshi A Jilani B Zehra N Knowledge and ethical perception regarding organ donation among medical students BMC Medical Ethics 2013 14 38 doi:10.1186/1472-6939-14-38 24070261 16 Da Silva KX Dsouza DB Mascarenhas VR Kankonkar PN Vaz FS Kulkarni MS Perceptions and attitude toward cadaveric organ donation among healthcare professionals at a tertiary health-care setting: Across-sectional study Indian J Transplant 2021 15 56 61 17 Reddy AV Guleria S Khazanchi RK Bhardwaj M Aggarwal S Mandal S Attitude of patients, public, doctors and nurses towards organ donation Transplant Proc 2003 35 18 12591287 18 Bapat U Kedlaya PG Gokulnath Organ donation, awareness, attitudes and beliefs among post graduate medical students Saudi J Kidney Dis Transpl 2010 21 174 80 20061720 19 Ghaffari M Latifi M Najafizadeh K Rakhshanderou S Courtney R Ramezankhani A Effects of interventions on organ donation among adults: A systematic review from 2000-2016 Transplantation 2017 101 S36 20 Soylu D Özdemir A Soylu A Does religious attitude affect organ donation and transplantation? Transpl Immunol 202271:101555 doi: 10.1016/j.trim. 2022.101555 21 Samata Srinivasula AS Doshi D Reddy BS Kulkarni S Influence of health education on knowledge, attitude, and practices toward organ donation among dental students J Educ Health Promot 2018 7 157 30693294 22 Naveena JH Margaret B Effectiveness of STP on knowledge and attitude of nursing students on eye donation Hindu 2019 20 33 40 23 Poreddi V Katyayani BV Gandhi S Thimmaiah R Badamath S Attitudes, knowledge, and willingness to donate organs among Indian nursing students Saudi J Kidney Dis Transpl 2016 27 1129 38 27900957

PMC010xxxxxx/PMC10353687.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-382 10.4103/ijcm.ijcm_480_22 Review Article Quest for Biomarkers of Positive Health: A Review Indrayan Abhaya Vishwakarma Gayatri 1 Verma Saumya 2 Sarmukaddam Sanjeev 3 Tyagi Asha 4 Department of Clinical Research, Max Healthcare, New Delhi, India 1 Department of Biostatistics, George Institute of Global Health India, New Delhi, India 2 Department of Statistics, University of Lucknow, Lucknow, Uttar Pradesh, India 3 Department of Community Medicine, BJ Medical College, Pune, Maharashtra, India 4 Department of Anesthesia, University College of Medical Sciences, Delhi, India Address for correspondence: Dr. Abhaya Indrayan, Department of Clinical Research, Max Healthcare, Saket, New Delhi 110 017, India. E-mail: a.indrayan@gmail.com May-Jun 2023 30 5 2023 48 3 382389 04 6 2022 19 9 2022 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. The positive health of a person can be defined as the ability to live long in good health, possibly with no activity limitation. No method is yet available for its objective assessment in individuals, and we propose a framework in this communication that can operationalize this concept. Instead of distal factors, such as diet and lifestyle because these are subjective and difficult to measure, we concentrate on the objectively measurable biomarkers such as immunity level, endorphins, and handgrip strength. The focus is on the major parameters that may protect from diseases and infirmity and can be assessed by noninvasive methods. A combination of such parameters may signify positive health. This may be a novel way to measure positive health at the individual level. In this communication, we briefly review the literature and identify a few major biomarkers that provide a protective shield and could determine the status of positive health at the individual level. This exercise demonstrates that the assessment of the positive health of a person is feasible. A scale based on these and other relevant parameters can be developed later that could quantitatively measure the exact level of positive health. As the exact combination of the parameters that protects from ailments is not fully known yet, a framework such as this may help in identifying the data gaps that require attention in this context. The proposed framework may initiate a discussion on indicators of positive health and characterize the parameters for intervention that could increase a healthy life. Biomarkers health assessment health promotion positive health protection of health review ==== Body pmcINTRODUCTION Health as a “complete physical, mental and social well-being” is an unattainable ideal and may be a mirage as the goalposts shift when an achievement is made. Public health professionals have been working on this concept at the population level,[1] but we focus in this communication on the positive health of the individuals. The concept of positive health of individuals and its domains have been described as the capability to remain healthy in the face of adversities and be able to successfully fight the ailment before its onset if it strikes.[2] In this communication, we dilate this concept with details of the domains and the specification of major biomarkers within each domain that can objectively measure the positive health of the individuals. We do not consider distal factors such as genes, stress, nutrition, exercise, sleep, and social interactions for assessing positive health because these are mostly complex, volatile, and defy direct measurement. There is enough evidence that such distal factors modulate proximal parameters that can be objectively measured. For example, genes are involved in hormonal signaling,[3] exercise reduces the level of the body’s stress hormones such as cortisol and adrenaline and stimulates the production of endorphins,[4] nutrition modulates immune functions,[5] and sleep influences cardiovascular endocrine and thermoregulatory system.[6] Thus, the positive health of a person can be directly measured by a battery of biomarkers, and such measurements will provide a much more objective assessment in a novel way. The identification and measurement of these biomarkers can help develop an effective decision-making support system for improving the health of the people and may help develop a scoring system. At the very least, this may initiate a discussion on individual-level indicators of positive health and characterize the parameters for intervention that could increase the healthy life. Positive health The concept of positive health has been around for a century although not in the form we now propose to present. The earliest reference we could locate appeared in 1924 when “positive health” was described as a double positive since health itself is a positive term.[7] The first use of the term in nearly the same sense we now discuss appeared in 2001 that described positive health as “the ability to cope with biological, psychological, and social stress” with parameters such as pain tolerance and vital capacity as the example of its indicators.[8] In 2008, it was proposed as a combination of ‘excellent’ status on biological, subjective, and functional measures.[9] In 2011, positive health was described as one’s ability to adapt and self-manage in the face of social, physical, and emotional challenges.[10] The latter two proposals include difficult-to-measure psychological and social parameters whereas our focus is on measurable medical parameters. Perhaps the most direct measure of positive health is the years lived without any physical, mental, or social limitation. At the community level, this is measured by the healthy life expectancy. Whether at the individual level or the community level, this can be measured only after death. Biomarkers could be more exact and can be measured when the person is alive. More than a thousand medical parameters can be measured. While the role of most of them is well-documented for causing disease, their physiological role in health protection is mostly obscure. Only few studied their positive features. We extensively searched nearly a thousand publications and reputed websites for the measurable parameters that have a major identifiable protective role in place of their role in causing diseases. For this, an extensive internet search was done for terms such as positive health, health protection, health maintenance, and prevention, with terms such as biomarkers, markers, features, parameters, etc., Documents for individual biomarkers were also studied to locate a relevant document for our review. These include articles, papers, websites, books, and other material. Once identified, the concerned biomarkers may look well-known but a-priori this information is rare. The biomarkers of positive health would be easy to study when divided into domains and items on the pattern of quality of life (QoL) measurement, although there may be some overlap as in the case of QoL. Each domain has many parameters, but we restrict to a selected set of few parameters that seem major to us on review of the literature so that the assessment of positive health becomes practically feasible. Optimal levels can be proposed later once a consensus is developed on the major parameters of positive health. Domains and items of positive health Our review suggests the following five major domains and items have a protective role and could be primarily considered for assessing positive health. They are all interdependent and together contribute to positive health [Figure 1]. Figure 1 Major domains of positive health and their inter-dependence Neurological parameters Neurological parameters perhaps are the most important domain of positive health. These possibly are the most intricate too as they easily transgress to abstracts. However, science is about accepting the challenges and making a beginning that could lead to a clear path. Our review of the literature suggests that the brain and nervous system parameters that bolster the mechanism to fight the ailments are yet to be investigated. In the absence of exactly measurable parameters, perhaps Montreal Cognitive Assessment Scale[11] can be used to assess brain health at a gross level. P3 amplitude can also be considered as it reflects neural activity related to the attentional and working memory processes.[12] Studies have demonstrated a negative correlation between gamma-aminobutyric acid (GABA) activity and anxiety.[13] The main physiological effects of GABA include neuronal modulation to improve nervous system disorders, regulation of anxiety, sleeplessness, mood disorders, protection against hypertension, diabetes, nephrotoxicity, and cancer.[14] Several other parameters are implicated for neural health although they may not be as major. These include growth hormone-releasing hormone with cognition-enhancing effects, nerve growth factor and brain-derived neurotrophic factor as biomarkers of neural health, and Tau protein that maintains functions of nerve cells and structure. Low levels of CRP, IL, and TNF are also implicated in determining cognitive functions and amyloidb peptides help in preventing cognitive decline in older adults. Prostaglandins regulate brain blood flow and can have a neuroprotective effect by acting on G-protein-coupled receptors. Salivary aamylase, cortisol, secretory immunoglobulin A, and chromogranin A have been investigated as biomarkers of pain and a lower level of norepinephrine indicates lesser anxiety. In view of the above, among several neurological parameters, the following can be considered major for assessing positive health: Montreal Cognitive Assessment Scale P3 amplitude GABA Others too are important but possibly not as major. Endocrinological substances The endocrine system has a profound effect on the regulation of stress response, growth and development, behavior, immunity, and energy and metabolism. Whereas the implication of hormones in diseases such as thyroidism, Cushing syndrome, and diabetes is evident, these are important for maintaining and protecting health also. Thyroid hormones help the body to use energy, stay warm, and keep the brain, heart, muscles, and other organs working as they should.[15] They are the key regulators of metabolism and development.[16] Thus, their adequate level is necessary for positive health. The critical role of insulin in maintaining a healthy level of blood sugar and preventing diabetes is well known. Estrogen and testosterone have a well-known role in sexual functions as discussed later but they have other protective roles too. Estrogen protects women against heart disease[17] and brain disorders such as Alzheimer’s.[18] Systemic estrogen helps protect against bone-thinning disease.[19] The protective role of testosterone is particularly related to pain and inflammation.[20] Melatonin controls the sleep and wake cycle, which may substantially contribute to positive health. Serotonin is believed to regulate anxiety, happiness, mood, sleep, appetite, memory, and learning ability.[21] and also regulates biological processes including cardiovascular functions, metabolic rate, and temperature control.[22] Oxytocin plays a critical role in social and emotional behavior[23] and can induce antistress-like effects and increase pain threshold.[24] Dopamine is a “feel good” hormone associated with reinforcement and helps in motivation.[25] Thus, these hormones have a vital role in maintaining good health. Endorphins control the perception of pain and fight stress.[26] They are the body’s natural pain killers and mood elevators.[4] Others that have a role, but probably not as major, are the following. Amino acids stimulate muscle synthesis, and peptides in milk modulate blood pressure, inflammation, and oxidation. Leucine is required for protein synthesis. Hydroxymethylbutyrate enhances and strengthens muscles. Besides digestion, enzymes are needed to support, maintain, and repair the body and its functions. Amylase and lipase breakdown carbohydrates and pepsin helps in the digestion of proteins – they together play a pivotal role in completing the digestive process. Amidst so many, the following endocrinological parameters can be considered major candidates for measuring positive health: Thyroid hormones Insulin Estrogen Melatonin, serotonin, oxytocin, dopamine Endorphins. Nutritional elements It is well-known that nutritional elements such as vitamins, folate, trace elements, omega-3 fatty acids, and probiotics, when present in an optimal combination, play a dominant role in maintaining health. But the specific role of various nutrients in preventing and fighting a disease seems still under investigation. Vitamins and minerals play an essential role in a variety of basic metabolic pathways that support fundamental cellular functions, particularly in energy-yielding metabolism, DNA synthesis, oxygen transport, and neuronal functions.[27] Specifically, vitamin A boosts the immune response in the elderly,[28] vitamin B-complex helps prevent infections and supports or promotes energy level, eyesight, brain functions, digestion, and muscle tone.[29] Vitamin C is an antioxidant that protects from infections and damage to body cells and produces collagen that helps bones and muscles to bond together.[30] It also helps reduce both the physiological and psychological effects of stress.[31] Vitamin D helps regulate the amount of calcium and phosphate – thus keeps bone, teeth, and muscles healthy,[32] and has immunomodulatory effect.[33] Vitamin E has a beneficial effect against oxidative stress.[34] Vitamin K is important for maintaining healthy blood vessels, blood coagulation, bone metabolism, and for regulating blood calcium levels.[35] Epidemiologic data on the relationship of trace elements with health and disease are incomplete,[36] but it is known that trace elements participate in oxygen-reduction reactions in energy metabolism.[37] Potassium, magnesium, calcium, and other nutrients help in keeping bones together and prevent osteoporosis. Selenium can increase life span and can prevent cancer proliferative. Perhaps the most important is zinc which has a crucial role in multiple biological processes, particularly in regulating immune cell signaling,[38] reducing inflammation, boosting the immune system, reducing the risk of age-related diseases, speeding wound healing, and improving acne.[39] It catalyzes enzyme activity, contributes to protein structure, and regulates gene expression.[40] Probiotics regenerate the digestive system with good microbes that neutralize the harmful ones. They prevent diarrhea and may boost the immune system. Probiotics can also improve anxiety, depression, autism, and memory[41] and can reduce the severity of certain allergies and eczema.[42] Omega-3 fatty acids build membranes around each cell in the body, including brain cells, and are associated with increased blood flow in the brain and lower the risk of premature death.[43] A low level of oxidative stress helps prevent cancer, cardiovascular diseases, neurological diseases, respiratory diseases, rheumatoid arthritis, kidney diseases, and sexual dysfunction.[34] The basic ingredient of oxidative stress is free radicals, which should be low to prevent gout, rheumatism, renal, calculi, etc.[44] but their presence is needed to synthesize some cellular structures and to be used to fight pathogens.[34] One of the most powerful antioxidants is glutathione that combats free radicals such as malondialdehyde (MDA) and has several other health benefits. Several other nutritional elements have a protective role although probably not as major. Lactic acid improves the nutritional value of food, controls intestinal infections, improves digestion of lactose, controls some types of cancer, and controls serum cholesterol levels. Glutamine can reduce many kinds of inflammation, fever, joint pain, allergies, and skin irritation. Beta-carotene delays cataract, prevents macular degeneration, and promotes good eye health. To choose a few, the major protective nutritional elements seem to be: Various vitamins (A, B, C, D, E, and K) Trace elements, particularly zinc Probiotics Fatty acids (omega-3, glutamine), lactic acid Oxidative stress, particularly glutathione. Immunity markers Good immunity is an indisputable asset to fight the microbes and eliminate antigens. Of various immunoglobulins, perhaps IgG and IgM are more important as IgG neutralizes toxins, viruses, and bacteria, and IgM is specialized to activate complement efficiently upon binding antigen.[45] Lymphocytes, which produce antibody molecules and macrophages that kill microorganisms and remove dead cells, are important biomarkers of immunity. One can go down to neutrophils, dendritic cells, natural killer cells, B and T lymphocytes but we keep them out of our preview so that a practical and easy-to-adopt proposal can be forwarded. Similarly, parameters such as tumor necrosis factor, which plays a critical role in defining intracellular organisms against invasion, are excluded despite their role in enhancing defense mechanisms in the body. The following items can be considered as the major measurable biomarkers of positive health in the immunity domain: IgG and IgM Lymphocytes Macrophages Physiological functions Among many, we restrict to those that are generally considered more important to maintain health and fight diseases. Because of the interdependence, an overlap is imminent. (i) Markers of Respiratory Functions Diseases such as lung cancer, respiratory tract infections, pulmonary embolism, asthma, and chronic obstructive pulmonary disease are directly related to the health of the lung. Lung functions such as forced vital capacity (FVC), forced expiratory volume (FEV), and peak expiratory flow rate, can be measured by spirometry. Generally, FEV1/FVC ratio is considered useful to evaluate lung functions.[46] Pulse oximetry estimates the oxygen saturation in the blood that helps replace worn-out cells, supplies energy, and supports the immune system. It also feeds our brain and helps in visual, cognitive, and electroencephalographic function.[47] The following can be regarded as the major lung function parameters for assessing positive health: FEV1/FVC ratio Oxygen saturation (ii) Markers of Reproductive Functions (applicable to reproductive age) It was observed that the mortality decreased in a dose-response manner as the percentage of motile and morphologically normal spermatozoa and semen volume increased.[48] Estrogen regulates the menstrual cycle, maintains pregnancy, and aids in sperm production.[49] It protects against cell death and stimulates the birth of new neurons.[50] In men, estradiol is essential for modulating libido, erectile functions, and spermatogenesis.[51] and testosterone triggers puberty and contributes to sex drive, bone mass, fat distribution, muscle mass and strength, and production of red blood cells and sperm.[52] Testosterone is also responsible for regulating sex differentiation, producing male sex characteristics, spermatogenesis, and fertility.[53] High testosterone and low estrogen help in erection in men.[51] Female reproductive health can be assessed by the levels of follicle-stimulating hormone (FSH), luteinizing hormones (LH), and 17-β estradiol.[54] In women, FSH is responsible for the growth of ovarian follicles that help maintain the menstrual cycle. It may have a direct relation with bone health in women by enhancing bone resorption. LH, in synergy with FSH, stimulates normal follicular growth and ovulation. In men, FSH helps in sperm reproduction. Thus, the major reproductive indicators for positive health could be: Males: Semen quality and volume, and testosterone Females: FSH, LH, and estradiol. (iii) Markers of Gastrointestinal Functions (including Liver) We have not been able to find a measurable indicator that can holistically assess the health of the digestive system. Besides protecting from liver-specific diseases such as cirrhosis, hepatitis, and liver cancer, adequate liver functions help fight infection and remove toxins from the body’s blood. The liver also keeps fluids in the bloodstream from leaking into surrounding tissue and carries hormones, vitamins, and enzymes through the body. Healthy liver slows down aging, increases energy, keeps skin healthy, and helps in hormone balancing.[55] Although all liver function parameters have importance in maintaining good health, albumin has multiple physiological functions such as maintaining colloidal osmotic pressure, binding of a wide variety of compounds, and provision of the bulk of plasma antioxidant activity,[56] synthesis of purine/pyrimidine bases, urea and protein synthesis, and gluconcogenesis.[57] Alanine transaminase (ALT) may be a good indicator of overall health, particularly in the context of obesity, metabolic syndrome, and cardiovascular disease.[58] Among lipids, cholesterol is important for the synthesis of hormones and vitamin D. Specific levels of LDL and HDL cholesterol provide some protection from heart disease. Lipoprotein (a) promotes vascular repair, and the lowest CVD mortality has been observed with a low level of triglycerides. The major liver functions parameters that can be used as biomarkers of positive health are: Albumin, ALT Lipid profile (iv) Markers of Circulatory Functions Cardiac output determines oxygen delivered to the cells and is an important component of how effectively the heart can meet the body’s demands for the maintenance of adequate tissue perfusion. It is related to the left ventricular ejection fraction that assesses global and segmental left ventricular function. We excluded both as these are difficult to measure objectively. However, parameters such as heart rate and blood pressure are completely non-invasive, easy, and amenable to the exact measurement. It is well known that a relatively lower resting heart rate has a protective role against cardiovascular diseases and all-cause mortality, and relatively low blood pressure reduces the risk of heart and kidney diseases. Among many hematological components, hemoglobin is essential for transferring oxygen in the blood from the lungs to the tissues. Glycosylated Hb (HbA1c) indicates long-term glycemic control. Whereas the adverse effects of high HbA1c are well known, low levels are also suggested to be associated with an increased risk of all-cause mortality.[59] Platelets help the body to control bleeding and they must be present in healthy numbers. A low hs-CRP level can protect from cardiovascular events. Creatine phosphokinase is important for muscle function, although a high level indicates muscle injury, including possibly the heart. Cardiac troponin regulates contractile function in skeletal and cardiac muscles. There may be several others with protective functions, but the following can be considered as the major circulatory function markers of positive health: BP, Heart rate Hb and HbA1c level Platelet count (v) Markers of Musculoskeletal Functions Bone mineral density (BMD) is a common measure of bone strength. A good BMD can save from fractures in the cases of falls or other accidents. The primary task of skeletal muscle is maintaining posture, breathing, and locomotion but it also represents important nutrient storage and metabolic regulator.[60] A low functional aging index based on the sensory and pulmonary parameters, grip strength, and gait speed can indicate positive health.[61] A healthy body mass index (BMI) indicates good health and long life, is a cornerstone in the prevention of chronic diseases, and is a promotional parameter of healthy aging.[62] It also means more energy, better regulation of body fluids and blood pressure, better sleep, and a reduced risk of several diseases. It naturally implies a reduced burden on the heart and circulation system and a reduced risk of heart disease. Low BMI has been found an independent predictor of pregnancy in women.[63] Healthy BMI also raises life satisfaction.[64] Thus, the major musculoskeletal parameters that can be considered as candidates for assessing positive health are: BMD Functional aging index, including handgrip strength BMI (vi) Markers of Urinary Functions Good kidney functions help in regulating blood pressure, the make-up of the blood, maintain hormones, and producing hormones for bones to make more blood cells. They metabolize vitamin D and maintain a proper balance between water, electrolytes, acids, and bases. They release the hormone that directs production of RBC and keeps blood minerals in balance. Glomerular filtration rate (GFR) is considered the optimal way to measure kidney function.[65] There is evidence that creatine might prevent skin aging, helps muscles gain strength, and may help minimize neurogenerative disorders[66] but important for the health of the kidney is a relatively low creatinine level in serum. A good indicator of health is a low level of protein-creatinine ratio because its level is associated with all-cause mortality.[67] A balanced pH keeps our systems operating the way they should.[68] Human life requires a tightly controlled pH level to survive.[69] Thus, the major kidney parameters that can indicate positive health are: GFR Protein/creatinine ratio pH value DISCUSSION The notion of positive health provides a context for methodological and theoretical debate that can enrich the theory and practice of health assessment.[70] For Harrison et al.,[71] the concept of positive health included harmonious family balance and mental and emotional stability. An editorial in the Lancet explained health as the ability to adapt.[72] Previous endeavors[9,10] are predominantly based on psychological and social factors such as positive emotions, life satisfaction, and happiness. These indeed could be important determinants, but they are subjective and difficult to measure. Evidence suggests that such factors mostly culminate in physiological changes such as better immunity levels and more balanced hormonal levels that can be exactly measured. This communication targets such biomarkers. The information regarding the protective role of these factors is mostly lacking and we dug the literature to identify the major protective parameters. Their link to positive health outcomes can help design interventions to build and sustain these assets and help people to increase their chance of living healthier and longer life.[73] To achieve parsimony and feasibility, we have restricted to a few biomarkers that seemed major to us. Although no medical parameter is absolute, our review suggests that endorphins, semen quality in men, FEV1/FVC ratio, oxygen saturation, BMD, P3 amplitude, and handgrip strength are examples of parameters whose higher levels generally indicate increasingly positive health. Measurement of most of these are cost-effective and the remaining will become cheaper when the idea of positive health catches up and used on a large scale. Among thousands of medical parameters, the initial identification of nearly 50 major biomarkers of positive health in this communication shows that it is feasible to objectively assess positive health. Once a consensus is developed on the relevant parameters, an exercise can be undertaken to identify the optimal levels of the parameters that signify positive health. Our effort may be ahead of time because many of the body’s defense mechanisms are not properly understood yet. The link between physiological processes and positive outcomes seems to have not been fully investigated. We believe it is time to think of approaches that can contribute to our understanding of how body functions work to defend from ailments and fill up this epistemic uncertainty. Our proposal may initiate a discussion on biomarkers of positive health and characterize the parameters for intervention that could increase the healthy life. This may also help in identifying the data gaps that require attention in this context. Limitations We have been quite brief since the endeavor here is just to propose an idea and show that it is useful and feasible although we agree that much work is required to make it operational in the actual setup. The references are quoted only for the major biomarkers but the source of all the statements is available from the corresponding author. The domains and items included in this article may give the impression that psychological and spiritual aspects have been ignored but we believe that most of these also ultimately affect biomarkers such as hormones. Nonetheless, those psychological and spiritual aspects have been excluded from our framework that do not affect biomarkers. CONCLUSION A novel proposal to assess the positive health of individuals has been developed with five domains and nearly 50 items. Despite being preliminary, this communication possibly covers most major measurable biomarkers and could be adequate to proceed ahead with operationalizing the concept of positive health and its assessment. The framework may also help in identifying the data gaps that can be used to develop policies for improving the health of the people. Contribution of the authors AI proposed the idea, developed it, and prepared the draft. GC helped in expanding the idea, searching the literature, and contributing to the manuscript. SV and SS searched the literature and proposed changes. AT provided inputs for improvement in the medical content of the manuscript. All the authors reviewed the draft and approved it. Highlights The concept of positive health as the capability to counter ailments before their onset and to live a long healthy life has so far mainly concentrated on lifestyle factors. This article identifies objectively measurable major biomarkers of positive health that can help operationalize this concept at the individual level. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. ==== Refs REFERENCES 1 Wojcik O Miller Mshp CE Plough AL Aligning health and social systems to promote population health, well-being, and equity Am J Public Health 2020 110 (S2) S176 7 32663089 2 Indrayan A Vishwakarma G Sarmukaddam S Verma S Beyond “Normal”: Optimal levels of medical parameters for assessing positive health Am J Int Med 2021 9 166 70 3 Harden KP Klump KL Introduction to the special issue on gene-hormone interplay Behav Genet 2015 45 263 7 25903987 4 Harvard Health Endorphins: The Brain's Natural Pain Reliever July 20 2021 Available from: https://www.health.harvard.edu/mind-and-mood/endorphins-the-brains-natural-pain-reliever 5 Childs CE Calder PC Miles EA Diet and immune function Nutrients 2019 11 1933 31426423 6 Schwimmer H Stauss HM Abboud F Nishino S Mignot E Zeitzer JM Effects of sleep on the cardiovascular and thermoregulatory systems: A possible role for hypocretins J Appl Physiol (1985) 2010 109 1053 63 20705949 7 Anonymous Why positive health? Cal State J Med 1924 22 28 8 Indrayan A Sarmukaddam S Medical Biostatistics 1st ed New York Marcel Dekker 2001 164 9 Seligman MEP Positive health Applied Psychol Int Rev 2008 57 3 18 10 Huber M Knottnerus JA Green L van der Horst H Jadad AR Kromhout D How should we define health? BMJ 2011 343 d4163 21791490 11 MoCA. Cognitive Assessment https://www.mocatest.org/- [Last accessed on 2021 Dec 18] 12 Polich J Kok A Cognitive and biological determinants of P300: An integrative review Biol Psychol 1995 41 103 46 8534788 13 Krishnakumar D Hamblin MR Lakshmanan S Meditation and yoga can modulate brain mechanisms that affect behavior and anxiety: A modern scientific perspective Anc Sci 2015 2 13 9 26929928 14 Grewal J Gamma-Aminobutyric Acid (GABA): A versatile bioactive compound Eur J Molecular Clin Med 2020 7 3068 75 15 American Thyroid Association Thyroid Function Tests Available from: https://www.thyroid.org/thyroid-function-tests/- [Last accessed on 2021 Dec 18] 16 Boelaert K Franklyn JA Thyroid hormone in health and disease J Endocrinol 2005 187 1 15 16214936 17 Paranjape SG Turankar AV Wakode SL Dakhale GN Estrogen protection against coronary heart disease: Are the relevant effects of estrogen mediated through its effects on uterus--such as the induction of menstruation, increased bleeding, and the facilitation of pregnancy? Med Hypotheses 2005 65 725 7 15950396 18 Simpkins JW Perez E Wang X Yang S Wen Y Singh M The potential for estrogens in preventing Alzheimer's disease and vascular dementia Ther Adv Neurol Disord 2009 2 31 49 19890493 19 Gambacciani M Levancini M Hormone replacement therapy and the prevention of postmenopausal osteoporosis Prz Menopauzalny 2014 13 213 20 26327857 20 Vodo S Bechi N Petroni A Muscoli C Aloisi AM Testosterone-induced effects on lipids and inflammation Mediators Inflamm 2013 2013 183041 23606790 21 Dfarhud D Malmir M Khanahmadi M Happiness and health: The biological factors –systematic review article Iran J Public Health 2014 43 1468 77 26060713 22 Berger M Gray JA Roth BL The expanded biology of serotonin Annu Rev Med 2009 60 355 66 19630576 23 Olff M Frijling JL Kubzansky LD Bradley B Ellenbogen MA Cardoso C The role of oxytocin in social bonding, stress regulation and mental health: An update on the moderating effects of context and interindividual differences Psychoneuroendocrinology 2013 38 1883 94 23856187 24 Uvnas-Moberg K Petersson M Oxytocin, einVermittler von Antistress, Wohlbefinden, sozialer Interaktion, Wachstum und Heilung [Oxytocin, a mediator of anti-stress, well-being, social interaction, growth and healing Z Psychosom Med Psychother 2005 51 57 80 German 15834840 25 Bromberg-Martin ES Matsumoto M Hikosaka O Dopamine in motivational control: Rewarding, aversive, and alerting Neuron 2010 68 815 34 21144997 26 Sprouse-Blum AS Smith G Sugai D Parsa FD Understanding endorphins and their importance in pain management Hawaii Med J 2010 69 70 1 20397507 27 Tardy AL Pouteau E Marquez D Yilmaz C Scholey A Vitamins and minerals for energy, fatigue and cognition: A narrative review of the biochemical and clinical evidence Nutrients 2020 12 228 31963141 28 Rumore MM Vitamin A as an immunomodulating agent Clin Pharm 1993 12 506 14 8354037 29 Healthline. Why Is Vitamin B Complex Important, and Where Do I Get It? Available from: https://www.healthline.com/health/food-nutrition/vitamin-b-complex- [Last accessed on 2021 Nov 15] 30 DePhillipo NN Aman ZS Kennedy MI Begley JP Moatshe G LaPrade RF Efficacy of vitamin C supplementation on collagen synthesis and oxidative stress after musculoskeletal injuries: A systematic review Orthop J Sports Med 2018 6 2325967118804544 30386805 31 Staff P “Vitamin C: Stress buster-Psychology Today” 2003 Available from: https://www.psychologytoday.com/intl/articles/200304/vitamin-c-stress-buster- [Last accessed on 2021 Oct 11] 32 NHS 2019 “Vitamin D.” Available from: https://www.nhs.uk/conditions/vitamins-and-minerals/vitamin-d/- [Last accessed on 2021 Oct 11] 33 Sassi F Tamone C D'Amelio P Vitamin D: Nutrient, hormone, and immunomodulator Nutrients 2018 10 1656 30400332 34 Pizzino G Irrera N Cucinotta M Pallio G Mannino F Arcoraci V Oxidative stress: Harms and benefits for human health Oxid Med Cell Longev 2017 2017 8416763 28819546 35 Gröber U Reichrath J Holick MF Kisters K Vitamin K: An old vitamin in a new perspective Dermatoendocrinol 2015 6 e968490 26413183 36 National Research Council (US) Committee on Diet and Health Diet and Health: Implications for Reducing Chronic Disease Risk Washington (DC) National Academies Press (US) 1989 Available from: https://pubmed.ncbi.nlm.nih.gov/25032333/ 37 Chojnacka K Saeid A Recent Advances in Trace Elements Wiley-Blackwell 2018 Available from: https://www.wiley.com/en-au/Recent+Advances+in+Trace+Elements-p-9781119133803 38 Kim B Lee WW Regulatory role of zinc in immune cell signaling Mol Cells 2021 44 335 41 33986184 39 Kubala J Zinc: Everything You Need to Know: Healthline November 14, 2018 Available from: https://www.healthline.com/nutrition/zinc- [Last accessed on 2021 Nov 18] 40 Trumbo P Yates Aa Schlicker S Poos M Dietary references intakes J Acad Nutri Dietetics 2001 101 294 301 41 Wang H Lee IS Braun C Enck P Effect of probiotics on central nervous system functions in animals and humans: A systematic review J Neurogastroenterol Motil 2016 22 589 605 27413138 42 Makrgeorgou A Leonardi-Bee J Bath-Hextall FJ Murrell DF Tang ML Roberts A Probiotics for treating eczema Cochrane Database Syst Rev 2018 11 CD006135 30480774 43 Harris WS Tintle NL Imamura F Qian F Korat AVA Marklund M Blood n-3 fatty acid levels and total and cause-specific mortality from 17 prospective studies Nat Commun 2021 12 2329 33888689 44 Pham-Huy LA He H Pham-Huy C Free radicals, antioxidants in disease and health Int J Biomed Sci 2008 4 89 96 23675073 45 Janeway CA Jr Travers P Walport M Shlomchik MJ Immunobiology: The Immune System in Health and Disease 5th ed New York Garland Science 2001 Available from: https://www.ncbi.nlm.nih.go v/books/NBK10757/ 46 Torén K Schiöler L Lindberg A Andersson A Behndig AF Bergström G The ratio FEV1/FVC and its association to respiratory symptoms-A Swedish general population study Clin Physiol Funct Imaging 2021 41 181 91 33284499 47 Hafen BB Sharma S Oxygen Saturation 2021 Aug 12 StatPearls Treasure Island (FL) StatPearls Publishing 2021 Jan Available from: https://pubmed.ncbi.nlm.nih.gov/30247849/ 48 Jensen TK Jacobsen R Christensen K Nielsen NC Bostofte E Good semen quality and life expectancy: A cohort study of 43,277 men Am J Epidemiol 2009 170 559 65 19635736 49 Dostalova P Zatecka E Dvorakova-Hortova K Of oestrogens and sperm: A review of the roles of oestrogens and oestrogen receptors in male reproduction Int J Mol Sci 2017 18 904 28441342 50 Wise PM Suzuki S Brown CM Estradiol: A hormone with diverse and contradictory neuroprotective actions Dialogues Clin Neurosci 2009 11 297 303 19877497 51 Schulster M Bernie AM Ramasamy R The role of estradiol in male reproductive function Asian J Androl 2016 18 435 40 26908066 52 Wein. Understanding How Testosterone Affects Men. NIH. 23 September, 2013 Available from: https://www.nih.gov/newsevents/nih-research-matters/understanding-how-testosterone-affectsmen- [Last accessed on 2021 Nov 08] 53 Nassar G Leslie S Physiology: Testosterone Treasure Island (FL) Stat Pearls Publishing Sept 26, 2018 Available from: https://europepmc.org/article/NBK/nbk526128 54 Navot D Rosenwaks Z Margalioth EJ Prognostic assessment of female fecundity Lancet 1987 2 645 7 2887939 55 Foreman D Why Liver Health is so Important?Algatech July 7, 2019 Available from: https://www.algatech.com/why-liver-health-is-so-important/- [Last accessed on 2021 Dec 03] 56 Levitt DG Levitt MD Human serum albumin homeostasis: A new look at the roles of synthesis, catabolism, renal and gastrointestinal excretion, and the clinical value of serum albumin measurements Int J Gen Med 2016 9 229 55 27486341 57 Ndrepepa G Aspartate aminotransferase and cardiovascular disease—A narrative review J Lab Prec Med 2020 6 doi:10.21037/jlpm-20-93 58 Kim W Serum activity of alanine aminotransferase (ALT) as an indicator of health and disease Hepatology 2008 47 1363 70 18366115 59 Carson AP Fox CS McGuire DK Levitan EB Laclaustra M Mann DM Low hemoglobin A1c and risk of all-cause mortality among US adults without diabetes Circ Cardiovasc Qual Outcomes 2010 3 661 7 20923991 60 Wolfe RR The underappreciated role of muscle: Am J ClinNutr 2006 84 475 82 61 Finkel D Sternäng O Jylhävä J Bai G Pedersen NL Functional aging index complements frailty in prediction of entry into care and mortality J Gerontol A Biol Sci Med Sci 2019 74 1980 6 31222213 62 Fontana L Hu FB Optimal body weight for health and longevity: Bridging basic, clinical, and population research Aging Cell 2014 13 391 400 24628815 63 Bond RT Nachef A Adam C Couturier M Kadoch IJ Lapensée L Obesity and infertility: A metabolic assessment strategy to improve pregnancy rate J Reprod Infertil 2020 21 34 41 32175263 64 Habibov N Auchynnikava A Luo R Fan L A healthy weight improves life satisfaction Int J Health Plann Manage 2019 34 396 413 30272382 65 Cimmino CJ Benefits of direct glomerular filtration rate (GFR) determination versus creatinine-based tests for evaluating renal function J Am Osteopath Assoc 1998 98 437 44 9747056 66 Riesberg LA Weed SA McDonald TL Eckerson JM Drescher KM Beyond muscles: The untapped potential of creatine Int Immunopharmacol 2016 37 31 42 26778152 67 Fulks M Stout RL Dolan VF Urine protein/creatinine ratio as a mortality risk predictor in non-diabetics with normal renal function J Insur Med 2012 43 76 83 22876411 68 Waugh A Grant A Anatomy and Physiology in Health and Illness 10th ed London Churchill Livingstone Elsevier 2007 69 Schwalfenberg GK The alkaline diet: Is there evidence that an alkaline pH diet benefits health? J Environ Public Health 2012 2012 727630 22013455 70 Locker D Gibson B The concept of positive health: A review and commentary on its application in oral health research Community Dent Oral Epidemiol 2006 34 161 73 16674748 71 Harrison GG Kagawa-Singer M Foerster SB Lee H Pham Kim L Nguyen TU Seizing the moment: California's opportunity to prevent nutrition-related health disparities in low-income Asian American population Cancer 2005 104 (12 Suppl) 2962 8 16276535 72 Editorial What is health?The ability to adapt Lancet 2009 373 781 19269498 73 Positive Health “What is positive health?” Available from: https://positivehealthresearch.org/ [Last accessed on 2021 Dec 11]

PMC010xxxxxx/PMC10353688.txt

==== Front Indian J Community Med Indian J Community Med IJCM Indian J Community Med Indian Journal of Community Medicine: Official Publication of Indian Association of Preventive & Social Medicine 0970-0218 1998-3581 Wolters Kluwer - Medknow India IJCM-48-379 10.4103/ijcm.ijcm_902_22 View Point Contributory Parenting: A ”Priceless Shift” from Indirect to Direct Parenting Taranikanti Madhuri Gaur Archana Ganji Vidya Taranikanti Sai Shriya 1 Department of Physiology, All India Institute of Medical Sciences, Bibinagar, Hyderabad, Telangana, India 1 Department of Obstetrics and Gynaecology, Agartala Government Medical College, Agartala, Tripura, India Address for correspondence: Dr. Madhuri Taranikanti, Department of Physiology, All India Institute of Medical Sciences, Bibinagar, Hyderabad, Telangana – 508126, India. E-mail: madhuri.tarani@gmail.com May-Jun 2023 30 5 2023 48 3 379381 05 11 2022 26 4 2023 Copyright: © 2023 Indian Journal of Community Medicine 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms. Parenting is a valuable investment that determines the quality of future independent life. From an evolutionary aspect, it has been well ingrained in the minds of humans as to how much resource each parent should contribute to this energy and time-consuming task. To encourage father’s contribution towards parenting and reduce the stress on mother, the concept of paid paternal leave has been implemented. Mere presence of the father in terms of the quantity of time spent without much qualitative value has no benefit, but the assumption that fathers are less competent based on their lower performance might also not be acceptable. An intriguing finding has demonstrated that prolonged contact with the infant triggers a change in previously absent male parenting behavior. With incentives on one hand and associated societal stigma on the other hand, it is to be analyzed whether the purpose of true parenting by fathers is being achieved. Hence, in the concept of contributory parenting it is necessary to recognize and respect each parenting style with the ultimate benefit being passed on to the child. Contributory parenting incentives paternal social stigma ==== Body pmcParental care is a collection of actions that adults take to care for their children. These actions may directly affect the child’s survival, growth, and psychological development. It is a valuable investment that determines the quality of future independent life in all living beings. From an evolutionary aspect, each species has strategically divided parenting tasks into maternal and paternal. Since long, it has been well ingrained in the minds of humans as to how much resource each parent should contribute to this energy and time-consuming task. In the past, several hypotheses have been proposed to describe this pattern of task division between the two sexes.[1,2] Classically, fathers have been the breadwinners and mothers have been the caretakers. However, in present circumstances, where both parents are earnestly engaged in their respective professions and equally contributing to the financial well-being of the family, parenting conflicts are on the rise. The government has recognized the importance of the father’s contribution toward parenting and made provisions to make him more sensitive toward the needs of both the newborn and the mother. While not an obligation, a few major private companies have chosen to follow the government policy of paternity leave. An additional hand in taking care of not only the newborn but also the household chores relieves the mother of much stress, which may already be present due to the postpartum effects. The realization of a combined and contributory effort will not only produce more bonding with the child and the rest of the family but also has given rise to the concept of paternity leave. This provision, if followed in both letter and spirit, can prove to be of great significance as the child gets the opportunity to approach either parent equally in an unbiased manner for all the needs. Employers have created the facility of providing paid paternal leave in addition to the already existing maternal leave facility to care for their newborns. Such an incentive is expected to retain quality staff and improve the productivity and morale of the employees.[3] While the incentives are beneficial, it is to be analyzed whether the purpose of true parenting by fathers is being achieved using this facility. In some countries, the stigma associated with “paternity leaves” exists with men fearing the criticism of taking leave for parenting.[4-6] The “hunter–gatherer” mentality still dominates, making it difficult for fathers to take the necessary steps to provide their children with proper parenting.[7] Along with the criticism from some societies of involvement in parental duties, men are also expected to be the main or sole breadwinner for the family, which is aptly explained by the hunter–gatherer mentality. As a result, mothers are being overburdened both physically and mentally with overconsumption of thought. In this entire scenario of parenting, the mere presence of the father in terms of the quantity of time spent without much qualitative value described as the “father effect” would not provide sufficient benefit. It is generally assumed that they have lower performance standards in this task, but it may not necessarily translate as less competent than mothers. While each of the parenting styles needs to be recognized and respected, it is yet to be deciphered whether it is due to lack of opportunity, circumstances, or an innate style of parenting that sometimes apparently projects fathers as less competent. In the present time, two versions of modern males have evolved, one that blindly follows the so-called tradition of fathering feels an internal barrier in assuming the new role and the other that makes every effort toward contributory parenting. Those involved in parenting tend to pave the way for similar actions in future generations. The environment so created encourages high career aspirations in their female offspring along with more likely engagement in gender-equal behaviors in their male offspring.[8] Animal models to study parenting under controlled conditions have been performed to answer some critical questions including what makes a good parent. The element of lactation has made uniparental female care the norm in humans, although many species including humans display alloparenting where parental care is provided by related family members and unrelated individuals.[9] The neurophysiology of mothering clearly demonstrates the hormonal changes in late pregnancy and the involvement of the mesocorticolimbic motivation system.[10] The release of oxytocin and dopamine leads to the long-term persistence of maternal behavior also referred to as “maternal memory.”[11] Despite new research demonstrating the role of epigenetic pathways in maternal behavior, mothering itself alters the behavior of women through altering cognition, spatial memory, and hippocampus plasticity.[12] Hormonal studies in relation to the neurobiology of fathering have been inconsistent and inconclusive.[13] An intriguing finding has demonstrated that prolonged contact with the infant triggers a change in previously absent male parenting behavior. Stimulation of the central medial preoptic area neurons and activation of galanin in the medial preoptic area induce involvement in fathering.[14] Studies of exclusion by removing the father from the biparental species have shown to have numerous and lifelong effects on offspring.[15] Behavioral studies in animals through collaboration between researchers involved in animal and human studies are needed to explore the deeper concepts in this field. A more comprehensive understanding of the evolution of parental care necessitates a closer look at basic life history on how males and females influence the evolution of maternal, paternal, and biparental care from an ancestral state of no care.[16] Situations like the coronavirus disease 2019 (COVID-19) in the last two years have provided opportunities for children and parents, particularly fathers to establish connections and build a strong bond. It is yet to be assessed whether COVID-19 has changed the perspectives and perceptions of parents toward parenting. Conclusion: Contributory parenting is a valuable investment that is beneficial to the offspring to lead a quality future independent life that would translate into a more responsible individual of the society. Irrespective of whether parenting is direct or indirect, when done with equal involvement and responsibility by both parents, it leads to the molding of a child into a holistic human being who is emotionally, physically, and mentally strong. Studying the interactions between environment, genetics, and epigenetic factors associated with parenting will help us understand the significance of each parent toward contributory parenting. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. ==== Refs REFERENCES 1 Gonzalez-Voyer A Kolm N Parental care and investment Encyclopedia of Life Sciences (ELS) Chicheste John Wiley &Sons, Ltd 2010 doi:10.1002/9780470015902.a0021907 2 Kokko H Jennions MD Parental investment, sexual selection and sex ratios J Evol Biol 2008 21 919 48 18462318 3 Suskind D Parent Nation: Unlocking Every Child's Potential, Fulfilling Society's Promise Penguin 2022 4 Kaufman G Barriers to equality: Why British fathers do not use parental leave Community Work Family 2017 21 1 16 5 Miyajima T Yamaguchi H I want to but I won't: Pluralistic ignorance inhibits intentions to take paternity leave in Japan Front Psychol 2017 8 1508 28979216 6 Petts RJ Knoester C Li Q Paid paternity leave-taking in the United States Community Work Fam 2020 23 162 83 32076386 7 Apicella CL Crittenden AN Hunter-gatherer families and parenting The handbook of evolutionary psychology;In D. Buss (Ed.) New Jersey John Wiley &Sons 2015 578 97 8 Fleming PJ McCleary-Sills J Morton M Levtov R Heilman B Barker G Risk factors for men's lifetime perpetration of physical violence against intimate partners: Results from the international men and gender equality survey (images) in eight countries PLoS One 2015 10 e0118639 25734544 9 Hrdy S Variable postpartum responsiveness among humans and other primates with “cooperative breeding”: A comparative and evolutionary perspective Horm Behav 2016 77 272 83 26518662 10 Numan M Young LJ Neural mechanisms of mother-infant bonding and pair bonding: Similarities, differences, and broader implications Horm Behav 2016 77 98 112 26062432 11 D’Cunha TM King SJ Fleming AS Lévy F Oxytocin receptors in the nucleus accumbens shell are involved in the consolidation of maternal memory in postpartum rats Horm Behav 2011 59 14 21 20932839 12 Stolzenberg DS Champagne FA Hormonal and non-hormonal bases of maternal behavior: The role of experience and epigenetic mechanisms Horm Behav 2016 77 204 10 26172856 13 Saltzman W Ziegler TE Functional significance of hormonal changes in mammalian fathers J Neuroendocrinol 2014 26 685 96 25039657 14 Tsuneoka Y Tokita K Yoshihara C Amano T Esposito G Huang AJ Distinct preoptic-BST nuclei dissociate paternal and infanticidal behavior in mice EMBO J 2015 34 2652 70 26423604 15 Bales KL Saltzman W Fathering in rodents: Neurobiological substrates and consequences for offspring Horm Behav 2016 77 249 59 26122293 16 Kenkel WM Perkeybile AM Carter CS The neurobiological causes and effects of alloparenting Dev Neurobiol 2017 77 214 32 27804277

PMC010xxxxxx/PMC10353710.txt

==== Front Plast Reconstr Surg Glob Open Plast Reconstr Surg Glob Open GOX Plastic and Reconstructive Surgery Global Open 2169-7574 Lippincott Williams & Wilkins Hagerstown, MD 00029 10.1097/GOX.0000000000005125 3 Breast Original Article Sarcopenia Best Predicts Complications in Free Flap Breast Reconstruction Jain Nirbhay S. MD Bingham Elijah BS Luvisa B. Kyle MD Frydrych Lynn M. MD Chin Madeline G. BA Bedar Meiwand MD, MSc Da Lio Andrew MD Roostaeian Jason MD Crisera Christopher MD Slack Ginger MD Tseng Charles MD Festekjian Jaco H. MD Delong Michael R. MD From the Division of Plastic Surgery, University of California Los Angeles, Los Angeles, Calif.; and David Geffen School of Medicine, University of California Los Angeles, Los Angeles, Calif. Michael R. Delong, MD, Division of Plastic Surgery, Department of Surgery, University of California at Los Angeles Health System, 200 Medical Plaza Drive, Suite 460, Los Angeles, CA 90095, E-mail: mdelong@mednet.ucla.edu 7 2023 18 7 2023 11 7 e51254 4 2023 1 6 2023 Copyright © 2023 The Authors. Published by Wolters Kluwer Health, Inc. on behalf of The American Society of Plastic Surgeons. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open-access article distributed under the terms of the Creative Commons Attribution-Non Commercial-No Derivatives License 4.0 (CCBY-NC-ND), where it is permissible to download and share the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal. Background: Breast reconstruction remains a major component of the plastic surgeon’s repertoire, especially free-flap breast reconstruction (FFBR), though this is a high-risk surgery in which patient selection is paramount. Preoperative predictors of complication remain mixed in their utility. We sought to determine whether the sarcopenia score, a validated measure of physiologic health, outperforms the body mass index (BMI) and modified frailty index (mFI) in terms of predicting outcomes. Methods: All patients with at least 6-months follow-up and imaging of the abdomen who underwent FFBR from 2013 to 2022 were included in this study. Appropriate preoperative and postoperative data were included, and sarcopenia scores were extracted from imaging. Complications were defined as any unexpected outcome that required a return to the operating room or readmission. Statistical analysis and regression were performed. Results: In total, 299 patients were included. Patients were split into groups, based on sarcopenia scores. Patients with lower sarcopenia had significantly more complications than those with higher scores. BMI and mFI both did not correlate with complication rates. Sarcopenia was the only independent predictor of complication severity when other factors were controlled for in a multivariate regression model. Conclusions: Sarcopenia correlates with the presence of severe complications in patients who undergo FFBR in a stronger fashion to BMI and the mFI. Thus, sarcopenia should be considered in the preoperative evaluation in patients undergoing FFBR. OPEN-ACCESSTRUE COUNTRYUNITED STATES ==== Body pmcTakeaways Question: How do various measurements of overall health, such as the sarcopenia score, the body mass index, and the modified frailty index, compare when looking at outcomes after free flap breast reconstruction? Findings: Sarcopenia outperforms the body mass index and the modified frailty index in predicting complications. Meaning: Sarcopenia may be superior for preoperative risk stratification and, if available, should be discussed with patients. INTRODUCTION Breast cancer remains among the most common cancers in women in the United States, with over 300,000 cases diagnosed each year.1,2 Given the immense physical, social, and emotional burden of breast cancer treatment, breast reconstruction is a critical phase of overall patient care.3 Although reconstruction using a tissue expander and prosthetic implant remains the most common approach, free-flap breast reconstruction (FFBR) can provide ample tissue and natural shape without requiring the implantation of a foreign body in appropriate patients. However, given the length of surgery and donor site morbidity, FFBR is considered a higher risk option, and strategic patient selection is important.4–6 The categorization of patient risk factors to predict postoperative outcomes is an important counseling tool in advising patients on the potential risks and benefits of FFBR. Consistent with well-established surgical risk factors, patient comorbidities such as obesity, diabetes, vascular disease, and tobacco use are associated with increased complications after autologous breast reconstruction.7–10 However, there is increasing interest in assessing risk for postoperative complications beyond these comorbidities. Particularly, some patients without the aforementioned risk factors may still have an increased degree of physiologic frailty that may be difficult to quantify with traditional metrics. Although our group and others have established the modified frailty index (mFI), which is a global picture of health from preoperative comorbidities as a predictor of short-term complications in FFBR,11,12 it has been demonstrated to be inferior to sarcopenia in other fields in the long term.13–15 Sarcopenia is the progressive and generalized depletion of skeletal muscle mass and strength associated with aging of physiologic decline.16 Fortunately, sarcopenia can be assessed with routine preoperative abdominal computed tomography (CT), which is often already obtained for preoperative cancer staging or surgical planning in FFBR. The sarcopenia score has already gained significant recognition as a prognostic factor for complications in surgical oncology procedures, and in patients undergoing transplant, abdominal, and vascular surgery.17–20 More recently, an emerging body of literature has suggested a relationship between preoperative sarcopenia and complication rates among reconstructive surgery patients, though the information is conflicted in FFBR patients.21 Further, sarcopenia has not been compared directly with mFI to assess if one measurement is superior. Therefore, we performed a retrospective review of our patients to determine if (1) sarcopenia was a predictor of postoperative complications and (2) if sarcopenia was a superior predictor of complications when compared with traditional risk factors, such as comorbid conditions like diabetes and obesity, as well as the mFI. METHODS Patient Selection and Data Extraction All patients who underwent an FFBR at our institution from 2013 to 2022 were eligible for screening for inclusion in the study. Patients were found by querying the operating room records for any patient with CPT code 19364. Over 1000 patients were found. Once patients were identified, a corresponding CT scan of the abdomen was then identified. If the CT scan was performed at most 6 months before the surgery or 1 month after, the patient was selected for inclusion (n = 299). All forms of CT imagery (including CT angiography, CT with or without contrast, and positron emission tomography with associated CT scan) were included. The sarcopenia measurement requires only muscular outline and can be equally determined in each of these types of CT scans. Once the patients were identified, they were screened to ensure at least 6-months follow-up was done before inclusion in the study (n = 299). Selected patients were then reviewed, and preoperative characteristics, including comorbidities, cancer stage, cancer care, and surgical history and body mass index (BMI), were extracted. The mFI was calculated based on history of hypertension, heart failure, diabetes, functional status, and chronic obstructive pulmonary disease. Intraoperative data (including operative time, vein size, artery size, and ischemia time) were collected. Postoperative complications were defined by any complication that required operative intervention or readmission. These complications included infection, hematoma, fat necrosis requiring debridement, flap failure, wound dehiscence, bulge deformity, seroma requiring evacuation, and delayed wound healing. Additionally, preoperative and postoperative characteristics measured as a continuous variable were normalized for statistical analysis, based on the average. This included a BMI split at 30 kg per m2, age at 50 years, and operative time at 500 minutes. Sarcopenia Measurement Sarcopenia scores were assessed using the ImageJ platform (National Institutes of Health, Madison, Wis.). All measurements were done by the lead author. CT scans were taken at the L3 body, and the cross-sectional area of the left psoas major muscle was normalized to the cross section of the vertebral body at L3 (Fig. 1). This is an accepted measurement of sarcopenia if cross-sectional area (in cm2) is not available in the image source program, as muscle degradation is more prominent when compared with bone degradation in patients with sarcopenia.18 The left side is chosen by convention. This ratio was then recorded for each patient. Ratios were then divided into groups as less than 0.2, 0.2–0.4, 0.4–0.6, 0.6–0.8, and more than 0.8. This further allows us to normalize sarcopenia scores against the population being studied rather than a healthy standard. Fig. 1. CT scans showing measurement of Sarcopenia. A, Sarcopenia measurements in a patient with an average sarcopenia score with cross-sectional area of the psoas muscle in red and the L3 vertebral body in green. The ratio of red to green, as measured by our PACS softtware, is used to calculate the sarcopenia score. B, Patient with poor sarcopenia score. Statistical Analysis All analysis was performed on SPSS (IBM, Armonk, N.Y.). ANOVA tests were performed to compare continuous variables and chi-squared test for discrete variables. Multivariate regression was also performed to control for covariates. All preoperative and intraoperative characteristics (including age, BMI, sarcopenia score, ischemia time, mFI, tracked comorbidities, vein size, and artery size) were included and controlled for during analysis. BMI, mFI, and sarcopenia score were run independently with the others as covariates. All additional factors found to be correlates with outcomes were also run to test for independence. RESULTS Demographic Data A total of 299 patients were included in this study. The average age of these patients was 52 years, with an average BMI of 27.8 kg per m2 and average sarcopenia score of 0.544. Sixty-four patients experienced complications (21%). Neoadjuvant chemotherapy was required in 127 patients (43%), and 184 patients (62%) required preoperative radiation. Sixty-two patients (21%) had an mFI of 1, and 13 (4%) had a score of 2. An estimated 172 patients (58%) had delayed reconstruction, and 20 (7%) had nipple sparing mastectomy. The high number of delayed reconstruction patients is due to the broad time period of our study, including patients from when immediate reconstruction was less popular. One hundred sixty patients had bilateral reconstruction (54%). The average ischemia time was 61 minutes, artery size was 2.6 mm, and operative time was 498 minutes. The average vein size was 2.8 mm. Full preoperative characteristics are summarized in Table 1, and complications are included in Table 2. Table 1. Demographics of Included Patients, Divided by Sarcopenia Score Factor Overall (n = 299) <0.2 (n = 3) 0.2–0.4 (n = 47) 0.4–0.6 (n = 157) 0.6–0.8 (n = 72) >0.8 (n = 20) P Age (y) 52 53 55 53 48 47 <0.001 BMI (kg/m2) 27.8 25.3 26.5 26.9 29.2 32.2 <0.001 Chemotherapy (%) 0.924  Preoperative 42.5 33 30 28 24 20  Postoperative 30.8 33 36 43 42 55  None 26.8 33 34 29 35 25 Radiation (%) 61.5 0 70 62 56 70 0.092 Diabetes (%) 6 0 6 5 8 5 0.885 Hypertension (%) 23 33 26 24 18 25 0.824 Poor function (%) 0 0 0 0 0 0 — COPD (%) 0 0 0 0 0 0 — Heart failure (%) 0.3 0 0 0 1 0 0.531 mFI (%) 0.989  0 75 67 72 75 78 75  1 21 33 23 22 17 20  2 4 0 5 3 5 5 Steroid use (%) 16 0 9 15 21 20 0.390 Delayed reconstruction (%) 58 33 62 57 58 55 0.883 Nipple sparing (%) 7 0 9 6 10 0 0.527 Bilateral reconstruction (%) 54 33 55 46 69 65 0.007 Axillary dissection (%) 32 33 30 36 25 35 0.596 Stage (%) 0.500  0 13 33 6 15 11 20  1 22 0 9 25 19 15  2 43 0 36 28 42 25  3 30 67 36 28 25 40  4 3 0 2 4 3 0 ASA score (%) 0.011  1 2 33 0 3 0 0  2 64 33 68 62 68 60  3 33 33 32 34 31 40 Abdominal surgery (%) 63 33 66 62 67 55 0.668 Breast surgery hx (%) 46 33 45 50 43 25 0.267 Operative time (min) 497 559 490 484 523 528 0.131 Ischemia time (min) 61 61 61 59 65 62 0.453 Vein size (mm) 2.8 2.8 2.8 2.8 2.7 2.9 0.379 Artery size (mm) 2.6 2.7 2.6 2.6 2.6 2.6 0.911 Complications (%) 21 100 32 20 14 20 0.002 Table 2. Complication Categories Studied, with Percentage of Total Complications and Percentage of Total Patients Included Complication Category Percentage of Total Complications Percentage of Total Patients Wound healing issue 39.1 8.4 Hematoma 17.2 3.7 Infection/abscess 17.2 3.7 Diastasis requiring mesh 7.8 1.7 Flap failure 10.9 2.3 Other medical issues 3.1 0.7 Death 4.7 1.0 Sarcopenia Scores Predicting Postoperative Complications Sarcopenia scores were split into five groups: less than 0.2, 0.2–0.4, 0.4–0.6, 0.6–0.8, and more than 0.8. A higher sarcopenia score correlates with less muscle wasting and signifies healthier physiology. Preoperative and postoperative characteristics are also summarized in Table 1. All groups were statistically equivalent, except for age (higher scores had younger patients), BMI (higher scores had heavier patients), laterality (higher scores have bilateral reconstruction), and ASA score (higher scores had worse ASA scores). When comparing the complication rate across different sarcopenia groups, the rate of complications decreases as the sarcopenia score increases (P = 0.002). Furthermore, when examining the sarcopenia score based on complication, the patients with no complications had a score of 0.556 compared with 0.499 for those with complications (P = 0.005). Other Predictors of Complications Secondary outcomes in this study included comparing the performance of the sarcopenia score with the performance of the traditional risk stratification measures, BMI and mFI, to determine superiority of one method over the others. As demonstrated in Table 1, sarcopenia does have a positive correlation with BMI; specifically, patients with a higher sarcopenia score have a higher BMI, and a negative correlation with age (lower age in higher sarcopenia), but the mFI score is equivalent across all sarcopenia groups. When investigating the average sarcopenia score in each mFI group, the values were statistically insignificant (0.5 for mFI of 0, 0.54 for mFI 1, 0.54 for mFI 2; P = 0.921). In contrast, patients with a BMI less than 30 kg per m2 had a sarcopenia score of 0.515 when compared with higher BMIs with a score of 0.603 (P < 0.001), and patients younger than 50 years of age had a score of 0.584 when compared with a score of 0.510 for older patients (P < 0.001). When assessing the predictive ability of mFI and BMI for complications, an mFI of 0 had 23% of patients with complications, an mFI of 1 with 19%, and an mFI of 2 with 0% (P = 0.127). Similarly, a BMI of less than 30 kg per m2 had 23% with unexpected complications, and a BMI more than 30 kg per m2, 17% (P = 0.217). Patients with complications had mean BMI of 27.1 kg per m2, and those without complications, a BMI of 28.3 kg/m2 (P = 0.136). Additionally, patients without complications had an average age of 52 years, whereas those with complications had an average age of 50 years (P = 0.043). However, those with an age younger than 50 had a complication rate of 24%, whereas those who were older had a rate of 18% (P = 0.108). Of all other preoperative characteristics, a history of abdominal surgery correlated with more complications, mainly wound dehiscence and bulge deformity, but a history of breast surgery correlated with lower rates of complication (Table 3). Table 3. Preoperative Characteristics Categorized by Complications Factor Overall (n = 299) No Complication (n = 235) Complication (n = 64) P Age (y) 52 52 50 0.043 BMI (kg/m2) 27.8 28 27 0.136 Chemotherapy (%) 0.554  Preoperative 42.5 27 27  Postoperative 30.8 44 38  None 26.8 29 36 Radiation (%) 61.5 64 52 0.064 Diabetes (%) 6 7 3 0.272 Hypertension (%) 23 25 16 0.110 Poor function (%) 0 0 0 — COPD (%) 0 0 0 — Heart failure (%) 0.3 0.4 0 0.601 mFI (%) 0.127  0 75 73 81  1 21 22 19  2 4 6 0 Steroid use (%) 16 17 13 0.425 Delayed reconstruction (%) 58 57 58 0.958 Nipple sparing (%) 7 6 9 0.332 Bilateral reconstruction (%) 54 54 50 0.525 Axillary dissection (%) 32 33 18 0.442 Stage (%) 0.406  0 13 11 18  1 22 23 17  2 43 31 34  3 30 31 25  4 3 3 5 ASA score (%) 0.723  1 2 2 3  2 64 64 66  3 33 34 31 Abdominal surgery (%) 63 60 73 0.049 Breast surgery hx (%) 46 49 33 0.018 Operative time (min) 497 503 478 0.079 Ischemia time (min) 61 61 59 0.228 Vein size (mm) 2.8 2.8 2.8 0.472 Artery size (mm) 2.6 2.6 2.6 0.352 Sarcopenia ratio 0.544 0.556 0.499 0.005 Multivariate Regression Multivariate regression was also performed to identify factors that were significant when controlling for all preoperative variables collected. This was performed for preoperative factors identified as statistically significant (age, sarcopenia score, breast surgery, abdominal surgery). Only sarcopenia score was found to be significant (odds ratio, 0.52; P = 0.010) when controlling for other variables (Table 4). Table 4. Multivariate Regression outcomes Parameter Odds Ratio P Age 1.49 0.300 Sarcopenia 0.52 0.010 History of abdominal surgery 1.56 0.776 History of breast surgery 0.67 0.653 DISCUSSION Breast reconstruction is a critical component of comprehensive breast cancer care. FFBR is an increasingly popular modality of breast reconstruction due to use of autologous tissues and avoidance of chronically implanted devices. However, FFBR is a complex procedure with increased perioperative risk when compared with implant-based reconstruction due to the extent of surgery involved. Many attempts have been made at identifying preoperative risk factors for FFBR, including BMI, history of diabetes, laterality of surgery, and the mFI. Sarcopenia is a relatively newer modality for measuring preoperative risk in FFBR. We sought to assess its utility in risk stratification and to determine whether it outperforms other preoperative characteristics such as age, BMI, and mFI. As demonstrated by our data, the sarcopenia score was the only independent predictor of complications in our patient cohort. Although the absolute difference of the sarcopenia score between the patients with and without complications is small, when splitting the patients by sarcopenia score it is obvious that patients with lower scores had a higher rate of complications than patients with higher sarcopenia scores. Importantly, BMI and mFI showed no significant correlation with complications in this dataset and thus, had no predictive power for complications. Factors such as age, history of breast surgery, and history of abdominal surgery did have correlations with complications that disappeared when controlling for other variables. This finding related to sarcopenia is supported by several studies in the literature, though it does stand in contrast to those of others. Kim et al22 described a series of patients who underwent FFBR, with preoperative sarcopenia associated with significantly higher rates of complication and BMI not associated with complications, though they did not include the mFI. This was also supported by the findings of Pittelkow et al, who found similar results in 103 patients, again not including the mFI.23 On the other hand, Broyles et al, Yoshino et al, and Sadok et al found that sarcopenia was not a strong predictor of complications and that BMI outperformed sarcopenia.24–26 These three studies that found negative results in sarcopenia also suffered from smaller sample sizes than used in our study. Furthermore, each of these studies evaluated specific subsets of patients, not the overall breast reconstruction population. In the study by Broyles, only patients who underwent delayed reconstruction after radiation (those classified as “high risk” by the authors) were analyzed. These patients are predisposed to complications, which may explain the equivalent rates. Broyles and Yoshino also defined sarcopenia as a static value, describing sarcopenia as a present or absent state based on an arbitrary value. In our study, we describe sarcopenia as a continuum; sarcopenia values are not necessarily translatable on a present/absent scale due to variance in patients due to ethnicity, height, and weight. Thus, a simple binary compared with a “standard” value is not necessarily reliable; comparing the sarcopenia score with the population standard, as we did with the quintile division, is a more reliable form of analysis. Sadok et al did not even study sarcopenia as an independent variable; they studied sarcopenic obesity, looking at sarcopenic patients with an elevated BMI, which again evaluates different factors than those in our study. Importantly, our study also demonstrates that BMI and mFI are not reliable predictors of complications. BMI has been demonstrated frequently as an unreliable predictor of complications across various surgical disciplines27–29; this study fits in with the prevailing trend in the literature. In our study, we demonstrate that the BMI in patients with complications and without complications are approximately the same, suggesting that having a high BMI alone does not result in having more complications. This is because BMI has been repeatedly demonstrated as a poor measure of overall physical health leading to complications,29 though it does correlate with specific complication types such as bulges and hernias. However, this is also true of abdominal surgical history, due to multiple violations of fascia, suggesting that BMI is not physiologically reliable. A more comprehensive measurement of overall health, such as the sarcopenia score, is more applicable. The mFI aspect, however, is more interesting. Again, mFI does have detractors in the literature due to the lack of strength of prediction, but recent studies have demonstrated that the mFI has predictive value.30–33 Interestingly, many of the mFI articles are database studies, which are limited in the scope and the assessment of individual complications, only presenting recent complications as well as the presence, not the severity. Our study was able to review longer term outcomes to assess procedure-associated complication rates more accurately, providing more granular and clinically meaningful data. With sarcopenia being identified as a risk factor, the question remains as to why sarcopenia is a predictor of complications and how it can be utilized in clinical practice. The physiologic relationship of sarcopenia and complications has not been elucidated. Sarcopenia is considered a global measure of physical health. Muscular catabolism is a sign of declining physiologic reserve, regenerative capacity, and impaired immunologic competence, all of which predispose patients to poor wound healing and complications.34 However, recent studies have demonstrated that regular exercise can boost sarcopenia scores.35 It is possible that improvements in preoperative nutrition and exercise may improve psoas mass and sarcopenia scores, and thus, outcomes. This study does have limitations. The sarcopenia measurements were taken by hand on individual scans, introducing error. Patients were followed up for a minimum of only 6 months after surgery in a retrospective fashion; this limits the detail that can be obtained regarding postoperative complications and preoperative characteristics. Specific complications that arise in the long term, such as bulges and hernias, could not be addressed. Complications were also grouped into general groups; flap takebacks and admissions for infections were treated equivalently, which limits the specificity of analysis. Further, CT scans were not reliably performed on the day of surgery, which would provide the optimal assessment of perioperative health. We attempted to control for this by performing our scans within a certain preoperative time period to best limit variability between scan and surgery. Various changes in health between date of scan and date of surgery could affect the relationship between sarcopenia score and complications. Neoadjuvant chemotherapy can also affect the sarcopenia score measured and the complications that occur postoperatively. Although the broad nature of the dataset is an asset in many fashions, allowing us to capture aspects of all types of patients, it can also limit subgroup analysis and more specific outcomes. CONCLUSIONS Breast cancer remains a common condition, and breast reconstruction is an important component of a comprehensive treatment strategy. Autologous breast reconstruction with free flaps is an increasingly popular form of reconstruction. Using CT scans obtained for surgical planning and/or cancer staging, we demonstrated that the sarcopenia score, a measure of muscle mass at the L3 vertebra, was a superior predictor of complications when compared with both the BMI and the mFI. DISCLOSURE The authors have no financial interest to declare in relation to the content of this article. Published online 18 July 2023. Presented as a poster at ASRM 2023, Miami, Fl., and AAPS 2023, Chicago, Ill. Disclosure statements are at the end of this article, following the correspondence information. ==== Refs REFERENCES 1. Momenimovahed Z Salehiniya H . Epidemiological characteristics of and risk factors for breast cancer in the world. Breast Cancer. 2019;11 :151–164.31040712 2. Pollom EL Qian Y Chin A . Rising rates of bilateral mastectomy with reconstruction following neoadjuvant chemotherapy. Int J Cancer. 2018;143 :3262–3272.29992582 3. Wilkins EG Cederna PS Lowery JC . Prospective analysis of psychosocial outcomes in breast reconstruction: one-year postoperative results from the Michigan Breast Reconstruction Outcome Study. Plast Reconstr Surg. 2000;106 :1014–1025; discussion 1026–1027.11039373 4. Dibbs R Trost J DeGregorio V . Free tissue breast reconstruction. Sem Plast Surg 2019;33 :59–66. 5. Garvey PB Villa MT Rozanski AT . The advantages of free abdominal-based flaps over implants for breast reconstruction in obese patients. Plast Reconstr Surg. 2012;130 :991–1000.23096600 6. Pirro O Metask O Vindigni V . Comparison of patient-reported outcomes after implant versus autologous tissue breast reconstruction using the BREAST-Q. Plast Reconstr Surg. 2017;5 :1. 7. Vyas RM Dickinson BP Festekjian JH . Risk factors for abdominal donor-site morbidity in free flap breast reconstruction. Plast Reconstr Surg. 2008;121 :1519–1526.18453973 8. Jandali S Nelson JA Sonnad SS . Breast reconstruction with free tissue transfer from the abdomen in the morbidly obese. Plast Reconstr Surg. 2011;127 :2206–2213.21617454 9. Prantl L Moellhoff N Fritschen UV . Impact of smoking status in free deep inferior epigastric artery perforator flap breast reconstruction: a multicenter study. J Reconstr Microsurg. 2020;36 :694–702.32726819 10. Masoomi H Clark EG Paydar KZ . Predictive risk factors of free flap thrombosis in breast reconstruction surgery. Microsurgery. 2014;34 :589–594.24665051 11. Jain NS Vuong LN Hickman LB . Using the modified frailty index to predict negative outcomes in free-flap breast reconstruction. Microsurg. 2021;41 :709–715. 12. Ali B Choi EE Barlas V . Modified frailty index (mFI) predicts 30-day complications after microsurgical breast reconstruction. J Plast Surg Hand Surg. 2022;56 :229–235.34431755 13. Dodds R Sayer AA . Sarcopenia and frailty: new challenges for clinical practice. Clin Med. 2016;16 :455–458. 14. Mori H Tokuda Y . Differences and overlap between sarcopenia and physical frailty in older community-dwelling Japanese. Asia Pac J Clin Nutr. 2019;28 :157–165.30896427 15. Shen Y Hao Q Zhou J . The impact of frailty and sarcopenia on postoperative outcomes in older patients undergoing gastrectomy surgery: a systematic review and meta-analysis. BMC Geriatr. 2017;17 :188.28826406 16. Cruz-Jentoft AJ Baeyens JP Bauer JM ; European Working Group on Sarcopenia in Older People. Sarcopenia: European consensus on definition and diagnosis: report of the European Working Group on Sarcopenia in Older People. Age Ageing. 2010;39 :412–423.20392703 17. Miyamoto Y Baba Y Sakamoto Y . Sarcopenia is a negative prognostic factor after curative resection of colorectal cancer. Ann Surg Oncol. 2015;22 :2663–2668.25564158 18. Zakaria HM Wilkinson BM Pennington Z . Sarcopenia as a prognostic factor for 90-day and overall mortality in patients undergoing spine surgery for metastatic tumors: a multicenter retrospective cohort study. Neurosurgery. 2020;87 :1025–1036.32592483 19. Englesbe MJ Patel SP He K . Sarcopenia and mortality after liver transplantation. J Am Coll Surg. 2010;211 :271–278.20670867 20. Barnes LA Li AY Wan DC . Determining the impact of sarcopenia on postoperative complications after ventral hernia repair. J Plast Reconstr Aesthet Surg. 2018;71 :1260–1268.30173713 21. Nakamura H Makiguchi T Yamaguchi T . Impact of skeletal muscle mass on complications following expander breast reconstruction. J Plast Reconstr Aesth Surg. 2020;73 :1285–1291. 22. Kim S Lee KT Jeon BJ . Association of preoperative sarcopenia with adverse outcomes of breast reconstruction using deep inferior epigastric artery perforator flap. Ann Surg Oncol. 2022;29 :3800–3808.35128597 23. Pittelkow EM DeBrock WC McLaughlin BE . Preoperatively identified sarcopenia leads to increased postoperative complications, hospital and ICU length of stay in autologous microsurgical breast reconstruction. J Reconstr Microsurg. 2020;36 :59–63.31470457 24. Yoshino M Oda G Nakagawa T . Higher body mass index is a more important risk factor than sarcopenia for complications in reconstruction of the deep inferior epigastric perforator. Asian J Surg. 2022;45 :360–366.34340895 25. Broyles JM Smith JM Phillips BT . The effect of sarcopenia on perioperative complications in abdominally based free-flap breast reconstruction. J Surg Oncol. 2020;122 :1240–1246.32673425 26. Sadok N Hartmans ME de Bock GH . The effect of sarcopenic obesity and muscle quality on complications after DIEP-flap breast reconstruction. Heliyon. 2022;8 :e09381.35600454 27. Ri M Aikou S Seto Y . Obesity as a surgical risk factor. Ann Gastroenterol Surg. 2018;2 :13–21.29863119 28. Chen H-N Chen X-Z Zhang WH . The impact of body mass index on the surgical outcomes of patients with gastric cancer. Medicine (Baltimore). 2015;95 :e1769. 29. De Santo LS Moscariello C Zebele C . Implications of obesity in cardiac surgery: pattern of referral, physiopathology, complications, prognosis. J Thorac Dis. 2018;10 :4532–4539.30174906 30. Kravchenko TV Ciaramella MA Ady J . Frailty index is a poor predictor of postoperative morbidity and mortality after ruptured abdominal aortic aneurysm. J Vasc Surg. 2020;72 :e94. 31. Khan MA Elsayed N Naazie I . Modified frailty index as a predictor for outcomes after transcarotid artery revascularization. J Vasc Surg. 2021;74 :e147–e148. 32. Pulik L Jaskiewicz K Sarzynska S . Modified frailty index as a predictor of the long-term functional result in patients undergoing primary total hip arthroplasty. Reumatologia. 2020;58 :213–220.32921828 33. Elsamadicy AA Freedman IG Koo AB . Modified-frailty index does not independently predict complications, hospital length of stay or 30-day readmission rates following posterior lumbar decompression and fusion for spondylolisthesis. Spine J. 2021;21 :1812–1821.34010683 34. Argilés JM Busquets S Stemmler B . Cachexia and sarcopenia: mechanisms and potential targets for intervention. Curr Opin Pharmacol. 2015;22 :100–106.25974750 35. Adams SC Segal RJ McKenzie DC . Impact of resistance and aerobic exercise on sarcopenia and dynapenia in breast cancer patients receiving adjuvant chemotherapy: a multicenter randomized controlled trial. Breast Cancer Res Treat. 2016;158 :497–507.27394134

PMC010xxxxxx/PMC10353711.txt

==== Front Plast Reconstr Surg Glob Open Plast Reconstr Surg Glob Open GOX Plastic and Reconstructive Surgery Global Open 2169-7574 Lippincott Williams & Wilkins Hagerstown, MD 00028 10.1097/GOX.0000000000005123 3 Cosmetic Original Article CONTOUR Australia: Condition of Submental Fullness and Treatment Outcomes with Belkyra Registry Boxley Sarah G. MBBS, FRACGP, FACAM * Lin Frank MBBS, FRACS † Lee See Neville MBBS ‡ St. Rose Suzanne PhD, MBA § Battucci Simona MD ¶ Simonyi Susan BScN, RN ‖ From * SkinBox Clinics, Perth, WA, Australia † Eastern Plastic Surgery, Box Hill North, VIC, Australia ‡ Kiora Medical Spa, Hawthorn, VIC, Australia § Global Epidemiology Pharmacovigilance and Patient Safety, AbbVie, Marlow, United Kingdom ¶ Pharmacovigilance and Patient Safety, AbbVie, Rome, Italy ‖ Clinical Program Development, AbbVie, Singapore. Sarah G. Boxley, MBBS, FRACGP, FACAM, SkinBox Clinics, 4/6 Short Street, Fremantle, WA 6160, Australia, E-mail: sboxley@me.com 7 2023 18 7 2023 11 7 e512316 6 2022 1 6 2023 Copyright © 2023 The Authors. Published by Wolters Kluwer Health, Inc. on behalf of The American Society of Plastic Surgeons. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open-access article distributed under the terms of the Creative Commons Attribution-Non Commercial-No Derivatives License 4.0 (CCBY-NC-ND), where it is permissible to download and share the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal. Background: Submental fat (SMF) contributes to an aged or overweight appearance that may negatively impact an individual’s psychological well-being. Deoxycholic acid (ATX-101) is an injectable formulation of deoxycholic acid approved to treat SMF. The Condition of Submental Fullness and Treatment Outcomes Registry (CONTOUR) Australia study was designed to understand treatment patterns and outcomes with ATX-101 in Australia. Methods: CONTOUR Australia was a phase 4, prospective, observational, multicenter registry that enrolled adults considering treatment for SMF reduction. Results: The registry enrolled 86 patients from six sites. Significant changes from baseline through the end of treatment indicated improvement in mild to moderate fullness associated with SMF on the Clinician-Reported SMF Rating Scale and the Patient-Reported SMF Rating Scale, improvement in SMF-associated psychological impact after treatment on the Patient-Reported SMF Impact Scale, no overall worsening in skin laxity based on Submental Skin Laxity Grade, and increased patient satisfaction with the face/chin on the Subject Self-Rating Scale after receiving treatment. Adverse events were all mild and mostly related to the injection site (ie, bruising and swelling). Conclusion: CONTOUR Australia observed clinically meaningful and significant outcomes and further supports ATX-101 as a well-tolerated and effective treatment for SMF reduction. OPEN-ACCESSTRUE COUNTRYAUSTRALIA SDCT ==== Body pmcTakeaways Question: What are the treatment patterns and clinical outcomes of the injectable form of deoxycholic acid in Australia? Findings: The Condition of Submental Fullness and Treatment Outcomes Registry (CONTOUR) Australia study (a phase 4, prospective, observational, multicenter registry) showed improvement in mild to moderate fullness associated with submental fat and increased patient satisfaction with face/chin on clinician- and patient-assessed rating scales after treatment with deoxycholic acid, with no overall worsening in skin laxity. Adverse events were mild and mostly related to the injection site. Meaning: CONTOUR Australia demonstrated clinically meaningful outcomes that further support deoxycholic acid as a well-tolerated and effective treatment for submental fat reduction. INTRODUCTION Genetics or lifestyle factors can result in fat accumulation in the preplatysmal submental area, also known as submental fat (SMF).1,2 During aging, excess SMF and weakening of the mandible septum, which holds fat compartments in place, obscure jawline contour/definition, thus contributing to an aged or overweight appearance, which may negatively impact an individual’s psychological well-being.1,3,4 Minimally invasive techniques for SMF reduction have become increasingly popular because they require less recovery time and can be used as part of a multimodal approach to create customizable treatment plans.5 ATX-101 [Belkyra in Australia, Canada, Europe, and South Korea, and Kybella in the United States; Kythera Biopharmaceuticals, Inc. (an affiliate of Allergan)] is an injectable formulation of deoxycholic acid, a secondary bile acid involved in dietary fat emulsification.2,6 ATX-101 causes adipocyte lysis when injected into subcutaneous fat tissue.2,6 Clinical trials have demonstrated the safety and efficacy of ATX-101 for reducing SMF and for improving patient satisfaction with the chin/jawline.7–11 Real-world clinical practice setting data from the Condition of Submental Fullness and Treatment Outcomes Registry (CONTOUR) study North America provided insights on the types of patients who seek SMF reduction, patient and physician treatment goals, and posttreatment perspectives.12,13 After the approval of ATX-101 in Australia in July 2016 for the treatment of submental fullness due to SMF,14 we conducted the current registry study (CONTOUR Australia) to develop a comprehensive understanding of how ATX-101 is utilized in clinical practice in Australia and assess the risks and benefits associated with its treatment. METHODS Patients Eligible patients were adults aged 18 years or older presenting with submental fullness due to accumulation of unwanted SMF and considered by their treating physician to be a candidate for SMF reduction. Key exclusion criteria included severe skin laxity, any other cause of fullness in the submental region (eg, thyroid enlargement, thyromegaly, cervical adenopathy, cervical lymphadenopathy, pronounced submandibular glands, lymph nodes, and muscles), and current or previous participation in an interventional clinical study involving ATX-101. Study Design CONTOUR Australia was a phase 4, prospective, observational, multicenter registry for patients with SMF treated with ATX-101 to help understand treatment patterns and outcomes in Australia. Study approval was obtained from each center’s independent ethics committee, and all patients provided written informed consent. No treatment assignment was performed in the study. Physicians and patients must have agreed on ATX-101 treatment before study enrollment. Patients who elected treatment were followed up until their SMF reduction treatment was completed or discontinued. Enrolled patients were assessed regularly by their treating physician according to usual clinical practice. Data collection was anticipated to continue for approximately 15 months after enrollment. The duration of individual patient participation (treatment and retreatment) varied depending on treatment characteristics, individual requirements, and the discretion of the treating physician. The registry included four phases: enrollment visit, baseline visit, follow-up visit, and end-of-treatment (EOT) visit. The EOT visit occurred either at the last scheduled follow-up visit after completion of ATX-101 treatment, within 3 months of the last treatment session, when a patient elected to discontinue treatment, or before study closure. Study Procedures ATX-101 was injected into the subcutaneous fat tissue in the submental area using an area-adjusted dose of 2 mg per cm2. A single treatment consisted of up to 50 injections, 0.2 mL per injection supplied in single-use vials (up to a total of 10 mL), spaced 1 cm apart. Up to six single treatments could be administered at intervals no less than 1 month apart. Assessments and End Points The primary effectiveness end points have been used in previous clinical studies15,16 and in the CONTOUR North America study.12,13 SMF severity was assessed using the validated Clinician-Reported Submental Fat Rating Scale (CR-SMFRS; 0 = absent, 1 = mild, 2 = moderate, 3 = severe, and 4 = extreme) and validated Patient-Reported Submental Fat Rating Scale (PR-SMFRS; 0 = no chin fat, 1 = slight amount, 2 = moderate amount, 3 = large amount, 4 = very large amount); lower values indicate improvement/less SMF.15,17 The psychological impact of SMF was evaluated using the validated Patient-Reported Submental Fat Impact Scale (PR-SMFIS), with scores ranging from 0 (no impact) to 10 (greatest negative impact).15,17 Skin laxity was assessed using the Submental Skin Laxity Grade (SMSLG; 1 = none, 2 = mild, 3 = moderate, 4 = severe).15 Patient satisfaction with the appearance of the face and chin was evaluated using the seven-point Subject Self-Rating Scale (SSRS), ranging from 0 (extremely dissatisfied) to 6 (extremely satisfied).15 The patient’s self-perception of age (SPA),18 which assesses whether the patient perceives oneself to look younger or older than actual age, was also evaluated. With the exception of the SPA scale, which was assessed during enrollment and at the EOT visit only, these end points were assessed at enrollment, follow-up visits, and the EOT visit. Other effectiveness end points collected during the EOT were the patient global questions (PGQ), as well as patient EOT and physician EOT questionnaires, which have also been used in previous clinical studies7,8 and the CONTOUR North America study.13 The PGQ included patients’ ratings of the fat under the chin, definition between the chin and neck, and satisfaction with treatment. Pain scores based on the pain numeric rating scale were collected at each treatment session, follow-up visit, and EOT and were categorized into four severity categories: none (0), mild (1–3), moderate (4–6), and severe (7–10). The worst category of pain experienced at patient level throughout the study was reported. Additional assessments included the Physician Practice Setting Questionnaire and the Patient and Physician Treatment Goals questionnaires. Exposure to study treatment was also assessed. Treatment details and treatment procedures questionnaires were completed by the physician at the enrollment visit, each treatment session, and all follow-up visits. (See document, Supplemental Digital Content 1, which describes these questionnaires in more detail. http://links.lww.com/PRSGO/C659.) The incidence, severity, and duration of adverse events (AEs) were assessed throughout the study in the safety population, which included all patients who received at least 1 ATX-101 injection. Statistical Analyses Effectiveness analyses were performed on the evaluable population, which consisted of all enrolled patients who received at least one ATX-101 injection, as well as at least one effectiveness assessment at the enrollment visit (baseline) and at least one effectiveness assessment at the follow-up or EOT visit. SMF assessments were summarized using descriptive statistics. For the follow-up and EOT visits, the mean change from baseline for each SMF assessment was analyzed using a paired t test. The number and proportion [and its associated 95% Clopper-Pearson confidence interval (CI)] of patients with at least a one-grade improvement in the assessment were reported for CR-SMFRS, PR-SMFRS, SMSLG, and SSRS. SPA assessments were summarized as frequency counts. For the categorical responses in the PGQ, the 95% CIs for the proportion of the response categories of interest were reported. Data from the Physician Practice Setting Questionnaire, the Patient and Physician Treatment Goals Questionnaires, exposure to study treatment, as well as treatment procedures and details questionnaires were summarized using descriptive statistics. AEs were summarized using descriptive statistics. Statistical analyses were performed using SAS software, version 9.3 (SAS Institute, Cary, N.C.). RESULTS Site and Physician Characteristics A total of six Australian sites/practices were involved in the study. The mean percentage of practices that focused on facial aesthetics was 72.5%. The sites provided aesthetic treatments for a mean of 13.2 years and had a median number of two physicians per site who focused on aesthetics. All six sites used injectable treatments. The mean percentage of patients presenting to the practices in the past week who had submental fullness due to SMF was 31.5%; only 0.8% of patients were referred by other physicians for treatment. The most frequently reported treatment options by sites were energy devices (laser- and radiofrequency-specified) and injectables (ATX-101), which were reported in five sites each (83.3%). Patients Of the 86 patients enrolled in the registry study, 79 (91.9%) were in the safety population, whereas 77 patients comprised the evaluable population. The majority of patients (75/86, 87.2%) completed the study. Among the 11 patients who did not complete the study, seven (8.1%) patients were lost to follow-up, three (3.5%) decided to discontinue, and one (1.2%) did not complete for other reasons (pregnancy). The majority of enrolled patients were women (95.3%) and White (72.1%), with a mean age of 43.2 years, a mean weight of 67.9 kg, and a mean body mass index of 25.2 kg per m2. Most patients were nonsmokers (Table 1). At the end of treatment, the change from baseline (mean [SD]) in weight (n = 56) and body mass index (n = 55) was 0.1 (3.55) kg and −0.0 (1.4) kg/m2, respectively. Table 1. Baseline Demographic Parameters Parameter Enrolled Population (N = 86) Mean (SD; range) age, y 43.2 (10.2; 24–66) Women, n (%) 82 (95.3) Mean (SD) weight, kg 67.9 (10.8) Mean (SD; range) BMI, kg/m2 25.2 (3.7; 16.6–34.8) Race, n (%)  White 62 (72.1)  Asian (Chinese) 23 (26.7)  Not reported 1 (1.2) Lifestyle and medical history, n (%)  Smoker 6 (7.0)  Diabetes 1 (1.2)  Hypertension 2 (2.3)  Other 1 (1.2) Patient and Physician Treatment Goals Questionnaires According to physicians, the top two goals for patients were to achieve a more defined jawline (34/86 patients, 39.5%) and to achieve an ideal submental contour (31/86 patients, 36.0%). The top four treatment goals endorsed by patients were to achieve a more defined jawline (40/86, 46.5%), to look younger (18/86, 20.9%), to look thinner (12/86, 14.0%), and to feel more confident (6/86, 7.0%). The mean amount of time patients had concerns about their double chin was 65.0 months. The majority of patients (76/86, 88.4%) did not receive previous SMF treatment, and most patients (61/86, 70.9%) did not report weight change in the past 12 months. A total of 37 patients (43.0%) reported being lightly active (1–3 days/week), 31 (36.0%) reported being moderately active (3–5 days/week), and 16 (18.6%) reported being sedentary. Treatment Procedures and Details A summary of treatment procedures and details for the baseline and first follow-up visits are presented in Table 2. Similar results were observed for follow-up visits two to four. The majority of patients did not receive other treatments in combination with ATX-101. However, for patients who received concomitant aesthetic treatments, the most frequently coadministered treatments were injectables, specifically botulinum toxin. Table 2. Treatment Procedures and Details (Safety Population) Parameter Baseline Follow-up Visit 1 Skin marking grid used  n/N (%) 54/79 (68.4) 31/33 (93.9)  Mean application time, min 2.8 5.2 Pretreatment comfort regimen used  n/N (%) 68/79 (86.1) 29/33 (87.9) Effectiveness, n/N (%) 63/68 (92.6) 29/29 (100.0) Patients who received other treatments, n/N (%) 10/79 (12.7) 3/33 (9.1) Mean number of other treatments given 1.5 1.5 Other treatment categories  Energy-based devices (eg, laser, ultrasound), n/N (%) 3/79 (3.8) 0/33 (0.0)  Injectables (eg, botulinum toxin, dermal fillers) 8/79 (10.1) 3/33 (9.1)  Other (eg, skin booster) 1/79 (1.3) 0/33 (0.0) Mean time spent for patient consultation, min 20.3 Mean time to administer treatment, min 10.0 Mean time to posttreatment, min 12.1 Extent of Exposure Table 3 summarizes the extent of exposure. The average treatment duration was 104.5 days with an average number of 1.6 treatment sessions, an average interval between sessions of 73.4 days, and an average volume of ATX-101 administered per injection site of 0.2 mL (Table 3). The mean (SD) injection volume across all treatment sessions was 3.6 (1.5) mL. The total volume of ATX-101 administered per treatment decreased from a mean (SD) of 3.7 (1.4) mL at treatment 1 (n = 79) to 2.7 (1.2) mL at treatment 4 (n = 3). Table 3. Exposure to Treatment over the Study Period (Safety Population) Parameter Result Treatment duration, d* (N = 79)  Mean (SD) 104.5 (120.6)  Median (range) 58.0 (1.0–372.0) Number of sessions (N = 79)  Mean (SD) 1.6 (0.9)  Median (range) 1.0 (1.0–6.0) Total volume of injection, mL  Mean (SD) 3.6 (1.5)  Median (range) 4.0 (2.0–8.0) Volume of treatment per injection site, mL (N = 78)  Mean (SD) 0.2 (0.1)  Median (range) 0.2 (0.1–0.5) Interval between treatments, days† (N = 33)  Mean (SD) 73.4 (39.5)  Median (range) 65.0 (24.0–220.0) Injection instrument used  Needle, n/N (%) 78/79 (98.7)  Cannula, n/N (%) 1/79 (1.3) Off-label use, n/N (%)‡ 7/79 (8.9) Total volume injected outside of submentum (mL)  Mean (SD) 1.6 (1.3)  Median (range) 0.8 (0.4–3.6) * Treatment duration in days was calculated as date of last treatment session – (date of index treatment + 1). † Patients who had only one treatment session were excluded. The interval between treatments was calculated as treatment duration in days/(number of treatment sessions during the treatment period – 1). ‡“ Off-label use” refers to the use of ATX-101 outside the approved submental area. Primary Effectiveness End Points Mean CR-SMFRS, PR-SMFRS, PR-SMFIS, and SMSLG scores decreased from baseline through the EOT (Figs. 1 and 2). Mean SSRS scores increased from baseline through the EOT (Fig. 2). Compared with baseline, the mean changes from baseline in CR-SMFRS, PR-SMFRS, PR-SMFIS, SMSLG, and SSRS scores were significantly different across follow-up visits one to three and EOT. By the EOT visit, the majority of patients had a 1-grade or more improvement in CR-SMFRS (Fig. 1), PR-SMFRS, and SSRS scores, whereas 41.7% (30/72) of patients had a 1-grade or more improvement in SMSLG scores (Table 4). The mean (SD) number of treatment sessions for patients with a 1-grade or more improvement in CR-SMFRS and PR-SMFRS scores was 1.6 (1.02) and 1.7 (0.98) sessions, respectively. In the SPA assessment, 27.3% (21/77) of patients at the EOT visit thought that they looked younger than their actual age (median number of years younger, 4; range, 2–10) compared with 14.3% (11/77) of patients at baseline (Fig. 3). Similarly, the number of patients who thought that they looked older than their actual age decreased from 28.6% to 1.3%. Representative patient photographs taken before and after treatment are shown in Figures 4 and 5. Table 4. Responder Rates for Primary Effectiveness End Points (Evaluable Population) End Point Visit Result 95% CI PR-SMFRS ≥1-grade improvement, n/N (%) 1 27/53 (50.9) 36.8, 64.9 2 14/22 (63.6) 40.7, 82.8 3 6/7 (85.7) 42.1, 99.6 4 0/1 (0.0) NA* EOT 54/72 (75.0) 63.4, 85.4 SMSLG ≥1-grade improvement, n/N (%) 1 17/52 (32.7) 20.3, 47.1 2 11/22 (50.0) 28.2, 71.8 3 4/7 (57.1) 18.4, 90.1 4 0/1 (0.0) NA* EOT 30/72 (41.7) 30.2, 53.9 SSRS ≥1-grade improvement, n/N (%) 1 37/53 (69.8) 55.7, 81.7 2 18/22 (81.8) 59.7, 94.8 3 7/7 (100.0) 59.0, 100.0 4 0/1 (0.0) NA* EOT 64/73 (87.7) 77.9, 94.2 PGQ responder rate for fat under chin, n/N (%) EOT  A great deal better 27/77 (35.1)  Moderately better 28/77 (36.4) PGQ responder rate for definition between chin and neck, n/N (%) EOT  A great deal better 25/77 (32.5)  Moderately better 29/77 (37.7) PGQ responder rate for satisfaction with treatment, n/N (%) EOT  Extremely satisfied 35/77 (45.5)  Moderately satisfied 23/77 (29.9) * No CIs available because n/N = 0/1. Fig. 1. Clinician-reported end points. A, Mean CR-SMFRS score by visit. B, Percentage of patients showing no. 1-, 2-, or 3-grade improvements, or a 1-grade deterioration, in CR-SMFRS by visit. C, Mean SMSLG score by visit. CFB, change from baseline; NA, not available. *No statistics available because n = 1. †CR-SMFRS 1-grade or more improvement. Fig. 2. Patient-reported end points. A, Mean PR-SMFRS score by visit. B, Mean PR-SMFIS score by visit. C, Mean SSRS score by visit. CFB, change from baseline; NA, not available. *No statistics available because n = 1. Fig. 3. Summary of patient SPA by visit. Fig. 4. Representative patient photographs. Before (A) and after treatment (B) photographs of a 31-year-old, nonsmoking White female patient who underwent two treatment sessions with ATX-101 (total of four ATX-101 vials) and achieved a 1-grade improvement in the CR-SMFRS, with slight worsening of skin laxity as measured by SMSLG scale 4 months after the last treatment session. Fig. 5. Representative patient photographs. Before (A) and after treatment (B) photographs of a 50-year-old, nonsmoking White male patient who underwent four treatment sessions with ATX-101 (total of 11 ATX-101 vials) and achieved a 1-grade or more improvement in the CR-SMFRS and no change in skin laxity as measured by the SMSLG scale 2 months after the last treatment session. Additional End Points Based on PGQ responses at the EOT, more than 70% of patients rated the fat under their chin and the definition between chin and neck as moderately or a great deal better, and more than 75% of patients were moderately or extremely satisfied with their treatment (Table 4). At baseline, the mean (SD) pain numeric rating scale score was 4.1 (2.5), which decreased to 3.1 (2.2) at postbaseline treatment visit 1 and 2.6 (2.6) at the EOT visit. An equal number of patients reported worst pain experienced as mild and moderate (28/77, 36.4% for each category), whereas 26.0% (20/77) of patients reported the worst pain experienced as severe. On the patient EOT questionnaire, 49 of 77 patients (63.6%) reported that they achieved their treatment goals, 17 (22.1%) indicated that they partially achieved their treatment goals, and only seven patients (9.1%) reported that they did not achieve their treatment goals. The most frequently reported reason for patients ending SMF reduction treatment was meeting their treatment goal (38/77, 49.4%), followed by treatment being too expensive (11/77, 14.3%). More than half of patients (53.2%) reported that meeting their treatment goal was the main reason for ending treatment. Additionally, 77.9% of patients indicated that after ending treatment they would undergo treatment again. Responses to the physician EOT questionnaire (See table, Supplemental Digital Content 2, which shows the physician end-of-treatment questionnaire. http://links.lww.com/PRSGO/C660) showed that physicians achieved treatment goals and were satisfied with the SMF outcome in the majority of patients (65/77, 84.4% for both responses). More than 90% of physicians believed that the injection training adequately prepared them to administer treatment and for patient side effects. Physicians reported that the decision to end SMF treatment was the patients in 51.9% of cases. Safety Of the 79 patients in the safety population, 18 (22.8%) reported 31 treatment-emergent AEs, all of which were mild in severity, were considered related to treatment, and resolved by the end of the study. Seventeen patients (21.5%) reported AEs during the first treatment session. Of the 31 reported AEs, 20 (64.5%) were reported during the first treatment session. Most AEs were related to the injection site (eg, bruising and swelling). A summary of AEs with corresponding average durations is reported in Table 5. Table 5. Summary of AEs* Safety Population (N = 79) Summary of AEs Patients with at least 1 AE, n (%) 18 (22.8)  Total number of AEs† 31 Patients discontinued from treatment due to AE, n (%) 1 (1.3)  Total number of AEs resulting in discontinuation from treatment† 2 Classification of AEs, n (%) Injection site 17 (21.5)  Bruising 3 (3.8)  Swelling 17 (21.5) Arterial injury 1 (1.3) Heart rate decreased 1 (1.3) Lethargy 1 (1.3) * All AEs were mild in severity and resolved by the end of the study. † The count is at event level (ie, if a patient has multiple AEs, all of his/her AEs are counted). One patient experienced an AE of special interest, and a vascular injury at or near the injection site. In this patient, injury to a small-caliber vessel may have occurred during injection of her left jowl area (not SMF). Postinjection, immediate blanching (diameter: ≈10 mm) was observed in an area away from but adjacent to the injection site; no hematoma was observed. The blanched skin led to a superficial skin ulcer (diameter: 2 mm) that healed spontaneously with no sequelae. The AE of vascular injury was mild in severity. No treatment of the AE was undertaken, and the patient continued in the study. The AE was assessed as being treatment related. Another patient experienced lethargy and a decreased heart rate that were considered related to treatment. Although both AEs were mild in severity and no medications were administered, the patient discontinued from treatment. DISCUSSION The primary effectiveness end points demonstrated significant changes from baseline through the EOT visit, which indicate an improvement in mild to moderate fullness associated with SMF (CR-SMFRS and PR-SMFRS), improvement in SMF-associated psychological impact (PR-SMFIS), no overall worsening in skin laxity (SMSLG), and increased patient satisfaction with face/chin (SSRS). All 31 AEs reported in 18 patients were mild in severity and resolved by the end of the study. The safety profile of ATX-101 in this study was consistent with the known safety profile of this treatment, and no new safety findings were reported. Nearly 80% of patients indicated that, after ending treatment with ATX-101, they would undergo treatment again, highlighting the satisfaction with and acceptable tolerability profile of this treatment modality. Improvements in SMF reduction, patient satisfaction, and psychological impact observed in the current study were comparable to those reported in the REFINE-1 and REFINE-2 clinical trials,7,8,15 as well as in the CONTOUR North America study.12,13 As with the current study, these studies also reported a lack of overall worsening of skin laxity despite reductions in SMF.13,15 Lower overall rates of AEs were observed in the registry studies compared with the clinical studies (13.0% and 22.8% of patients reporting AEs in the CONTOUR North America and Australia studies, respectively, versus 97.3% in the pooled data from REFINE-1 and 2).12,13,15 This finding may be a result of patients undergoing fewer treatments with lower total ATX-101 volumes used in clinical practice compared with the clinical trials, wherein investigators apply the dose as per protocol.12 Similar to previous studies, most AEs were related to the injection site (eg, swelling and bruising).12,13,15 Unlike previous studies,13,15 there were no reports of mandibular nerve injury/paresis, alopecia, or dysphagia in the current study. One patient experienced an arterial injury at or close to the injection site, and another patient experienced both decreased heart rate and lethargy; these three AEs were assessed as treatment related. This prospective, observational registry study provides valuable insights on the use of ATX-101 to treat SMF in a real-world setting. In general, patient registries not only collect postmarketing surveillance data for approved drugs, but they also characterize the experiences of patients and physicians.19 Data from patient registries can be used to improve clinical care guidelines and prompt deeper engagement among the healthcare community. CONTOUR North America has helped guide clinical practice patterns, such as understanding the importance of patient selection and adequate treatment volumes and intervals, the use of combination therapies (eg, cryolipolysis or liposuction for severe SMF, injectables for lower face rejuvenation), and the need to educate patients on treatment course and expected AEs.12,13 A key limitation of the observational study design is the potential loss of data attributed to patients being lost to follow-up or having irregular follow-up visits. Additionally, compared with CONTOUR North America and the phase 3 REFINE trials,13,15,16 CONTOUR Australia enrolled considerably fewer patients, creating challenges in drawing parallels between these studies. Due to the limited sample size and patient demographic (the majority were women and White), the data may not represent the full population of Australian clinics that treat submental fullness due to SMF and also may not reflect the different perspectives of other populations toward SMF. The current study design did not quantify changes in SMF volume; future studies can use photographic imaging analysis to quantify these and other changes, such as changes in the submental angle. CONCLUSIONS The primary effectiveness end points used in the current study suggest that in an Australian population, ATX-101 treatment reduced SMF severity and increased patient satisfaction in the appearance of the chin. Overall, the study reports clinically meaningful and significant outcomes and further supports ATX-101 as a well-tolerated and effective treatment for SMF reduction. DISCLOSURES Simona Battucci is an employee of AbbVie. Sarah G. Boxley is paid consultant and advisory board member for Allergan Aesthetics, an AbbVie Company. Neville Lee See is investigator for Allergan Aesthetics, an AbbVie Company. Susan Simonyi is an employee of AbbVie. Suzanne St. Rose is employee of AbbVie at the time the study was conducted. Frank Lin has no financial interest to declare in relation to the content of this article. This study was supported by Allergan (before its acquisition by AbbVie). Writing and editorial assistance was provided to the authors by Adrienne Drinkwater, PhD, and Maria Lim, PhD, of Peloton Advantage, LLC, an OPEN Health company, and was funded by AbbVie. Neither honoraria nor other form of payment were made for authorship. ACKNOWLEDGMENT Study approval was obtained from each center’s independent ethics committee, and all patients provided written informed consent. Supplementary Material Published online 18 July 2023. Disclosure statements are at the end of this article, following the correspondence information. Related Digital Media are available in the full-text version of the article on www.PRSGlobalOpen.com. ==== Refs REFERENCES 1. Baumann L Shridharani SM Humphrey S . Personal (self) perceptions of submental fat among adults in the United States. Dermatol Surg. 2019;45 :124–130.30234657 2. Ascher B Fellmann J Monheit G . ATX-101 (deoxycholic acid injection) for reduction of submental fat. Expert Rev Clin Pharmacol. 2016;9 :1131–1143.27457304 3. Goodman GJ Subramanian M Sutch S . Beauty from the neck up: introduction to the special issue. Dermatol Surg. 2016;42 :S260–S262.27787265 4. Swift A Liew S Weinkle S . The facial aging process from the “inside out.” Aesthet Surg J. 2021;41 :1107–1119.33325497 5. Shamban AT . Noninvasive submental fat compartment treatment. Plast Reconstr Surg Glob Open. 2016;4 :e1155.28018773 6. Walker PS Lee DR Toth BA . Histological analysis of the effect of ATX-101 (deoxycholic acid injection) on subcutaneous fat: results from a phase 1 open-label study. Dermatol Surg. 2020;46 :70–77.30883481 7. Humphrey S Sykes J Kantor J . ATX-101 for reduction of submental fat: a phase III randomized controlled trial. J Am Acad Dermatol. 2016;75 :788–797.e7.27430612 8. Jones DH Carruthers J Joseph JH . REFINE-1, a multicenter, randomized, double-blind, placebo-controlled, phase 3 trial with ATX-101, an injectable drug for submental fat reduction. Dermatol Surg. 2016;42 :38–49.26673433 9. Rzany B Griffiths T Walker P . Reduction of unwanted submental fat with ATX-101 (deoxycholic acid), an adipocytolytic injectable treatment: results from a phase III, randomized, placebo-controlled study. Br J Dermatol. 2014;170 :445–453.24147933 10. Ascher B Hoffmann K Walker P . Efficacy, patient-reported outcomes and safety profile of ATX-101 (deoxycholic acid), an injectable drug for the reduction of unwanted submental fat: results from a phase III, randomized, placebo-controlled study. J Eur Acad Dermatol Venereol. 2014;28 :1707–1715.24605812 11. Dover JS Kenkel JM Carruthers A . Management of patient experience with ATX-101 (deoxycholic acid injection) for reduction of submental fat. Dermatol Surg. 2016;42 :S288–S299.27787269 12. Behr K Kavali CM Munavalli G . ATX-101 (deoxycholic acid injection) leads to clinically meaningful improvement in submental fat: final data from CONTOUR. Dermatol Surg. 2020;46 :639–645.31517654 13. Palm MD Schlessinger J Callender VD . Final data from the Condition of Submental Fullness and Treatment Outcomes Registry (CONTOUR). J Drugs Dermatol. 2019;18 :40–48.30681793 14. Australian Public Assessment Report for Deoxycholic Acid. Woden, ACT, Australia: Australian Government Department of Health Therapeutic Good Administration; 2017. 15. Dayan SH Schlessinger J Beer K . Efficacy and safety of ATX-101 by treatment session: pooled analysis of data from the phase 3 REFINE trials. Aesthet Surg J. 2018;38 :998–1010.29401213 16. McDiarmid J Ruiz JB Lee D . Results from a pooled analysis of two European, randomized, placebo-controlled, phase 3 studies of ATX-101 for the pharmacologic reduction of excess submental fat. Aesthetic Plast Surg. 2014;38 :849–860.24984785 17. Lupo M Dover JS Baradaran S . Development and validation of the patient-reported submental fat rating scale and the patient-reported submental fat impact scale. Dermatol Surg. 2021;47 :522–525.33306491 18. Carruthers J Carruthers A . Botulinum toxin type a treatment of multiple upper facial sites: patient-reported outcomes. Dermatol Surg. 2007;33 :S10–S17.17241408 19. de Souza MP Rangel Miller V . Significance of patient registries for dermatological disorders. J Invest Dermatol. 2012;132 :1749–1752.22695283

PMC010xxxxxx/PMC10353712.txt

==== Front Int J Womens Dermatol Int J Womens Dermatol JW9 International Journal of Women's Dermatology 2352-6475 Lippincott Williams & Wilkins Hagerstown, MD 00007 10.1097/JW9.0000000000000097 3 Research Letters Vulvar allergens in topical preparations recommended on social media: a cross-sectional analysis of Facebook groups for lichen sclerosus Luu Yen BA a* Admani Shehla MD b a Department of Dermatology, University of Missouri – Kansas City School of Medicine, Kansas City, Missouri b Department of Dermatology, Stanford University School of Medicine, Stanford, California * Corresponding author. E-mail address: ytlx9d@umsystem.edu (Y. Lunn). 18 7 2023 10 2023 9 3 e097e097 13 2 2023 02 6 2023 Copyright © 2023 The Authors. Published by Wolters Kluwer Health, Inc. on behalf of Women’s Dermatologic Society. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open-access article distributed under the terms of the Creative Commons Attribution-Non Commercial-No Derivatives License 4.0 (CCBY-NC-ND), where it is permissible to download and share the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal. allergic contact dermatitis Facebook lichen sclerosus social media vulva OPEN-ACCESSTRUE ==== Body pmcWhat is known about this subject in regard to women and their families? Vulvar allergic contact dermatitis (vACD) frequently complicates underlying vulvar dermatoses, affecting over 50% of women with preexisting vulvar lichen sclerosus (LS). Fragrances, preservatives, and topical medicaments are the most implicated sources of allergens in vACD. Women with LS seeking support via social media encounter a variety of topical preparations that may contain allergens. What is new from this article as messages for women and their families? The overwhelming majority of topical preparations encountered by women with LS on social media support groups contained vulvar allergens. Our findings highlight the high prevalence of allergenic botanical extracts/spices within topical preparations used by LS patients on social media. As social media represents an increasingly important source of health information, knowledge on prevalent allergens encountered by patients with LS on social media may inform patch-testing selection to identify culprits of vACD and patient counseling by physicians on therapies. Vulvar allergic contact dermatitis (vACD) complicates preexisting vulvar lichen sclerosus (LS) in over 50% of cases, leading to misdiagnosis, treatment delays, and severe declines in quality of life.1,2 Allergenic fragrances, preservatives, and topical medicaments are major causes of vACD via local or systemic immune activation.1,3 Over time, social media (SM) has become an increasingly popular source of health information.4 Women with LS seeking support via SM encounter a variety of topical preparations that may contain allergens. We evaluated the prevalence of vulvar allergens within topical preparations recommended by women with LS on SM using a cross-sectional analysis of Facebook groups. Facebook was systematically searched using the keywords “lichen sclerosus/sclerosis,” “vulva,” “vagina,” and “genital.” Groups containing at least 1,000 members were included. Groups targeting multiple/other vulvar dermatoses were excluded. Up to 500 consecutive posts were analyzed per group (April–September 2022). Ingredient lists of topical preparations recommended by members were analyzed to identify allergens implicated in vACD (Table 1). Table 1 Allergens implicated in vulvar and anogenital allergic contact dermatitis Allergen class Allergen name Fragrances Balsam of Peru (M. pereirae) Fragrance mix I and II Cinnamal/cinnamic aldehyde Eugenol Isoeugenol Hydroxyisohexyl 3-cyclohexene carboldehyde Hydroxy citronellal Preservatives Quaternium-15 Paraben mix MCI/MI 2-Bromo-2-nitropropane-1,3-diol Ethylenediamine dihydrochloride Stearyl alcohol Imidazolidinyl urea 2-Phenoxyethanol Chloroacetamide Diazolidinyl urea MDBGN Sodium metabisulfite Thimerosal Kathon Formaldehyde Bronopol Antibiotics Framycetin sulfate Streptomycin sulfate Neomycin Bacitracin Polymyxin Anesthetics Caine mix Antiseptics Chlorhexidine Povidone-iodine Cetrimide Mercuric chloride Gentian violet Benzethonium chloride Phenylmercuric salts Potassium dichromate Botanical extracts/spices Arnica montana Compositae mix Jasmine Oak moss Jojoba oil Eucalyptus oil Peppermint oil Emollients Propylene glycol Lanolin alcohol Glycerin Miscellaneous Nonoxynol Acrylates Thymol Toluene-sulfonamide formaldehyde resin Mercaptobenzothiazole Carba mix Thiuram mix p-Phenylenediamine Thymol Tectol Latex Hexylresorcinol Oxyquinoline sulfate Quinine hydrochloride Phenylmercuric acetate, butyrate Colophony (rosin) MBDGN, methyldibromoglutaronitrile; MCI/MI, methylchloroisothiazolinone/methylisothiazolinone. A total of eight Facebook groups (median number of members: 4,759; range: 1,330–8,680) and 4,000 posts were analyzed. In total, 78.7% (118/150) topical preparations contained at least one known anogenital allergen (Table 2). Emollients, namely glycerin, lanolin, and propylene glycol, were the most prevalent allergens, present in 62.7% of topical preparations. Preservatives (52%), fragrances (20.7%), botanical extracts/spices (20.7%), and medicaments (14.6%) followed. Allergenic preservatives included stearyl alcohol, 2-phenoxyethanol, paraben mix, and methylchloroisothiazolinone/methylisothiazolinone. Fragrance mix I/II was the most common fragrance (71%). Jojoba was the most prevalent botanical extract/spice (58.1%), followed by Arnica montana, peppermint, eucalyptus, jasmine, and oak moss. Table 2 Frequency of vulvar allergens in topical preparations recommended to lichen sclerosus patients on Facebook groups Allergen Frequency (No. %) Fragrances 31 (20.7)  Balsam of Peru 1 (0.7)  Fragrance mix I and II 22 (14.7)  Cinnamal 2 (1.3)  Eugenol 1 (0.7)  Hydroxy citronellal 5 (3.3) Preservatives 78 (52)  Quaternium-15 1 (0.7)  Paraben mix 15 (10)  MCI/MI 12 (8)  Ethylenediamine dihydrochloride 1 (0.7)  Stearyl alcohol 20 (13.3)  Imidazolidinyl urea 8 (5.3)  2-Phenoxyethanol 18 (12)  Diazolidinyl urea 3 (2) Antibiotics 11 (7.3)  Neomycin 2 (1.3)  Bacitracin 6 (4)  Polymyxin 3 (2) Anesthetics 9 (6)  Caine mix 9 (6) Antiseptics 2 (1.3)  Chlorhexidine 1 (0.7)  Benzethonium chloride 1 (0.7) Botanical extracts/spices 31 (20.7)  Arnica montana 5 (3.3)  Jasmine 1 (0.7)  Oak moss 1 (0.7)  Jojoba oil 18 (12)  Eucalyptus oil 2 (1.3)  Peppermint oil 4 (2.6) Emollients 94 (62.7)  Propylene glycol 21 (14)  Lanolin alcohol 20 (13.3)  Glycerin 53 (35.3) Miscellaneous 12 (8)  Nonoxynol 1 (0.7)  Acrylates 9 (6)  Thymol 1 (0.7)  Toluene-sulfonamide formaldehyde resin 1 (0.7) MCI/MI, methylchloroisothiazolinone/methylisothiazolinone. The overwhelming majority of topical preparations encountered by women with LS on SM contained known vulvar allergens, highlighting the importance of patient counseling by physicians before the application of topical therapies. We demonstrate the high prevalence of allergenic emollients and botanical extracts/spices within natural topical preparations. Older adults and those unfamiliar with SM platforms may have been excluded. The keywords for inclusion criteria may not have captured all LS Facebook groups. The narrow timeframe studied may have limited the types of topical preparations identified, which may not reflect current trends. To prevent the exclusion of allergens known by multiple names, we analyzed ingredient lists using all known names of allergens. Our findings highlight the high prevalence of allergenic botanical extracts/spices within topical preparations encountered by LS patients on SM. Over 60% of women with chronic vulvar disease use botanical products, and this percentage is increasing, which may be due to the belief that natural preparations are safer than their synthetic counterparts.1,5 Tea tree oil and compositae mix are allergenic sensitizers. Additionally, fragrance mix is an indicator allergen for several spices.3 Women with LS on SM forums are often keen to trying alternative treatments:5 thus, botanical extracts/spices may represent important culprits of vACD due to increased topical exposures in the context of underlying vulvar barrier impairment.1,3 As SM represents an increasingly important source of health information for patients with LS, our findings may inform patient counseling by physicians on therapies and patch testing selection to identify culprits of vACD. Conflicts of interest None. Funding None. Study approval The authors confirm that any aspect of the work covered in this manuscript that involved human patients has been conducted with the ethical approval of all relevant bodies. Author contributions SA and YL contributed to research design, performance of research, and writing of the paper. YL contributed to data analysis. Published online 18 July 2023 ==== Refs REFERENCES 1. Woodruff CM Trivedi MK Botto N Kornik R . Allergic contact dermatitis of the vulva. Dermatitis. 2018;29 :233–43. doi: 10.1097/DER.0000000000000339.30179968 2. Trivedi MK Woodruff CM Kornik R Botto N . Patch testing in vulvar allergic contact Dermatitis. Dermatitis 2018;29 :95–6. doi: 10.1097/DER.0000000000000345.29494390 3. Vandeweege S Debaene B Lapeere H Verstraelen H . A systematic review of allergic and irritative contact dermatitis of the vulva: the most important allergens/irritants and the role of patch testing. Contact Dermatitis 2023;88 (4 ):249–62. doi: 10.1111/cod.14258.36458568 4. Szeto MD Mamo A Afrin A Militello M Barber C . Social media in dermatology and an overview of popular social media platforms. Curr Dermatol Rep 2021;10 :97–104. doi: 10.1007/s13671-021-00343-4.34692234 5. Bentham GL Manley K Halawa S Biddle L . Conversations between women with vulval lichen sclerosus: a thematic analysis of online forums. BMC Womens Health 2021;21 :71. doi: 10.1186/s12905-021-01223-6.33596903

PMC010xxxxxx/PMC10353713.txt

==== Front Hemasphere Hemasphere HS9 HemaSphere 2572-9241 Lippincott Williams & Wilkins Philadelphia, PA 00005 10.1097/HS9.0000000000000929 3 002 Article Context-dependent T-cell Receptor Gene Repertoire Profiles in Proliferations of T Large Granular Lymphocytes Assmann Jorn L.J.C. 1 Vlachonikola Elisavet 2 Kolijn Pieter M. 1 Agathangelidis Andreas 2 Pechlivanis Nikolaos 2 Papalexandri Apostolia 3 Stamatopoulos Kostas 2 Chatzidimitriou Anastasia 2 Langerak Anton W. 1 1 Laboratory for Medical Immunology, Department of Immunology, Erasmus MC, Rotterdam, Netherlands 2 Institute of Applied Biosciences, Centre for Research and Technology Hellas, Greece 3 Hematology Department and HCT Unit, Papanicolaou Hospital, Thessaloniki, Greece Correspondence: Anton W. Langerak (a.langerak@erasmusmc.nl). 8 2023 17 7 2023 7 8 e92926 1 2023 12 6 2023 Copyright © 2023 the Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the European Hematology Association. 2023 https://creativecommons.org/licenses/by-nc-sa/4.0/ This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as the author is credited and the new creations are licensed under the identical terms. T cell large granular lymphocyte (T-LGL) lymphoproliferations constitute a disease spectrum ranging from poly/oligo to monoclonal. Boundaries within this spectrum of proliferations are not well established. T-LGL lymphoproliferations co-occur with a wide variety of other diseases ranging from autoimmune disorders, solid tumors, hematological malignancies, post solid organ, and hematopoietic stem cell transplantation, and can therefore arise as a consequence of a wide variety of antigenic triggers. Persistence of a dominant malignant T-LGL clone is established through continuous STAT3 activation. Using next-generation sequencing, we profiled a cohort of 27 well-established patients with T-LGL lymphoproliferations, aiming to identify the subclonal architecture of the T-cell receptor beta (TRB) chain gene repertoire. Moreover, we searched for associations between TRB gene repertoire patterns and clinical manifestations, with the ultimate objective of discriminating between T-LGL lymphoproliferations developing in different clinical contexts and/or displaying distinct clinical presentation. Altogether, our data demonstrates that the TRB gene repertoire of patients with T-LGL lymphoproliferations is context-dependent, displaying distinct clonal architectures in different settings. Our results also highlight that there are monoclonal T-LGL cells with or without STAT3 mutations that cause symptoms such as neutropenia on one end of a spectrum and reactive oligoclonal T-LGL lymphoproliferations on the other. Longitudinal analysis revealed temporal clonal dynamics and showed that T-LGL cells might arise as an epiphenomenon when co-occurring with other malignancies, possibly reactive toward tumor antigens. OPEN-ACCESSTRUE SDCT ==== Body pmcINTRODUCTION T cell large granular lymphocyte (T-LGL) leukemia is a rare hematological malignancy characterized by chronic (oligo)clonal proliferation of cytotoxic T cells.1 Although often asymptomatic, patients may eventually develop symptoms, for example, neutropenia, thrombocytopenia, or anemia.2 In recent years, T-LGL leukemia has emerged as a paradigmatic example of antigen-driven leukemogenesis, with transitions from poly or oligoclonally reactive responses to monoclonal T-LGL leukemia occurring in a gradual manner.3 Notably, boundaries between polyclonal, oligoclonal, and monoclonal T-LGL lymphoproliferations are not well-established. In fact, T-LGL clones expand due to persistent antigenic stimulation and characteristic T-LGL cells are seen in various disease contexts, ranging from autoimmune disorders, depletion of B cells through monoclonal anti-CD20 antibodies, solid tumors, and hematologic malignancies to solid organ and allogeneic hematopoietic stem cell transplantation (allo-HSCT).4–7 Therefore, T-LGL cells may arguably be triggered by various antigens. Eventually, profound immune dysregulation leads to persisting T-LGL clones, with the constitutive activation of the STAT3 transcription activator as a hallmark.8 Koskala et al first identified activating mutations in exon 21 of the STAT3 gene, coding for the SH2 domain, whereas other mutations were later identified, all in or around exon 21.9,10 Although only 30%–40% of all patients with T-LGL leukemia bear mutations in the STAT3 gene, all patients carry cells that show hyperactive STAT3 signaling.11 Based on the concept that antigens drive T-LGL cells to proliferate, much research has been conducted regarding the T-cell receptor beta (TRB) gene repertoire looking for evidence of antigen selection.12,13 Most of this work has been performed with traditional spectratyping methods.14 Additionally, subcloning of individual TRBV-TRBD-TRBJ gene rearrangements followed by Sanger sequencing has also been performed, leading to low-throughput and low-resolution datasets.12 However, firm conclusions regarding the precise impact of antigenic drive on T-LGL proliferations could not be drawn and, thus, it remains uncertain if T-LGL cells create an environment for immune-dysregulation disorders to thrive or, instead, arise as a reactive population.15 Here, we sought to obtain comprehensive insight into the TRB gene repertoire in a series of well-characterized patients with T-LGL lymphoproliferations employing next-generation sequencing (NGS), thus allowing not only to evaluate dominant clonotypes but also to unravel the subclonal architecture. Furthermore, by serial sampling over time and in different disease contexts, we aimed to elucidate if T-LGL lymphoproliferations emerge as reactive populations and how they evolve. We show that the TRB gene repertoire of patients with T-LGL lymphoproliferations is profoundly context-dependent, displaying distinct clonality patterns under different circumstances, suggesting that T-LGL proliferations might be triggered by a broad range of antigens. In addition, we argue that large monoclonal populations of T-LGL cells might be the causative factor of neutropenia in patients with T-LGL proliferations (with or without STAT3 mutations), and that oligoclonal T cell populations might be the consequence of associated malignancies that are often co-occurring in T-LGL patients. MATERIALS AND METHODS Study group Our study group included 27 patients with TRαβ+CD8+ T-LGL proliferations and 22 age-matched healthy controls. The biobanks of the Department of Immunology of Erasmus MC, University Medical Center, Rotterdam, The Netherlands and the Hematology Department and HCT Unit of the G. Papanicolaou Hospital, Thessaloniki, Greece were retrospectively inspected to include persistent TRαβ+CD8+ T-LGL proliferations in peripheral blood (PB). Cases were included based on a combination of clinical, immunophenotypical, and molecular data (Table 1). Five TCRαβ+CD8+ T-LGL leukemia cases were specifically selected for longitudinal analyses. Control samples were obtained from Sanquin Blood Bank (Amsterdam, The Netherlands) and the Institute of Applied Biosciences, CERTH, Thessaloniki, Greece, upon informed consent and anonymized before use. The study was approved by the institutional review boards of the participating institutions (MEC-2015-0617; MEC-2011–0409; CERTH: EHT.COM-45, 21/03/2019). Of note, samples TLGL5 and TLGL15 were removed from all analyses, except the HLA analysis, because of minor laboratory contamination. Table 1 Patient Characteristics of the Present Cohort of Cases With T-LGL Lymphoproliferations Patient Age Gender % LGL Within CD3 Circulating LGLs (× 109/L) Immunophenotype Associated Disease Symptoms Degree of `Neutropenia Therapy Sampling Pre/Post Tx Impact of Tx on LGL STAT3 Mutation LGL1 70 M 75 3.5 CD8+CD4-CD5-/CD57+ None Neutropenia, B-symptoms Mild None Pre Unknown No LGL2 73 M 48 4.4 CD8+CD4-CD57+ Chronic NK-cell leukemia Neutropenia, recurrent infections Mild MTX Pre Unknown D661Y LGL3 67 M 55 2.0 CD8+CD4-CD2-/CD57+ RA Neutropenia, B symptoms, thrombocytopenia Mild Plaquenil Pre Unknown Y640F LGL4 72 M 72 - CD8+CD4- None Thrombocytopenia No None Pre Unknown Y640F LGL5 35 F 52 3.7 CD8+CD4-CD57+ None Neutropenia Mild None Pre Unknown No LGL6 74 F 41 0.8 CD8+CD4-CD7-/CD57+ RA, splenomegaly Neutropenia Severe MTX Pre Low-level presence Y640F LGL7 58 M 6 0.6 CD8+CD4-CD57+ ITP, MGUS Asymptomatic (with respect to T-cell clone) No None (with respect to T-cell clone) Pre Unknown No LGL8 63 M 50 8.2 CD8+CD4-CD57+ M. Waldenström (2005); T-LGL post allo-Tx Asymptomatic (with respect to T-cell clone) No None (with respect to T-cell clone) Pre Unknown No LGL9 58 M 29 6.1 CD8+CD4-CD57+ HCL Asymptomatic (with respect to T-cell clone) No None (with respect to T-cell clone) Pre Unknown No LGL10 64 F 24 0.6 CD8+CD4-CD57+ Unknown Neutropenia Mild Unknown Pre Unknown Y640F LGL11 38 F 8 0.3 CD8+CD4-CD5-/CD57+ None Neutropenia Severe MTX Pre Unknown No LGL12 78 M 20 0.8 CD8+CD4-CD7-/CD57+ AML Asymptomatic (with respect to T-cell clone) No None (with respect to T-cell clone) Pre Low-level presence No LGL13 76 F 52 12.0 CD8+CD4-CD57+ None B symptoms No None Pre Unknown Y640F LGL14 41 F 40 2.5 CD8+CD4-CD7-/CD57+ PRCA Anemia, B symptoms No CSA Pre Unknown No LGL15 37 M 40 1.0 CD8+CD4-CD5-CD57+ M. Crohn Agranulocytosis Severe Unknown Pre Unknown Y640F LGL16 48 F 7 0.3 CD8+CD4-CD57+ DLBCL (2003, CNS loc 2005) Neutropenia, anemia Mild None (with respect to T-cell clone) Pre Unknown No LGL17 27 M 44 4.2 CD3+/8+/CD57+ None Pancytopenia Mild Yes Post Yes No LGL18 59 M 24 2.3 CD3+/8+/CD57+ None Neutropenia, thrombocytopenia Mild Yes Pre No Tx No LGL19 15 M 46 1.1 CD3+/8+/CD57+ Allo-Tx AIHA No Yes Pre Yes No LGL20 51 F 41 2.2 CD3+/8+/CD57+ None Neutropenia Mild Yes Pre Yes No LGL21 44 M 69 5.0 CD3+/8+/CD57+ None None No None Pre No Tx D661Y LGL22 70 6.3 CD3+/8+/CD57+ None None No None Pre No Tx Y640F LGL23 85 M 0.6 CD3+/8+/CD57+ None None No None Pre No Tx No LGL24 65 M 17 0.4 CD3+/8+/CD57+ None Neutropenia Severe Yes Pre Yes No LGL25 35 M 29 0.8 CD3+/8+/CD57+ None None No None Pre No Tx No LGL26 32 M 23 1.8 CD3+/8+/CD57+ None None No None Pre No Tx No LGL27 73 M 40 1.6 CD3+/8+/CD57+ Smoldering Myeloma None Νo None Pre No Tx No aAll T-LGL patients were diagnosed based on the cell morphology/immunophenotype and/or molecular genetics. AIHA = Autoimmune hemolytic anemia; AML = acute myeloid leukemia; CSA = cyclosporine A; DLBCL = diffuse large B-cell lymphoma; HCL = hairy cell leukemia; Het = heterogeneous; ITP = immune thrombocytopenia; MGUS = monoclonal gammopathy of unknown significance; MTX = methotrexate; PRCA = pure red cell aplasia; RA = rheumatoid arthritis; T-LGL = T cell large granular lymphocyte; Tx = transplantation. PCR amplification of TRBV-TRBD-TRBJ gene rearrangements and library preparation Genomic DNA was isolated from PB mononuclear cells and 500 ng was used for multiplex polymerase chain reaction (PCR) amplification of TRBV-TRBD-TRBJ gene rearrangements following the BIOMED-2 protocol (35 cycles).16 Library preparation was performed as described with an index PCR of 12 cycles.17 Bioinformatics Initial NGS data assessment was performed by the Illumina signal-processing software, including base-calling, quality control, adapter trimming, and demultiplexing, resulting in the exclusion of low-quality and erroneous sequences. A purpose-built, validated bioinformatics algorithm was applied to the raw NGS reads in order to merge and further filter the paired-end reads based on the strict length and quality criteria, as previously described.18 Merged reads were further annotated using the IMGT/HighV-QUEST tool and the annotated sequences were further characterized by the T-cell Receptor/Immunoglobulin Profiler software. For additional details, see Supplemental Materials and Methods. Data are available under ENA Project ID: PRJEB47814. Definitions and interpretation of results In this study, clonotypes were defined as TRBV-TRBD-TRBJ gene rearrangements with identical TRBV gene usage and TRB CDR3 amino acid (aa) sequence. Clonotypes were characterized as expanded when they comprised ≥2 nucleotide sequences. The most expanded clonotype within a sample is referred to as immunodominant. Clonotypes representing only a single read were characterized as singletons. For TRB gene repertoire analysis, clonotypes rather than single rearrangements were taken into consideration in order to avoid potential biases. Individual TRBV gene frequencies within a sample were calculated based on the number of clonotypes using the particular TRBV genes over the total number of clonotypes. For the connectivity analysis as well as the comparisons and clustering of clonotypes based on their sequence identity, only expanded clonotypes were evaluated whereas singletons were excluded. This decision was intentionally taken in order to avoid potentially false estimates of clonotype sharing, as it is very difficult to discriminate whether clonotypes called by a single read represent a true finding or instead arise from sequencing errors or mapping biases (index hopping effect). The latter is an inherent limitation of NGS technology that could erroneously assign sequencing reads to the wrong index during demultiplexing, hence leading to an artificial increase in similarity between samples. Subclonal connectivity We interrogated the subclonal architecture of the TRB gene repertoire in a given sample by characterizing clonotypes that are clonally related with the immunodominant one that is those with identical CDR3 length as the dominant one yet differing in a single aa position. Pairwise distance calculation was performed and only CDR3s with 1 difference within their aa sequence were connected as clonally related. Clonotype comparisons and clustering of TRBV-TRBD-TRBJ rearrangements based on TRB CDR3 amino acid sequence restriction Clonotype comparisons were also performed between the different T-LGL cases and cross-comparisons against a dataset composed of TR clonotype sequences either retrieved from the VDJdb (a curated database of TCR sequences with known antigen specificities19 or available to our group from previous studies in benign ethnic neutropenia (BEN),20 chronic idiopathic neutropenia (CIN),19 chronic lymphocytic leukemia (CLL),18 monoclonal B lymphocytosis ([MBL], a potentially precursor state to CLL, identified in otherwise healthy individuals),18 and healthy controls. Additionally, comparisons were undertaken with a published dataset of TRB gene rearrangement sequences from patients with T-LGL leukemia reported by Kerr et al.21 Furthermore, common immunogenetic signatures between the samples were identified by clustering of TRBV-TRBD-TRBJ rearrangements based on the shared TRB CDR3 aa sequence motifs using a modified version of the Teiresias algorithm adapted to the immune repertoire analyses.22,23 For additional information see Supplemental Materials and Methods. HLA-A, HLA-B, HLA-C low-resolution typing Typing of HLA-A, HLA-B, and HLA-C in low resolution was performed with the Olerup SSP HLA typing kits according to the manufacturer’s protocol (CareDx, Vienna, Austria). STAT3 mutation analysis STAT3 exon 21 was amplified from genomic DNA. Thereafter, PCR products were analyzed using bidirectional Sanger sequencing to assess STAT3 hotspot mutations. All cases that were found negative by Sanger sequencing were assessed for the presence of STAT3 mutations in exon 21 by NGS with a total coverage >16,000× for each sample. Statistical analysis Descriptive statistics for discrete parameters included counts and frequency distributions. For quantitative variables, statistical measures included medians and min-max values. Mann-Whitney test was used to test differences between groups. For all comparisons, P values <0.05 were considered statistically significant. Statistical tests and levels of significance are indicated in the figure legends. Shannon diversity index was calculated using the vegan package in R, based on the equation H = -Σ [(pi) × ln(pi)], with pi being the proportion of each species. Data visualization tools Data visualization was performed in the R environment, using the open source data visualization framework RawGraphs. Aligned TRB CDR3 amino acid (aa) sequences were visualized using WebLogo. CDR3 aa positions are shown according to the IMGT numbering for the V domain.24 RESULTS The TRB gene repertoire of patients with T-LGL proliferations is largely (oligo)clonally restricted We sequenced 65 samples from 27 patients with T-LGL proliferations and 22 controls. A general overview on the acquisition of productive TRBV-TRBD-TRBJ gene rearrangement sequences and clonotypes is presented in Suppl. Table S1. Both patients with T-LGL proliferations and healthy controls displayed skewing of the TRBV gene repertoire (Suppl. Table S2A). However, cross-comparisons of the respective repertoires revealed 13 differentially used TRBV genes (Suppl. Table S2A; P < 0.05). Significant differences were also observed between T-LGL patients and healthy controls regarding TRBJ genes (Suppl. Table S2B). Regarding TRBD genes, the TRBD1 gene predominated over the TRBD2 gene in both sample groups (Suppl. Table S2C). Deep sequencing of the TRB gene repertoire allowed us to assess the clonal composition and the clonality of all investigated T-LGL cases. We observed that the immunodominant clonotype had a median frequency of 19.3% (0.4%–52.0%) in T-LGL proliferations versus 6.9% (1.3%–21.5%) in healthy controls (Figure 1A; P < 0.001). The median cumulative frequency of the 10 most expanded clonotypes per sample in T-LGL patients accounted for 40.2% (3.5%–64.3%) of the total TRB gene repertoire, thus highlighting a monoclonal to oligoclonal repertoire. In contrast, healthy controls were characterized by an oligoclonal to polyclonal profile with a much lower median cumulative frequency of the 10 most expanded clonotypes/sample of 23.7% (4.5%–37.6%) (Figure 1B; P < 0.01). Both the immunodominant and the top 10 clonotypes differed greatly in size between T-LGL patients and healthy controls. These repertoire differences were further quantified using the Shannon index, which is used to mathematically calculate the diversity within the TRB gene repertoire.25,26 The median Shannon diversity score of patients with T-LGL proliferations was 6.3, whereas in healthy controls it was 7.8. (Figure 1C; P < 0.01). Figure 1. TRB gene repertoire diversity in patients with T-LGL lymphoproliferations. (A) Clonotype frequencies of patients with T-LGL lymphoproliferations and % of immunodominant clones. (B) Clonotype frequencies of healthy controls and % of top 10 clonotypes (B). Shannon diversity scores of patients with T-LGL lymphoproliferations and healthy controls. (C) Graphs indicate the mean with SD. Statistical significance was tested using the Mann Whitney U test. Level of significance indicated in the plots: ***P < 0.001; ****P < 0.0001. T-LGL = T cell large granular lymphocyte; TRB = T-cell receptor beta. The TRB (sub)clonal architecture of T-LGL proliferations is context-dependent We then examined whether the TRB gene repertoire of patients with T-LGL proliferations might be context-dependent by performing comparative analyses on samples derived from patients with various comorbidities and/or specific clinic-biological features. Shannon diversity scores of T-LGL patients with neutropenia (P < 0.05), STAT3 mutations (P < 0.001), and associated malignancies (P < 0.001) were all lower compared with healthy controls, but did not differ from T-LGL counterparts without these characteristics (Figure 2A). Further in-depth analysis of TRBV gene usage revealed no preference for certain TRBV genes in the groups with neutropenia, STAT3 mutations, or patients with T-LGL proliferations and associated malignancies either, as compared with the other T-LGL cases. Figure 2. Context-dependent TRB gene repertoire of patients with T-LGL lymphoproliferations. (A) Shannon diversity index of subgroups of patients with T-LGL lymphoproliferations (red dots indicate STAT3-mutated cases). (B) Clonotype frequencies of neutropenic patients with T-LGL lymphoproliferations, patients with T-LGL proliferations harboring STAT3 mutations and patients with T-LGL lymphoproliferations with associated malignancies. Comparison of % immunodominant clonotypes and top 2–5 clonotypes of different T-LGL subgroups. (C) Graphs indicate the mean with SD. Statistical significance was tested using the Mann Whitney U test. Level of significance indicated in the plots: *P < 0.05; **P < 0.001. **Patients with STAT3 mutations are depicted in red (STAT3 mutated patients without clinical associations are indicated with an X), neutropenic patients are depicted in grey, patients with associated malignancies are depicted in gold, and patients with autoimmune phenomena (especially RA) are depicted in blue. T-LGL = T cell large granular lymphocyte; TRB = T-cell receptor beta. Next, we assessed the clonality profiles between the different patient groups. Because there was considerable overlap between patients with neutropenia or STAT3 mutations, these patient categories were grouped in this analysis. Notably, while virtually all STAT3 mutant/neutropenic T-LGL cases exhibited pronounced monoclonal expansions, the profile of patients with associated malignancies was frequently oligoclonal (Figure 2B). The median frequency of the immunodominant clonotype in neutropenic and/STAT3 mut T-LGL cases was 25.6% of the TRB gene repertoire (0.6%–52.0%) versus 15.5% (8.2%–21.6%) in T-LGL cases with associated malignancies (Figure 2C; P =0.26). Of note, in this latter group, the top 2–5 clonotypes (ie, top 5 clonotypes minus the immunodominant one) represented a median of 28.6% (6.1%–34.9%) of the total repertoire, whereas in the former group (ie, neutropenic/STAT3mut), a median value of only 6.1% (1.6%–26.2%) was observed (Figure 2C; P < 0.001), strengthening the argument that the subclonal architecture of T-LGL proliferations is context-dependent. In the Dutch part of our cohort for which HLA data were available, no clear restrictions were observed toward a certain HLA allele or haplotype in neutropenic patients or patients with coexisting malignancies (Table 2). That said, we did observe that 71% of T-LGL patients bearing STAT3 mutations (5/7 patients) used the HLA-B15 allele (odds ratio = 21.42; P = 0.04); however, the small size of the cohort hinders more definitive conclusions at this point. Table 2 HLA-A, HLA-B, and HLA-C Profiles in Patients With T-LGL Lymphoproliferations Patient HLA A HLA B HLA C LGL1 A01 A26 B38 B44 C05 C12 LGL2 A02 B15 B35 C03 C04 LGL3 A01 B15 B39 C03 C07 LGL4 A03 A32 B41 B55 C03 C17 LGL5 A01 A11 B27 B35 C02 C04 LGL6 A01 A02 B15 B37 C01 C06 LGL7 A01 A02 B08 B44 C07 LGL8 A01 B08 C07 LGL9 A02 A68 B38 B51 C12 C15 LGL10 A02 A24 B15 B27 C02 C03 LGL11 A01 A24 B18 B51 C07 C16 LGL12 A02 A29 B44 B51 C05 C14 LGL13 A02 A03 B07 B15 C03 C07 LGL14 A02 B08 C07 LGL15 A01 A02 B08 B40 C03 C07 LGL16 A30 A68 B39 B44 C07 C12 T-LGL = T cell large granular lymphocyte. Temporal dynamics of the (sub)clonal TRB gene repertoire architecture in T-LGL proliferations Using NGS, we assessed the clonal dynamics of the TRB gene repertoire over time (median of 4 time points; range, 2–6) in representative cases of the various groups of T-LGL proliferations in our cohort. In patients with STAT3-mutated T-LGL proliferations, clonal drift of smaller clonotypes was observed (Table 1; Figure 3A). Such clonal drift was also seen in neutropenic T-LGL patients, as exemplified by the case of patient LGL11 (Table 1; Figure 3B). Figure 3. Patients with T-LGL lymphoproliferations display a context-dependent (sub)clonal TRB gene repertoire architecture that may shift over time. A. Longitudinal analysis of the subclonal TRB gene architecture of a patient showing a STAT3-mutated T-LGL leukemia. (B) Clonal dynamics of the TRB gene repertoire of a neutropenic patient with T-LGL leukemia under therapy. (C) Complex interplay between the TRB gene repertoire and M-protein levels in a case with T-LGL lymphoproliferation and an associated plasma cell malignancy. (D) Dynamics of the TRB gene repertoire in a patient developing a T-LGL lymphoproliferation post allo-HSCT. Allo-HSCT = allogeneic hematopoietic stem cell transplantation; T-LGL = T cell large granular lymphocyte; TRB = T-cell receptor beta.. Regarding T-LGL cases with coexisting clonal conditions, we evaluated multiple time points in patient LGL7 (Table 1) bearing a T-LGL proliferation concomitant with a monoclonal gammopathy of undetermined significance. While the overall shift of the subclonal composition of the TRB gene repertoire over the course of 8 years was relatively modest, combined analysis of M-protein levels and TRB gene repertoire sequencing provided evidence of correlations between TRB repertoire dynamics and M-protein fluctuations (Figure 3C). Another intriguing scenario concerns the development of a T-LGL proliferation in the context of allo-HSCT performed in a patient with Ph+ acute lymphoblastic leukemia (LGL19; Table 1), where we observed significant clonal drift temporally connected with various incidents posttransplantation (including Epstein-Barr virus reactivation, neutropenia, and episodes of autoimmune hemolytic anemia), while also documenting expansion of clonotypes present in the repertoire of the donor (Figure 3D). Finally, we studied a father and a son presenting with cytopenias and splenomegaly in a context of T-LGL proliferation; the clinical features and Sanger-based immunoprofiling of these cases have been reported previously.27 The cumulative frequency of the 10 most expanded clonotypes in the diagnostic sample of the father was 20.2% with 2 dominant clonotypes at relative frequencies of 6.2% and 5.1%. In the son, the 10 most expanded clonotypes of his diagnostic samples accounted for almost 19.5% of the total repertoire, rendering it oligoclonal with a dominant clone at a relative frequency of 8.9%. After 5 years, the clonality pattern remained stable; however, expansion of multiple clonotypes was observed. In fact, clonal drift was evident in the son with expansion of certain minor clonotypes over time, as exemplified by the clonotype TRBV29-1 – CSASTGDRSGANVLTF, which started with a relative frequency of 1% at the time of diagnosis and eventually expanded to 6% of the repertoire, representing the second immunodominant clonotype of that particular follow-up sample. Various degrees of connectivity for the dominant clonotype of patients with T-LGL lymphoproliferations Seeking to obtain a comprehensive view of the subclonal architecture of the TRB repertoire in the cases under study, next we focused on the related clonotypes to the most immunodominant one. The relation between the immunodominant clonotype and the highly similar clonotypes was visualized as connectivity graphs where CDR3 aa sequences differing in a particular position were connected with a line. The frequency of the most immunodominant clonotype was re-estimated summing up the frequencies of the highly similar clonotypes. Based on that metric, we observed various degrees of connectivity, ranging from low-level, where the frequency of the dominant clonotype including the highly similar clonotypes was slightly changed, to high-level, where the change observed in the relative frequency of the dominant clonotype was significant (Figure 4A-D). Noteworthy was the pattern in case LGL7 over time, where the connectivity observed for the dominant clonotype was particularly high with a median change in its relative frequency of 10.1% (range, 5.7%–10.9%) (Figure 4E). Figure 4. Connectivity networks of the dominant clonotype of patients with T-LGL lymphoproliferations. (A and B) Graphs depicting the connectivity networks formed between the dominant clonotype and other clonotypes of the same length of 2 representative cases with T-LGL lymphoproliferations taking into account the highly similar clonotypes present in each sample. The sequence logo represents the total repertoire of highly similar clonotypes identified. The sequence logo of the CDR3 region of one representative case and the graph depicting low-level connectivity of the dominant clonotype. (C and D) The sequence logo of the CDR3 region of one representative case and the graph depicting high-level connectivity of the dominant clonotype. The large circle depicts the dominant clonotype of the sample and only the CDR3s with one difference are connected. (E) Overtime kinetics and clonal dynamics of highly similar clonotypes in a representative case. T-LGL = T cell large granular lymphocyte. Public clonotypes in the repertoire of T-LGL lymphoproliferations Combinatorial pattern discovery analysis was performed using the TEIRESIAS algorithm and the clonotypes assigned to the clusters formed were compared to the VDJdb,20 which allows making predictions regarding the possible antigenic specificity of the clustered clonotypes. Overall, 3675 of 9073 (40.5%) of all clonotypes with a relative frequency ≥0.1%, were assigned to 1436 distinct clusters. In many cases, the existence of multiple shared patterns within TRB CDR3 sequences led to their concurrent assignment in clusters characterized by more broadly shared sequence patterns, hence, greater individual cluster size. At the end of the process, 1250 clusters were formed including clonotypes that could not be further clustered: the number of cases in each of these final clusters ranged from 2 to 23. Two distinct types of clusters were identified, characterized by usage of the same or diverse TRBV genes, respectively. In both cluster types, the overall TRB CDR3 similarity was very high, that is, most positions in the TRB CDR3 region were extremely, if not entirely, conserved, while only few positions displayed variability (Figure 5). Figure 5. Clusters of cases with restricted TRB CDR3. Two representative clusters are depicted including a sequence logo of the CDR3 regions that clustered together, the TRB gene usage and the presumed antigen specificity when a clonotype of those clustered was also found in the VDJdb. One representative cluster including 23 clonotypes with relative frequency ranging between 0.1% and 5.6% characterized by homogeneity regarding the TRBV gene usage. The clustered clonotypes derived from 18 different cases (T-LGL, n = 6; EN, n = 1; CIN, n = 3; CLL, n = 2; MBL, n = 4; healthy, n = 2). The clustered clonotypes were identified in the VDJdb and different antigenic specificities can be speculated based on the literature (A). One representative cluster including 20 clonotypes (reactive frequency, 0.1%–3.7%) with heterogeneity regarding the TRBV genes used. The clustered clonotypes derived from 17 different cases (T-LGL, n = 7; EN, n = 2; CIN, n = 2; CLL, n = 2; MBL, n = 2; healthy, n = 2). No hits with the VDJdb were found. In brackets the number of clonotypes using the particular gene is indicated (B). CIN = chronic idiopathic neutropenia; CLL = chronic lymphocytic leukemia; EN = benign ethnic neutropenia; MBL = monoclonal B lymphocytosis; T-LGL = T cell large granular lymphocyte; TRB = T-cell receptor beta. Finally, the comparison of the clustered clonotypes against the VDJdb database highlighted a potential antigenic specificity for only a minor fraction of clonotypes from the T-LGL dataset (Figure 5). Cross-entity comparisons highlight the disease-biased nature of the TRB gene repertoire in T-LGL lymphoproliferations We performed cross-entity comparisons between the dataset of expanded clonotypes from the present cohort of patients with T-LGL lymphoproliferations (n = 191,372), a published study of patients with T-LGL leukemia (n = 486,540),21 and a dataset of 2,079,944 unique TRB clonotypes from various entities. When comparing the 2 T-LGL datasets, 1457 of 677,912 (0.21%) clonotypes were found to be uniquely shared between cases with T-LGL lymphoproliferation from our present series and that of Kerr et al (Suppl. Figure S1A). Within each individual series, the incidence of clonotypes shared by at least 2 cases was overall similar: 7.94% (38,660/486,540) in the Kerr study versus 7.15% (13,697/191,372) in the present study. Furthermore, we used TEIRESIAS algorithm for the combined datasets and we additionally identified 2 high-frequency clusters of shared TRB CDR3 aa sequence motifs: the first connected a case from the Kerr study and case LGL23 from the present cohort, while the second connected 2 cases from the Kerr study. Next, in order to assess the relatedness of these connected clonotypes with other clonotypes comprising the LGL repertoire, we used the CDR3 amino acid sequences of these 2 clusters as baits and searched for CDR3s of the same length within the complete repertoires of all samples of the present cohort that differ in a single amino acid. This led to the identification of a number of additional low-frequent clusters with various degrees of connectivity, further implying shared immunogenetic signatures within different cases with T-LGL lymphoproliferations (Suppl. Figure S2). Cross-entity comparisons between the 2 datasets of T-LGL lymphoproliferations versus the datasets from other entities revealed 40,741 of 677,912 (6%) shared clonotypes. The largest group included 16,885 of 40,741clonotypes (41.5%) shared between T-LGL lymphoproliferations and healthy controls; closely followed by a group of 11,574 of 40,741 clonotypes (28.6%) shared between T-LGL lymphoproliferations and otherwise healthy individuals with MBL (Suppl. Figure S1B). Relevant to mention, 1587 of 40,741 (3.9%) of these clonotypes have been previously characterized to be specific for viral epitopes (Epstein-Barr virus, cytomegalovirus, human immunodeficiency virus 1, hepatitis C virus, dengue virus, influenza A, Molluscum contagiosum virus, Herpes simplex virus, and human T-lymphotropic virus). DISCUSSION We profiled by NGS a cohort of 27 well-established patients with T-LGL proliferations and a publicly available dataset by Kerr et al and aimed at identifying the subclonal architecture of the TRB gene repertoire. Moreover, we searched for associations between TRB gene repertoire patterns and clinical manifestations, with the ultimate objective of discriminating between T-LGL proliferations developing in different clinical contexts and/or displaying distinct clinical presentation. Clonotype connectivity analysis and combinatorial pattern discovery analysis were used, depicting the disease-biased nature of the TRB gene repertoire in T-LGL lymphoproliferations. Patients with expanded T-LGL cells can remain asymptomatic for prolonged periods of time. Within this time frame, T-LGL cells may eventually shift toward monoclonality and this can be linked to and/or accompanied by the presence of genomic aberrations and/or the emergence of clinical manifestations.11 This scenario is supported by our present findings, where a clearly monoclonal repertoire was seen exclusively in neutropenic patients and patients carrying STAT3 mutations. Thus, neutropenic patients typically present at the far end of the spectrum of T-LGL lymphoproliferations; arguably, these larger monoclonal populations secrete higher amounts of FAS-ligand (FASL), resulting in FAS-mediated apoptosis of neutrophils.28 Nonetheless, patients may develop severe neutropenia without profound (mono)clonal expansions of T-LGL cells due to mechanisms not yet characterized. Overactivation of STAT3 has been implicated as a central hub in T-LGL lymphoproliferations, driving them toward a more malignant phenotype.29 Unsurprisingly, therefore, STAT3-mutated patients harbor large monoclonal populations, as proliferation and apoptosis resistance is driven by constitutive STAT3 activation.30 Moreover, T-LGL lymphoproliferations with STAT3 mutations display more clinical symptoms as well. Neutropenia cooccurs in a large fraction of the STAT3-mutated patients, which might be explained by the fact that STAT3-mutated T-LGL cells are highly active, thus secreting higher amounts of FASL and proinflammatory cytokines that would eventually lead to symptoms such as neutropenia. Taken together, it seems that the larger, monoclonal T-LGL lymphoproliferations establish an environment in which neutrophils go into apoptosis, thus representing the more malignant T-LGL variant. Both solid and hematological tumors can be immunogenic, although the adaptive immune system is mostly hindered from mounting an effective response and clearing the tumor due to multiple, frequently coexisting mechanisms of immune escape operating in individual patients.31 Considering the above, it is not paradoxical that a subgroup of patients with T-LGL lymphoproliferations and associated malignancies in our study displayed clear oligoclonal expansions. Similar clonality patterns have been disclosed with NGS in tumor infiltrating lymphocytes and circulating T cells of patients with both solid tumors32 and hematological malignancies.18,33 Hence, the presence of oligoclonality supports that T-LGL lymphoproliferations might emerge in response to tumor-associated antigens; it remains to be elucidated whether these T-LGL cells participate in immune surveillance but also how they might be functionally debilitated. Collectively, our data strongly suggest that, in the context of other malignancies, T-LGL lymphoproliferations are more likely a consequence, acting in (ineffective) tumor surveillance. That said, it cannot be a priori excluded that both these B- and T-cell lymphoproliferations are acting toward common viral or autoantigens.34 Allo-HCT can create a microenvironment with abundant antigenic triggers for T-LGL clones to expand.35–37 This is attested by the correlation of clinical events and longitudinal TRB gene repertoire analysis in a patient from our study who underwent allo-HSCT and developed clonally expanded T-LGL cells, gradually evolving into overt T-LGL leukemia. This highlights the thin line that segregates reactive T-LGL lymphoproliferations from their leukemic counterparts, while raising the question why the full blown T-LGL leukemia occurred so many years posttransplantation. Arguably, this outcome could reflect a form of profound immune dysregulation or very late onset chronic graft versus host disease, with the T-LGL cells being hyper-reactive to an alloantigen or autoantigen due to failing thymic selection posttransplantation.38 Increased sequence identity of different TR CDR3 supports affinity for the same antigens and common evolutionary forces, without, however, excluding the possibility that distinct CDR3 could share common specificity.39 On the evidence presented herein, the case for (auto)antigenic stimulation is very strong in patients with T-LGL lymphoproliferations, a claim supported by the clonotype connectivity analysis, which revealed clusters of highly related clonotypes coexisting in the same patient and, moreover, displaying over time drift. An observation that was also described by Huuhtanen and colleagues,40 who demonstrated through single-cell TCR sequencing considerable overlap between the nonleukemic and leukemic part of the repertoire in patients with T-LGL leukemia, with 72% of the leukemic T-LGL clonotypes sharing TCR similarities with their nonleukemic repertoire. This phenomenon of repertoire skewing and TCR similarity within the nonleukemic repertoire was also observed in patients with CD4+ T-LGL leukemia.41 Along these lines, cross-entity comparisons highlighted the uniqueness of the TRB gene repertoire in patients with T-LGL lymphoproliferations, given that only a minor fraction of the total unique clonotypes was found to be shared with other entities, even those presenting with neutropenia (eg, CIN or BEN). This conclusion is also supported by the results from the predicted specificity analysis, where the clonotypes shared between T-LGL lymphoproliferations and other entities were postulated to be specific against viruses. Along these lines, one could speculate that interactions with exogenous antigens and/or auto/neoantigens arising in the context of viral infections (but also cancer or autoimmune disorders) could provide the initial antigenic drive for CD8+ T cells, kicking off the process that eventually leads to the emergence of T-LGL lymphoproliferations. Considering the fact that Huuhtanen and Bhattacharya demonstrated with single-cell sequencing that repertoire skewing is present within the nonleukemic repertoire and, more importantly, that the nonleukemic repertoire also shows TCR similarity with the leukemic repertoire, it seems very plausible that antigenic stimulation provides the initial drive for T-LGL cells to expand.40,41 In future studies, we will try to solidify this hypothesis by sequencing the repertoires of sorted T-LGL cells and their healthy CD8+ counterparts within the same patient, as it cannot be a priori excluded that some effects are observed by presence of a public repertoire or the non-LGL T-cell population. In summary, we report that the TRB gene repertoire of patients with T-LGL lymphoproliferations is extremely context-dependent, displaying distinct clonality patterns in different disease contexts. We also illustrate that there is a thin line between malignant T-LGL cells causing symptoms such as neutropenia and reactive T-LGL lymphoproliferations and that the diagnosis of T-LGL leukemia has to be established with great caution, as T-LGL lymphoproliferations may or may not give rise to clear symptoms. In addition, our longitudinal analyses revealed profound temporal clonal dynamics, raising the possibility that T-LGL lymphoproliferations might represent an epiphenomenon when co-occurring with other malignancies, being reactive toward, for example, tumor antigens. Author contributions JLJCA, AC, KS, and AWL designed the study. JLJCA and EV performed the experiments. EV, KG, and AA performed the bioinformatics analyses. JLJCA, EV, KG, NP, PMK, AC, KS, and AWL analyzed data. JLJCA, EV, KG, AC, KS, and AWL interpreted results and wrote the article. All authors approved the article. AWL and KS were responsible for financial support. AP provided well-defined patient samples. DISCLOSURES The authors have no conflicts of interest to disclose. SOURCES OF FUNDING The authors declare no sources of funding related to the work described in his manuscript. Supplementary Material JLJCA and EV shared first authorship and have contributed equally to this work. KS, AC, and AWL shared last authorship and have contributed equally to this work. Supplemental digital content is available for this article. ==== Refs REFERENCES 1. Lamy T Moignet A Loughran TP Jr . LGL leukemia: from pathogenesis to treatment. . Blood. 2017;129 :1082–1094.28115367 2. Mohan SR Maciejewski JP . Diagnosis and therapy of neutropenia in large granular lymphocyte leukemia. Curr Opin Hematol. 2009;16 :27–34.19057202 3. Langerak AW Sandberg Y van Dongen JJM . Spectrum of T-large granular lymphocyte lymphoproliferations: ranging from expanded activated effector T cells to T-cell leukaemia. Br J Haematol. 2003;123 :561–562.14617025 4. Bockorny B Dasanu CA . Autoimmune manifestations in large granular lymphocyte leukemia. Clin Lymphoma Myeloma Leuk. 2012;12 :400–405.22999943 5. Lamy T Loughran TP . How I treat LGL leukemia. Blood. 2011;117 :2764–2774.21190991 6. Papadaki T Stamatopoulos K Kosmas C . Clonal T-large granular lymphocyte proliferations associated with clonal B cell lymphoproliferative disorders: report of eight cases. Leukemia. 2002;16 :2167–2169.12357377 7. Papalexandri A Karypidou M Stalika E . Skewing of the T-cell receptor repertoire in patients receiving rituximab after allogeneic hematopoietic cell transplantation: what lies beneath? Leuk Lymphoma. 2019;60 :1685–1692.30652530 8. Zhang D Loughran TP Jr . Large granular lymphocytic leukemia: molecular pathogenesis, clinical manifestations, and treatment. Hematology Am Soc Hematol Educ Program. 2012;2012 :652–659.23233648 9. Koskela HLM Eldfors S Ellonen P . Somatic STAT3 mutations in large granular lymphocytic leukemia. N Engl J Med. 2012;366 :1905–1913.22591296 10. Rajala HLM Porkka K Maciejewski JP . Uncovering the pathogenesis of large granular lymphocytic leukemia—novel STAT3 and STAT5b mutations. Ann Med. 2014;46 :114–122.24512550 11. Teramo A Barila G Calabretto G . STAT3 mutation impacts biological and clinical features of T-LGL leukemia. Oncotarget. 2017;8 :61876–61889.28977911 12. Sandberg Y Kallemeijn MJ Dik WA . Lack of common TCRA and TCRB clonotypes in CD8+/TCRαβ+ T-cell large granular lymphocyte leukemia: a review on the role of antigenic selection in the immunopathogenesis of CD8+ T-LGL. Blood Cancer J. 2014;4 :e172–e172.24413066 13. Clemente MJ Wlodarski MW Makishima H . Clonal drift demonstrates unexpected dynamics of the T-cell repertoire in T-large granular lymphocyte leukemia. Blood. 2011;118 :4384–4393.21865345 14. O’Keefe CL Plasilova M Wlodarski M . Molecular analysis of TCR clonotypes in LGL: a clonal model for polyclonal responses. J Immunol. 2004;172 :1960–1969.14734782 15. Langerak AW Assmann JLJC . Large granular lymphocyte cells and immune dysregulation diseases–the chicken or the egg? Haematologica. 2018;103 :193–194.29386374 16. Van Dongen JJM Langerak AW Brüggemann M . Design and standardization of PCR primers and protocols for detection of clonal immunoglobulin and T-cell receptor gene recombinations in suspect lymphoproliferations: report of the BIOMED-2 Concerted Action BMH4-CT98-3936. Leukemia. 2003;17 :2257–2317.14671650 17. Vlachonikola E Vardi A Stamatopoulos K Hadzidimitriou A . High-throughput sequencing of the T-cell receptor beta chain gene repertoire in chronic lymphocytic leukemia. Methods Mol Biol. 2019;1881 :355–363.30350216 18. Vardi A Vlachonikola E Karypidou M . Restrictions in the T-cell repertoire of chronic lymphocytic leukemia: high-throughput immunoprofiling supports selection by shared antigenic elements. Leukemia. 2017;31 :1555–1561.27904140 19. Bagaev DV Vroomans RM Samir J . VDJdb in 2019: database extension, new analysis infrastructure and a T-cell receptor motif compendium. Nucleic Acids Res. 2020;48 (D1 ):D1057–D1062.31588507 20. Shugay M Bagaev DV Zvyagin IV . VDJdb: a curated database of T-cell receptor sequences with known antigen specificity. Nucleic Acids Res. 2018;46 (D1 ):D419–D427.28977646 21. Kerr CM Clemente MJ Chomczynski PW . Subclonal STAT3 mutations solidify clonal dominance. Blood Adv. 2019;3 :917–921.30898763 22. Rigoutsos I Floratos A . Combinatorial pattern discovery in biological sequences: the TEIRESIAS algorithm. Bioinformatics (Oxford, England). 1998;14 :55–67.9520502 23. Darzentas N Hadzidimitriou A Murray F . A different ontogenesis for chronic lymphocytic leukemia cases carrying stereotyped antigen receptors: molecular and computational evidence. Leukemia. 2010;24 :125–132.19759557 24. Lefranc M-P . IMGT, the international ImMunoGeneTics database®. Nucleic Acids Res. 2003;31 :307–310.12520009 25. Pielou EC . Shannon’s formula as a measure of specific diversity: its use and misuse. Am Naturalist. 1966;100 :463–465. 26. Stewart JJ Lee CY Ibrahim S . A Shannon entropy analysis of immunoglobulin and T cell receptor. Mol Immunol. 1997;34 :1067–1082.9519765 27. Stalika E Papalexandri A Iskas M . Familial CD3+ T large granular lymphocyte leukemia: evidence that genetic predisposition and antigen selection promote clonal cytotoxic T-cell responses. Leuk Lymphoma. 2014;55 :1781–1787.24180333 28. Liu JH Wei S Lamy T . Chronic neutropenia mediated by fas ligand. Blood. 2000;95 :3219–3222.10807792 29. Steinway SN LeBlanc F Loughran TP Jr . The pathogenesis and treatment of large granular lymphocyte leukemia. Blood Rev. 2014;28 :87–94.24679833 30. Epling-Burnette PK Liu JH Catlett-Falcone R . Inhibition of STAT3 signaling leads to apoptosis of leukemic large granular lymphocytes and decreased Mcl-1 expression. J Clin Invest. 2001;107 :351–362.11160159 31. Smyth MJ Dunn GP Schreiber RD . Cancer immunosurveillance and immunoediting: the roles of immunity in suppressing tumor development and shaping tumor immunogenicity. Adv Immunol. 2006;90 :1–50.16730260 32. Chen Z Zhang C Pan Y . T cell receptor β-chain repertoire analysis reveals intratumour heterogeneity of tumour-infiltrating lymphocytes in oesophageal squamous cell carcinoma. J Pathol. 2016;239 :450–458.27171315 33. Vlachonikola E Vardi A Kastritis E . Longitudinal t cell immunoprofiling of patients with relapsed and/or refractory myeloma who receive daratumumab monotherapy: A subanalysis of a phase 2 study (the rebuild study). Washington, DC: American Society of Hematology; 2019. 34. Wlodarski MW O’Keefe C Howe EC . Pathologic clonal cytotoxic T-cell responses: nonrandom nature of the T-cell–receptor restriction in large granular lymphocyte leukemia. Blood. 2005;106 :2769–2780.15914562 35. Ram R Ben-Bassat I Shpilberg O . The late adverse events of rituximab therapy–rare but there!. Leuk Lymphoma. 2009;50 :1083–1095.19399690 36. Horowitz MM Gale RP Sondel PM . Graft-versus-leukemia reactions after bone marrow transplantation. Blood. 1990;75 :555–562.2297567 37. Zambello R Loughran TP Trentin L . Serologic and molecular evidence for a possible pathogenetic role of viral infection in CD3-negative natural killer-type lymphoproliferative disease of granular lymphocytes. Leukemia. 1995;9 :1207–1211.7630196 38. Socié G Ritz J . Current issues in chronic graft-versus-host disease. Blood. 2014;124 :374–384.24914139 39. Huisman W Hageman L Leboux DA . Public t-cell receptors (TCRs) revisited by analysis of the magnitude of identical and highly-similar TCRs in virus-specific t-cell repertoires of healthy individuals. Front Immunol. 2022;13 :851868.35401538 40. Huuhtanen J Bhattacharya D Lönnberg T . Single-cell characterization of leukemic and non-leukemic immune repertoires in CD8+ T-cell large granular lymphocytic leukemia. Nat Commun. 2022;13 :1981.35411050 41. Bhattacharya D Teramo A Gasparini VR . Identification of novel STAT5B mutations and characterization of TCRβ signatures in CD4+ T-cell large granular lymphocyte leukemia. Blood Cancer J. 2022;12 :31.35210405

PMC010xxxxxx/PMC10353717.txt

==== Front Europace Europace europace Europace 1099-5129 1532-2092 Oxford University Press US 10.1093/europace/euad196 euad196 Letter to the Editor AcademicSubjects/MED00200 Eurheartj/1 Eurheartj/7 Ablation for long QT syndrome: local or global repolarization effects? https://orcid.org/0000-0002-0448-5143 Wichterle Dan Department of Cardiology, Institute for Clinical and Experimental Medicine (IKEM), Prague, Czechia https://orcid.org/0009-0008-1555-562X Segeťová Markéta Department of Cardiology, Institute for Clinical and Experimental Medicine (IKEM), Prague, Czechia https://orcid.org/0000-0001-8319-1460 Krebsová Alice Department of Cardiology, Institute for Clinical and Experimental Medicine (IKEM), Prague, Czechia Corresponding author. Tel: +420 602 848 364. E-mail address: wichterle@hotmail.com Conflict of interest: None declared. 7 2023 10 7 2023 10 7 2023 25 7 euad196© The Author(s) 2023. Published by Oxford University Press on behalf of the European Society of Cardiology. 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (https://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com ==== Body pmcWe have read with interest the paper by Pappone et al.1 in the recent issue of the Journal. The respectable group of authors should be congratulated for suggesting the novel concept of arrhythmogenesis in patients with long QT syndrome and for introducing the ablation strategy that may expand the therapeutic armamentarium in those who suffer from recurrent malignant ventricular arrhythmias. It is amazing to see a localized substrate in the epicardium of the right ventricular outflow tract. Although the ablation of this epicardial layer with clearly abnormal electrical activities appeared highly effective in the prevention of recurrent arrhythmias, we would like to discuss the author’s observation that ablation was accompanied by a shortening of corrected QT (QTc) interval. Accurate and precise QT interval measurement is challenging even in healthy subjects, let alone in patients with T- and U-wave abnormalities. As an example, in Figure 2 (patient #4), the authors presented the procedural recordings of 12-lead electrocardiogram (ECG) before and after ablation with presumable shortening of QT interval from 489 to 439 ms and QTc interval from 525 to 482 ms. The measurement was performed on a single cardiac cycle, but the use of electronic callipers and the paper speed of 50 cm/s guaranteed the reliable detection of a potential shortening of repolarization by >10 ms and definitely by ∼40 ms as suggested. We re-assessed these readings in magnified and digitized Figure 2. The original callipers were placed correctly in the baseline ECG recording, so that the baseline QT data appear realistic. In the post-ablation ECG, however, the QRS-onset calliper was placed clearly late as is evident in almost all ECG leads. On the other hand, the T-wave-offset calliper was placed too early which can be best seen in leads V5 and V6. After the elimination of this ‘bilateral’ bias of ∼20 ms in favour of QT interval shortening by 40 ms, the final correct QT interval could be 479 ms with QTc interval of 526 ms, indicating virtually no change in the duration of repolarization induced by epicardial substrate ablation. In Supplementary material online, Figure S4 for the same patient, a 9-month follow-up ECG is presented with an automated reading of QT interval (382 ms) and QTc interval (404 ms) that appear correct by the naked eye, but the comparison to former ECGs, which were acquired in the anaesthetized patient, is not relevant and standard pre-ablation ECG is not provided. We thus wonder whether Pappone et al. would admit that proposed substrate ablation may indeed modify or even shorten the repolarization of ventricular myocardium, but that this works locally, not globally. Such a conclusion would be more plausible than unsupported speculation that ‘ablation over these abnormal regions causes distal denervation that impairs repolarization of the entire heart’. In our opinion, relatively limited ablation cannot substantially affect the multiple neural inputs into the heart as well as the overall functioning of the intrinsic autonomic nervous system at the ventricular level. ==== Refs Reference 1 Pappone C , CiconteG, AnastasiaL, GaitaF, GrantE, MicaglioEet al Right ventricular epicardial arrhythmogenic substrate in long-QT syndrome patients at risk of sudden death. Europace 2023;25 :948–55.36610790

PMC010xxxxxx/PMC10353721.txt

==== Front PNAS Nexus PNAS Nexus pnasnexus PNAS Nexus 2752-6542 Oxford University Press US 10.1093/pnasnexus/pgad216 pgad216 Biological, Health, and Medical Sciences Plant Biology AcademicSubjects/MED00010 AcademicSubjects/SCI00010 AcademicSubjects/SOC00010 Ethylene-mediated metabolic priming increases photosynthesis and metabolism to enhance plant growth and stress tolerance https://orcid.org/0000-0003-1100-0535 Brenya Eric Department of Biochemistry & Cellular and Molecular Biology, University of Tennessee, Knoxville, TN 37996, USA https://orcid.org/0000-0002-2038-8052 Dutta Esha Genome Science and Technology Program, University of Tennessee, Knoxville, TN 37996, USA https://orcid.org/0000-0003-2120-3010 Herron Brittani Department of Biochemistry & Cellular and Molecular Biology, University of Tennessee, Knoxville, TN 37996, USA https://orcid.org/0000-0002-9696-129X Walden Lauren H Department of Biochemistry & Cellular and Molecular Biology, University of Tennessee, Knoxville, TN 37996, USA https://orcid.org/0000-0002-0780-7056 Roberts Daniel M Department of Biochemistry & Cellular and Molecular Biology, University of Tennessee, Knoxville, TN 37996, USA Genome Science and Technology Program, University of Tennessee, Knoxville, TN 37996, USA https://orcid.org/0000-0002-8172-2034 Binder Brad M Department of Biochemistry & Cellular and Molecular Biology, University of Tennessee, Knoxville, TN 37996, USA Genome Science and Technology Program, University of Tennessee, Knoxville, TN 37996, USA Bayer Edward Editor To whom correspondence should be addressed: Email: bbinder@utk.edu Competing Interest: The authors declare no competing interest. 7 2023 18 7 2023 18 7 2023 2 7 pgad21623 1 2023 13 6 2023 20 6 2023 18 7 2023 © The Author(s) 2023. Published by Oxford University Press on behalf of National Academy of Sciences. 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs licence (https://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Abstract Enhancing crop yields is a major challenge because of an increasing human population, climate change, and reduction in arable land. Here, we demonstrate that long-lasting growth enhancement and increased stress tolerance occur by pretreatment of dark grown Arabidopsis seedlings with ethylene before transitioning into light. Plants treated this way had longer primary roots, more and longer lateral roots, and larger aerial tissue and were more tolerant to high temperature, salt, and recovery from hypoxia stress. We attributed the increase in plant growth and stress tolerance to ethylene-induced photosynthetic-derived sugars because ethylene pretreatment caused a 23% increase in carbon assimilation and increased the levels of glucose (266%), sucrose/trehalose (446%), and starch (87%). Metabolomic and transcriptomic analyses several days posttreatment showed a significant increase in metabolic processes and gene transcripts implicated in cell division, photosynthesis, and carbohydrate metabolism. Because of this large effect on metabolism, we term this “ethylene-mediated metabolic priming.” Reducing photosynthesis with inhibitors or mutants prevented the growth enhancement, but this was partially rescued by exogenous sucrose, implicating sugars in this growth phenomenon. Additionally, ethylene pretreatment increased the levels of CINV1 and CINV2 encoding invertases that hydrolyze sucrose, and cinv1;cinv2 mutants did not respond to ethylene pretreatment with increased growth indicating increased sucrose breakdown is critical for this trait. A model is proposed where ethylene-mediated metabolic priming causes long-term increases in photosynthesis and carbohydrate utilization to increase growth. These responses may be part of the natural development of seedlings as they navigate through the soil to emerge into light. ethylene photosynthesis metabolism growth stress tolerance National Science Foundation 10.13039/100000001 MCB-1716279 Hunsicker Research Incentive Award Dr. Donald L. Akers Jr. Faculty Enrichment ==== Body pmcSignificance Statement The ability of seedlings to transition from darkness to light as they emerge from the soil is critical for plant survival. Here, we demonstrate that ethylene is an important factor early in seedling development that has long-lasting effects on plant growth and tolerance to stresses after they transition from darkness into light. Our study illustrates that transient exposure to ethylene in darkness results in long-term increases in photosynthesis and carbohydrates upon transition to light. These changes lead to increased growth and stress tolerance. This response is likely to be widespread in angiosperms since several angiosperm species show growth enhancement under these conditions. Introduction Enhancing plant vigor is a major challenge because of an increasing human population and reduction in arable land. Methods to increase growth and stress resistance are key to addressing this challenge. However, the success of these approaches is challenging, since improvement in growth often leads to compensation through a decrease in stress tolerance and vice versa. This trade-off can have profound implications on strategies to enhance both plant growth and stress tolerance. Plant growth is regulated by a variety of hormones including ethylene, which is a gaseous hormone that has wide-ranging effects on plants that impact growth, development, and responses to various stresses that reduce crop yield and postharvest storage (1, 2). Because of its complex and widespread signaling role in growth and stress responses, ethylene biosynthesis and signaling are often targeted for genetic or chemical control to improve agricultural outcomes and postharvest storage (3). The current model for ethylene signaling posits that ethylene receptors signal to constitutive triple response 1 (CTR1), which functions as a negative regulator of the pathway (4–6). Downstream of CTR1 is ethylene-insensitive 2 (EIN2) which is a central regulator of ethylene signaling (7). When bound to ethylene, the receptors are inhibited leading to a reduction in CTR1 activity resulting in EIN2 dephosphorylation and cleavage to release the C-terminal portion of EIN2 (8). This leads to stabilization of EIN3 and EIN3-like1 (EIL1) transcription factors that bind to target gene promoters causing changes in other ethylene-responsive genes, including other transcription factors, leading to ethylene responses (9–14). Light affects some responses to ethylene and alters the networks of transcription factors regulated by ethylene downstream of EIN3 and EIL1 (15–18). Ethylene also promotes greening of seedlings during photomorphogenesis (19). These changes have been linked to seedling survival as they emerge from under the soil into the light. Many of the core elements of the ethylene signaling pathway were discovered using the “triple response” assay of dark-grown Arabidopsis thaliana seedlings characterized by a shorter root and hypocotyl, a thicker hypocotyl, and an exaggerated apical hook (4, 20). These responses help the seedlings to grow out of the soil (18). While conducting a “triple response” screen on dark-grown Arabidopsis seedlings, we surprisingly observed that after removal of ethylene and transfer to light, seedlings pretreated with ethylene grew larger and were more stress tolerant than control seedlings. This long-lasting increase in growth included longer primary and lateral roots, a higher density of lateral roots, and an increase in aerial tissue fresh weight. We show that these changes in response to ethylene in the dark are caused by an increase in photosynthesis and carbohydrate levels after transfer to light and removal of ethylene. The growth enhancement response occurs in multiple angiosperm species suggesting that this is a general feature of ethylene signaling in angiosperms transitioning from darkness (underground) to light during germination. The mechanisms for these responses are likely to be different from the effects of continuous ethylene exposure, which can also affect photosynthesis and growth (2, 21). Results Seedlings grow larger after transient treatment with ethylene Ethylene typically inhibits the growth of dark-grown eudicot seedlings (1). As expected, Arabidopsis seedlings treated for 3 days in the dark with ethylene displayed a typical “triple response” with shorter hypocotyls and roots, exaggerated apical hooks, and thicker hypocotyls compared with controls (Fig. S1). Unexpectedly, when ethylene was removed and the seedlings were grown under a 16-h photoperiod, the seedlings that had been pretreated with ethylene developed longer primary and lateral roots, a higher number of lateral roots, and more aerial tissue fresh weight than control seedlings that were not exposed to ethylene (Fig. 1A–D). Hypocotyls remained shorter than air controls days after ethylene removal (Fig. S2). Using end-point analysis, the faster primary root growth became evident 3 days after transfer to light and persisted throughout the observation period (Fig. 1B). The effects of ethylene were long-lasting and resulted in an increase in rosette and plant height after transfer to light compared with controls (Fig. 1E and F). Fig. 1. Phenotypes of plants after exposure to ethylene in the dark in the presence and absence of added sugar. Germinating Arabidopsis seeds were exposed to 0.7 ppm ethylene or ethylene-free air in the dark. At this time (day 0), they were transferred to ethylene-free conditions and grown under a 16-h photoperiod. In some cases, exogenous sugar was included as indicated. A) Images of wild-type (Col) seedlings were acquired 10 days after transfer to light. Scale bar = 5 mm. B, C) The effects of ethylene pretreatment at the indicated concentrations of sucrose on wild-type B) new primary root growth and C) lateral root density at the indicated times after transfer to light and ethylene-free conditions. D) The effects of ethylene pretreatment on leaf fresh weight of wild-type and ein2-5 seedlings in the presence or absence of 0.8% (w/v) sucrose at 12 days after transfer to light and ethylene-free conditions. E, F) Images of plants grown in soil for D) 22 days or E) 33 days after transfer to light and ethylene-free conditions. Scale bars = 1 cm. G) Images of growing roots were captured every 15 min for the first 22 h after transfer to light. The average growth rate ± SEM for each time interval is plotted (n ≥ 10). H) The effects of ethylene pretreatment on new root growth of wild-type, ein2-5, and ein3-1;eil1-1 seeds in the presence or absence of 0.8% (w/v) sucrose at 9 days after transfer to light and ethylene-free conditions. I) The effects of ethylene pretreatment on new primary root growth in wild-type seedlings in the presence or absence of 0.8% (w/v) sucrose, glucose, or fructose at 9 days after transfer to light and ethylene-free conditions. J) The effects of ethylene pretreatment on new primary root growth in wild-type seedlings in the presence or absence of 0.02% (w/v) glucose or mannose at 9 days after transfer to light and ethylene-free conditions. Data in B–D) and H–J) represent the mean ± SEM (n ≥ 15), the data were analyzed by ANOVA, and the different letters indicate significant difference (P < 0.05). To determine how rapidly plants respond with increased primary root growth, we used time-lapse imaging to examine the rate of root growth for the first 22 hours after transfer to light (Fig. 1G). We estimated the latent time for an effect of ethylene pretreatment with two methods to provide a range of times where ethylene pretreatment may begin to affect growth. The time at which there was first a statistical difference (Students t test, P ≤ 0.05) in the mean value was 2.25 hours, whereas a value of 3.75 hours was obtained using cumulative sum control chart analysis to determine when the trends diverge. The enhancement of growth is not limited to the Columbia (Col) ecotype (Fig. S3), and other flowering plant species also exhibit the ethylene-induced enhanced growth response (Fig. 2). Ethylene-pretreated tomato (Solanum lycopersicum, cultivar Floridade) plants sown in soil were taller and had more leaf area than controls, whereas ethylene-pretreated cucumber (Cucumis sativus, cultivar Beit Alpha Burpless) plants sown in soil had an accelerated production of leaves and wheat (Triticum aestivum) seedlings responded with faster primary root growth. Thus, the ethylene-induced growth stimulation signaling pathway may be widespread in the angiosperms. Fig. 2. Ethylene enhances the growth of several plant species. Germinating tomato (S. lycopersicum, cultivar Floridade) A–D) and cucumber (C. sativus, cultivar Beit Alpha Burpless) E) seeds sown in soil were treated with ethylene or ethylene-free air in the dark for 4 days and wheat (T. aestivum) seeds grown on agar F, G) for 3.5 days. At this time, the seedlings were transferred to light and ethylene-free conditions. Photos show representative plants A) 7 days, B) 9 days, E) 11 days, and F) 1.5 days after transfer to light. E) Arrows point to first true leaves. Seedlings on the left in each panel are ethylene-free controls and on the right pretreated with 0.7 ppm ethylene. Scale bars = 1 cm. C, D) Quantification of tomato height and leaf area of tomato seedlings 9 days after transfer to light. G) Quantification of wheat primary root length 1.5 days after transfer to light. Data in C), D), and G) are the mean ± SEM (n ≥ 6) and statistical differences from the untreated controls determined with Student's t test (*P < 0.05; ***P < 0.001). One possible explanation for enhanced growth is that pretreatment with ethylene might lead to a negative feedback loop to reduce ethylene biosynthesis or responsiveness. However, this does not appear to be the case since there was no statistical difference in ethylene production of the control and ethylene-pretreated Arabidopsis seedlings after transfer to ethylene-free conditions (Fig. S4A). Furthermore, the mRNA levels of several ethylene-inducible genes, which were upregulated immediately after treatment with ethylene for 3 days in the dark, returned to basal expression levels 5 days after removal of ethylene and transfer to light. This is consistent with no long-term changes in ethylene responses (Fig. S4B). The ethylene-induced growth response was not simply a delayed response to ethylene since seedlings that were administered continuous ethylene treatment after transfer to light lacked the growth enhancement phenotype and showed shorter roots (Fig. S5) in agreement with the earliest studies on Arabidopsis (4, 20). Additionally, the hypocotyls were taller in the seedlings kept in continuous ethylene upon transfer to light, which is consistent with prior reports documenting that ethylene stimulates the growth of hypocotyls in light-grown Arabidopsis seedlings (22). Thus, the removal of ethylene is required for growth enhancement. To determine if ethylene pretreatment in darkness is required, we conducted experiments in continuous light. This showed that ethylene treatment in darkness is required for growth stimulation since ethylene pretreatment of seedlings in light (rather than darkness) followed by ethylene-free conditions did not cause stimulation of growth (Fig. S6). These results indicate that there is a critical period during seedling development during which transient ethylene exposure in darkness causes lasting growth stimulation upon subsequent exposure to light. Ethylene signaling via EIN2 and EIN3/EIL1 mediates ethylene-induced growth enhancement The above experiments indicate that ethylene pretreatment in darkness is having profound and long-lasting effects on plants after removal of ethylene and transfer to light. To determine if the primary ethylene signaling pathway is involved, we examined ein2-5 and ein3-1;eil1-1 ethylene-insensitive Arabidopsis seedlings. Neither mutant responded to the ethylene pretreatment with longer primary roots (Fig. 1H) and ein2-5 mutants did not respond with increased leaf growth (Fig. 1D). These data demonstrate that the observed growth phenotypes require the primary ethylene signaling pathway. Signaling downstream of EIN3/EIL1 varies depending on light conditions and involves the phytochrome-interacting factors (PIFs) which are important for the transition from darkness to light, and, depending on lighting conditions, they can determine the levels of and binding sites for EIN3 to regulate which genes are regulated (15–19, 23–27). PIF1, PIF3, PIF4, and PIF5 have overlapping gene targets with EIN3 (28). Because of this, and our observation that a transition from darkness to light is important for the ethylene-induced growth stimulation, we examined the effects of ethylene on pif1-1;pif3-7;pif4-2;pif5-3 (pifq) seedlings. The pifq Arabidopsis seedlings still responded to ethylene pretreatment with enhanced growth comparable to wild-type (Fig. S7) indicating these transcription factors are not involved in this response. Sucrose, glucose, and fructose phenocopy ethylene pretreatment The basipetal flow of auxin and photosynthesis-derived sugars are important for root development (29, 30). Therefore, we explored the potential roles for each of these in ethylene-induced stimulation of growth. Application of N-1-naphthylphthalamic acid (NPA) to Arabidopsis seedlings, which blocks auxin transport, reduced primary root growth by ∼40%, but did not block stimulation of primary root growth or leaf growth by ethylene (Fig. S8A and B). In contrast, NPA prevented changes in lateral root density (Fig. S8C) suggesting that auxin transport is required for lateral root formation, but not the other two traits. Auxin transport is known to be critical for lateral root initiation and emergence (30). We also monitored the expression of DR5::GUS in control and experimental seedlings 5 days after transfer to light. The DR5 promoter is sensitive to auxin and provides a measure of auxin responsiveness (31). Ethylene pretreatment had little or no measurable effect on GUS expression in either leaves or root tips indicating auxin responses are not measurably altered by ethylene pretreatment under these conditions (Fig. S8D and E). We typically carry out “triple response” assays in the absence of added sugar. To explore if the enhanced growth is controlled by sugar, we conducted ethylene treatment experiments in the presence of different concentrations of sucrose. A typical ethylene-induced growth inhibition response of hypocotyls and roots (3 days ethylene in darkness) was observed for all seedlings regardless of sucrose concentration (Fig. S1B and C). After subsequent transfer to light, the length of the primary roots of control seedlings increased as a function of sucrose concentration, with maximum growth observed with 0.8% (w/v) sucrose (Fig. 1B). Ethylene pretreatment followed by transfer to light led to increased root growth compared with the controls at all concentrations of exogenous sucrose tested. Sucrose also regulates lateral root formation (32). Increasing sucrose concentrations caused an increase in lateral root density in the air controls and ethylene led to a further increase at all sucrose concentrations, although it took longer for a statistical difference to occur at 0.8% (w/v) sucrose (Fig. 1C). Exogenous sucrose also led to higher leaf fresh weight (Fig. 1D). Unlike stimulation of growth by ethylene, ethylene signaling is not required for increased root or leaf growth in response to exogenous sucrose (Fig. 1D and H). Since sucrose is broken down to fructose and glucose, we next tested whether or not these monosaccharides also phenocopy the growth enhancement phenotype caused by ethylene pretreatment, or whether the effect is specific for sucrose. Addition of 0.8% (w/v) glucose or fructose led to increased primary root growth similar to the effect of sucrose (Fig. 1I). Pretreatment with ethylene led to a further enhancement of growth. Thus, like sucrose, the application of either glucose or fructose phenocopies the effects of ethylene pretreatment suggesting that the enhanced growth in ethylene-pretreated plants might be mediated by an increase in photosynthesis-derived sugars. Sugars are acting as a carbon source to affect growth after ethylene pretreatment Our observations suggest that the growth enhancement phenotype is due to higher levels of sugars. Sugars can affect growth both as signaling molecules and as carbon sources. To first determine whether or not changes in sugar sensing are important, we examined the effects of ethylene pretreatment on a double mutant in Arabidopsis lacking two proteins involved in sugar sensing, hexokinase1 (HXK1) and regulator of G protein signaling 1 (RGS1) (Fig. S9A). The hxk1-3;rgs1-2 double mutant still responded to ethylene with more growth. Consistent with HXK1 not being involved, the transcript abundance of HXK1 was not altered 5 days after ethylene treatment (Fig. S9B). To explore this more, we compared the effects of glucose with mannose. Mannose and glucose are epimers that both signal via HXK1; however, unlike glucose, mannose is poorly metabolized via glycolysis (33–35). If sugar signaling has a major role in the growth enhancement from ethylene pretreatment, then mannose and glucose should have similar effects. However, unlike glucose, mannose inhibited root growth enhancement caused by ethylene (Fig. 1J). These data indicate that the HXK1 sugar sensing pathway is not essential for ethylene-induced growth enhancement. Rather, sugars act as carbon sources to enhance growth after ethylene treatment. Ethylene pretreatment increases photosynthesis In the absence of added sucrose, primary growth inhibition, rather than growth enhancement, was observed in seedlings kept in darkness after removal of ethylene (Fig. S10A). This combined with the above results with sucrose, glucose, fructose, and mannose suggests that photosynthesis-derived sugars are important for the growth stimulation. We therefore examined which wavelengths of light cause growth enhancement since transition to light is important for the growth phenotype. Arabidopsis seedlings transferred to monochromatic blue, green, or red light had enhanced growth after ethylene pretreatment compared with controls suggesting each contributes to the growth response. However, blue light resulted in growth enhancement similar to white light, whereas less growth enhancement occurred with green and red light. Transfer to far-red light led to no growth enhancement. When these experiments were conducted in the presence of 0.2% (w/v) sucrose, growth enhancement by ethylene pretreatment was observed under all wavelengths of light tested and a small, but statistically significant, increase in darkness also occurred (Fig. S10B). To elucidate the role of photoreceptors in the growth enhancement phenotype, we examined mutants lacking various photoreceptors including mutants that lack the phototropin (Phot) 1 or 2 blue light photoreceptors, a mutant lacking both Phot1 and 2 (phot1;phot2), and a quadruple mutant (phyA,B/cry1,2) lacking the phytochrome (Phy) A and B photoreceptors for red/far-red light and the cryptochrome (Cry) 1 and 2 photoreceptors for blue light. All of these mutants responded to ethylene pretreatment with growth enhancement (Fig. S11) indicating these receptors are not critical for this trait under these conditions. Based on these results, we hypothesized that photosynthesis-derived sugars are increased by ethylene pretreatment and are required for this trait. To test this, we evaluated Arabidopsis seedling growth in the presence of norfluorazone or lovastatin to block chlorophyll biosynthesis, or in the presence of 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU) which blocks photosynthesis by disrupting electron transfer in photosystem II (36–39). None of these chemicals blocked root growth inhibition caused by application of ethylene for 3 days in darkness (Fig. S12A and B), but, in the absence of added sugar, all three chemicals prevented stimulation of growth by ethylene pretreatment (Figs. 3A and S12C). However, these chemicals also greatly reduced root growth in the control plants making it difficult to determine if they are blocking the effects of ethylene or are having toxic effects on the seedlings. To address this, we examined mutants that affect photosynthesis. Constitutive photomorphogenic 1 (COP1) affects photomorphogenesis, and cop1 mutants are pale in the light (40). Arabidopsis root growth of cop1-4 seedlings was inhibited by a 3-day treatment with ethylene in darkness (Fig. S12D) showing that this mutant responds to ethylene. However, ethylene pretreatment in the absence of added sucrose failed to stimulate growth of the cop1-4 roots (Fig. 3B). Ethylene pretreatment did not affect COP1 transcript abundance 5 days after transfer to light (Fig. S12E). We also examined the effects of several higher order mutants of members of the family of small subunits of ribulose 1,5-bisphosphate carboxylase/oxygenase (Rubisco) which have been shown to greatly reduce carbon fixation and growth (41). Neither rbcs1a2b nor rbcs1a3b double mutants affected responses to ethylene pretreatment (Fig. 3C). However, in the absence of exogenous sucrose, the rbcs1a2b3b triple mutants, which fix much less carbon than the double mutants (41), failed to have growth stimulation after ethylene pretreatment. These results show that interfering with either the light reactions of photosynthesis or carboxylation reactions of the Calvin cycle blocks the effects of ethylene on growth. Addition of exogenous sucrose to cop1-4, rbcs1a2b3b, or DCMU-treated wild-type seedlings partially rescued the enhancement of growth from ethylene pretreatment (Fig. 3A–C). Thus, sucrose is necessary and when sucrose is available, ethylene pretreatment enhances growth, perhaps by enhancing sucrose breakdown. Fig. 3. Ethylene pretreatment increases carbon assimilation. Germinating Arabidopsis seeds were treated with 0.7 ppm ethylene or ethylene-free air for 3 days in the dark and then transferred to ethylene-free air and light. Unless otherwise indicated, no exogenous sugar was added. A) Wild-type seeds were sown in the presence or absence of 0.2% sucrose in the presence of 2.5 µm DCMU to block electron transport. Solvent-treated samples are included as controls. B, C) Wild-type and cop1-4 B) or rbcs C) mutants were sown in the absence or presence of 0.2% (w/v) sucrose. A–C) New root primary root growth was measured 9 days after transfer to light. Data are the average ± SEM of at least 15 seedlings. Different letters denote statistical difference (P < 0.05) using ANOVA. D) The Fv/Fm was determined in tissue of wild-type seedlings at the indicated times after transfer to white light. Data represent the mean ± SEM (n ≥ 9). E) Chlorophyll was extracted from excised cotyledons of wild-type seedlings and quantified at different times after transfer to white light as indicated. Data were normalized to tissue fresh weight and represent the mean ± SEM (n ≥ 6). D, E) Data were analyzed by Student's t test and found to be statistically different from seedlings not treated with ethylene with a *P < 0.05 and **P < 0.005. F) Transcript levels of selected genes that encode proteins involved in photosynthesis and chlorophyll metabolism were evaluated by qPCR as described in the Materials and methods at 5 days after transfer to light. Each gene was normalized to its levels in the control condition and to housekeeping genes. G) Measurements of carbon assimilation and stomatal conductance in individual leaves 3 weeks after transfer to light were made in three separate experiments and normalized to the amount in control samples. Different symbols represent individual data points from the different experiments. The mean ± SEM is plotted (n ≥ 29). In D), F), and G), data were analyzed by Student's t test and found to be statistically different from seedlings not treated with ethylene with a *P < 0.05 or ***P < 0.001. Since photosynthesis is the source of sugars in plants, we examined various aspects of photosynthesis in more detail to determine whether or not ethylene pretreatment affects photosynthesis. The cotyledons of seedlings treated with ethylene were greener than control seedlings for several hours after transfer to light. This combined with our results with sucrose and the fact that photosynthesis is most rapid in blue light where we see the largest growth enhancement led us to question whether or not increased chlorophyll content and photosynthesis underlie the increased growth we observe. Both the optimum quantum efficiency (Fv/Fm) (Fig. 3D) and chlorophyll content (Fig. 3E) of ethylene-pretreated seedlings were higher for several hours after transfer to light and ethylene-free conditions. However, these differences were not long-lasting. Thus, it seems unlikely that these early, transient changes cause the enduring growth changes, but may be involved in the early stages of growth enhancement after transfer to light. We hypothesized that genes involved in the growth enhancement would be altered several days after removal of ethylene. Therefore, we analyzed the transcript levels of several genes that encode proteins related to photosynthesis. Of the six genes examined, ethylene pretreatment caused an increase in the transcript abundance of four genes 5 days after transfer to light (Fig. 3F). This included chlorophyll a/b binding protein 2 (CAB2), which encodes a protein that is part of the light-harvesting complex associated with photosystem II, glutamyl-tRNA reductase (HEMA1) involved in chlorophyll biosynthesis, and RBCS1A and RBCS2B. Correlating with these increases in RBCS1a and RBCS2B, ethylene pretreatment led to a 23% increase in CO2 assimilation 3 weeks after transfer to light (Fig. 3G) showing that photosynthetic capacity is increased long-term. In contrast, stomatal conductance was not significantly affected. Together, these results indicate that ethylene pretreatment leads to long-lasting increases in expression of select transcripts encoding proteins related to photosynthesis which are likely involved in the long-term increase in carbon fixation. Ethylene pretreatment leads to increased levels of starch and glucose To explore further the link between ethylene pretreatment, enhanced growth, and photosynthesis-derived sugars, we grew Arabidopsis seedlings in the absence of added sugars and compared the levels of starch and glucose in ethylene-pretreated seedlings compared with controls 5 days after transfer to light. As seen in Fig. 4A and B, ethylene pretreatment led to an increase in starch levels in cotyledons, leaves, and roots after transfer to ethylene-free conditions. Consistent with this increase in starch, the transcript abundance of starch synthase 3 (SS3) that encodes a starch biosynthesis enzyme was increased in the seedlings pretreated with ethylene (Fig. 4C). The sucrose phosphate synthase 3F (SPS3F) gene involved in sucrose biosynthesis and the glucuronic acid substitution of xylan 2 (GUX2) gene involved in cell wall biosynthesis were also up-regulated by ethylene pretreatment. Fig. 4. Ethylene pretreatment increases starch and glucose levels. Germinating Arabidopsis seeds in the absence of added sugar were treated with 0.7 ppm ethylene or ethylene-free air for 3 days in the dark and then transferred to ethylene-free air and white light with a 16-h photoperiod. A) Control (left) and ethylene-treated (right) seedlings stained for starch as described in the Materials and methods 9 days after transfer to light. Scale bar is 3 mm. B) Quantification of starch levels normalized to fresh weight of the tissue at 9 days after transfer to light (n = 6). C) Transcript levels of genes that encode enzymes involved in starch and sucrose biosynthesis were evaluated 5 days after transfer to ethylene-free air and light by qPCR as described in the Materials and methods. Each gene was normalized to its levels in the control condition and to housekeeping genes. D) Quantification of glucose levels normalized to fresh weight of the tissue at the indicated days after transfer to light (n = 5). E) Transcript levels of CINV1 and CINV2 were evaluated 5 days after transfer to ethylene-free air and light by qPCR as described in the Materials and methods. Each gene was normalized to its levels in the control condition and to housekeeping genes. In B–E), Student's t test was used to determine statistically significant change from the control samples (*P < 0.05; **P < 0.005; ***P < 0.001). F) The amount of new primary root growth was compared in Col (wild-type) and cinv1;cinv2 double mutants 9 days after transfer to light in the presence or absence of 0.8% (w/v) sucrose. Different letters denote statistically significant differences (P < 0.05) as determined by ANOVA. Data in B–F) are the mean ± SEM (n ≥ 15). Ethylene pretreatment also results in 266% elevation of seedling glucose levels compared with controls (Fig. 4D). We also observed a concomitant increase in the transcript abundance of cytosolic invertase 1 (CINV1) and CINV2 which encode enzymes that degrade sucrose to glucose and fructose and are critical for normal growth (42) (Fig. 4E). Furthermore, mutant seedlings with the loss of both genes (cinv1;cinv2) did not respond to ethylene pretreatment with enhanced root growth in either the presence or absence of sucrose (Fig. 4F). Root and hypocotyl growth of cinv1;cinv2 seedlings were inhibited by a 3-day treatment with ethylene in darkness (Fig. S13) showing that this mutant responds to ethylene. These results are consistent with a model where ethylene pretreatment is leading to an increase in the degradation of sucrose by these invertases to form glucose and fructose which leads to more growth. The above results show that transient ethylene treatment of dark-grown seedlings grown in the absence of added sugar results in sustained changes in the accumulation of carbohydrates upon subsequent growth under light conditions. To determine how widespread the effects of ethylene pretreatment are on cellular metabolism, we carried out an untargeted metabolomic analysis on seedlings grown in the absence of added sugar 6 days after removal of ethylene and transfer to light. Partial least squares discriminant analysis of the data shows a clear separation of control versus ethylene-pretreated samples (Fig. S14A). Based on variable importance in the projection (VIP) scores, 15 metabolites, including trehalose/sucrose, are contributing the most to this separation (Fig. S14B). Of the 131 metabolites detected, 13 were significantly reduced and 97 were significantly increased (P ≤ 0.05, t test) by pretreatment with ethylene (Table S1 and Fig. S14C). Pathway analysis shows that carbon fixation was up-regulated supporting our observations above for an increase in carbon assimilation (Fig. S15 and Table S2). Additionally, amino acid metabolism, purine and pyrimidine metabolism, glyoxylate/dicarboxylate metabolism, glutathione metabolism, and nicotinate/nicotinamide metabolism were affected by ethylene pretreatment. The map of altered metabolism illustrates the interconnectedness of the pathways involved and the widespread up-regulation of metabolism. These data show that ethylene pretreatment results in prolonged changes in many metabolites in multiple metabolic pathways. We refer to this as ethylene-mediated metabolic priming. To further explore processes affected by ethylene-mediated metabolic priming, we conducted RNA sequencing analyses on seedlings 5 days after transfer to light. Under these conditions, ethylene pretreatment resulted in 1,938 gene transcripts being altered 2x or more with an adjusted P ≤ 0.05. Of these, 1,102 were up-regulated and 836 were down-regulated (Fig. S16 and Spreadsheet S1). We examined these genes for enrichments in Gene Ontology (GO) biological processes (43). This revealed that processes associated with up-regulated genes include development, differentiation, morphogenesis, cell cycle, cell division, and cell wall organization (Table S3). In contrast, almost all GO biological processes associated with down-regulated genes have to do with responses to different environmental signals and stress (Table S4). Thus, there appears to be a shift toward growth-related processes and away from defense-related processes. Many of the altered gene transcripts mapped onto the same metabolic pathways as the altered metabolites including carbohydrate metabolism, carbon fixation, amino acid metabolism, and purine and pyrimidine metabolism (Fig. S17). However, additional processes altered at the transcript level by ethylene pretreatment became apparent from the RNA-seq results including fatty acid metabolism, plastoquinone cycling, and the biosynthesis of tocopherol, isoprenoids, monolignol, flavanol, and flavone. These data indicate that transient treatment of plants with ethylene results in up-regulation of processes involved in plant growth and development. Seedlings are more stress tolerant after transient treatment with ethylene Ethylene is a critical hormone that coordinates many stress responses in plants (1). Ethylene pretreatment is known to “prime” plants to survive better upon subsequent challenge with various stresses (44–54). Therefore, we asked whether or not ethylene pretreatment under the conditions reported here also affects tolerance to various abiotic stresses. Since exogenous sugar can increase stress tolerance (55, 56), unless otherwise noted, no sugar was added to the media for these stress experiments. Arabidopsis plants were exposed to each stress 5 days after transfer to light and removal of ethylene. In the absence of ethylene pretreatment, most seedlings exposed to 22 min of high temperature (43°C) or 7 days of 150 mm NaCl, or during recovery from 12 h of hypoxia in darkness, became bleached and died (Figs. 5 and S18). Ethylene priming led to more seedlings surviving exposure to the stress. In seedlings exposed to either high temperature or salt stress, nearly 100% of seedlings survived the stress after exposure to ethylene. By 2 days after hypoxia, the ethylene-pretreated seedlings showed a clear and statistically significant higher survival rate compared with the control seedlings. This trend continued to day 3 of recovery from hypoxia. In all three stress conditions, ethylene pretreatment had no measurable effect on the survival of seedlings not challenged with the stress. Loss of both CINV1 and CINV2 resulted in seedlings that did not respond to ethylene pretreatment with enhanced heat tolerance (Fig. 5A) and application of sucrose phenocopied ethylene with increased heat tolerance (Fig. 5D) suggesting that carbohydrate levels are important for the effects of ethylene priming on stress tolerance. Fig. 5. Ethylene pretreatment increases tolerance to abiotic stresses. Germinating Arabidopsis seeds in the absence of added sugar were treated with 0.7 ppm ethylene or ethylene-free air for 3 days in the dark and then transferred to ethylene-free air and white light with a 16-h photoperiod. A) Five days after transfer to light, Col (wild-type) and cinv1;cinv2 seedlings were exposed to high temperature (43°C) for 22 min. Survival rates 2 days after exposure to high temperature are shown compared with seedlings not stressed with high temperature (22°C). B) Five days after transfer to light, seedlings were exposed to 150 mm NaCl stress and survival assessed 1 week later compared with no stress controls. **P < 0.01 indicates statistical difference from the no ethylene condition using Student's t test. C) Five days after transfer to light, the seedlings were exposed to hypoxia stress for 12 h in darkness and then transferred back to white light and normoxia and allowed to recover. Normoxia controls are shown for comparison. Survival was assessed each day for 3 days. Different letters denote statistically significant differences (P < 0.05) on each day as determined by ANOVA. D) Wild-type seedlings were treated as in A) in the presence or absence 0.2% (w/v) sucrose. In A) and D), different letters denote statistically significant differences (P < 0.05) as determined by ANOVA. Discussion Enhancing plant growth and vigor is an important challenge that needs to be met to feed a growing human population. Methods to increase growth and stress resistance are key to addressing this challenge. However, there is often an inverse relationship between growth and stress tolerance (57). This trade-off can have profound implications on bioengineering strategies to enhance plant yield. Despite these challenges, bioengineering has led to plants with 25% or more photosynthetic efficiency, growth, and yield (58–60). Here, we show that treating dark-grown Arabidopsis seedlings with ethylene for several days followed by transfer to ethylene-free conditions and light results in robust increases in the growth of leaves and roots and increased tolerance to abiotic stresses. Correlating with these results, seedlings pretreated with ethylene have almost a 25% increase in carbon fixation and large increases in carbohydrate levels. Although auxin is often involved in growth-related phenotypes, our results indicate that ethylene pretreatment is not altering auxin responses to stimulate leaf and primary root growth. NPA blocks the enhanced formation of lateral roots by ethylene pretreatment. More experiments will be needed to determine if this simply reflects a requirement for auxin transport or if ethylene pretreatment is altering auxin responses for this trait. In contrast, our studies reveal that transient ethylene treatment results in large and long-term increases in both photosynthesis and carbohydrate abundance resulting in metabolic priming to impact growth and stress tolerance that is comparable to existing bioengineering strategies. Photosynthetically derived sugars control root growth early in seedling development (61), and hexoses and sucrose can affect Arabidopsis root growth and architecture (32, 62, 63). Previous work showed that ethylene affects chloroplast development and chlorophyll levels and photosynthesis in diverse ways depending on factors such as plant species, age of the plant, lighting conditions, and stress conditions (2, 21), and treating etiolated Arabidopsis seedlings with ethylene promotes greening when they are transferred to light (18, 19). Here, we show an interesting effect of ethylene on photosynthesis that is induced in darkness, but is sustained long-term after removal of ethylene and growth in the light. Similar to a prior study on cucumber (64), an increase in chlorophyll levels and optimum quantum efficiency occurred in etiolated Arabidopsis seedlings pretreated with ethylene compared with controls. However, these differences disappeared by 12 h after transfer to light making it unlikely that these changes are important for the long-lasting changes in growth. However, the increase in root growth occurs within several hours after transfer to light and these early, transient changes in photosynthesis may have a role in the early stages of growth stimulation. Long-lasting increases in RBCS1a, RBCS2b, CINV1, and CINV2 occurred after ethylene treatment which correlates with the higher carbon assimilation and carbohydrate levels we measured and increased growth. Our results with monochromatic light indicate that blue light is important for growth enhancement after priming. Although photosynthesis is stimulated by blue light and photosynthesis is important for this trait, plants also respond to blue light via photoreceptors such as the Cry and Phot (65, 66). Higher order phot1;phot2 and phya,b;cry1,2 mutants still respond to ethylene priming with enhanced growth. However, both mutant combinations slightly reduce the growth enhancement response. One interpretation of this is that these photoreceptors are involved, but not required, for this response. For instance, phot mutants affect chloroplast movement which could alter photosynthesis (67) to reduce the effects of priming. Alternatively, these photoreceptors may be required but have overlapping functions that will only become evident with higher order mutant combinations that eliminate both sets of blue light receptors. It is also possible that other blue light receptors not tested in this study are required (68, 69). The growth enhancement from ethylene pretreatment was blocked in cop1 and rbcs1a2b3b mutants and by chemical inhibitors of photosynthesis and chlorophyll synthesis. Thus, in the absence of exogenous sucrose, proper chloroplast development and photosynthesis are required to observe growth stimulation after treatment with ethylene. Addition of sucrose to cop1, rbcs1a2b3b, and DCMU-treated wild-type seedlings partially rescues the effects of ethylene pretreatment. These observations coupled with the fact that ethylene pretreatment causes a long-lasting up-regulation of RBCS1a, RBCS2b, and carbon fixation are consistent with the idea that increases in photosynthetically derived sugars are involved in the growth enhancement phenotype. The fact that such enhancement is not observed in rbcs1a2b3b seedlings shows that the increases in RBCS transcripts are required for the effects of ethylene pretreatment. Addition of sucrose to seedlings maintained in darkness after ethylene treatment results in ethylene-mediated root growth stimulation supporting the idea that ethylene is leading to changes in sugar utilization independently of photosynthesis. Ethylene pretreatment leads to large increases in sucrose, glucose, and starch which correlate with increases in several genes related to carbohydrate metabolism including CINV1 and CINV2 that encode invertases that break down sucrose and are critical for carbon allocation for cellulose biosynthesis and normal plant growth (42, 70, 71). We show that CINV1 and CINV2 are required for the growth enhancement caused by ethylene pretreatment, and addition of sucrose does not rescue this trait in cinv1;cinv2 seedlings. Thus, these invertases are regulated by ethylene and are required for the long-term growth effects caused by ethylene-mediated metabolic priming. However, our data do not distinguish between direct regulation of CINV1 and CINV2 by ethylene and indirect regulation via other factors such as increased sucrose levels from higher photosynthesis. Taken together, our results support a model where transient treatment of dark-grown seedlings with ethylene results in a factor or factors that, upon transfer to light and ethylene-free conditions, result in the long-lasting up-regulation of genes that lead to increased photosynthesis and carbon assimilation that lead to higher levels of glucose, sucrose, and starch and other metabolites to enhance growth (Fig. 6). Whether or not any of these effects are directly due to ethylene signaling or are secondary effects caused by other changes has yet to be determined. Although the metabolomic data indicate many metabolites are up-regulated by ethylene, the results with cinv1;cinv2 suggest that it is the breakdown of sucrose that is critical for growth enhancement. Our experiments demonstrated that the growth stimulation phenotype is probably due to increased carbon availability, rather than sugar signaling via HXK1. However, we cannot rule out that sugar signaling via other pathways that control growth and stress responses, such as via the target of rapamycin (TOR) and sucrose non-fermenting 1-related protein kinase 1 (SnRK1) pathways (72), are important for this response. Fig. 6. Model for enhanced plant vigor from ethylene pretreatment in darkness. In this model, exposure to ethylene while seeds are germinating in darkness activates ethylene signaling resulting in higher EIN3/EIL1 activity. In the dark, this results in the “triple response.” Also, an unknown factor or factors downstream of EIN3 and EIL1 lead to long-lasting changes that, upon illumination and removal of ethylene, result in increased photosynthesis and sugar metabolism resulting in more stress tolerance and growth. It is currently unclear if ethylene is directly affecting all of these pathways or if some effects are secondary to a primary response. In either case, increased photosynthesis leads to higher glucose levels in the leaves to increase starch accumulation. The synthesis of sucrose is also increased which is transported from source to sink tissues where it is broken down to glucose (and fructose). Metabolomic data indicate other metabolic pathways are also affected. Although COP1 is known to affect EIN3 levels in the dark, our data suggest that COP1 affects priming via its role in photomorphogenesis. The changes in photosynthesis and carbohydrate metabolism might occur naturally when seeds germinate underground in darkness and ethylene levels are high due to mechanostimulation from the soil. Upon emergence into light aboveground, the seedlings are exposed to less ethylene because of less mechanostimulation and diffusion away from the aboveground parts of the plant leading to long-lasting developmental changes. RNA-seq revealed that ethylene pretreatment leads to the long-term up-regulation of genes involved in development and morphogenesis and down-regulation of genes annotated to be involved in signaling for environmental stresses. This is consistent with the fact that stress tolerance and growth often show an inverse relationship. However, physiologically, we observe an increase in both growth and tolerance to several abiotic stresses. Given the large increases in carbohydrate levels, the increase in both growth and stress tolerance is consistent with prior research showing that exogenous application of sugars can increase growth (61) and stress tolerance (55, 56), and starch stores are often key to survival of plants under stress (73). Furthermore, ethylene pretreatment increases the levels of both CINV1 and CINV2 and cinv1;cinv2 loss-of-function mutants fail to have increased growth or heat stress tolerance indicating that regulation of sucrose breakdown is a key process affected by ethylene pretreatment. Ethylene has previously been shown to prime plants for specific stresses (44, 45, 47–54), but in most cases, this has not been linked to photosynthesis or carbohydrate levels. However, pretreatment of rice seedlings with the ethylene precursor 1-aminocyclopropane-1-carboxylic acid to increase ethylene levels caused enhanced submergence tolerance and led to higher chlorophyll levels (46) supporting the idea that metabolic priming by ethylene is widespread across angiosperm species. Our results showing enhanced growth of tomato, cucumber, and wheat support this. It is likely that other factors are influenced by ethylene to affect stress tolerance. For instance, the metabolomic analysis indicates that several pathways that affect stress tolerance, such as glyoxylate, nicotinate, and glutathione metabolism, are affected by ethylene-mediated metabolic priming. It is likely that ethylene signaling in darkness is leading to long-term changes that are indispensable for the growth phenotype. Our data do not identify which factor(s) downstream of EIN3/EIL1 mediate the ethylene effects in darkness that are responsible for these long-term changes upon transfer to light. However, distinct light and dark transcriptional networks regulated by EIN3 and EIL1 could provide a clue to the mechanism for this trait. Two such candidates that are involved in the distinct light and dark responses to ethylene are the PIF transcription factors (15, 17–19, 24, 25) and COP1 (21, 64, 70). Our results with pifq plants indicate these PIFs are not involved; however, COP1 is required for the growth enhancement phenotype in the absence of exogenous sucrose. COP1 has previously been shown to regulate EIN3 since EIN3 levels are enhanced by COP1 in darkness (19, 74). However, given the importance of COP1 in the development of chloroplasts, it is more likely that COP1 is required for growth enhancement through its effects on photomorphogenesis and photosynthesis. In support of this, exogenous sucrose does not rescue ethylene-mediated growth enhancement of ein3;eil1, but does rescue cop1-4 suggesting that EIN3/EIL1 are not acting downstream of COP1 for this trait. During seed germination, mechanical stress caused by growing through soil leads to higher ethylene levels, which is important for seedlings to sense mechanical stress and successfully navigate through the soil from darkness into light (18, 74–78). Our studies show that germinating seeds may require this initial stress to increase ethylene levels to affect development for robust growth and tolerance to stresses. This is consistent with the idea that mechanical stress can prepare plants against future stresses by affecting hormone biology (79). The growth stimulation caused by ethylene seems to be a general trait of angiosperms since we see similar ethylene-induced growth enhancement in several angiosperm species. Thus, this represents a possible new approach to increase plant vigor, perhaps in greenhouse applications where plants can be easily pretreated with ethylene. Materials and methods Plant materials Unless otherwise specified, experiments were carried out on A. thaliana in the Col background. Col and Wassilewskija seeds are lab stocks. The ein2-5 seeds were from Anna Stepanova (7), ein3-1;eil1-1 seeds from Joseph Ecker (12), cop1-4 mutants from Peter Schopfer (61), cinv1;cinv2 seeds from Charles Anderson (70), mutants of the small subunits of Rubisco (rbcs2b3b, rbcs1a3b, and rbcs1a2b3b) were from Alistair McCormick (41), and hxk1-3;rgs1-2 (80), pif1-1;pif3-7;pif4-2;pif5-3 (81), and additional ecotypes and the phot1 and phot2 seeds were from the Arabidopsis Biological Resource Center. The DR5::GUS lines were from Elena Shpak. All mutants are in the Col background. Seeds of tomato were obtained from Zellajake Farm and Garden, cucumber from Isla's Garden Seeds, and wheat from Palouse Brand. Preparation of seeds and growth Arabidopsis seeds were surface sterilized and sown on 0.8% (w/v) agar containing half-strength Murashige and Skoog basal salt mixture, pH 6.0 with no added sugar. In some experiments, the medium was supplemented with the indicated concentration of other compounds. Prior to growth experiments, seeds were stratified for 2–3 days at 4°C, light-treated for 2 to 4 h under continuous fluorescent lights, and then wrapped in aluminum foil. Seeds were allowed to germinate and grow on vertically orientated plates for 3 days in darkness in chambers through which ethylene-free air or ethylene was passed at a flow rate of 50 ml min−1. Unless otherwise specified, ethylene treatment of seedlings was carried out with methods modified from Binder et al. (82) using a concentration of 0.7 ppm ethylene. At the end of this treatment (designated day 0), images were acquired by scanning on a flat-bed scanner. The position of each root tip was then marked, and the plates were transferred to ethylene-free chambers under a 16-h photoperiod (120 µmol m−2 s−1) at 21 to 23°C for up to 14 days. To examine the effects of monochromatic light, seedlings were transferred to ethylene-free conditions under continuous 40 µmol m−2 s−1 white, 53 µmol m−2 s−1 blue (λmax = 462 nm), 20 µmol m−2 s−1 green (λmax = 525 nm), 28 µmol m−2 s−1 red (λmax = 672 nm), 12 µmol m−2 s−1 far-red (λmax = 732 nm), or darkness for 9 days. Monochromatic light was delivered using light-emitting diode arrays from Quantum Devices. The amount of new primary root growth and number of lateral roots were determined using ImageJ (ver. 1.52E). Lateral root density was then calculated for each seedling from these parameters. For all analyses, the length of new root growth after transfer to light is measured. In some experiments, leaf biomass was determined 14 days after transfer to light by removing shoot tissue and weighing the tissue fresh weight for each seedling. To determine the longer-term effects of ethylene pretreatment, Arabidopsis seeds were prepared as above and sown in soil comprised of a 2:1 mixture of peat light:perlite. They were then treated in darkness for 4–5 days with either 0.7 ppm ethylene or ethylene-free air followed by ethylene-free conditions in a 16-h photoperiod. Images were taken at the times designated. Tomato and cucumber seeds were surface sterilized with 5% (v/v) bleach for 30 min. Tomato seeds were placed under white light for 2–3 h and then planted in soil mixture. Cucumber seeds were soaked in water for 1 h under white light, then excess water removed and light treated for 2 h. They were then transferred to soil. Wheat seeds were surface sterilized with 33% (v/v) bleach for 15 min, rinsed three times with sterile distilled water, and sown on 0.8% (w/v) agar plates. For all three species, the seeds were then treated in darkness for 3.5–4 days with either 0.7 ppm ethylene or ethylene-free air. They were transferred to ethylene-free and long-day conditions. At the indicated times, images were taken of the plants. The height and leaf area of tomato were determined using ImageJ. Wheat seedlings were imaged 1.5 days after transfer to light and the length of the primary root determined using ImageJ. Growth rate measurements of primary roots Seedlings were treated and grown on vertically orientated plates as described above. After 3 days in the dark in the presence or absence of 0.7 ppm ethylene, seedlings were transferred to continuous white light (75 µmol m−2 s−1) and ethylene-free conditions. At this time, high-resolution time-lapse imaging was carried out on the primary roots using imaging setups previously described (83). Imaging was started 15 min after transfer to light and images were acquired every 15 min for 22 h. New growth in each time interval was then determined and the growth rate calculated for each time interval. To estimate the latent period for an effect of ethylene pretreatment, we used two methods. One was to determine the time at which ethylene pretreatment led to a statistically significant change in the mean growth rate between the two conditions using Student’s t test (P ≤ 0.05). The other was using cumulative sum control chart analysis to determine the time at which the growth rate trends diverged between the control and experimental seedlings (84). Photosynthesis and chlorophyll measurements To determine optimum quantum efficiency (Fv/Fm) and chlorophyll levels, ∼50 Arabidopsis seeds were plated on agar plates and treated as above. For quantum efficiency measurements, plates of seeds were transferred to continuous white light and at the times indicated placed in a FluoroCam 800MF (Photon Systems Instruments, Czech Republic), dark acclimated for 3–5 min, and the Fv/Fm measured using a preprogramed protocol. To determine chlorophyll a and chlorophyll b levels, cotyledons were excised at the times indicated. Chlorophyll was extracted with acetone and quantified with a spectrophotometer according to the methods of Lichtenthaler (85). Data were normalized to the tissue fresh weight. To determine photosynthetic capacity and stomatal conductance, Arabidopsis plants were grown in soil and treated as above. After 24 days growth in a 16-h photoperiod, leaf 7 was clamped in a cuvette (LI6400XT, LI-COR, Lincoln, NE, USA) with a light emitting diode. Light intensity was 350 μmol m−2 s−1 (10% blue and 90% red) with a block temperature of 22°C, humidity of 55%, constant CO2 concentration maintained at 400 μmol mol−1, and a flow rate of 500 μmol s−1. When stomatal conductance and net assimilation were stable, two to three measurements were made to represent a technical replicate. All experiments were repeated three times. Ethylene measurements Arabidopsis seeds were prepared as above except they were placed on agar in 130-ml jelly jars covered with aluminum foil. Eight days after treatment with ethylene, the jars were sealed with gas-tight lids fitted with a septum, and 24 h later, the amount of ethylene in the headspace was measured using an ETD-300 ethylene detector (Sensor Sense). GUS reporter gene assay and microscopy Transgenic seedlings expressing DR5::GUS were grown and treated with ethylene or ethylene-free air as described above. GUS staining was carried out 5 days after transfer to light. Images of leaves and roots were taken using a dissecting microscope (Olympus SZH10) with a digital camera (Canon Rebel T1i). Starch and glucose analyses Arabidopsis seedlings were grown and treated with ethylene and ethylene-free air as described above. For starch staining, seedlings were harvested at 9 days after transfer to light and then were boiled in 95% ethanol to remove chlorophyll after which seedlings were rinsed twice in distilled water. Leaves were stained for 10 min in Lugol's iodine solution (Volu-Sol) in the dark and then rinsed in water to clear excess background. Images were acquired using Zeiss Stereo microscope Discovery V8 with a Canon camera. For starch quantification, seedlings were collected 9 days after transfer to light and for glucose quantification 5 and 9 days after transfer to light. For both assays, seedlings were frozen in liquid nitrogen after weighing and then ground into fine power in a 2-ml Eppendorf tube containing two steel balls using a homogenizer (Biospec). Starch quantification was performed as described in the protocol of the starch assay kit (SA20, Sigma) and glucose quantification as described in the glucose assay kit (GAHK20, Sigma). For all assays, seedlings were harvested at 10 h into the photoperiod. Untargeted metabolomic analysis Arabidopsis seedlings were grown and treated as described above. Seedlings were harvested 6 days after transfer to light and at 10 h into the photoperiod. Samples were weighed and quickly frozen in liquid nitrogen before untargeted metabolomic analysis using ultraperformance liquid chromatography coupled with hybrid quadrupole Orbitrap mass spectrometry (UPLC Orbitrap MS/MS) at the Biological and Small Molecule Mass Spectrometry Core Facility in the Department of Chemistry at the University of Tennessee. MetaboAnalyst 5.0 was used for partial least squares discriminant analysis and VIP scores and to conduct pathway analysis (86). Data represent the analysis of five biological replicates for controls and four replicates for ethylene-pretreated samples which were each analyzed with three technical replicates and normalized to tissue fresh weight. RNA isolation and real-time qRT-PCR Whole seedlings were collected and quickly frozen in liquid nitrogen. Total RNA was extracted using the PureLink plant RNA reagent (Ambion, TX, USA) and treated with turbo-DNase (Invitrogen). RNA was purified using acid phenol–chloroform (Invitrogen). RNA concentration was determined using a NanoDrop after which samples were normalized to the same concentration for cDNA synthesis. First-strand cDNA synthesis was done using SensiFAST cDNA synthesis kit (Meridian Bioscience). Primers were designed across exon junction using the online Primer3Plus program. Secondary structures were identified to eliminate primer dimers using the IDT UNAFold program, and primer quality and efficiency were determined to be between 90 and 100%. qPCR was performed using Sensifast SYBR No-Rox (Meridian Bioscience) as described in the manual with the following PCR conditions: 28 cycles of 98°C for 30 s, 98°C for 10 s, 60°C for 30 s, and 72°C for 30 s followed by 72°C for 5 min. Gene expression levels relative to the previously validated reference genes UBQ10 (At4g05320) and GAPDH (At1g13440) (87) were used for each sample following the method of Pfaffl (88). Specific primer sequences for these analyses are listed in Table S5. RNA sequencing Arabidopsis seeds were germinated in the presence or absence of 0.7 ppm ethylene in darkness for 3 days, followed by transfer to ethylene-free conditions and light for 5 days. At this time, seedlings were harvested and quickly frozen in liquid nitrogen. Total RNA was extracted using TRIzol and sent to Genewiz (Azenta Life Sciences, South Plainfield, NJ) for library preparation and sequencing using an Illumina HiSeq 4000. This produced over 25 million paired-end reads per sample. All data analyses including read trimming, GO analyses, differential gene analysis, and mapping reads to the A. thaliana reference genome were conducted by Genewiz. Three biological replicates were used for each condition. The RNA-seq data discussed in this publication are deposited in the NCBI's Gene Expression Omnibus (89) and are accessible through the GEO Series Accession Number GSE218645 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE218645). Salt stress assays Seeds were initially sown on sterile nylon filters placed on agar plates and treated as above. At 5 days after removal of ethylene and transfer to light, the seedlings were transferred to plates with either 150 mm NaCl or no added salt and survival rate determined 1 week later by monitoring chlorosis. All experiments were repeated 3 times. Heat tolerance and hypoxia assays For both assays, seeds were prepared as described above and stress administered 5 days after transfer to light. Heat tolerance assays were carried out based on published methods (90, 91) at either 22 or 43°C for 22 min. Seedlings were returned to 22°C, and any that turned white and had not grown new leaves by 3 days afterwards were scored as having died. Hypoxia stress treatments on Arabidopsis seedlings were administered by argon gas as previously described (92). Briefly, hypoxia was administered by purging a sealable chamber (9.5-l jar Gas Pak System, BBL) with argon gas (AR UHP300, Airgas) to a final O2 level <2% measured by using a Traceable Oxygen probe (Fisher Scientific). Plants were kept in hypoxic conditions in darkness for 12 h and then were returned to normal aerobic long-day growth chamber conditions. Plant survival was then assessed over a 3-day period by monitoring the absence of shoot chlorosis (93). Control (normoxic) plants were treated identically except that the argon purging step was omitted. Statistical analyses Pairwise comparisons were conducted with Student’s t tests using Excel (Office 2019). For multiple group comparisons, ANOVA tests were performed using GraphPad Prism ver. 9.3.l. Cumulative sum control chart analysis was conducted using Excel. Supplementary Material pgad216_Supplementary_Data Click here for additional data file. Acknowledgments We thank Jenny Zhang and Grace Phelan for technical assistance, Eric Schaller, Gloria Muday, Barry Bruce, K. Trout, Andreas Nebenführ, Rachel P. McCord, Ricardo Urquidi-Camacho, and Mariano Labrador for helpful conversations and advice, and the Susan Kalisz lab for use of their LI-COR. Supplementary material Supplementary material is available at PNAS Nexus online. Funding This project was supported by the United States National Science Foundation Grant MCB-1716279 and a Hunsicker Research Incentive Award and Dr. Donald L. Akers Jr. Faculty Enrichment Fellowship from the BCMB department to B.M.B. Author contributions E.B., D.M.R., and B.M.B. designed the research; E.B., B.M.B., L.H.W., E.D., and B.H. performed the research and analyzed the data; and E.B., D.M.R., and B.M.B. wrote the paper. Data availability All data needed to evaluate this study are present in the paper, supplementary materials, or the publicly available RNA-seq data set at GEO Series Accession Number GSE218645 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE218645). ==== Refs References 1 Abeles F , MorganP, SaltveitMJ. 1992. Ethylene in plant biology. 2nd ed. San Diego, CA: Academic Press. p. 414. 2 Khan NA , FerranteA, Munné-BoschS. 2022. The plant hormone ethylene: stress acclimation and agricultural applications. Cambridge (MA): Elsevier. p. 248. 3 Mahajan PV , CalebOJ, SinghZ, WatkinsCB, GeyerM. 2014. Postharvest treatments of fresh produce. Philos Trans A Math Phys Eng Sci. 372 :20130309. 4 Bleecker AB , EstelleMA, SomervilleC, KendeH. 1988. Insensitivity to ethylene conferred by a dominant mutation in Arabidopsis thaliana. Science 241 :1086–1089.17747490 5 Chang C , KwokSF, BleeckerAB, MeyerowitzEM. 1993. Arabidopsis ethylene-response gene ETR1: similarity of product to two-component regulators. Science 262 :539–544.8211181 6 Kieber JJ , RothenbergM, RomanG, FeldmanKA, EckerJR. 1993. CTR1, a negative regulator of the ethylene response pathway in Arabidopsis, encodes a member of the Raf family of protein kinases. Cell 72 :427–441.8431946 7 Alonso JM , HirayamaT, RomanG, NourizadehS, EckerJR. 1999. EIN2, a bifunctional transducer of ethylene and stress responses in Arabidopsis. Science 284 :2148–2152.10381874 8 Binder BM . 2020. Ethylene signaling in plants. J Biol Chem. 295 :7710–7725.32332098 9 Potuschak T , et al 2003. EIN3-dependent regulation of plant ethylene hormone signaling by two Arabidopsis F box proteins: EBF1 and EBF2. Cell 115 :679–689.14675533 10 Guo HW , EckerJR. 2003. Plant responses to ethylene gas are mediated by SCF (EBF1/EBF2)-dependent proteolysis of EIN3 transcription factor. Cell 115 :667–677.14675532 11 Gagne JM , et al 2004. Arabidopsis EIN3-binding F-box 1 and 2 form ubiquitin-protein ligases that repress ethylene action and promote growth by directing EIN3 degradation. Proc Natl Acad Sci U S A. 101 :6803–6808.15090654 12 Alonso JM , et al 2003. Five components of the ethylene-response pathway identified in a screen for weak ethylene-insensitive mutants in Arabidopsis. P Natl Acad Sci U S A. 100 :2992–2997. 13 Chao QM , et al 1997. Activation of the ethylene gas response pathway in Arabidopsis by the nuclear protein ETHYLENE-INSENSITIVE3 and related proteins. Cell 89 :1133–1144.9215635 14 Solano R , StepanovaA, ChaoQM, EckerJR. 1998. Nuclear events in ethylene signaling: a transcriptional cascade mediated by ETHYLENE-INSENSITIVE3 and ETHYLENE-RESPONSE-FACTOR1. Genes Dev. 12 :3703–3714.9851977 15 Liu X , et al 2017. EIN3 and PIF3 form an interdependent module that represses chloroplast development in buried seedlings. Plant Cell 29 :3051–3067.29114016 16 Zhang X , et al 2018. Integrated regulation of apical hook development by transcriptional coupling of EIN3/EIL1 and PIFs in Arabidopsis. Plant Cell 30 :1971–1988.30104405 17 Zhong S , et al 2012. A molecular framework of light-controlled phytochrome action in Arabidopsis. Curr Biol. 22 :1530–1535.22818915 18 Zhong S , et al 2014. Ethylene-orchestrated circuitry coordinates a seedling's response to soil cover and etiolated growth. Proc Natl Acad Sci U S A. 111 :3913–3920.24599595 19 Zhong S , et al 2009. EIN3/EIL1 cooperate with PIF1 to prevent photo-oxidation and to promote greening of Arabidopsis seedlings. Proc Natl Acad Sci U S A. 106 :21431–21436.19948955 20 Guzmán P , EckerJR. 1990. Exploiting the triple response of Arabidopsis to identify ethylene-related mutants. Plant Cell 2 :513–523.2152173 21 Ceusters J , Van de PoelB. 2018. Ethylene exerts species-specific and age-dependent control of photosynthesis. Plant Physiol. 176 :2601–2612.29438047 22 Smalle J , HaegmanM, KurepaJ, Van MontaguM, StraetenDVD. 1997. Ethylene can stimulate Arabidopsis hypocotyl elongation in the light. P Natl Acad Sci U S A. 94 :2756–2761. 23 Zhang Y , LiuZ, ChenY, HeJ-X, BiY. 2015. PHYTOCHROME-INTERACTING FACTOR 5 (PIF5) positively regulates dark-induced senescence and chlorophyll degradation in Arabidopsis. Plant Sci. 237 :57–68.26089152 24 Ueda H , et al 2020. Genetic interaction among phytochrome, ethylene and abscisic acid signaling during dark-induced senescence in Arabidopsis thaliana. Front Plant Sci. 11 :564.32508856 25 Wu Q , et al 2020. Allosteric deactivation of PIFs and EIN3 by microproteins in light control of plant development. Proc Natl Acad Sci U S A. 117 :18858–18868.32694206 26 Xie Y , et al 2021. Arabidopsis FHY3 and FAR1 function in age gating of leaf senescence. Front Plant Sci. 12 :770060. 27 Sakuraba Y , et al 2014. Phytochrome-interacting transcription factors PIF4 and PIF5 induce leaf senescence in Arabidopsis. Nat Commun. 5 :4636.25119965 28 Jeong J , et al 2016. Phytochrome and ethylene signaling integration in Arabidopsis occurs via the transcriptional regulation of genes co-targeted by PIFs and EIN3. Front Plant Sci. 7 :1055.27486469 29 van Gelderen K , KangC, PierikR. 2018. Light signaling, root development, and plasticity. Plant Physiol. 176 :1049–1060.28939624 30 Muday G , RahmanA, BinderBM. 2012. Auxin and ethylene: collaborators or competitors? Trends Plant Sci. 17 :181–195.22406007 31 Ulmasov T , MurfettJ, HagenG, GuilfoyleTJ. 1997. Aux/1AA proteins repress expression of reporter genes containing natural and highly active synthetic auxin response elements. Plant Cell. 9 :1963–1971.9401121 32 Macgregor DR , DeakKI, IngramPA, MalamyJE. 2008. Root system architecture in Arabidopsis grown in culture is regulated by sucrose uptake in the aerial tissues. Plant Cell 20 :2643–2660.18952782 33 Salas J , SalasM, ViñuelaE, SolsA. 1965. Glucokinase of rabbit liver. J Biol Chem. 240 :1014–1018.14284695 34 Jang JC , SheenJ. 1994. Sugar sensing in higher plants. Plant Cell 6 :1665–1679.7827498 35 Jang JC , LeónP, ZhouL, SheenJ. 1997. Hexokinase as a sugar sensor in higher plants. Plant Cell 9 :5–19.9014361 36 Jabben M , DeitzerGF. 1979. Effects of the herbicide san 9789 on photomorphogenic responses. Plant Physiol. 63 :481–485.16660752 37 Kobayashi K , et al 2007. LOVASTATIN INSENSITIVE 1, a novel pentatricopeptide repeat protein, is a potential regulatory factor of isoprenoid biosynthesis in Arabidopsis. Plant Cell Physiol. 48 :322–331.17213228 38 Rodriéguez-Concepcioén M , et al 2004. Distinct light-mediated pathways regulate the biosynthesis and exchange of isoprenoid precursors during Arabidopsis seedling development. Plant Cell 16 :144–156.14660801 39 Russell AW , et al 1995. Photosystem II regulation and dynamics of the chloroplast D1 protein in Arabidopsis leaves during photosynthesis and photoinhibition. Plant Physiol. 107 :943–952.12228414 40 Deng XW , CasparT, QuailPH. 1991. Cop1: a regulatory locus involved in light-controlled development and gene expression in Arabidopsis. Genes Dev. 5 :1172–1182.2065972 41 Khumsupan P , et al 2020. Generating and characterizing single- and multigene mutants of the Rubisco small subunit family in Arabidopsis. J Exp Bot. 71 :5963–5975.32734287 42 Barratt DHP , et al 2009. Normal growth of Arabidopsis requires cytosolic invertase but not sucrose synthase. Proc Natl Acad Sci U S A. 106 :13124–13129.19470642 43 Ashburner M , et al 2000. Gene Ontology: tool for the unification of biology. Nat Genet. 25 :25–29.10802651 44 van Veen H , et al 2013. Two Rumex species from contrasting hydrological niches regulate flooding tolerance through distinct mechanisms. Plant Cell 25 :4691–4707.24285788 45 Vwioko E , AdinkwuO, El-EsawiMA. 2017. Comparative physiological, biochemical, and genetic responses to prolonged waterlogging stress in okra and maize given exogenous ethylene priming. Front Physiol. 8 :632.28993735 46 Huang Y-C , YehT-H, YangC-Y. 2019. Ethylene signaling involves in seeds germination upon submergence and antioxidant response elicited confers submergence tolerance to rice seedlings. Rice 12 :23.30972510 47 Kępczyński J . 2021. Gas-priming as a novel simple method of seed treatment with ethylene, hydrogen cyanide or nitric oxide. Acta Physiologiae Plantarum. 43 :117. 48 Hussain S , et al 2020. Ethylene response of salt stressed rice seedlings following Ethephon and 1-methylcyclopropene seed priming. Plant Growth Reg. 92 :219–231. 49 Hartman S , et al 2019. Ethylene-mediated nitric oxide depletion pre-adapts plants to hypoxia stress. Nat Commun. 10 :4020.31488841 50 Peng J , et al 2014. Salt-induced stabilization of EIN3/EIL1 confers salinity tolerance by deterring ROS accumulation in Arabidopsis. PLoS Genet. 10 :e1004664. 51 Tscharntke T , ThiessenS, DolchR, BolandW. 2001. Herbivory, induced resistance, and interplant signal transfer in Alnus glutinosa. Biochem Syst Ecol. 29 :1025–1047. 52 Bailey BA , DeanJF, AndersonJD. 1990. An ethylene biosynthesis-inducing endoxylanase elicits electrolyte leakage and necrosis in Nicotiana tabacum cv Xanthi leaves. Plant Physiol. 94 :1849–1854.16667926 53 Nascimento WM , CantliffeDJ, HuberDJ. 2004. Ethylene evolution and endo-β-mannanase activity during lettuce seed germination at high temperature. Sci Agric. 61 :156–163. 54 Liu Z , et al 2022. Ethylene augments root hypoxia tolerance via growth cessation and reactive oxygen species amelioration. Plant Physiol. 190 :1365–1383.35640551 55 Wang LH , et al 2019. Effects of exogenous glucose and sucrose on photosynthesis in triticale seedlings under salt stress. Photosynthetica 57 :286–294. 56 Hernández-Madrigal F , et al 2018. Sucrose protects Arabidopsis roots from chromium toxicity influencing the auxin–plethora signaling pathway and improving meristematic cell activity. J Plant Growth Regul. 37 :530–538. 57 da Silva AC , et al 2020. The Yin and Yang in plant breeding: the trade-off between plant growth yield and tolerance to stresses. Biotechnol Res Innovation. 3 :73–79. 58 Kromdijk J , et al 2016. Improving photosynthesis and crop productivity by accelerating recovery from photoprotection. Science 354 :857–861.27856901 59 Wei S , et al 2022. A transcriptional regulator that boosts grain yields and shortens the growth duration of rice. Science 377 :eabi8455. 60 De Souza AP , et al 2022. Soybean photosynthesis and crop yield are improved by accelerating recovery from photoprotection. Science 377 :851–854.35981033 61 Kircher S , SchopferP. 2012. Photosynthetic sucrose acts as cotyledon-derived long-distance signal to control root growth during early seedling development in Arabidopsis. Proc Natl Acad Sci U S A. 109 :11217–11221.22733756 62 Freixes S , ThibaudM-C, TardieuF, MullerB. 2002. Root elongation and branching is related to local hexose concentration in Arabidopsis thaliana seedlings. Plant Cell Environ. 25 :1357–1366. 63 Mishra BS , SinghM, AggrawalP, LaxmiA. 2009. Glucose and auxin signaling interaction in controlling Arabidopsis thaliana seedlings root growth and development. PLoS One 4 :e4502.19223973 64 Alscher RG , CastelfrancoPA. 1972. Stimulation by ethylene of chlorophyll biosynthesis in dark-grown cucumber cotyledons. Plant Physiol. 50 :400–403.16658183 65 Hart JE , GardnerKH. 2021. Lighting the way: recent insights into the structure and regulation of phototropin blue light receptors. J Biol Chem. 296 :100594. 66 Lin C . 2002. Blue light receptors and signal transduction. Plant Cell 14 :S207–S225.12045278 67 Łabuz J , SztatelmanO, HermanowiczP. 2022. Molecular insights into the phototropin control of chloroplast movements. J Exp Bot. 73 :6034–6051.35781490 68 Losi A , GärtnerW. 2012. The evolution of flavin-binding photoreceptors: an ancient chromophore serving trendy blue-light sensors. Ann Rev Plant Biol. 63 :49–72.22136567 69 Li X , LiangT, LiuH. 2021. How plants coordinate their development in response to light and temperature signals. Plant Cell 34 :955–966. 70 Barnes WJ , AndersonCT. 2018. Cytosolic invertases contribute to cellulose biosynthesis and influence carbon partitioning in seedlings of Arabidopsis thaliana. Plant J. 94 :956–974.29569779 71 Pignocchi C , et al 2020. Restriction of cytosolic sucrose hydrolysis profoundly alters development, metabolism, and gene expression in Arabidopsis roots. J Exp Bot. 72 :1850–1863. 72 Margalha L , ConfrariaA, Baena-GonzálezE. 2019. SnRK1 and TOR: modulating growth–defense trade-offs in plant stress responses. J Exp Bot. 70 :2261–2274.30793201 73 Thalmann M , SanteliaD. 2017. Starch as a determinant of plant fitness under abiotic stress. New Phytol. 214 :943–951.28277621 74 Shi H , et al 2016. Seedlings transduce the depth and mechanical pressure of covering soil using COP1 and ethylene to regulate EBF1/EBF2 for soil emergence. Curr Biol. 26 :139–149.26748855 75 Pandey BK , et al 2021. Plant roots sense soil compaction through restricted ethylene diffusion. Science 371 :276–280.33446554 76 Harpham NVJ , et al 1991. The effect of ethylene on the growth and development of wild-type and mutant Arabidopsis thaliana (L.) Heynh. Ann Bot. 68 :55–61. 77 Hussain A , BlackC, TaylorI, RobertsJA. 1999. Soil compaction. A role for ethylene in regulating leaf expansion and shoot growth in tomato? Plant Physiol. 121 :1227–1237.10594109 78 Goeschl JD , RappaportL, PrattHK. 1966. Ethylene as a factor regulating the growth of pea epicotyls subjected to physical stress. Plant Physiol. 41 :877–884.16656334 79 Brenya E , et al 2022. Mechanical stress acclimation in plants: linking hormones and somatic memory to thigmomorphogenesis. Plant Cell Environ. 45 :989–1010.34984703 80 Huang J-P , Tunc-OzdemirM, ChangY, JonesAM. 2015. Cooperative control between AtRGS1 and AtHXK1 in a WD40-repeat protein pathway in Arabidopsis thaliana. Front Plant Sci. 6 :851.26528314 81 Leivar P , et al 2008. Multiple phytochrome-interacting bHLH transcription factors repress premature seedling photomorphogenesis in darkness. Curr Biol. 18 :1815–1823.19062289 82 Binder BM , RodriguezFI, BleeckerAB, PattersonSE. 2007. The effects of group 11 transition metals, including gold, on ethylene binding to the ETR1 receptor and growth of Arabidopsis thaliana. FEBS Lett. 581 :5105–5109.17931631 83 Binder BM , et al 2004. Arabidopsis seedling growth response and recovery to ethylene. A kinetic analysis. Plant Physiol. 136 :2913–2920.15466220 84 Duncan AJ . 1974. Cumulative sum control charts. Quality control and industrial statistics. New York: Wiley. p. 464–482. 85 Lichtenthaler HK . 1987. Chlorophylls and carotenoids: pigments of photosynthetic biomembranes. In: PackerL, DouceR, editors. Methods in enzymology. Cambridge (MA): Academic Press. p. 350–382. 86 Pang Z , et al 2022. Using MetaboAnalyst 5.0 for LC–HRMS spectra processing, multi-omics integration and covariate adjustment of global metabolomics data. Nat Protoc. 17 :1735–1761.35715522 87 Jin Y , LiuF, HuangW, SunQ, HuangX. 2019. Identification of reliable reference genes for qRT-PCR in the ephemeral plant Arabidopsis pumila based on full-length transcriptome data. Sci Rep. 9 :8408.31182737 88 Pfaffl MW . 2001. A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Res. 29 :e45.11328886 89 Edgar R , DomrachevM, LashAE. 2002. Gene expression omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res. 30 :207–210.11752295 90 Silva-Correia J , FreitasS, TavaresRM, Lino-NetoT, AzevedoH. 2014. Phenotypic analysis of the Arabidopsis heat stress response during germination and early seedling development. Plant Methods 10 :7.24606772 91 Li S , et al 2014. HEAT-INDUCED TAS1 TARGET1 mediates thermotolerance via HEAT STRESS TRANSCRIPTION FACTOR A1a-directed pathways in Arabidopsis. Plant Cell 26 :1764–1780.24728648 92 Lokdarshi A , ConnerWC, McClintockC, LiT, RobertsDM. 2016. Arabidopsis CML38, a calcium sensor that localizes to ribonucleoprotein complexes under hypoxia stress. Plant Physiol. 170 :1046–1059.26634999 93 Beamer ZG , et al 2021. Aquaporin family lactic acid channel NIP2; 1 promotes plant survival under low oxygen stress in Arabidopsis. Plant Physiol. 187 :2262–2278.34890456

PMC010xxxxxx/PMC10353740.txt

==== Front Invest Ophthalmol Vis Sci Invest Ophthalmol Vis Sci IOVS Investigative Ophthalmology & Visual Science 0146-0404 1552-5783 The Association for Research in Vision and Ophthalmology 37440261 10.1167/iovs.64.10.14 IOVS-23-37549 Review Review Determining a Worldwide Prevalence of Oculocutaneous Albinism: A Systematic Review Worldwide Prevalence of Albinism Kromberg Jennifer G. R. 1 Flynn Kaitlyn A. 1 Kerr Robyn A. 1 1 Division of Human Genetics, National Health Laboratory Service and School of Pathology, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa * Correspondence: Jennifer G. R. Kromberg, Division of Human Genetics, School of Pathology, Faculty of Health Sciences, University of the Witwatersrand and National Health Laboratory Service, PO Box 1038, Johannesburg 2000, South Africa; jengrace130@gmail.com. 13 7 2023 7 2023 64 10 1412 6 2023 01 5 2023 Copyright 2023 The Authors 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. Purpose The aim of this systematic review was to investigate the available data on the epidemiology of oculocutaneous albinism (OCA) around the world, and to determine whether a generalizable, worldwide prevalence figure could be proposed. Methods Extensive literature search strategies were conducted, interrogating PubMed, Scopus, and Web of Science, to locate relevant literature. Ultimately 34 studies reporting original data were included for analysis. Results Findings showed that most data were outdated, and only 6 of 34 articles (18%) were published after 2010. There were few good studies with sound methodology and large, clearly defined population samples. Only a small proportion of countries worldwide (26/193 [13%]) have produced prevalence figures for OCA. By continent, African studies were disproportionately represented (15/34 [44%]). The highest prevalence rates (range, 1 in 22 to 1 in 1300; mean, 1 in 464) were reported in population isolates. The mean prevalence from four African countries was 1 in 4264 (range, 1 in 1755 to 1 in 7900). Prevalence for three countries in Europe (mean, 1 in 12,000; range, 1 in 10,000 to 1 in 15,000) may be underestimated, as the phenotype, in fair-skinned populations, may be missed or misdiagnosed as ocular albinism or isolated visual impairment. Population rates may vary depending on local cultural factors (e.g., consanguineous matings) and may change over time. Conclusions The prevalence of OCA varies widely between continents and population groups, and it is often influenced by local factors. It was not possible, therefore, to determine a single, generalizable worldwide prevalence rate for OCA, although continental rates for Africa and Europe are useful. albinism Africa epidemiology frequency prevalence worldwide ==== Body pmcThe use of data, including prevalence data, is an essential component for decision-making within health care systems. Prevalence data, specifically, can be used to assess the burden of a disorder, influence health care policy, encourage dialogue and raise public awareness, in any country, so appropriate research is both necessary and worthwhile. Oculocutaneous albinism (OCA) is a group of inherited pigment disorders in which melanin biosynthesis is decreased or absent.1 OCA occurs in populations all over the world at varying frequencies,2,3 but there are few good prevalence studies.4 Several types are described (OCA1–7), each being caused by mutations in a different gene.5 Phenotypically, the eyes6 and skin7 are affected, resulting in poor visual acuity8 and a predisposition to skin cancer. Intellectual maturity has been shown to be within the normal range.9 Squamous and basal cell carcinomas occur, mostly on the head and neck areas,10 and are the cause of mortality in some individuals with OCA. In Africa, myths and superstitions surround the condition and this has led to human rights violations and erroneous beliefs regarding the cause of the disorder with, in certain instances, fatal consequences.11 OCA, specifically OCA2, is the most common Mendelian autosomal recessive condition in southern Africa, occurring with a prevalence of approximately 1 in 4000 African individuals.12,13 Prevalence figures for OCA for the rest of sub-Saharan Africa (central, East, and West Africa) show a relatively similar picture (however, in these regions, where malaria is endemic, sickle cell anemia is the most common autosomal recessive condition, owing to positive selection for the HbS beta globin mutation). Only one case of OCA1 has been described in Africa14; however, this type of OCA occurs relatively frequently in Europe (e.g., in Denmark 44% of tested cases had OCA1).15 Two European studies have found that OCA occurs in approximately 1 in 14,000 people in Denmark15 and 1 in 12,000 in the Netherlands.16 Although much of the worldwide prevalence data for OCA were generated some time ago, their publication led to an appreciation of the frequency of albinism in Africa. This in turn led to the establishment of clinics, parent support groups and advocacy groups, and, further, directly focused academic research, over several decades, including research on prevalence rates. One worldwide figure that could be applied to countries without their own data might be useful. Presently many countries, where no prevalence studies have been performed, sometimes rely on the old estimate of 1 in 17,000.17 This figure was obtained from a study carried out on a mixed ancestry population (mostly of European and African descent) living in North Carolina and may not accurately reflect the prevalence rate of populations from different ancestries. Examining the worldwide prevalence data will help to clarify the epidemiology of OCA, avoid under- or over-estimating its frequency in specific populations, and can lead to the provision of improved and more reliable information to local health departments. The present study aims to establish the prevalence rates for OCA worldwide. The objectives are to investigate relevant data published in the literature over time, to establish for which countries and populations information exists, to assess and compare findings in and between countries and, finally, to determine whether it is possible to arrive at a useful, broad, worldwide population-based estimate of prevalence. Methods The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines were followed for this systematic review.18 Protocol and Registration At the outset, this project was not registered with the international Prospective Register of Systematic Reviews (PROSPERO), and there were no registered reviews on PROSPERO relating to the epidemiology of albinism at the time. The project protocol was developed by the three authors together during the planning phase of the project. Search Strategy The search of the literature to retrieve relevant articles was conducted on 26 March 2021 across three databases: PubMed, Scopus, and Web of Science. No filters applied restrictions to language or date of publication. Search Builder version 1.0 was used to set-up the search query.19 This query included the following terms: “albinism,” “oculocutaneous albinism,” “prevalence,” “frequency,” “rate of occurrence,” “epidemiology,” “incidence,” “birth rate,” and “community rate.” Study Selection Rayyan software20 was used for article management and study selection. Duplicated articles were scanned for and deleted by KF. Abstracts were reviewed and assessed by KF, RK, and JK, and studies that were not relevant were removed. All remaining articles were assessed through full text by KF, RK, and JK. Articles found to provide evidence relating to the investigation of this study were approved for data extraction. Additional articles referenced in the approved articles, that were not found in the initial data search, and relevant chapters in books, were identified and subsequently included into the study if they contained original data (e.g., Kromberg, 2018,3 Lund and Roberts, 201821). During the data extraction process articles were subjected to the Joanna Briggs Institute (JBI) critical appraisal checklist for studies reporting prevalence data.22 Data Collection Process and Data Items The following data were extracted (where possible) for all articles included in the study by KF and JK: The first author and year of publication; the country and/or city in which the study was conducted; the sample or population size; the sex ratio of the samples; the infrastructure through which the study was conducted (e.g., hospital, school, survey); any reported mating patterns in the population or community; the setting of the population: rural or urban; any reported statistics or observations regarding the lifespan of the participants; the type of albinism reported; prevalence as reported by the authors; frequency as reported by the authors; any reported cultural incidences of note, such as occupation, age distribution, stigma, and treatment of participants within their communities; and any additional comments. This project followed all the ethical standards for research without direct contact with human or animal subjects. Results Altogether, 1680 articles were identified during the literature search of the three chosen databases. The article selection steps are illustrated in the Figure. Filtering included the removal of duplicates; removal based on abstracts (i.e., the article abstract indicated the absence of prevalence data); removal based on full text reads (i.e., the article did not contain relevant prevalence data upon reading the full text of the article); and removal based on poor article quality as identified by the JBI critical appraisal checklist. Figure. Flow chart illustrating the procedure of source selection. After selection, 32 articles remained, and these were screened for the required data on prevalence. In addition to these 32 articles, 2 book chapters containing relevant and original prevalence data were identified, and these were included for downstream data extraction. Ultimately, the total number of included sources was 34. Information extracted from the 34 sources provided data across 6 continents with some better represented than others (Table). The sources, dated between 1913 and 2018, were mostly outdated and only 6 of the 34 (18%) were published after 2010. Table. Albinism Prevalence Data Categorized by Continent Location Population Description Prevalence Reference (Authors, Year) Australia: Oceania  Polynesia Polynesians in Tuvalu 1 in 669 Johanson et al. 201024 Africa  Botswana Tswana village (isolate) 1 in 1307 Kromberg 20183  Cameroon Bamileke group 1 in 11,900 Aquaron 198025  Bamileke group 1 in 7900 Aquaron 199026,*  Cameroon 1 in 28,000 Aquaron 199026  Egypt Pediatric Hospital patients 1 in 5843 Mohamed et al. 201027  Namibia National Census 1 in 1755 Lund & Roberts 201821,*  Nigeria Lagos city 1 in 2858 Barnicot 195228 Benin city 1 in 7000 Barnicot 195228 East Central State 1 in 15,000 Okoro 197529 Ibo (population isolate) 1 in 1100 Okoro 197529  South Africa Eastern Cape province 1 in 3759 Oettle 196330 Eastern Cape province 1 in 3000 Rose 197431 Soweto, Johannesburg 1 in 3900 Kromberg & Jenkins 198212 Vhavhenda population 1 in 2239 Lund et al. 200732  Swaziland Hhohho district 1 in 1900 Kromberg 20183  Tanzania Hospital survey 1 in 1400 Luande 198533 National Census 1 in 2673 Lund & Roberts 201821,*  Zimbabwe School children in Harare 1 in 2883 Kagore & Lund 199534 School children 1 in 4728 Lund 199613,* Tonga (population isolate) 1 in 1000 Lund et al. 199723 Asia  Borneo Iban population 1 in 5−10,000 Abrahams 197235  Japan Old estimate 1 in 47,000 Neel et al. 194936 Europe  Denmark National register 1 in 14,000 Gronskov et al. 200915,*  France Old estimate (1910) 1 in 10−20,000 Pearson et al. 19132  Germany Old estimate (1910) 1 in 20,000 Pearson et al. 19132  Italy Old estimate 1 in 23−26,000 Pearson et al. 19132  The Netherlands General population 1 in 15,000 Van Dorp 198737,* Medical Centre patients 1 in 12,000 Kruijt et al. 201816  Northern Ireland Whole population 1 in 10,000 Froggatt 196038,*  Norway Old estimate (1898) 1 in 25,000 Pearson et al. 19132  Russia Old estimate (1909) 1 in 100,000 Pearson et al. 19132  Scotland Old estimate 1 in 15−25,000 Pearson et al. 19132  Spain Spanish gypsies/Gutanos, 1 in 1200 Gamella & Nunez- Negrillo 201339 SE Spain (isolate) Spanish gypsies (isolate) 1 in 3300 Gamella & Nunez-Negrillo 201339 Non-gypsy population 1 in 50,000 Martinez-Frias & Bermejo 199240 North America  Canada British Columbia 1 in 20,600 McLeod & Lowry 197541 United States of America  Arizona Hopi Indians (isolate) 1 in 227 Woolf 196542 Hopi Indians (isolate) 1 in 231 Woolf & Dukepoo 196943 Navajo 1 in 3750 Woolf 196542 1 in 1500−2000 Yi et al. 200344  New Mexico Jemez (isolate) 1 in 140 Woolf 196542 Laguna (isolate) 1 in 2000 Woolf 196542 San Juan (isolate) 1 in 500 Woolf 196542 Zuni (isolate) 1 in 247 Woolf 196542  North Carolina State population 1 in 17,000 Witkop et al. 197017 Central America  Panama San Blas, Cuna (isolate) 1 in 210 Stout 194645 San Blas, Cuna (isolate) 1 in 167 Keeler 196446 South America  Brazil Caingang Indians (isolate) 1 in 27 Salzano 196147 North Brazilian Island (isolate) 1 in 22 Freire-Maia et al. 197848 * African and European studies with baseline population samples >1,000,000. Altogether 15 of the 34 sources (44%), including the 2 book chapters,3,21 were from the African continent. These 15 sources provided prevalence statistics across 9 African countries (6 were in southern Africa). The second highest output was from Europe with seven articles reporting rates across 10 countries, followed by North America with 5 articles, South and Central America with 4 articles, Asia with 2 articles, and Australia-Oceania with only 1 article. Lack of any original published prevalence data from India and China was noted. Population isolates accounted for a large proportion (15/34 [44%]) of the extracted sources. These isolates were defined by the authors as populations that were geographically and culturally isolated from neighboring populations, within or between countries, and were therefore highly unlikely to interact and/or reproduce with individuals from neighboring areas. Data from such isolates showed that they had higher rates of albinism than any of the countries reporting prevalence rates. Where both a large general population sample and a population isolate was studied in the same country, exemplified by two Zimbabwe studies,13,23 the isolate showed a significantly higher prevalence than the general population, as expected. In the four African countries with studies that had adequate methodology and large population samples (>1,000,000), the mean prevalence rate was 1 in 4264, with a range of 1 in 1755 to 1 in 7900. In the three European countries with studies with adequate methodology and similar large samples (>1,000,000), the mean rate was 1 in 13,000 (range, 1 in 10,000 to 1 in 15,000). The Table shows that prevalence data are only available for a few countries (26/193 [13%]) and many areas around the world have no published prevalence data at all (with the limitation that filtering removed non-English articles). The findings also show that the reported prevalence rates of OCA vary widely in different populations, for example, 1 in 14,000 in Denmark,15 versus 1 in 1755 in Namibia,21 and even in the same population over time, for example, over 31 years in the Netherlands.16,37 Discussion Countries With Data The data obtained in this review from the different countries were based on research of varying quality, reliability, and validity. In some cases, information was obtained from general observations in hospital clinics,27 whereas in others, data were based on clinical research and counting of health center cases with an educated guess as to the value of the denominator,33 and only a few were based on sound research principles and large community-based samples.12,21 Nevertheless, what can be deduced is that albinism (usually OCA2)49 has a higher reported prevalence rate in individuals of African ancestry than in those of European origin. Further, albinism is more common in isolates than in large population groups, as might be expected for a recessive condition in small and inbred populations. There is some debate that the low rates reported in European countries might be inaccurate50,51 because the phenotype is not as easily distinguishable in the general population and therefore not as well-recognized. Affected children might be diagnosed as having ocular albinism (OA) or low vision rather than OCA. For example, the findings in two Irish studies, completed 50 years apart, when compared, show the prevalence in Northern Ireland to be higher (1 in 10,000 reported in 1960 vs. 1 in 6600 reported in 2014) once the children with albinism in eye clinics and schools for the partially sighted are diagnosed and considered.38,50 The outdated statistics presented in 1913, in a monograph by Pearson et al.,2 for countries such as Russia and Norway, are educated guesses at best. They are based on estimates made by doctors working in various countries who were contacted by Pearson and his colleagues and asked to estimate the prevalence of albinism. These estimates are presented here because no research has been undertaken, and published, to produce better prevalence figures in those countries. The highest rates of albinism in the world are found in population isolates. Two Brazilian isolates showed particularly high rates (1 in 22 on a small coastal island,48 and 1 in 27 among the Caingang Indians living in reservations in the South-East of Brazil),47 followed by the isolates in New Mexico (1 in 140)42 and Arizona (1 in 227 in the Hopi Indians).42 These isolated populations exist within geographical barriers and have been living and reproducing (with endogamous mating patterns) within these confines for several generations. For example, the 5000 Hopi Indians in Arizona live on the flat tops of three mesas, where they settled several centuries ago. There are data from two isolated populations in Africa: one group live in a secluded Zambezi River valley in Northern Zimbabwe (Tonga population of 11,000 people, rate 1 in 1000)23 and the other in a large remote village in Botswana (Tswana population of 18,000, rate 1 in 1300).3 The population size of these African isolates is larger than those found in the Americas, and consequently the OCA rates are lower. There is more prevalence information from countries in Africa than from countries on other continents. This is partly because albinism is a more obvious disorder in darker skinned populations, than in lighter skinned peoples living elsewhere in the world. Second, albinism is surrounded by myths and superstitions in Africa, which made it of interest, not only to the early explorers, medical doctors, geneticists, scientists, and epidemiologists, but also to social scientists such as psychologists and anthropologists. Third, a large-scale, long-term, wide-ranging research project on albinism was carried out over four decades in southern Africa, resulting in many novel findings and publications on the condition.52 The data presented in most studies in this review were collected from urban based samples. However, rates derived from urban and rural studies often differ. For example, the Namibia census data showed a rural rate of 1 in 1459, whereas the urban rate was 1 in 2409.21 Factors such as cultural habits, mating patterns, migration, founder effect, and gene frequencies, as well as community attitudes, can affect prevalence rates, so that country-wide data cannot be extrapolated from data collected at a specific local site. Changes in Prevalence Over Time Research suggests that reported prevalence figures can change over time. Two studies from the Netherlands, undertaken 31 years apart, support this finding.16,37 Similarly, Grønskov et al.15,53 suggest that there has been an increased prevalence of albinism (their samples include both OCA and some OA cases) in Denmark over time. It is suggested that changes in prevalence rates are actually due to better and more stringent diagnostic criteria, improved case finding and medical care, and (most likely) better methods of selecting cases (Kruijt, personal communication, 2022), determining prevalence and data collection. Undertaking molecular studies on suspected cases, that might have been misdiagnosed or missed on clinical examination, shows that more individuals are affected with albinism than initially thought.51 Determining a Worldwide Prevalence Rate It is seemingly not possible to determine a worldwide prevalence rate that could be applied in those countries where rates are not available. Rates seem to be markedly different between populations of African versus non-African descent. Further, local psychosocial and cultural factors differ and can affect prevalence rates. Such factors include mating patterns, negative attitudes to albinism (leading to infanticide, rejection, stigmatization, and neglect),54 positive attitudes (leading to preferential treatment; e.g., not being obliged to work),42 and geographical factors, such as numbers of sunny days, limited cloud cover and high UV risks. For example, in Israel, the OCA prevalence rate of 1 in 7000 (higher than in many other European populations) has been associated with local cultural practices that favor consanguineous marriages (Blumenfeld, personal communication, 2022). Furthermore, the prevalence could be affected by local health management infrastructure and the availability of services. Where these services are adequate and efficient, as in first-world countries, the complications of the condition can be managed effectively, and it becomes less disabling. Continental prevalence differences are apparent in the findings from this systematic review. Africa has the highest prevalence, with an average of approximately 1 in 4000 for OCA in the scattered countries in which studies have been performed. From the few available European studies, it can be estimated that the rate is about three times lower, averaging at approximately 1 in 13,000. It has been proposed that this rate is an underestimate and molecular studies are supporting this suggestion.51 Further, OA may be diagnosed in some European children rather than OCA, unless molecular studies are performed to confirm the actual diagnosis. The data from the Far East are limited, but the estimates from China and Japan36,55 suggest that the rates there are lower than those in Europe. In the abstract (in English) of Gong et al.55 (the paper was excluded from the present review owing to language of publication), the authors suggested that the prevalence rate in the Han population in the Shandong province of China was 1 in 18,000. Neel et al.36 studied consanguineous mating in Japan and from the data, on “induced recessive mutations” in their sample, they estimated the rate of albinism in that population to be 1 in 47,000. Findings Highlight Limitations of the Data The main limitation on the data used in this review is the lack of good, reliable, published studies on the prevalence of albinism in various countries around the world. Most of the articles identified for attention were outdated (28/34 or 82%, being published before 2010) and very few recent studies have been undertaken. Prevalence data that are ≥10 years old may no longer be valid (prevalence rates were found to be higher by approximately 25% over 31 years in the Netherlands).16 Also, case finding in some European studies included both individuals with OCA and OA (no cases of OA have been reported in African populations, as far as the authors are aware), so that rates in different countries may not be strictly comparable. A further limitation is associated with the fact that only English language publications were searched. Reports in other languages may have been missed (e.g., Gong et al.55 in Chinese), as may those published as a letter to a journal editor (e.g., Healey et al.50), those not indexed in PubMed, Scopus, or Web of Science, and those where the key words used in the present study were not represented (e.g., Rajab et al.56). Implications of Findings The findings from this review draw attention to the fact that individuals with OCA occur in all populations. However, if countries in Africa, with no prevalence data, wish to use an estimated prevalence figure they should use the range of 1 in 4000 to 7000, rather than the previously used rate of 1 in 17,000.17 Also, in European countries for which there are no data, using a range of 1 in 12,000 to 15,000 would be more appropriate than using the prevalence rate of 1 in 17,000. The data sourced for this project were mostly published many years ago. New and better research studies on prevalence rates, with stringent methodology, improved diagnostic criteria, and large samples, are essential if the epidemiology of OCA in countries around the world is to be better understood. Further, health care professionals worldwide need to acknowledge that OCA occurs and that individuals with OCA have special needs that must be met. Offering specific and appropriate health services, such as skin and eye care, as well as genetic counseling, would improve the quality of life for people with OCA, allowing them to manage their condition and reach their potential. Further, community education on prevalence rates is required to increase awareness and acceptance of the condition. Providing good diagnostic services, early screening and treatment, and regular monitoring for visual and skin problems, would decrease the burden on health departments of costly ophthalmology and, particularly, oncology treatment, which could otherwise be required later. Such services, particularly in developing countries, would also prevent persons with OCA becoming disabled (and unemployable) and a burden on their families and communities. Conclusions OCA needs to be recognized, wherever it occurs, and not left untreated. This is especially so in Africa with its high prevalence rates, high risks of skin cancer, and significant human rights violations in certain countries. Skin damage and compromised vision have significant health, social, educational, and financial consequences for the person with OCA, as well as for the family and community. As discussed in the United Nations by Ero in 2019,57 it is important, in Africa, that updated prevalence data be collected and used to justify the need for appropriate health services and community education around the condition.57 In this systematic review, the prevalence of OCA worldwide has been investigated. The comprehensive literature search shows that very few reliable and valid studies have been carried out, rates of albinism in many populations are only estimates, or not available at all, and further research is necessary. The findings reinforce the previously suspected fact that albinism is more prevalent in Africa than elsewhere, and health departments need to be aware of prevalence rates and the need to provide for appropriate health services, all over the continent. Further, albinism is probably underdiagnosed in European communities and more accurate diagnostics are required if people with OCA are to be identified and benefit from the appropriate health and educational services. Appreciating that prevalence figures can be used to leverage health policy, create public awareness, and generally stimulate dialogue, the value of generating good and current data cannot be overemphasized. Acknowledgments The assistance of the National Health Laboratory Service (NHLS) and the University of the Witwatersrand in providing the first author with an Honorary Associate Professorship and an office in the Division of Human Genetics, and the meticulous work of research assistant, Dylan J. McCarthy, are acknowledged. Funding Sources: This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. Disclosure: J.G.R. Kromberg, None; K.A. Flynn, None; R.A. Kerr, None ==== Refs References 1. Summers CG. Albinism: classification, clinical characteristics, and recent findings. Optom Vis Sci. 2009; 86 (6 ): 659–662.19390472 2. Pearson K, Nettleship E, Usher CH. A monograph on albinism in man. In: Research Memoirs Biometric Series. Vol VIII . London: Drapers Co; 1913. 3. Kromberg J . Epidemiology of albinism. In: Kromberg J, Manga P, eds. Albinism in Africa: historical, geographical, medical, genetic, and psychosocial aspects. San Diego: Elsevier, Academic Press; 2018: 57–79. 4. Hong ES, Zeeb H, Repacholi MH. Albinism in Africa as a public health issue. BMC Public Health. 2006; 6 (1 ): 212.16916463 5. Aamir A, Kuht HJ, Grønskov K, Brooks BP, Thomas MG. Clinical utility gene card for oculocutaneous (OCA) and ocular albinism (OA)—an update. Eur J Hum Genet. 2021; 29 (10 ): 1577–1583.33504991 6. Williams SEI. Albinism and the eye. In: Kromberg J, Manga P, eds. Albinism in Africa: historical, geographical, medical, genetic, and psychosocial aspects. San Diego: Elsevier Academic Press; 2018: 135–149. 7. Hartshorne S, Manga P. Dermatological aspects of albinism. In: Kromberg J, Manga P, eds. Albinism in Africa: historical, geographical, medical, genetic, and psychosocial aspects. San Diego: Elsevier Academic Press; 2018: 121–134. 8. Kammer R. Visual rehabilitation and albinism. In: Kromberg J, Manga P, eds. Albinism in Africa: historical, geographical, medical, genetic and psychosocial aspects. San Diego: Elsevier, Academic Press; 2018: 151–170. 9. Manganyi NC, Kromberg JGR, Jenkins T. Studies on albinism in the South African Negro I. Intellectual maturity and body image differentiation. J Biosoc Sci. 1974; 6 (1 ): 107–112.4828335 10. Kromberg JGR, Castle D, Zwane EM, Jenkins T. Albinism and skin cancer in Southern Africa. Clin Genet. 1989; 36 (1 ): 43–52.2766562 11. Kajiru I, Mubangizi JC. Human rights violations of persons with albinism in Tanzania: the case of children in temporary holding shelters. African Human Rights Law Journal;. 2019; 19 (1 ): 246–266. 12. Kromberg JG, Jenkins T. Prevalence of albinism in the South African negro. S Afr Med J. 1982; 61 (11 ): 383–386.7064008 13. Lund PM. Distribution of oculocutaneous albinism in Zimbabwe. J Med Genet. 1996; 33 (8 ): 641–644.8863154 14. Badens C, Courrier S, Aquaron R. A novel mutation (delAACT) in the tyrosinase gene in a Cameroonian black with type 1A oculocutaneous albinism. J Dermatol Sci. 2006; 42 (2 ): 121–124.16517127 15. Grønskov K, Ek J, Sand A, et al . Birth prevalence and mutation spectrum in Danish patients with autosomal recessive albinism. Invest Ophthalmol Vis Sci. 2009; 50 (3 ): 1058.19060277 16. Kruijt CC, de Wit GC, Bergen AA, Florijn RJ, Schalij-Delfos NE, van Genderen MM. The phenotypic spectrum of albinism. Ophthalmology. 2018; 125 (12 ): 1953–1960.30098354 17. Witkop CJ, Nance WE, Rawls RF, White JG. Autosomal recessive oculocutaneous albinism in man. Evidence for genetic heterogeneity. Am J Hum Genet. 1970; 22 (1 ): 55–74.4983623 18. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred Reporting Items for Systematic Reviews and Meta-Analyses: the PRISMA statement. PLoS Med. 2009; 6 (7 ): e1000097.19621072 19. Kamdar BB, Shah PA, Sakamuri S, Kamdar BS, Oh J. A novel search builder to expedite search strategies for systematic reviews. Int J Technol Assess Health Care. 2015; 31 (1-2 ): 51–53.25989817 20. Ouzzani M, Hammady H, Fedorowicz Z, Elmagarmid A. Rayyan—a web and mobile app for systematic reviews. Syst Rev. 2016; 5 (1 ): 210.27919275 21. Lund PM, Roberts M. Prevalence and population genetics of albinism. In: Kromberg J, Manga P, eds. Albinism in Africa: historical, geographical, medical, genetic, and psychosocial aspects. San Diego: Elsevier, Academic Press; 2018: 81–98. 22. Moola S, Munn Z, Tufanaru C, et al . Systematic reviews of etiology and risk. In: Aromataris E, Munn Z, eds. JBI manual for evidence synthesis. Nashville, TN: JBI; 2020. 23. Lund PM, Puri N, Durham-Pierre D, King RA, Brilliant MH. Oculocutaneous albinism in an isolated Tonga community in Zimbabwe. J Med Genet. 1997; 34 (9 ): 733–735.9321758 24. Johanson HC, Chen W, Wicking C, Sturm RA. Inheritance of a novel mutated allele of the OCA2 gene associated with high incidence of oculocutaneous albinism in a Polynesian community. J Hum Genet. 2010; 55 (2 ): 103–111.20019752 25. Aquaron R . L'albinisme oculo-cutane au Cameroun, A propos de 216 observations. Rev Epidemiol Sante Publique. 1980; 28 (1 ): 81–88.7465914 26. Aquaron R. Oculocutaneous albinism in Cameroon. A 15-year follow-up study. Ophthalmic Paediatr Genet. 1990; 11 (4 ): 255–263.2096353 27. Mohamed AF, El-Sayed NS, Seifeldin NS. Clinico-epidemiologic features of oculocutaneous albinism in northeast section of Cairo – Egypt. EJMHG. 2010; 11 (2 ): 167–172. 28. Barnicot CA. Albinism in South-Western Nigeria. Ann Eugen. 1953; 18 (1 ): 38–73. 29. Okoro AN. Albinism in Nigeria. A clinical and social study. Br J Dermatol. 1975; 92 (5 ): 485–492.1174464 30. Oettle AG. Skin cancer in Africa. National Cancer Institute Monograph. 1963; 10 : 197–214. 31. Rose EF. Pigment anomalies encountered in the Transkei. S Afr Med J. 1974; 48 : 2345–2347.4432185 32. Lund PM, Maluleke TG, Gaigher I, Gaigher MJ. Oculocutaneous albinism in a rural community of South Africa: a population genetic study. Ann Hum Biol. 2007; 34 (4 ): 493–497.17620156 33. Luande J, Henschke CI, Mohammed N. The Tanzanian human albino skin. Natural history. Cancer. 1985; 55 (8 ): 1823–1828.3978567 34. Kagore F, Lund PM. Oculocutaneous albinism among schoolchildren in Harare, Zimbabwe. J Med Genet. 1995; 32 (11 ): 859–861.8592327 35. Abrahams PH. Albinos in Borneo. Lancet. 1972; 1 : 101–102. 36. Neel J V, Kodani M, Brewer R, Anderson RC. The incidence of consanguineous matings in Japan, with remarks on the estimation of comparative gene frequencies and the expected rate of appearance of induced recessive mutations. Am J Hum Genet. 1949; 1 (2 ): 156–178.17948392 37. Van Dorp DB. Albinism, or the NOACH syndrome. Clin Genet. 1987; 31 (4 ): 228–242.3109790 38. Froggatt P. Albinism in Northern Ireland. Ann Hum Genet. 1960; 24 (3 ): 213–238.13825323 39. Gamella JF, Carrasco-Muñoz EM, Núñez Negrillo AM. Oculocutaneous albinism and consanguineous marriage among Spanish Gitanos or Calé–a study of 83 cases. Coll Antropol. 2013; 37 (3 ): 723–734.24308209 40. Martínez-Frías ML, Bermejo E. Prevalence of congenital anomaly syndromes in a Spanish gypsy population. J Med Genet. 1992; 29 (7 ): 483–486.1640427 41. McLeod R, Lowry RB. Incidence of albinism in British Columbia (B.C.). Separation by hairbulb test. Clin Genet. 1976; 9 (1 ): 77–80.1248166 42. Woolf CM. Albinism among Indians in Arizona and New Mexico . Am J Hum Genet. 1965; 17 (1 ): 23–35.14255554 43. Woolf CM, Dukepoo FC. Hopi Indians, inbreeding, and albinism. Science (1979). 1969; 164 (3875 ): 30–37. 44. Yi Z, Garrison N, Cohen-Barak O, et al . A 122.5-kilobase deletion of the P gene underlies the high prevalence of oculocutaneous albinism type 2 in the Navajo population. Am J Hum Genet. 2003; 72 (1 ): 62–72.12469324 45. Stout DB. Further notes on albinism among the San Blas Cuna, Panama. Am J Phys Anthropol. 1946; 4 (4 ): 483–490.20281561 46. Keeler C. The incidence of the Cuna Moon-Child Albinos. J. Hered. 1964; 55 (3 ): 115–120.14170400 47. Salzano FM. Rare genetic conditions among the Gaingang Indians. Ann Hum Genet. 1961; 25 (2 ): 123–130.14496533 48. Freire-Maia N, de Andrade FL, de Athayde-Neto A, et al . Genetic investigations in a Northern Brazilian island. II. Random drift. Hum Hered. 1978; 28 (6 ): 401–410.680702 49. Stevens G, van Beukering J, Jenkins T, Ramsay M. An intragenic deletion of the P gene is the common mutation causing tyrosinase-positive oculocutaneous albinism in southern African Negroids. Am J Hum Genet. 1995; 56 (3 ): 586–591.7887411 50. Healey N, McLoone E, Saunders KJ, Jackson AJ, McClelland JF. Are worldwide albinism prevalence figures an accurate reflection? An incidental finding from a Northern Ireland study. BJO. 2014; 98 (7 ): 990.1–990. 51. Arveiler B, Michaud V, Lasseaux E. Albinism: an underdiagnosed condition. J Invest Dermatol. 2020; 140 (7 ): 1449–1451.31883962 52. Kromberg J, Manga P, (eds). Albinism in Africa. historical, geographical, medical, genetic, and psychosocial aspects. San Diego: Elsevier Academic Press; 2018. 53. Grønskov K, Ek J, Brondum-Nielsen K. Oculocutaneous albinism. Orphanet J Rare Dis. 2007; 2 (1 ): 43.17980020 54. Kromberg JGR, Kerr R. Oculocutaneous albinism in southern Africa: Historical background, genetic, clinical and psychosocial issues. Afr J Disabil. 2022; 11 : 877.36353393 55. Gong Y, Shao C, Zheng H, Chen B, Guo Y. Study on genetic epidemiology of albinism. Yi Chuan Xue Bao. 1994; 21 (3 ): 169–172.7917429 56. Rajab A, Al Rashdi I, Al Salmi Q. Genetic services and testing in the Sultanate of Oman. Sultanate of Oman steps into modern genetics. J Community Genet. 2013; 4 (3 ): 391–397.23821042 57. Ero IK. Report of the Independent Expert on the Enjoyment of Human Rights by Persons with Albinism. Geneva: United Nations; 2019.

PMC010xxxxxx/PMC10353742.txt

==== Front Transl Vis Sci Technol Transl Vis Sci Technol TVST Translational Vision Science & Technology 2164-2591 The Association for Research in Vision and Ophthalmology 37450282 10.1167/tvst.12.7.16 TVST-23-5455 Retina Retina Retinal Vessel Caliber Measurement Bias in Fundus Images in the Presence of the Central Light Reflex Retinal Vessel Caliber and Central Light Reflex Pappelis Konstantinos 1 2 Jansonius Nomdo M. 1 2 1 Department of Ophthalmology, University of Groningen, University Medical Center Groningen, Groningen, the Netherlands 2 Graduate School of Medical Sciences, Research School of Behavioural and Cognitive Neurosciences, University of Groningen, Groningen, the Netherlands * Correspondence: K. Pappelis, Department of Ophthalmology, University Medical Center Groningen, PO Box 30.001, 9700 RB, Groningen, the Netherlands. e-mail: k.pappelis@rug.nl 14 7 2023 7 2023 12 7 1616 6 2023 17 1 2023 Copyright 2023 The Authors 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. Purpose To investigate the agreement between a fundus camera and a scanning laser ophthalmoscope in retinal vessel caliber measurements and to identify whether the presence of the central light reflex (CLR) explains potential discrepancies.  Methods For this cross-sectional study, we obtained fundus camera and scanning laser ophthalmoscope images from 85 eyes of 85 healthy individuals (aged 50–65 years) with different blood pressure status. We measured the central retinal artery equivalent (CRAE) and central retinal artery vein equivalent (CRVE) with the Knudtson–Parr–Hubbard algorithm and assessed the CLR using a semiautomatic grading method. We used Bland–Altman plots, 95% limits of agreement, and the two-way mixed effects intraclass correlation coefficient for consistency [ICC(3,1)] to describe interdevice agreement. We used multivariable regression to identify factors associated with differences in between-device measurements. Results The between-device difference in CRAE (9.5  µm; 95% confidence interval, 8.0–11.1  µm) was larger than the between-device difference in CRVE (2.9  µm; 95% confidence interval, 1.3–4.5  µm), with the fundus camera yielding higher measurements (both P < 0.001). The 95% fundus camera–scanning laser ophthalmoscope limits of agreement were –4.8 to 23.9  µm for CRAE and –12.0 to 17.8  µm for CRVE. The corresponding ICCs(3,1) were 0.89 (95% confidence interval, 0.83–0.92) and 0.91 (95% confidence interval, 0.86–0.94). The between-device CRAE difference was positively associated with the presence of a CLR (P = 0.002). Conclusions Fundus cameras and scanning laser ophthalmoscopes yield correlated but not interchangeable caliber measurements. The CLR induces bias in arteriolar caliber in fundus camera images, compared with scanning laser ophthalmoscope images. Translational Relevance Refined measurements could yield better estimates of the association between retinal vessel caliber and ophthalmic or systemic disease. fundus imaging scanning laser ophthalmoscope retinal vessel caliber central light reflex hypertension ==== Body pmcIntroduction The retinal blood vessels have long been known to be implicated in major ophthalmic pathologies, such as glaucoma and age-related macular degeneration, as well as in diseases with ocular involvement, namely, diabetes and arterial hypertension.1–4 Some of their properties, such as retinal vessel caliber, have also been suggested as promising biomarkers in the evaluation of various nonophthalmic pathologies, including atherosclerosis, coronary heart disease, and neurological diseases.5–8 However, the assessment of retinal vessels is not entirely standardized and may be influenced by technical factors and different modalities, thus compromising potential applications in the clinical setting. Recent advances in optical coherence tomography angiography have enabled the detailed, noninvasive visualization of downstream microcirculation to the capillary level, but the evaluation of the larger retinal vessels is still based predominantly on fundus imaging.9–12 Retinal vessel caliber is often assessed qualitatively in the routine clinical evaluation of fundus images.13 However, several clinical and epidemiological studies have adopted a quantitative approach to retinal vessel caliber assessment, predicting the diameters of the central retinal vessels from the diameters of their visible daughter branches in the en face images.14–17 Retinal vessel caliber quantification is performed by various semiautomatic image analysis software programs that have been shown to not yield interchangeable measurements.18–20 Nevertheless, even when the same image processing algorithms are used, vessel tracking could still be influenced by the choice of imaging modality.21,22 Indeed, the numerous devices used for fundus imaging differ in terms of optical principles, properties of the light source, scattering, image acquisition, and image processing, among other aspects.23,24 These factors could affect the vessel intensity profiles and, consequently, the detection of vessel edges. Another factor that could significantly affect profile-based edge detection is the presence of the central light reflex (CLR), that is, the hyper-reflective region running along the center of the blood column. The CLR appears mostly in arterioles, likely owing to the thickening and hardening of their walls, and is exacerbated in arterial hypertension, leading to phenotypes known as copper and silver wiring.25 Biased estimates of retinal vessel caliber owing to the CLR could have an effect on the interpretation of clinical and epidemiological findings. Therefore, the aim of this study was to investigate the agreement of retinal vessel caliber measurements between a color fundus camera and a scanning laser ophthalmoscope and to identify whether the presence of the CLR explains potential discrepancies. For this purpose, we collected high-resolution images of the retinal vasculature from ophthalmologically healthy individuals with different blood pressure (BP) statuses, we quantified the retinal vessel caliber with standard image processing methods, and we assessed the CLR using objective grading. Methods Study Design and Population This is a cross-sectional study. We prospectively invited subjects between 50 and 65 years of age from the Lifelines Biobank, an ongoing cohort study in the northern Netherlands (n = 167,000). Selection was a priori based on BP profile, placing particular emphasis on including subjects consistently belonging to the tails of the BP distribution during their follow-up. The exact criteria, rationale, and power specifications of the recruitment process have been described extensively in our recent studies on the same population.26,27 In short, the subjects included were required to fall into one of four predefined groups: low BP, normal BP, treated high BP, and untreated high BP. Subjects in the low BP group were invited based on documented recordings of systolic BP (SBP) and diastolic BP (DBP) that were below the respective 10th percentile of the population on two or more separate occasions. Subjects in the untreated high BP group were invited based on such recordings above the 90th percentile of the population. Subjects in the normal BP group were required to have no record of antihypertensive treatment, as well as both SBP and DBP within 1 standard deviation from the mean of the age-matched population. Finally, subjects in the treated high BP group were invited based on documented uninterrupted use of antihypertensive treatment for at least 1 year. We obtained a medical history and screened all of the subjects who responded to our invitation. Regarding the ophthalmic screening, the inclusion criteria were best-corrected visual acuity of no less than 0.8 (20/25), spherical refractive error within –3 and +3 diopters, cylinder no more than 2 diopters, intraocular pressure no more than 21 mm Hg (noncontact tonometer Tonoref II; Nidek, Aichi, Japan), no reproducibly abnormal visual field test locations (Frequency Doubling Technology [C20-1 screening mode]; Carl Zeiss Meditec, Jena, Germany), no ophthalmic pathology or history of previous ophthalmic surgery, and no family history of glaucoma. Regarding general medical history, the exclusion criteria were: established diagnosis of diabetes, cardiovascular disease (except for arterial hypertension), hematologic disease, or lung disease.  All participants provided written informed consent. The ethics board of the University Medical Center Groningen approved the study protocol (#NL61508.042.17). The study followed the tenets of the Declaration of Helsinki. Fundus Imaging The imaging session was performed in the evening (between 5:00 pm and 6:30 pm) for all subjects. After screening, we applied 0.5% tropicamide eye drops and the participants rested in a dimly lit, quiet room for 20 minutes. Subsequently, we obtained brachial artery BP readings, in sitting position, with an automatic monitor (Omron M6 Comfort, Omron Healthcare, Kyoto, Japan). We recorded the average of two measurements unless there was a discrepancy of at least 10 mm Hg in SBP or 5 mm Hg in DBP, in which case we recorded the average of three measurements. We obtained images from the eye that fulfilled the inclusion criteria, or from a random eye, if both did. We used a color fundus camera (TRC-NW400; Topcon Corporation, Tokyo, Japan) and a scanning laser ophthalmoscope (Optomap 200Tx; Optos PLC, Dunfermline, UK) in random order. Two 45° fundus images centered at the optic disc were obtained with the fundus camera, and two 60° (ResMax) images were obtained with the scanning laser ophthalmoscope, with the laser gain adjusted at moderately pigmented iris. The scanning laser ophthalmoscope simultaneously acquires two images, one at a wavelength of 532 nm (green) and the other at a wavelength of 633 nm (red). The retinal vasculature attains better contrast at the green wavelength, so we used those images for subsequent analysis.28,29 All images were stored in an uncompressed format (.tiff). For a participant to be included in the analysis, we required all four images (two images per device) to be artifact free and of high quality. Image quality was judged subjectively, following the grading criteria proposed by Laurik-Feuerstein et al.30 In short, images were assessed for focus (sharp depiction of smaller order vessels and minor retinal alterations), illumination (absence of washed-out or dark areas interfering with grading), image field (proper centration allowing for grading of the region of interest), and absence of artifacts (camera reflexes, dust spots or fingerprints, eyelash images, or arc defects). Vessel Segmentation Analysis was performed with freely available, semiautomatic software (Automated Retinal Image Analyzer, Peter Bankhead), which was developed based on fundus images.31 The software uses grayscale equivalents of the fundus images, generated from the green channel information (Figs. 1A–1B). Before segmentation, wavelet thresholding was applied to enhance vessel contrast. All images were segmented using the same preset threshold settings, smoothing scales, and wavelet levels. Vessel centerlines were subsequently generated following morphological thinning and spline fitting. Figure 1. (A) Grayscale equivalent of the disc-centered 45° fundus image obtained with the TRC-NW400. (B) Grayscale equivalent of the ResMax image obtained with the 532-nm (green) laser of the Optomap 200Tx. (C, D) Disc-centered ring (inner and outer diameters equal to 2 and 3 optic disc diameters, respectively), within which the vessel diameter measurements were obtained for each imaging modality. The venule with the largest diameter in this region is marked in red. (E, F) Example of a delineated arteriole depicted with both modalities and the intensity profile plot at one of its cross-sections (marked in yellow). Edges appear further away in the image obtained by the TRC-NW400 (139  µm) compared with the Optomap 200Tx (110  µm), owing to the presence of the central light reflex in the former. (G, H) The same arteriole and the intensity profile plot at a different location, in absence of the central light reflex. Vessel diameter at this location is similar between the two devices (105 and 102  µm, respectively). The measurement region was the standard disc-centered ring with inner and outer diameters equal to 2 and 3 optic disc diameters, respectively (Figs. 1C–1D).15 The maximal distance d between two points at the border of the optic disc was marked by an experienced grader (K.P.), and a circle of diameter d was placed around the optic disc automatically. Because the shape of the optic disc can deviate from this circle and the area of the optic disc determines the exact location of the caliber measurements obtained, we also performed structural optical coherence tomography imaging of the optic nerve head (Canon HS100 SD-OCT; Canon, Inc., Tokyo Japan) to obtain better estimates of its area. The optical coherence tomography apparatus automatically locates the borders of the optic disc at the opening of Bruch's membrane, following a circular scanning pattern. Vessel edges were traced automatically within the region of interest. The software detects the edges in the direction perpendicular to the vessel centerlines. Edge detection is based on Gaussian-smoothed intensity profiles (Figs. 1E–1H), and the locations where the intensity gradient is maximal (i.e., zero-crossings of the second derivative) are used to define the vessel borders. All images were screened for obvious segmentation errors and were manually adjusted by an experienced observer (K.P.), when deemed necessary. Vessel Diameter Measurement The average diameter of each vessel within the region of interest was recorded in pixels. Pixels were transformed to units of length ( µm) based on manufacturer specifications and Gullstrand's schematic eye, that is, assuming a distance of 17 mm between the secondary nodal point and the retina and not accounting for variations in corneal curvature or axial length. We used the standard Knudtson–Parr–Hubbard algorithm, whose intragrader and intergrader reliability have been established previously as excellent.15 In short, the six largest arterioles and six largest venules within the region of interest were identified and selected for analysis (Fig. 2). Subsequently, the largest and smallest arteriole were paired and merged into one single value (representing the hypothetical width of a parent trunk) using the formula provided below: Dparent=0.88Dlarge2+Dsmall2, where D denotes the arteriolar diameter and the factor 0.88 is a branching coefficient. Figure 2. Cropped down fundus image (A) and scanning laser ophthalmoscope image (B) displaying the six largest arterioles and six largest venules within the measurement region. Vessel edges are marked in red and vessel centerlines are marked in blue. The calibers of these branches are iteratively used for the calculation of the central retinal artery and central retinal vein equivalents. The same was applied to the second largest and second smallest arterioles, followed by the third largest and third smallest arterioles. Of the resulting three new values, the largest and smallest were merged again (using the same formula) and their output was finally merged with the previously unpaired value, yielding a singular value. The same procedure was repeated for the venules, but with a branching coefficient of 0.95. These back-calculated singular values are known as the central retinal artery equivalent (CRAE) and central retinal vein equivalent (CRVE), and they represent the predicted calibers of the central retinal artery and vein, respectively. We also calculated their ratio (CRAE/CRVE), known as the arteriovenous ratio (AVR), which is often used to account for interindividual variability in vessel size and for errors in caliber measurements introduced owing to image magnification.32 For each device, we recorded the average CRAE, CRVE, and AVR of the two images obtained. CLR Quantification We used the grayscale equivalents of the color fundus images to quantify the CLR because it was much more prominent in those images than in the scanning laser ophthalmoscope images. It is likely that this is due to the color fundus images used containing intensity information from the red oxygen-sensitive wavelengths, resulting in increased contrast between the vessel walls and the blood column.33 Additionally, according to the confocal principle, the laser ophthalmoscope would theoretically prevent any scattering occurring in more superficial layers. As shown in Figures 1E–1H, the locations where the CLR is present are characterized by an elevation in the corresponding vessel pixel intensity profile. We quantified the CLR inside the same six arterioles used in the vessel diameter measurements and the same region of interest. Specifically, the CLR was classified as present in a vessel segment if at least 50% of its measured cross-sections displayed an elevation in their intensity profile. Vessel cross-sections were equidistant and each edge pixel belonged to exactly one cross-section. The CLR was classified as absent in all other cases. Based on the number of vessel segments with CLR, each fundus image was assigned a CLR score from 0 to 6. Statistical Analyses All normally distributed variables were described by the mean and standard deviation. Variables with a skewed distribution were described by the median and interquartile range. Comparisons between devices were performed with paired t-tests for normally distributed variables and with the Wilcoxon signed-rank test for variables with a skewed distribution. Pearson's r was used to establish correlations between normally distributed variables. We used Bland–Altman plots, the coefficient of repeatability, and the two-way mixed effects intraclass correlation coefficient for consistency [ICC(3,1)] to describe intradevice repeatability for CRAE, CRVE, and AVR. We used Bland–Altman plots, 95% limits of agreement, and the ICC(3,1) to describe the interdevice agreement and interchangeability for the same variables. We used multivariable regression to examine the association between CLR and the between-device difference in arteriolar caliber, adjusting for potential confounders (disc area and spherical equivalent). The between-device difference in venular caliber was also included in the model to account for unexplained variance originating from image acquisition. Because of the well-known relationship between CLR and BP, a second multivariable model was also built, replacing the CLR with BP. All analyses were performed using R (version 3.3.3; R Foundation for Statistical Computing, Vienna, Austria) and SPSS (version 28; IBM Corp., Armonk, NY). A P value of 0.05 or less was considered statistically significant. Results In total, 105 subjects satisfied the inclusion criteria. Owing to suboptimal quality or presence of artifacts in at least one image out of the four obtained, 20 subjects were excluded. Thus, 85 eyes from 85 patients were ultimately included in the analysis. The characteristics of the study sample are displayed in Table 1. The CLR was present in at least one vessel segment in 78.8% of the subjects. Table 1. Characteristics of the Study Population Age, Years 56.0 (52.5 to 60.5) Sex, % female 58.8 SBP, mm Hg 126 (113 to 147) DBP, mm Hg 81 (69 to 90) BMI, kg/m2 24.8 (22.1 to 28.0) IOP, mm Hg 13.8 ± 3.1 SEQ, D 0.00 (–0.94 to +0.94) Disc area, mm2 1.95 (1.70 to 2.30) CLR score, % per score 0: 21.2 1: 25.9 2: 29.4 3: 14.1 4: 4.7 5: 4.7 6: 0.0 BMI, body mass index; CLR, central light reflex; DBP, diastolic blood pressure; IOP, intraocular pressure; SBP, systolic blood pressure; SEQ, spherical equivalent. Values are median (interquartile range) or mean ± standard deviation unless otherwise indicated. Table 2 shows the CRAE, CRVE, and AVR, as measured by the fundus camera and scanning laser ophthalmoscope. The fundus camera yielded significantly larger values for all three variables. The between-device difference in CRAE (9.5  µm; 95% confidence interval, 8.0–11.1  µm) was larger than the between-device difference in CRVE (2.9  µm; 95% confidence interval, 1.3–4.5  µm). Scatterplots for absolute agreement and Bland–Altman plots are presented in Figure 3. As shown in these plots, the magnitude of the observed between-device difference (bias) did not depend on the magnitude of the vessel caliber, averaged over both devices (PCRAE = 0.80; PCRVE = 0.49; PAVR = 0.67), that is, the difference is a fixed offset rather than a factor. Table 2. Vessel Diameter Measurements FC SLO P Value CRAE,  µm 169.3 ± 15.2 159.7 ± 15.4 7.2 ⋅ 10−20 CRVE,  µm 240.9 ± 17.4 238.0 ± 18.0 7.9 ⋅ 10−4 AVR 0.704 ± 0.056 0.672 ± 0.057 2.3 ⋅ 10−19 AVR, arteriovenous ratio; CRAE, central retinal artery equivalent; CRVE, central retinal vein equivalent; FC, fundus camera; SLO, scanning laser ophthalmoscope. Values are mean ± standard deviation. Boldface entries indicate statistical significance. Figure 3. (Top) Absolute agreement between devices plotted for the central retinal artery equivalent (CRAE), the central retinal vein equivalent (CRVE), and the arteriovenous ratio (AVR). The y = x line depicted is the line of absolute agreement. The majority of data points for all three metrics lie above the y = x line, indicating that the TRC-NW400 yields larger diameter measurements than the Optomap 200Tx (top left and middle) and that this difference is larger in the arterioles than in the venules (top right). (Bottom) Bland–Altman plots showing the mean difference (dashed lines are 95% confidence intervals) of the measurements yielded by the two modalities and the bounds between which this difference is expected to be found 95% of the time. Differences between devices are not associated with mean values of devices, and the difference is larger in arterioles than in venules. Table 3 displays the between-device 95% limits of agreement and the between-device ICC for consistency, as well as the coefficient of repeatability and ICC for agreement within each device. Table 3. Intradevice and Interdevice Agreement and Consistency Metrics CoR (95% CI) FC CoR (95% CI) SLO 95% LoA (95% CI) FC - SLO CRAE ( µm) 10.0 9.8 Lower: –4.8 (8.9 to 11.0) (8.7 to 10.9) (–6.4 to –3.3) Upper: 23.9 (22.3 to 25.5) CRVE ( µm) 11.6 11.9 Lower: –12.0 (10.4 to 12.9) (10.6 to 13.2) (–13.6 to –10.4) Upper: 17.8 (16.1 to 19.4) AVR 0.044 0.041 Lower: –0.017 (0.039 to 0.049) (0.037 to 0.046) (–0.022 to –0.011) Upper:0.080 (0.075 to 0.085) ICC (95% CI) ICC (95% CI) ICC (95% CI) FC SLO between-device CRAE 0.945 0.948 0.885 (0.917 to 0.964) (0.922 to 0.966) (0.828 to 0.923) CRVE 0.944 0.945 0.908 (0.915 to 0.963) (0.917 to 0.964) (0.862 to 0.939) AVR 0.923 0.935 0.905 (0.884 to 0.949) (0.901 to 0.957) (0.857 to 0.937) AVR, arteriovenous ratio; CI, confidence interval; CoR, coefficient of repeatability; CRAE, central retinal artery equivalent; CRVE, central retinal vein equivalent; FC, fundus camera; ICC, intraclass correlation coefficient; LoA, limits of agreement; SLO, scanning laser ophthalmoscope. The left side of Figure 4 depicts the univariable association of the difference between the arteriolar measurements yielded by the two modalities and the CLR score. The difference increased significantly with increasing CLR (1.7  µm per unit change in CLR; P = 0.004). The right side of Figure 4 depicts the univariable association of the difference between arteriolar measurements and the difference between venular measurements. The differences were significantly associated (P = 1.5 ⋅ 10−8). Table 4 (Model A) displays the multivariable regression analysis with the difference between the arteriolar measurements as the dependent variable and the CLR score as the independent variable. When the CLR score is replaced in the model by SBP and DBP, higher SBP and lower DBP are associated with an increase in observed between-device CRAE difference (Table 4, Model B). Figure 4. (Left) Difference between the arteriolar measurements yielded by the two modalities as a function of the central light reflex (CLR) grade (quantified from the TRC-NW400 images). The CLR grade ranges from 0 to 6 and depicts the number of measured vessels in which the light reflex appears in ≥50% of the cross-sections that are assessed along each vessel segment between the measurement circles. Difference increases with increasing presence of the CLR. The regression line is depicted with 95% confidence intervals of the slope. (Right) Difference between arteriolar measurements depicted as a function of the difference between venular measurements. Arteriolar difference increases with increasing venular difference. The regression line is depicted with 95% confidence intervals of the slope. Table 4. Factors Associated With Observed Between-Device CRAE Difference: Multivariable Analysis CRAE Difference (FC - SLO;  µm), Model A Beta (95% CI) P Value CLR score 1.6 (0.6 to 2.6) .002 CRVE difference (FC - SLO;  µm) 0.5 (0.4 to 0.7) 1.6 ⋅ 10−8 Disc area (mm2) 0.6 (–2.1 to 3.3) .66 SEQ (D) 0.0 (–0.9 to 0.8) .92 CRAE Difference (FC - SLO;  µm), Model B Beta (95% CI) P Value SBP (mm Hg) 0.1 (0.0 to 0.2) .028 DBP (mm Hg) –0.2 (–0.4 to 0.0) .027 CRVE difference (FC – SLO;  µm) 0.5 (0.4 to 0.7) 1.9 ⋅ 10−8 Disc area (mm2) 0.4 (–2.4 to 3.1) .79 SEQ (D) 0.1 (–0.8 to 1.0) .77 CI, confidence interval; CRAE, central retinal artery equivalent; CRVE, central retinal vein equivalent; CLR, central light reflex; DBP, diastolic blood pressure; FC, fundus camera; SBP, systolic blood pressure; SEQ, spherical equivalent; SLO, scanning laser ophthalmoscope. Boldface entries indicate statistical significance. Discussion In this study, we reported on the extent of agreement between a color fundus camera and a scanning laser ophthalmoscope in quantifying retinal vessel caliber. We showed that, despite high intraclass correlation, vascular calibers derived from color fundus photography are larger compared with scanning laser ophthalmoscope images. This offset is more pronounced in arteriolar than venular calibers. Greater observed between-device difference in arteriolar caliber was associated with the presence of a more prominent CLR. The caliber magnitudes that we reported in this study are generally in agreement with those reported in population studies implementing the standard Knudtson–Parr–Hubbard algorithm taking into consideration age and BP-related differences.4,34–36 The vast majority of studies use color fundus cameras to quantify retinal vessel caliber; however, a few studies have used scanning laser ophthalmoscopes.28,37 Automated quantification and grading of the CLR inside blood vessels is a more challenging and less explored task, while subjective grading could significantly compromise reproducibility.38,39 To tackle these problems, we proposed an objective, semiautomatic method by making use of the vessel profiles obtained for caliber estimation. The difference between the fundus camera-derived and scanning laser ophthalmoscope-derived retinal vessel caliber values reported in our study was an offset rather than a factor, because it was found to be independent of the measured caliber. In addition, the difference was larger in arteries than in veins. Therefore, magnification effects are highly unlikely to be the underlying cause of this observation, which is also corroborated by the fact that the spherical equivalent was not significant in multivariable analysis. Edge location algorithms used for vessel delineation usually rely on the half-width at half-maximum principle or on zero-crossings of the second derivative of the intensity profile.31 The CLR renders the intensity dip less pronounced, introducing bias to the estimated edge location, thus resulting in a seemingly thicker vessel. In addition to the effect of the CLR on the intensity profile, it has been shown that edge location is also biased when low-pass filtered and nonlinearly transformed.40,41 However, in contrast with the CLR, this phenomenon is less likely to explain the observed difference between arteries and veins because of the use of green channel information, in which arteries and veins appear with similar contrast. The between-device mean difference observed in this study was equal or less than the within-device coefficients of repeatability that, from a clinical perspective, suggests that devices are sufficiently interchangeable. Nevertheless, color fundus images are likely to be more informative for assessing the CLR, whereas vessel diameter measurements derived from scanning laser ophthalmoscope images are less affected by the confounding effect of the CLR. From an epidemiological perspective, our results suggest that the effect of cardiovascular disease on retinal vessel caliber is likely to be underestimated, because the CLR makes arteriolar calibers appear wider in color fundus images. Indeed, models A and B in Table 4 show that, owing to the CLR, patients with higher pulse pressure (usually a sign of pronounced atherosclerosis and arteriosclerosis) are more likely to have higher retinal vessel caliber measurements in color fundus photographs. In all cases, it is important to introduce a standardized, objective protocol in retinal vessel caliber estimations.32,42,43 The main strengths of our study are the implementation of objective CLR quantification, as well as the fact that we were able to assess differences in vessel diameters in a wide range of BPs, from hypotensive individuals to treated or untreated hypertensives. One limitation of our study is the fact that vascular caliber was only quantified using one vessel processing software, but there may be disparities between different algorithms.18–20 To date, there have been no studies investigating the performance of any vessel processing software in the presence of the CLR, and it is likely that some algorithms may be less affected than others. However, since edge detection is usually based on half-width at half-maximum or inflection principles, we hypothesize that this effect will be present on most occasions. Additional studies are needed to confirm this speculation. In addition, our population was predominantly Caucasian; hence, our results cannot be generalized safely to other ethnicities, especially owing to the documented effect of pigmentation on fundus contrast.44 Finally, only the vessel lumen (blood column) is visible with conventional imaging techniques, which does not allow for assessment of the vessel wall. Adaptive optics have been used to visualize the vessel wall and quantify the wall-to-lumen ratio.45,46 In conclusion, we showed that the presence of a prominent CLR induces bias in retinal arteriolar caliber measurements in color fundus images, when compared with scanning laser ophthalmoscope images. Future population studies should use CLR-adjusted vessel diameters to obtain refined estimates of the relationship between retinal vessel caliber and ophthalmic or systemic diseases. Acknowledgments Financial Support: Stichting Oogfonds Nederland. The funding organization had no role in the design, conduct, analysis, or publication of this research. Meeting Presentation: Association for Research in Vision and Ophthalmology (ARVO), Baltimore, May 3–7, 2020 (held online). Disclosure: K. Pappelis, None; N.M. Jansonius, None ==== Refs References 1. Mitchell P, Leung H, Wang JJ, et al . Retinal vessel diameter and open-angle glaucoma: the Blue Mountains Eye Study. Ophthalmology. 2005; 112 : 245–250.15691558 2. Jeganathan VSE, Kawasaki R, Wang JJ, et al . Retinal vascular caliber and age-related macular degeneration: the Singapore Malay Eye Study. Am J Ophthalmol. 2008; 146 : 954–959.e1.18760764 3. Klein R, Klein BEK, Moss SE, Wong TY. Retinal vessel caliber and microvascular and macrovascular disease in type 2 diabetes: XXI: the Wisconsin Epidemiologic Study of Diabetic Retinopathy. Ophthalmology. 2007; 114 : 1884–1892.17540447 4. Ikram MK, Witteman JCM, Vingerling JR, Breteler MMB, Hofman A, de Jong PTVM. Retinal vessel diameters and risk of hypertension: the Rotterdam Study. Hypertension. 2006; 47 : 189–194.16380526 5. Klein R, Sharrett AR, Klein BE, et al . Are retinal arteriolar abnormalities related to atherosclerosis?: The Atherosclerosis Risk in Communities Study. Arterioscler Thromb Vasc Biol. 2000; 20 : 1644–1650.10845884 6. McGeechan K, Liew G, Macaskill P, et al . Prediction of incident stroke events based on retinal vessel caliber: a systematic review and individual-participant meta-analysis. Am J Epidemiol. 2009; 170 : 1323–1332.19884126 7. Baker ML, Marino Larsen EK, Kuller LH, et al . Retinal microvascular signs, cognitive function, and dementia in older persons: the Cardiovascular Health Study. Stroke. 2007; 38 : 2041–2047.17525385 8. Guo S, Yin S, Tse G, Li G, Su L, Liu T. Association between caliber of retinal vessels and cardiovascular disease: a systematic review and meta-analysis. Curr Atheroscler Rep. 2020; 22 : 16.32440852 9. Jia Y, Tan O, Tokayer J, et al . Split-spectrum amplitude-decorrelation angiography with optical coherence tomography. Opt Express. 2012; 20 : 4710–4725.22418228 10. Pappelis K, Jansonius NM. Quantification and Repeatability of vessel density and flux as assessed by optical coherence tomography angiography. Transl Vis Sci Technol. 2019; 8 : 3. 11. Pappelis K, Choritz L, Jansonius NM. Microcirculatory model predicts blood flow and autoregulation range in the human retina: in vivo investigation with laser speckle flowgraphy. Am J Physiol Heart Circ Physiol. 2020; 319 : H1253–H1273.32986964 12. Al-Nosairy KO, Prabhakaran GT, Pappelis K, Thieme H, Hoffmann MB. Combined multi-modal assessment of glaucomatous damage with electroretinography and optical coherence tomography/angiography. Transl Vis Sci Technol. 2020; 9 : 7. 13. Wong TY, Mitchell P. Hypertensive retinopathy. N Engl J Med . 2004; 351 : 2310–2317.15564546 14. Hubbard LD, Brothers RJ, King WN, et al . Methods for evaluation of retinal microvascular abnormalities associated with hypertension/sclerosis in the Atherosclerosis Risk in Communities Study. Ophthalmology. 1999; 106 : 2269–2280.10599656 15. Knudtson MD, Lee KE, Hubbard LD, Wong TY, Klein R, Klein BEK. Revised formulas for summarizing retinal vessel diameters. Curr Eye Res. 2003; 27 : 143–149.14562179 16. Wong TY, Klein R, Nieto FJ, et al . Retinal microvascular abnormalities and 10-year cardiovascular mortality: a population-based case-control study. Ophthalmology. 2003; 110 : 933–940.12750093 17. Amerasinghe N, Aung T, Cheung N, et al . Evidence of retinal vascular narrowing in glaucomatous eyes in an Asian population. Invest Ophthalmol Vis Sci. 2008; 49 : 5397–5402.18719076 18. French C, Heitmar R. Comparison of static retinal vessel caliber measurements by different commercially available platforms. Optom Vis Sci. 2021; 98 : 1104–1112.34570034 19. Yip W, Tham YC, Hsu W, et al . Comparison of common retinal vessel caliber measurement software and a conversion algorithm. Transl Vis Sci Technol. 2016; 5 : 11. 20. Mautuit T, Cunnac P, Cheung CY, et al . Concordance between SIVA, IVAN, and VAMPIRE software tools for semi-automated analysis of retinal vessel caliber. Diagnostics (Basel). 2022; 12 : 1317.35741127 21. Mautuit T, Semecas R, Hogg S, et al . Comparing measurements of vascular diameter using adaptative optics imaging and conventional fundus imaging. Diagnostics (Basel). 2022; 12 : 705.35328258 22. Heitmar R, Kalitzeos AA. Reliability of retinal vessel calibre measurements using a retinal oximeter. BMC Ophthalmol. 2015; 15 : 184.26705024 23. Terasaki H, Sonoda S, Tomita M, Sakamoto T. Recent advances and clinical application of color scanning laser ophthalmoscope. J Clin Med Res. 2021; 10 : 718. 24. DeHoog E, Schwiegerling J. Fundus camera systems: a comparative analysis. Appl Opt. 2009; 48 : 221–228.19137032 25. Kaushik S, Tan AG, Mitchell P, Wang JJ. Prevalence and associations of enhanced retinal arteriolar light reflex. Ophthalmology. 2007; 114 : 113–120.17070582 26. Pappelis K, Jansonius NM. Retinal oxygen delivery and extraction in ophthalmologically healthy subjects with different blood pressure status. Transl Vis Sci Technol. 2022; 11 : 9. 27. Pappelis K, Jansonius NM. U-shaped effect of blood pressure on structural OCT metrics and retinal perfusion in ophthalmologically healthy subjects. Invest Ophthalmol Vis Sci. 2021; 62 : 5. 28. Blair NP, Wanek J, Felder AE, et al . Retinal oximetry and vessel diameter measurements with a commercially available scanning laser ophthalmoscope in diabetic retinopathy. Invest Ophthalmol Vis Sci. 2017; 58 : 5556–5563.29079858 29. Hoover AD, Kouznetsova V, Goldbaum M. Locating blood vessels in retinal images by piecewise threshold probing of a matched filter response. Proc AMIA Symp. 1998: 931–935.9929355 30. Laurik-Feuerstein KL, Sapahia R, Cabrera DeBuc D, Somfai GM. The assessment of fundus image quality labeling reliability among graders with different backgrounds. PLoS One. 2022; 17 : e0271156.35881576 31. Bankhead P, Scholfield CN, McGeown JG, Curtis TM. Fast retinal vessel detection and measurement using wavelets and edge location refinement. PLoS One. 2012; 7 : e32435.22427837 32. Heitmar R, Vonthein R. Clinically valid conclusions from retinal photographs need the best formulae. Graefes Arch Clin Exp Ophthalmol. 2021; 259 : 811–813.33394163 33. Ouyang Y, Shao Q, Scharf D, Joussen AM, Heussen FM. An easy method to differentiate retinal arteries from veins by spectral domain optical coherence tomography: retrospective, observational case series. BMC Ophthalmol. 2014; 14 : 66.24884611 34. Ikram MK, Kamran Ikram M, de Jong FJ, et al . Are retinal arteriolar or venular diameters associated with markers for cardiovascular disorders? The Rotterdam Study. Invest Ophthalmol Vis Sci. 2004; 47 : 189–194. 35. Gepstein R, Rosman Y, Rechtman E, et al . Association of retinal microvascular caliber with blood pressure levels. Blood Press. 2012; 21 : 191–196.22263560 36. Kawasaki R, Cheung N, Wang JJ, et al . Retinal vessel diameters and risk of hypertension: the Multiethnic Study of Atherosclerosis. J Hypertens. 2009; 27 : 2386–2393.19680136 37. Garg G, Venkatesh P, Chawla R, Takkar B, Temkar S, Damodaran S. Normative data of retinal arteriolar and venular calibre measurements determined using confocal scanning laser ophthalmoscopy system - Importance and implications for study of cardiometabolic disorders. Indian J Ophthalmol. 2022; 70 : 1657–1663.35502046 38. Bhuiyan A, Cheung CY, Frost S, et al . Development and reliability of retinal arteriolar central light reflex quantification system: a new approach for severity grading. Invest Ophthalmol Vis Sci. 2014; 55 : 7975–7981.25358734 39. Nguyen UTV, Bhuiyan A, Park LAF, et al . An automated method for retinal arteriovenous nicking quantification from color fundus images. IEEE Trans Biomed Eng. 2013; 60 : 3194–3203.23807422 40. Jansonius NM, Cervantes J, Reddikumar M, Cense B. Influence of coherence length, signal-to-noise ratio, log transform, and low-pass filtering on layer thickness assessment with OCT in the retina. Biomed Opt Express. 2016; 7 : 4490–4500.27895990 41. Jansonius NM, Stam L, de Jong T, Pijpker BA. Quantitative analysis of illusory movement: spatial filtering and line localization in the human visual system. Perception. 2014; 43 : 1329–1340.25669050 42. Heitmar R, Kalitzeos AA, Patel SR, Prabhu-Das D, Cubbidge RP. Comparison of subjective and objective methods to determine the retinal arterio-venous ratio using fundus photography. J Optom. 2015; 8 : 252–257.26386537 43. Liew G, Sharrett AR, Kronmal R, et al . Measurement of retinal vascular caliber: issues and alternatives to using the arteriole to venule ratio. Invest Ophthalmol Vis Sci. 2007; 48 : 52–57.17197515 44. Rochtchina E, Wang JJ, Taylor B, Wong TY, Mitchell P. Ethnic variability in retinal vessel caliber: a potential source of measurement error from ocular pigmentation? The Sydney Childhood Eye Study. Invest Ophthalmol Vis Sci. 2008; 49 : 1362–1366.18385051 45. Streese L, Brawand LY, Gugleta K, Maloca PM, Vilser W, Hanssen H. New frontiers in noninvasive analysis of retinal wall-to-lumen ratio by retinal vessel wall analysis. Transl Vis Sci Technol. 2020; 9 : 7. 46. Hillard JG, Gast TJ, Chui TYP, Sapir D, Burns SA. Retinal arterioles in hypo-, normo-, and hypertensive subjects measured using adaptive optics. Transl Vis Sci Technol. 2016; 5 : 16.

PMC010xxxxxx/PMC10353743.txt

==== Front Invest Ophthalmol Vis Sci Invest Ophthalmol Vis Sci IOVS Investigative Ophthalmology & Visual Science 0146-0404 1552-5783 The Association for Research in Vision and Ophthalmology 37450310 10.1167/iovs.64.10.15 IOVS-22-36631 Retina Retina Flow Heterogeneity and Factors Contributing to the Variability in Retinal Capillary Blood Flow Human Retinal Capillary Flow Heterogeneity Neriyanuri Srividya 1 Bedggood Phillip 1 Symons R. C. Andrew 1 2 3 4 Metha Andrew B. 1 1 Department of Optometry and Vision Sciences, The University of Melbourne, Parkville, Victoria, Australia 2 Department of Surgery, The University of Melbourne, Parkville, Victoria, Australia 3 Centre for Eye Research Australia, East Melbourne, Victoria, Australia 4 Department of Surgery, Alfred Hospital, Monash University, Melbourne, Victoria, Australia * Correspondence: Srividya Neriyanuri, Department of Optometry and Vision Sciences, The University of Melbourne, 200 Berkeley Street, Parkville, VIC 3010, Australia; drsrividyaneriyanuri@hotmail.com. 14 7 2023 7 2023 64 10 1517 6 2023 15 12 2022 Copyright 2023 The Authors 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. Purpose Capillary flow plays an important role in the nourishment and maintenance of healthy neural tissue and can be observed directly and non-invasively in the living human retina. Despite their importance, patterns of normal capillary flow are not well understood due to limitations in spatial and temporal resolution of imaging data. Methods Capillary flow characteristics were studied in the retina of three healthy young individuals using a high-resolution adaptive optics ophthalmoscope. Imaging with frame rates of 200 to 300 frames per second was sufficient to capture details of the single-file flow of red blood cells in capillaries over the course of about 3 seconds. Results Erythrocyte velocities were measured from 72 neighboring vessels of the parafoveal capillary network for each subject. We observed strong variability among vessels within a given subject, and even within a given imaged field, across a range of capillary flow parameters including maximum and minimum velocities, pulsatility, abruptness of the systolic peak, and phase of the cardiac cycle. The observed variability was not well explained by “local” factors such as the vessel diameter, tortuosity, length, linear cell density, or hematocrit of the vessel. Within a vessel, a moderate relation between the velocities and hematocrit was noted, suggesting a redistribution of plasma between cells with changes in flow. Conclusions These observations advance our fundamental understanding of normal capillary physiology and raise questions regarding the potential role of network-level effects in explaining the observed flow heterogeneity. blood flow retinal capillary erythrocyte velocity adaptive optics imaging human flow heterogeneity ==== Body pmcThe retina is exceptional for being the most metabolically active tissue in the body. Supporting the various cellular processes that enable visual function is a complex system of capillary networks with regional and corresponding functional specialization.1 The density of vascular supply in a normal human retina is not homogeneous but is continuously variable both tangentially across its surface and radially in depth.2 Supply difference requirements can be met by simple variations in anatomical densities; for example, the inner capillary supply at the macula is greater than is available to the peripheral retina, presumably to support the higher density of inner retinal cells nearer the fovea by way of increased areal density.3 Although such regional differences in capillary densities clearly exist, it is not known whether the distribution of actual blood flow follow these differences or not. Heterogeneity of red blood cell transit speed through nearby capillary segments has been observed4–7 and may imply a level of variability in capillary resistance, even within a single layer. Similar local mechanisms may give rise to observed heterogeneity in the flow of leukocytes, as well.8,9 This article provides detailed observations that might lead to insights into the differences in patterns of capillary flow and the local variables influencing the flow across the foveal vascular network. This understanding could help unveil interesting pathophysiological mechanisms behind selective regional (both laterally and in depth) disruptions seen in vasculopathies such as diabetic retinopathy10–15 and add to our understanding of capillary blood flow generally. In general, large retinal vessels are marked as healthy or diseased based on examining physical variables such as size, shape, color, and reflex of the vessel using standard ophthalmoscopic or color fundus photographic imaging. With the advent of adaptive optics (AO) imaging, structural variables at a much finer scale, such as vessel caliber, wall-to-lumen ratio, and tortuosity, have become discernable in small vessels such as capillaries and have been flagged as promising indicators for disease at ever earlier stages of the vasculopathy process.16–18 However, physiological variables such as the resting flow state and stimulus-evoked flow changes might provide greater insights into normal functioning of retinal vessels, particularly the capillaries.19 The flow measures have potential to be labeled as “biomarkers” of preclinical disease, especially if they can be shown to be significant in initiating structural damage that becomes visible only in a later clinical stage. These include flow measures such as cell speeds through a vascular segment and the variability (i.e., heterogeneity) of cell speeds in a small vascular bed region. Pulsatility is one such flow-derived functional measure that has been reported in both mouse and human retinal capillaries, where a clear time-dependent variability in flow is seen during systolic and diastolic phases of the cardiac pulse.4,5,20–22 AO imaging was used to identify the erythrocyte aggregates in parafoveal capillaries.23 It was shown that the presence of erythrocyte aggregates (seen as dark tails and suggesting increased hematocrit) significantly increased the average velocities compared to vessels without the aggregates. The network dynamics further influenced the velocities at the bifurcations, where the daughter vessels with dark tails had lower velocities compared to the parent vessels with tails.24 In addition to the role of erythrocyte aggregates, leukocytes also play a role in producing variability, as the lumen is blocked by the white blood cells, so erythrocytes bank up behind it (with plasma assumedly flowing past). Studies have shown specific leukocyte paths with variable flow speed and hematocrit resulting from the intermittent passage of leukocytes.9 As described above, retinal vascular diseases such as diabetes affect different regions of retina to varying degrees. Documenting the variations in flow patterns in smaller blood vessels may help in determining why some retinal locations (or some vessels in particular) are more or less prone to damage. Apart from the disease significance of lost capillary function in conditions such as stroke and dementia, studying the variability in normal capillary flow improves our understanding of the basic mechanisms of blood flow. Hence, the objectives of this study were as follows. First, we wanted to observe and characterize spatial heterogeneity in capillary flow, which can also be termed “inter-vessel variability”; this is the degree to which adjacent foveal capillary segments support similar or dissimilar flow patterns. Structural variables within and/or at the vessel site might influence the characteristics of cellular flow (e.g., average velocity, maximum and minimum velocities) through any given vessel segment. Structural measures of potential relevance are the diameter, length, and tortuosity of vessels. Resistance to flow is classically (in larger vessels) proportional to the fourth power of vessel diameter; in the capillaries, the resistance to flow may be even greater than expected by this relationship.25 Resistance is further proportional to the length of the vessel through which blood must traverse. Resistance is not expected to depend on vessel tortuosity at the microvascular scale due to the dominance of viscous rather than inertial forces in such vessels.26 In addition to structural factors pertaining to individual blood vessels, the composition of the blood column may alter flow dynamics. Prior adaptive optics work has shown that the presence of erythrocyte aggregates in slightly larger vessels than those considered here (diameter >7–8 µm) tends to show faster flow.23,24 Recent work from our laboratory has also shown that the separation between successive red blood cells is strongly linked to the shape of individual cells, suggesting that the forces experienced by a cell may depend on the composition of the blood column.25 Our second objective in the present work was to explore the extent to which the observed flow patterns may be affected by local (within the vessel) factors, including the vessel diameter, length, tortuosity, and composition of the blood column expressed by the linear cell density or by the hematocrit. The goal of comparing these variables is to try to form explanatory models of flow in the retinal capillary bed. The basic “rules” describing retinal capillary flow patterns are not well understood, and sophisticated computer simulations are required to predict flow.27–29 Even such work does not typically consider pulsatile input at all, which is now known to be a feature of flow in all retinal capillaries analyzed to date. In the absence of a simple model, the first step in the chain of inquiry ought to be consideration of the basic statistical patterns (e.g., covariance among available outcome measures), as well as normative data. Methods The study was approved by the Human Research Ethics Committee of The University of Melbourne (ethics approval number 1137234) and adhered to the tenets of the Declaration of Helsinki. Written informed consent was obtained from all study participants, and all study procedures were clearly explained prior to the study protocol being initiated. Three healthy individuals (one male and two females; age range, 22–23 years) who were free from any known ocular or systemic illnesses participated in this study. Participants with clear media and a refractive error of less than 4 diopters (D) spherical and 2 D astigmatic error were considered suitable for small vessel imaging. The left eye of each participant was chosen for vascular imaging using optical coherence tomography angiography (OCTA) (SPECTRALIS; Heidelberg Engineering, Heidelberg, Germany) and AO (custom “flood” illuminated system described below). Pupils were pharmacologically dilated using 0.5% tropicamide (Alcon, Geneva, Switzerland) at least 20 minutes prior to the AO imaging. The structural information from OCTA was used as a guide to image the vessels around the foveal avascular zone (FAZ) using AO, which formed the main investigative tool for the study. The principles and implementation of the AO imaging system have been described in detail in previously published articles7,30–32 and are described briefly here. AO Imaging Procedure The image capturing part of the AO setup consisted of a 2560 × 2160-pixel Andor NEO sCMOS camera (Andor Technology PLC, Belfast, UK), which enabled the fast image acquisition rates required for tracking blood flow (in this work, 200–300 fps). All pixels were exposed simultaneously using the Global Shutter mode of the camera, resulting in images that were free of intra-frame distortion due to eye movements. The ocular wavefront aberrations were measured using a Hartmann–Shack wavefront sensor (Adaptive Optics Associates, Devens, MA, USA). Measured wavefront slopes were corrected by a 97-channel deformable mirror (ALPAO, Montbonnot-Saint-Martin, France) in real time using a closed-loop control operating at 30 fps by custom MATLAB software (version R2017a; MathWorks, Natick, MA, USA). The best focus of the blood vessels was achieved by adjusting the defocus component of the adaptive optics correction in 0.05-D steps. The criterion was set such that the appearance of cellular flow within the maximum number of capillaries in the field was subjectively maximized (as opposed to, for example, vessel walls or larger sized vessels). The depth of focus of our system is such that this can be done reliably to the nearest 0.05 D (typical) or 0.10 D (in the case of poorer image quality). Even finer adjustments are discernible when using a (stationary) model eye of similar numerical aperture to a human eye. Imaging Procedure Small patches of the foveal microvascular network that were 1.25° (approximately 360 µm) in diameter were imaged using a 750-nm light (supercontinuum laser passed through a tunable SuperChrome transmission filter, SC480-8; Fianium, Southampton, UK) in a darkened laboratory. This light resulted in a power of 0.36 mW at the cornea, which adhered to the American National Standards Institute guidelines of maximum permissible exposure standards for continuous illumination over 3.4 seconds and was at least 10 times safer than the standard.33 The fixation target used was a diffusely back-illuminated fixation grid printed in black on white paper. The imaging area was centered 1° to 2.5° from fixation, around the FAZ, and the participant was asked to stare at the appropriate location on the fixation target to bring each region of interest into the field of view. The vertical extent of the imaging field was set to either 512 pixels (covering approximately 0.89° or 256 µm on the human retina) to achieve a frame rate of 200 fps or to 328 pixels (164 µm) to achieve a frame rate of 300 fps under the Global Shutter setting. The horizontal extent of the imaging field was not limited by frame-rate considerations and was set by a field stop to span 1.25° diameter (360 µm). The frame rate was adjusted to attempt maximizing the amount of useful information captured from a given capillary network. For example, in an area of slower flow, a slower frame rate could be chosen, permitting a taller field size and hence a larger number of vessels captured. If flow was too fast, however, the higher frame rate with fewer rows in the image would be required to capture more of the flow. This judgment was made on an ad hoc basis in each imaged field. The imaging was done using different frame rates of 200 or 300 frames per second (fps) for a maximum of 3.4 seconds to create a video sequence for the calculation of flow velocity, described below. Videos of greater duration could not be recorded due to memory bandwidth limitations in transferring data from the camera to our computer. However, much longer recording periods would not typically be desirable due to variations in tear film, associated blinking, and difficulty in maintaining steady fixation over longer periods. It should be noted that our acquisition was not synchronized to the activity of the heart (for example, triggered based on the electrocardiogram waveform). Although cardiac activity is an extremely important source of variability, a variable lag among subjects is expected, resulting from differences in the electrical activity of the heart, its subsequent muscular contraction, and propagation of the pulse wave along the vascular tree before finally reaching the retinal capillaries. To ensure we fully captured variability due to cardiac contractility, we recorded over ∼3 seconds (around three full cardiac cycles). For the comparison of the phase of velocity waveforms across vessels, we therefore restricted our analysis to just those vessels imaged at the same time within our field of view. Spatial Heterogeneity The variability in velocities between vessels across the network is referred to here as spatial heterogeneity, which was studied by examining the capillary vessel flow patterns from neighboring retinal fields surrounding the FAZ. Extracting Cell Velocities Figure 1 shows how an individual vessel segment was chosen and used for velocity analysis in one of the subjects. First, a suitable region for imaging was identified from the OCTA map (Fig. 1A). A motion contrast image was generated from AO video data in all chosen regions,34 and all such images were combined to generate a high-resolution montage (Fig. 1B). From the motion contrast montage, the complete vascular network skeleton was manually traced using Photoshop (Adobe, San Jose, CA, USA). As an initial step in data processing, raw video sequences were manually inspected to identify vessels with sufficiently high contrast, slow flow, and sparse hematocrit that cellular flow could be tracked unambiguously throughout the full cardiac cycle (an example selected vessel is shown in Figs. 1C and 1D). No vessels containing putative leukocyte flow were included for analysis. For vessels manually approved for analysis in this way, red cell velocity through time was computed within a rolling epoch 150 ms in duration from the AO video data using a kymograph approach.35 The epoch (100–150 ms) chosen was somewhat arbitrary. A sufficient number of frames are required to produce an accurate velocity estimate; however, if too many frames are used, this will compromise the ability to track sudden changes in velocity. Epochs of 100 ms (300 fps) and 150 ms (200 fps) were chosen as a trade-off among these factors and are comparable to epochs chosen in previous studies.5,20,21 Figure 1. Identifying a vessel segment of interest. (A) Retinal fundus scanning laser ophthalmoscope (SLO) and SPECTRALIS OCTA. The area in red indicates the region imaged using AO. (B) AO montage highlighting regions of interest (encircled; yellow arrows indicate flow direction). (C, D) Division image highlighting a vessel segment of interest (C) whose skeleton (D) was used for constructing a velocity plot. As shown in Figure 2, velocities (given by a slope) were estimated using three standard approaches: spatiotemporal (ST) plot and two-dimensional (2D) spatial and 2D temporal correlograms. The best estimate was taken from the approach that provided the highest signal-to-noise ratio.35 As reported in our previous study,35 the spatial and temporal correlograms were deemed more reliable than the ST plot in the majority of epochs analyzed. In this study, across several thousand epochs, the temporal correlogram was selected in 70% of epochs, the spatial correlogram in 26%, and the ST plot in 4% of epochs. If the slope given by an approach did not appear to visually match with the flow information presented in a window, the best of the three approaches was picked subjectively (i.e., by the human supervisor), and the velocity information was overwritten for that window. In rare cases where all three approaches failed to pick up a slope that matched the direction of the flow information determined subjectively, the velocity for that window was flagged as “NaN” (not a number) and excluded from the analysis (Fig. 2C). Epochs were excluded where the kymograph appeared to display flow in the reverse direction, which often indicates aliasing. Upon detailed analysis, of a total of 2956 epochs analyzed across 72 vessels, 10 individual epochs were excluded. About 10 to 15 neighboring vascular segments within each field were considered suitable for velocity analysis. It is worth noting that, for this study, we did not use a recent, more robust automated method published by our group21 because it was not available at the time of commencement of this study. Figure 2. Examples showing Kymograph analysis. (A, B) Good examples of kymographs from subjects 1 and 2, respectively, showing that all three approaches estimated reliable velocities (indicated by a slope in blue). The example shows an approach picked as “best” (indicated by red star) by the algorithm based on a high signal-to-noise ratio. (C) An example in which all three approaches failed to pick up a best slope and hence the velocity for this window was taken to be not a number (NaN). Parameters Derived From Velocity–Time Plots The velocity–time plots (i.e., average epochal velocities plotted as a function of time for each vessel, also described as velocity plots) were derived from included data for all the vessels. The following flow measures were directly extracted from the velocity–time plot (blue data points in Fig. 3) for each vessel segment: The average velocity (Vave), defined as the mean over the full time period observed (minimum 3 seconds), and the maximum or peak velocity (Vmax_raw), as well as minimum or trough velocity (Vmin_raw), were noted across all epochs and quantified for each vessel segment. The deviation or dispersion of cell speeds from the mean in the video of a given vessel was given by velocity standard deviation (Vstdev), and a normalized measure of velocity variability was given by the coefficient of variation (COV), calculated as Vstdev/Vave. Figure 3. Triangle-wave curve fit for velocity plot. Raw velocity estimates plotted against time are shown in blue points. Each data point shows cellular flow velocity through a given capillary segment within a single epoch (100 ms). A fitted curve is shown as a solid red line for raw velocity information plotted over 3 seconds of data (points in blue) for a single vessel. Dashed red lines indicates 95% bounds for the fitted data. To capture key information regarding flow dynamics within each vessel, we fitted a triangular-wave parametric curve to the data for velocity over time (blue data points in Fig. 3). This was performed using standard MATLAB curve fitting routines (we used the Fit function of the MATLAB R2017a Curve Fitting toolbox), and 95% confidence intervals (CIs) for the data were also returned, as shown in Figure 3. The fitted velocity traces were generally well modeled by this simple curve fit approach, with average R2 of 0.78 ± 0.17 (mean ± SD). The velocity data for each vessel were fitted for five essential key parameters: (1) maximum velocity (Vmax_curvefit); (2) minimum velocity (Vmin_curvefit); (3) peak time (time taken for the onset of first peak); (4) period (time between consecutive peaks); and (5) abruptness (defined as a proportion of the period with rising velocity and calculated as “rise time/period”). Pulsatility was defined by calculating (Vmax – Vmin)/(Vmax + Vmin). Measuring Local Vessel Structure Variables In order to study and predict the velocity variations within a network, we measured local vessel structural variables and other local flow variables such as linear cell density (LCD) and hematocrit. Vessel Diameters Using registered “division” images to represent our best information regarding the internal lumen diameters of vessel segments, vessel diameters were measured in MATLAB by applying the Otsu automatic image thresholding method for detecting edges.36 To threshold each vessel, pixels were removed from the image that were closer to another vessel (creating a region of interest containing only the vessel and pixels in the background). A maximum likelihood curve was fitted to the binary “above threshold or below threshold” image data, with the edge of the vessels chosen as the 50% probability point. The thresholded image data were bootstrapped by randomly selecting, with replacement, the pixels to be included in the likelihood curve. Confidence intervals were thus generated by sorting the bootstrapped diameter measures, with narrow intervals indicating robust measurement of internal vessel diameter. Vessel lengths were calculated in pixel units from the one-pixel skeletonized segments between bifurcations. The vessel lengths reported here represent the full length of the vessel between two bifurcations. Vessel tortuosity was defined as the ratio between the length of the vessel to the shortest distance between the start and end of the vessel segment. Vessel length was calculated using the following conversion: 1 pixel = 0.5 µm, based on the paraxial magnification expected for our system in imaging an emmetropic eye with a focal length of f = 17 mm in air; this is an approximate figure only, with refraction ranging up to 4 D in our subjects. Measuring Local Vessel Flow Variables The local vessel flow variables that were studied are LCD (number of cells per millimeter length of vessel) and hematocrit (fraction of blood column occupied by erythrocytes). The LCD for a vessel segment was defined as the number of cells per unit length of vessel segment. The hematocrit content of a vessel was measured as a ratio of the length of cells occupying a vessel to the length of vessel analyzed. Both measures require that the cells occupying a vessel segment are delineated. This was achieved by motion correct of the XT plot, which has been described elsewhere.37 In brief, the approach is to shift each row of the kymograph according to the flow velocity, producing a “stationary” XT plot in which vertical bright and dark vertical bands correspond to cells and plasma (Fig. 4B). Figure 4. Counting red blood cells. (A) Kymograph shows intensity along a vessel centerline as a function of time. Slope given by the blue line indicates the flow velocity (∼0.11 mm/s). (B) Motion-corrected kymograph obtained by cumulatively sliding each row of A according to the flow velocity. (C) As for B but with a binary mask overlaid to indicate the position of found cells (see text). (D) Binary mask from C projected onto the original kymograph from A. Scale bar: 20 µm horizontal and 100 ms vertical. The fundamental assumption underlying the shifting of data in horizontal rows is that the blood column is a fluid and should therefore be incompressible. Within a rigid tube we would expect the entire blood column to travel at the same speed even during the systolic transition. However, it may be noted that, if the capillary tube is not entirely rigid, then differences in speed along the blood column should manifest—for example, as acceleration lag due to the finite propagation speed of the pulse pressure wave. This is a subtle signal that can be revealed in many capillary vessels with our approach.38 Overall, however, the high speed of the pulse wave compared with the much slower speed of blood flow is such that the analysis method described in Figure 4 (based on incompressibility of fluid) is a good approximation to compensate for the motion of the blood constituents; that is, the entire blood column can be presumed to travel at the same speed. It should be noted that the point of compensating for cellular motion (i.e., rendering them as straight vertical bars) is that it allows data from each cell to easily be averaged over time. This dramatically improves the signal-to-noise ratio for image segmentation, greatly aiding computation of the hematocrit compared with (for example) simply applying standard automated thresholding approaches to the kymograph. Because either cells or plasma can appear with positive contrast depending on the plane of focus,35 video data were manually inspected to confirm the identity of cells; in cases where this judgment was difficult, the vessel was excluded from analysis. To delineate cells, a brute force approach was adopted in which binary image masks consisting of a series of bars were fit to the motion-corrected kymograph. The free parameters were the location, width, and number of bars. The mask giving the greatest statistical difference (P value of t-test) between the intensity of pixels inside and outside the mask was taken as the one most faithfully delineating cells from plasma. Statistical Analysis of the Data Further statistical analysis of the data was performed using SPSS Statistics 21.0 for Windows (IBM Corp., Chicago, IL, USA). The SPSS Statistics software was used to scrutinize the distribution of the parameters and discern correlations among them. Correlations among parameters were examined using the two-tailed Pearson's test at a significance level of 0.05. Correlation coefficients between 0.00 and 0.30 were considered to indicate negligible correlation; 0.30 to 0.50, low correlation; 0.50 to 0.70, moderate correlation; and 0.70 to 0.90, high correlation.39 Results This section presents data concerning the flow of red blood cells tracked through 72 different retinal capillary segments (13 different fields of interest) from three subjects. The numbers of vessels analyzed for each subject were 20, 34, and 18, respectively. As described in the Table, the variables are categorized as either flow parameters or as local structural or flow variables. Table. Descriptive Statistics of the Flow Parameters and Local Structural Variables N Mean ± SD 95% CI Range Flow Parameters V ave (mm/s) 72 1.18 ± 0.46 1.07–1.28 0.26–2.30 V stdev (mm/s) 72 0.33 ± 0.15 0.30–0.37 0.11–0.76 V max (mm/s), raw 72 1.87 ± 0.73 1.70–2.04 0.61–3.73 V min (mm/s), raw 72 0.65 ± 0.31 0.58–0.73 0.09–1.40 Coefficient of variation (Vstdev/Vave) 72 0.29 ± 0.65 0.27–0.30 0.20–0.50 V max (mm/s), curve fit 72 1.70 ± 0.66 1.55–1.86 0.45–3.45 V min (mm/s), curve fit 72 0.65 ± 0.28 0.59–0.72 0.08–1.35 Peak time (ms) 72 621 ± 253.90 562.34–681.91 103.97–1093.84 Period (ms) 72 908.51 ± 85.87 887.93–928.28 638.15–1236.46 Pulsatility contrast 72 0.45 ± 0.10 0.42–0.47 0.25–0.72 Abruptness 72 0.18 ± 0.13 0.15–0.21 0.00–0.74 Local Structural Variables Diameter (µm) 72 4.33 ± 0.90 4.11–4.54 3.00–7.75 Tortuosity 46 1.31 ± 0.40 1.19–1.43 1.06–2.73 Vessel length (µm) 46 230.36 ± 128.48 192.21–268.52 32.73–702.17 Local flow variables  Linear cell density (cells/mm) 59 48.06 ± 14.84 44.19–51.93 23.66–93.67  Hematocrit (%) 59 0.41 ± 0.08 0.39–0.43 0.24–0.57 N indicates number of capillary vessels. Low abruptness value indicates a quick rise time and vice versa. General Flow Characteristics Each vessel segment from three subjects (total of 72 vessels) was quantified for flow parameters (Table). For each AO video, a peak-time offset for each vessel was calculated by subtracting the individual peak times from the average peak time (of all the vessels within a given field). The peak-time offset serves as a dispersion measure for the relative phase of the cardiac cycle through vessels in the field. The peak-time offsets in a field ranged from 4.6 ms to −230.3 ms. Approximate heart rates (calculated as 1/period × 60 seconds) ranged between a minimum of 48 and a maximum of 94 beats per minute. Variability Within Individual Subjects When studied within single subjects, variations in the spread of flow parameters were observed across different retinal fields. Data from two subjects are presented here (Fig. 5). As an example, the average velocities ranged between 0.5 and 2.0 mm/s, the maximum velocities varied between 0.8 and 3.0 mm/s in a subject, and the peak-time offsets ranged from 13 ms to −128 ms. Within a subject, maximum velocity (noted across different retinal fields) was manifestly correlated with average velocity, as expected (Fig. 5A). However, no such significant relations were noted for abruptness, pulsatility and the peak-time offsets (Figs. 5B–5D). Figure 5. Scatterplots representing the spread of flow parameters as a function of average velocity. (A–C) Spread of Vmax (A), spread of abruptness (B), and spread of peak time offset (C). Points in different colors and symbols represent data from retinal locations 1.75° inferior, 0.75° temporal (in triangle, blue), 1.25° superior (in circle, orange), 2.25° inferior, and 0.5° temporal (in diamond, green) from subject 1. (D) Spread of pulsatility in subject 2. a.u., Arbitrary unit. Colored (and shaped) points represent data from retinal locations 0.5° inferior, 1.5° temporal (in diamond, green), 1° superior (in circle, yellow), 1° nasal, and 0.5° inferior (in square, purple) and 1° temporal, 1° superior (in cross, red), 2° inferior, and 0.5° temporal (in triangle, blue) from subject 2. Error bars indicate 95% CIs on curve-fit parameters. Spatial Heterogeneity Spatial heterogeneity in the characteristics of cellular flow through nearby capillaries (as described earlier) was observed by analyzing a minimum of three and a maximum of 10 adjacent vessel segments from each field that had been imaged contemporaneously. This was assessed in different retinal fields across our three study subjects. An example for spatial heterogeneity within a retinal field 1.25° superior (to the FAZ) is shown in Figure 6, where the variability in flow parameters among neighboring vessels is clearly shown. A quick rise time (low abruptness) is seen in vessel segment “e,” whereas in contrast segment “i” shows an abruptness as high as 0.38 (indicating a slow rise time). Similarly, a great variability in peak-time offset can be seen within neighboring vessels (in a retinal field 1.75° inferior and 0.5° temporal); one vessel has an offset of −34 ms whereas another vessel within the same field has an offset of −128 ms, which indicates a significant dispersion in peak times within a field. Figure 6. Spatial heterogeneity of velocity plots within an imaged field. Curve fitted velocity plots for neighboring vessels from a field 1.25° superior to the FAZ from subject 1 (unique vessels are indicated by an alphabetical label and their corresponding velocity plots are presented). Arrows indicate blood flow direction; the venous end is indicated by a blue star. Scale bar: 50 µm. We estimated the pulse wave velocity (PWV) of propagation by comparing the path difference and time difference in the peaks for neighboring vessels within a retinal field. For example, the PWV for the field presented in Figure 6 was calculated using feeder distance and the peak times for two sample vessels—namely, segment “g” and segment “a” as follows: (1) PWV=(feederdistance1-feederdistance2)/(timetopeak1-timetopeak2)PWV=(517.68-385.97)/(674.17-661.83)=11mm/s This is comparable to what has been reported by Bedggood and Metha.38 Within each retinal field across our three study subjects, adjacent vessel segments exhibited variable flow patterns. For example, the greatest variations in pulsatility ranged between 0.28 and 0.71 within a single field. Similarly, Vmax ranged between 0.7 and 2.7, and Vmin ranged between 0.3 and 1.2 mm/s within a single field. Correlations of Flow Parameters and Local Vessel Variables An exploratory data analysis was done to reveal any important associations in the dataset and the significant ones have been plotted here. Of a total of 15 independent correlations run between flow parameters and local vessel variables from a total of 72 vessel segments (n = 3 subjects), the following correlations were found to be statistically significant (Fig. 7). Figure 7. Scatterplot representing correlations between different flow and local vessel variables. (A, B) Correlations within the flow parameters: A (r = −0.28, P = 0.016) and B (r = −0.49, P = 0.0001). (C–F) Correlations between flow variables and local vessel structural variables: C (r = 0.30, P = 0.045), D (r = −0.30, P = 0.038), E (r = −0.04, P = 0.701), and F (r = −0.10, P = 0.503). Error bars represent 95% CIs on the curve-fit parameters of abruptness and Vmin. The COV data do not show any error bars, as these are computed from raw measures (Vave and Vstdev). Relation of Local Flow Variables to the Velocities LCDs and hematocrits were analyzed and could be reliably computed for 59 of the 72 vessel segments (across all three subjects). These LCDs averaged to 48.06 ± 14.7 cells/mm (mean ± SD), and the hematocrits averaged to 0.41 ± 0.08. Figure 8 depicts the general distribution of these parameters with respect to a Gaussian fit. The LCD and hematocrit values were analyzed for their associations with the flow measures and possible structural variables. Among all of the possible associations between the variable sets of LCDs and hematocrits with the flow parameters (i.e., a total of 10 independent correlations were run), LCD was found to be positively correlated with vessel length (r = 0.46; P = 0.004; n = 37). A low negative correlation was noted between the LCD and Vmin_raw (r = −0.27; P = 0.034; n = 59) (Fig. 9). Figure 8. Histogram representations of LCD and hematocrit data. (A) Distribution of LCD showing a left skew relative to the normal Gaussian fit indicated as a black bell-shaped curve. (B) Normal distribution of hematocrit. Figure 9. (A, B) Scatterplots representing the variability in LCD with vessel length (A) and Vmin_raw (B). (C, D) Scatterplot representing relations between instantaneous velocity estimates and vessel flow variables within a single vessel segment. Data presented were taken from multiple short windows (of 100-ms width) within a single vessel. LCDs in general were positively correlated with the hematocrits, as noted across 59 vessel segments (r = 0.51; P = 0.0001). Unlike LCDs, however, no significant associations were found for average hematocrits across all vessel segments with any of the structural variables and flow measures. In order to explore any relation of hematocrits with flow measures, instantaneous velocity and hematocrit (and LCD) variables were studied at the individual vessel level. Out of 59 vessels, nearly 50% of the vessels had a positive correlation between the hematocrit and the instantaneous velocity, and 37% of the vessels had a positive correlation between LCD and velocity. To further illustrate some of the above correlations, some examples of significant relations from individual vessels are presented in Figures 9C and 9D. In a vessel segment from the retinal field 1.75° inferior and 0.75° temporal to FAZ in subject 1, an inverse relation was noted between instantaneous velocities and LCD (r = −0.46; P < 0.001). In another vessel segment from the same field, a moderate positive relation was noted between the instantaneous velocity and hematocrit (r = 0.55; P < 0.001). Discussion This study provides data and analysis that potentially offer insights into capillary flow characteristics in normal human retinas. One of the interesting findings from the present study is that the flow of red blood cells through one vessel segment can be quite different from the flow through its immediate neighbors (i.e., there is marked spatial flow heterogeneity). This was observed for a number of measures such as the average velocity, minimum velocity, maximum velocity, pulsatility, peak time, and abruptness within individuals, and within each retinal field. Retinal Capillary Vessels Exhibit Pulsatility One of the important observations from this study was that pulsatility in cellular flow was observed within all of the vessels in our subjects. This agrees with the results of studies in human subjects assessing 22 and 41 vessels.4,5 We have confirmed this phenomenon in vessels (n = 72) exhibiting velocities ranging from a minimum of 0.1 mm/s to a maximum of 3.7 mm/s. The overall erythrocyte pulsatility averaged 0.45 (range, 0.25–0.72), which is lower compared to two other studies, which reported average pulsatility values of 0.79 (0.48–1.28)4 and 0.91 (0.63–1.57),5 similar to the leukocyte pulsatility range of 0.54–0.61 reported by Tam et al.9 and pulsatility of 0.45 (0.31–0.65) reported by Martin and Roorda.8 These studies measured pulsatility using (Vmax – Vmin)/Vmean. In the current study, we calculated pulsatility using the formula (Vmax – Vmin)/(Vmax + Vmin). This difference in the way pulsatility was measured would have resulted in slightly lower values in our study as compared to other studies. Capillary flow pulsatility is supported by the rhythmic action of the heart, where the alternating systolic and diastolic pressures of the cardiac cycle drives temporal variations in the velocities measured within the retinal capillaries.4,5,19,21,22 The shape of the waveforms in our study appears similar to those described in other reports.5 Traditionally, it was assumed that the pulse pressure wave arising from the contraction of the left ventricle in the heart is dampened within the arterial system, so that there is no evidence of the pressure variations remaining at the capillary level.40 If the pressure waves were damped in this way before reaching the capillary bed, we would not be able to measure rhythmic variations in capillary velocity, as reported here and elsewhere 4,5,19,21,22 In addition, direct measurement of pulse wave propagation through retinal capillaries has recently been confirmed by Bedggood and Metha.38 An outstanding question is whether the degree of pulsatility observed in flow velocity can be used to infer the compliance of retinal vessels, whereby stiffer vessels are known to be less compliant (and, hence, for example, propagate changes in pressure more rapidly). Measurement of microvascular compliance could provide a powerful tool to measure changes in the stiffness of small vessels that may occur in vascular disease such as stroke. Perhaps relevant to this goal is our observation that pulsatility values were directly related to the systolic rising phase proportions (indicated by quick “abruptness”). It was observed that vessels with steep rising-phase proportions were more pulsatile in nature. This is similar to the findings from nail-fold capillaries where the capillary pulse pressure amplitude (that is, the difference between maximum and minimum pressures) were strongly correlated to the slope during systolic proportion.41 Here, we speculate that the reason for observing a range of pulsatility values within a normal individual could be due to the different path lengths taken by blood traveling through different upstream routes to the same vessel bed. In retinal vascular disease, a range of pulsatility values within a capillary bed could indicate that some vessel paths were stiffer, as a result of pathological processes, while other paths remain relatively unaffected. In more advanced disease where all upstream vessels are very stiff (or non-compliant), one could imagine less variability in pulsatility (as well as faster propagation of the pulse wave) throughout the capillary bed. Thus, the pulsatility parameter could be explored in future studies to investigate its potential utility in studying diseased populations. Variability in Average Velocities The average erythrocyte velocities reported here are comparable to those of previously published studies using AO imaging. The Vave ± SD (range) values of 1.18 ± 0.46 (0.26–2.30) found here are in close agreement to those reported previously: 1.33 ± 0.28 (1.30–1.80),7 1.14 (0.30–2.26),4 1.22 (0.58–1.98),5 and 1.70 (0.93–3.32) mm/s.6 A mean Vmax of 1.87 (range, 0.61–3.73) observed here matched very well with previously reported Vmax values of 1.88 (0.80–3.35)4 and 1.88 (0.71–3.98).5 Figure 5 shows that the average velocity is correlated to the maximum. Although this may seem trivial, the very strong degree of correlation may be of use in that the average speed can be reliably predicted if the peak speed is known, both from the point of view of developing fluid mechanical flow models and from the perspective of developing biomarkers that do not require that every vessel has been interrogated for multiple cardiac cycles. In the current study, using frame rates of 200 to 300 fps, we were able to capture flow information to a maximum velocity of about 3.7 mm/s. We have not reported on vessels with velocities higher than this, as fast-flowing vessels give the appearance of flow reversals and were not suitable for our study. So, our analysis may not be representative of all vessels in a given network. However, it is the best we could do with the available methodology and since we have characterized all vessels with clear single corpuscular flow, the results presented in this study are the best attempt of characterizing flow in a subset of capillary vessels. Flow Heterogeneity Within a Network The initial results from our own group were the first, to the best of our knowledge, to indicate that capillary vessels exhibit a greater spatial heterogeneity (relative to the temporal heterogeneity) in a number of flow parameters such as Vmin, Vmax, Vave, abruptness, peak time, and period (Neriyanuri S, et al. IOVS 2019;60:ARVO E-Abstract 4596). Here, we also explored a number of local vessel factors that can possibly give rise to variability in flow across the vascular network, an investigation that has not previously been undertaken. Microvascular heterogeneity is not altogether unexpected in metabolically active tissues such as heart, brain, and the retina.42–44 In the retina, at least two fundamental mechanisms—namely, stochastic angiogenesis and the dimensional problem—have been postulated that might lead to the heterogeneity in microvascular distribution and heterogeneity in local blood flow on a small scale.45 Retinal tissue has marked structural and dimensional heterogeneity, with organizations of neurons into distinct cellular nuclear layers separated by synaptic layers. At the level of small vessels (such as capillaries), flow is expected to be very sensitive to even small changes in vessel structure. Flow heterogeneity has thus been postulated to arise as a consequence of heterogeneous intraretinal oxygen consumption to meet the metabolic demands of variably distributed retinal neural cells.46 Alternatively, Jespersen and Østergaard,47 modeling capillary transit time heterogeneity in brain, reported that if heterogeneity exists then there is greater spare capacity to shift to more homogeneous states in conditions of physiological stress such as hypoxia and hyperemia, thereby improving overall oxygen extraction efficiency. Effect of Local Vessel Structural Variables on Flow For the sample of capillaries included in this analysis, vessel diameters were found to be poorly associated with any of the flow variables in our subjects. This is similar to findings in mouse retina, where vessel diameters did not show any significant associations with the flow velocities in capillary vessels.22 This is in contrast to the larger vessels, where changes in velocity have been reported following changes in diameter in accordance with Newtonian predictions.22,48,49 Although we observed that vessel segments with shorter lengths exhibited higher COV (or pulsatility) compared to vessels with longer lengths, no significant relations were found between flow parameters and vessel tortuosity. The significant association between vessel length and pulsatility is in accord with basic flow mechanics of vessels for any caliber, where the amount of resistance is expected to accrue the greater the length of vessel through which the blood must travel. The lack of an association between vessel tortuosity and flow parameters is in accord with the predictions of fluid dynamics. Capillaries feature slow flow and narrow diameters, giving them a Reynolds number far less than 1, indicating that the viscous forces far outweigh the contribution of inertial forces acting on the fluid.26 This means that twists and bends in a vessel that might be expected to result in significant energy dissipation in larger vessels may not have a significant effect at the capillary level. Local Vessel Flow Variables and Flow Heterogeneity The local flow variables measured for vessels included LCD and hematocrit. Our findings on average hematocrits (0.41) and their range (0.24–0.57) are similar to those reported in the literature. Mouse retinal capillary hematocrits varied between 0.24 and 0.43.20 An average hematocrit of 0.49 was noted in mouse retina in another study.50 These values are comparable to the findings reported in choroid and the brain capillaries in rats.51 There is a paucity of information on the cell counts in human retinal capillaries, but the findings reported here on LCD and hematocrit should help establish normal ranges in the future. A proportional relationship between velocity and hematocrit was seen in at least 50% of the vessel segments in our subjects. In general, an increase in velocity with increasing hematocrits indicates adequate oxygen supply to meet the high oxygen demands of the inner retina.52 Within an individual vessel, as well, a consistent increase in velocity as a function of hematocrit/LCD content was noted, which suggests a plasma redistribution occurring locally with changing velocities. A possible explanation would be that an increased oxygen demand causes the retinal vessels to dilate; in this state, where the vessels have lower resistance, the cells can enter the vessel more readily and therefore the hematocrit can increase. Study Limitations The depth of focus of the imaging system may have influenced our diameter measurements, as a focus favoring superficial vessels would appear to blur slightly deeper vessels and vice versa. The depth of focus of our system is approximately 0.05 to 0.10 D (approximately 15–30 µm for an emmetropic eye). Our imaged locations were also close to the edge of the FAZ where there is not much stratification of the retinal capillary beds. Therefore, one would anticipate little error in vessel diameter because of geometric blur. However, the refractive and scattering properties of blood constituents imaged within the capillaries are known to be detectable well outside the vascular lumen, manifesting as secondary bright bands in the motion contrast image. This creates uncertainty in measured lumen diameters, which can be mitigated somewhat by considering the brightest band in the motion contrast image.25,35,53 Future Recommendations As our data were collected adjacent to the foveal avascular zone, we did not consider a sufficiently broad range of retinal locations to analyze any dependency of flow parameters on eccentricity in the retina. We speculate that the avascular zone could be a region of comparatively great oxygen demand because it is the first opportunity for the inner retinal neurons to receive any inner retinal blood supply and because of the high density of neurons in the fovea. Further work is needed to explore this. Our pilot analysis (n = 4 vessels) on temporal heterogeneity (i.e., variations in flow with time) in the same vessel imaged up to five times (spanning 2–38 minutes) showed that the flow parameters varied only minutely. The greatest ranges in Vmax, Vmin, abruptness, and pulsatility contrast were recorded as 1.75 to 2.14 mm/s, 0.31 to 0.48 mm/s, 0.19 to 0.33, and 0.38 to 0.52, respectively. These preliminary observations on the stability of flow parameters in the same vessel over time should be confirmed with more observational work. Conclusions The results from this study indicate spatial heterogeneity in the characteristics of normal red cell flow through parafoveal capillary networks. The average velocities within a vessel were moderately influenced by the linear cell density and hematocrit content within that vessel. The overall flow among retinal capillaries was not associated with vessel diameter or vessel tortuosity, whereas vessel length was weakly associated with velocity. Because local vessel factors did not significantly influence the flow within a vessel, it would be useful to consider the variables at the network level to further explore which factors drive the observed variations in capillary flow characteristics. Acknowledgments Supported by an Australian Government Research Training Program Scholarship (SN) and the Australian Research Council Discovery Project (DP180103393to AM, PAB). Figures 1, 3, 5 and 6 have been reproduced (under CC BY-NC-ND 4.0) from the material presented by the first author (SN) at ARVO 2019, with modifications to Figure 5 (replotted using complete set of data). Disclosure: S. Neriyanuri, None; P. Bedggood, None; R.C.A. Symons, None; A.B. Metha, None ==== Refs References 1. Wong-Riley MTT. Energy metabolism of the visual system. Eye Brain. 2010; 2 : 99–116.23226947 2. Campbell JP, Zhang M, Hwang TS, et al . Detailed vascular anatomy of the human retina by projection-resolved optical coherence tomography angiography. Sci Rep. 2017; 7 : 42201.28186181 3. Mutlu F, Leopold IH. The structure of human retinal vascular system. Arch Ophthalmol. 1964; 71 (1 ): 93–101.14066048 4. de Castro A, Huang G, Sawides L, Luo T, Burns SA. Rapid high resolution imaging with a dual-channel scanning technique. Opt Lett. 2016; 41 (8 ): 1881–1884.27082369 5. Gu B, Wang X, Twa MD, Tam J, Girkin CA, Zhang Y. Noninvasive in vivo characterization of erythrocyte motion in human retinal capillaries using high-speed adaptive optics near-confocal imaging. Biomed Opt Express. 2018; 9 (8 ): 3653–3677.30338146 6. Warner RL, Gast TJ, Sapoznik KA, Carmichael-Martins A, Burns SA. Measuring temporal and spatial variability of red blood cell velocity in human retinal vessels. Invest Ophthalmol Vis Sci. 2021; 62 (14 ): 29. 7. Bedggood P, Metha A. Direct visualization and characterization of erythrocyte flow in human retinal capillaries. Biomed Opt Express. 2012; 3 (12 ): 3264–3277.23243576 8. Martin JA, Roorda A. Pulsatility of parafoveal capillary leukocytes. Exp Eye Res. 2009; 88 : 356–360.18708051 9. Tam J, Tiruveedhula P, Roorda A. Characterization of single-file flow through human retinal parafoveal capillaries using an adaptive optics scanning laser ophthalmoscope. Biomed Opt Express. 2011; 2 (4 ): 781–793.21483603 10. Tang J, Mohr S, Du YP, Kern TS. Non-uniform distribution of lesions and biochemical abnormalities within the retina of diabetic humans. Curr Eye Res. 2003; 27 (1 ): 7–13.12868004 11. Kern TS, Engerman RL. Vascular lesions in diabetes are distributed non-uniformly within the retina. Exp Eye Res. 1995; 60 (5 ): 545–549.7615020 12. Hudson C, Flanagan JG, Turner GS, Chen HC, Rawji MH, McLeod D. Exaggerated relative nasal-temporal asymmetry of macular capillary blood flow in patients with clinically significant diabetic macular oedema. Br J Ophthalmol. 2005; 89 (2 ): 142.15665341 13. Haj Najeeb B, Simader C, Deak G, et al . The distribution of leakage on fluorescein angiography in diabetic macular edema: a new approach to its etiology. Invest Ophthalmol Vis Sci. 2017; 58 (10 ): 3986–3990.28796876 14. Holm K, Adrian ML. In diabetic eyes, multifocal ERG reflects differences in function between the nasal part and the temporal part of the macula. Graefes Arch Clin Exp Ophthalmol. 2012; 250 (8 ): 1143–1148.22331146 15. Kaizu Y, Nakao S, Yoshida S, et al . Optical coherence tomography angiography reveals spatial bias of macular capillary dropout in diabetic retinopathy. Invest Ophthalmol Vis Sci. 2017; 58 (11 ): 4889–4897.28973335 16. Ikram MK, Cheung CY, Lorenzi M, Klein R, Jones TLZ, Wong TY. Retinal vascular caliber as a biomarker for diabetes microvascular complications. Diabetes Care. 2013; 36 (3 ): 750–759.23431093 17. Burns SA, Elsner AE, Chui TY, et al . In vivo adaptive optics microvascular imaging in diabetic patients without clinically severe diabetic retinopathy. Biomed Opt Express. 2014; 5 (3 ): 961–974.24688827 18. Tam J, Dhamdhere KP, Tiruveedhula P, et al . Disruption of the retinal parafoveal capillary network in type 2 diabetes before the onset of diabetic retinopathy. Invest Ophthalmol Vis Sci. 2011; 52 (12 ): 9257–9266.22039250 19. Bedggood P, Metha A. Adaptive optics imaging of the retinal microvasculature. Clin Exp Optom. 2020; 103 (1 ): 112–122.31797452 20. Guevara-Torres A, Joseph A, Schallek JB. Label free measurement of retinal blood cell flux, velocity, hematocrit and capillary width in the living mouse eye. Biomed Opt Express. 2016; 7 (10 ): 4228–4249.27867728 21. Bedggood P, Metha A. Mapping flow velocity in the human retinal capillary network with pixel intensity cross correlation. PLoS One. 2019; 14 (6 ): e0218918.31237930 22. Joseph A, Guevara-Torres A, Schallek J. Imaging single-cell blood flow in the smallest to largest vessels in the living retina. eLife. 2019; 8 : e45077.31084705 23. Arichika S, Uji A, Hangai M, Ooto S, Yoshimura N. Noninvasive and direct monitoring of erythrocyte aggregates in human retinal microvasculature using adaptive optics scanning laser ophthalmoscopy. Invest Ophthalmol Vis Sci. 2013; 54 (6 ): 4394–4402.23716632 24. Shigeta A, Akihito U, Sotaro O, Kazuaki M, Nagahisa Y. Adaptive optics-assisted identification of preferential erythrocyte aggregate pathways in the human retinal microvasculature. PLoS One. 2014; 9 (2 ): e89679.24586959 25. Bedggood P, Ding Y, Metha A. Measuring red blood cell shape in the human retina. Opt Lett. 2023; 48 (7 ): 1554–1557.37221708 26. Fung Y-C . Biomechanics: Circulation, 2nd ed. New York: Springer-Verlag; 1996. 27. Balogh P, Bagchi P. Analysis of red blood cell partitioning at bifurcations in simulated microvascular networks. Phys Fluids. 2018; 30 (5 ): 051902. 28. Balogh P, Bagchi P. Direct numerical simulation of cellular-scale blood flow in 3D microvascular networks. Biophys J. 2017; 113 (12 ): 2815–2826.29262374 29. Bagchi P. Mesoscale simulation of blood flow in small vessels. Biophys J. 2007; 92 (6 ): 1858–1877.17208982 30. Duan A, Bedggood PA, Bui BV, Metha AB. Evidence of flicker-induced functional hyperaemia in the smallest vessels of the human retinal blood supply. PLoS One. 2016; 11 (9 ): 1–17. 31. Bedggood P, Metha A. Optical imaging of human cone photoreceptors directly following the capture of light. PLoS One. 2013; 8 (11 ): 1–10. 32. Duan A, Bedggood PA, Metha AB, Bui BV. Reactivity in the human retinal microvasculature measured during acute gas breathing provocations. Sci Rep. 2017; 7 (1 ): 2113.28522835 33. Delori FC, Webb RH, Sliney DH. Maximum permissible exposures for ocular safety (ANSI 2000), with emphasis on ophthalmic devices. J Opt Soc Am A Opt Image Sci Vis. 2007; 24 (5 ): 1250–1265.17429471 34. Tam J, Martin JA, Roorda A. Noninvasive visualization and analysis of parafoveal capillaries in humans. Invest Ophthalmol Vis Sci. 2010; 51 (3 ): 1691–1698.19907024 35. Bedggood P, Metha A. Analysis of contrast and motion signals generated by human blood constituents in capillary flow. Opt Lett. 2014; 39 (3 ): 610–613.24487878 36. Otsu N. A threshold selection method from gray-level histograms. IEEE Trans Syst Man Cybern Syst. 1979; 9 (1 ): 62–66. 37. Bedggood P, Metha A. Recovering the appearance of the capillary blood column from under-sampled flow data. Opt Lett. 2020; 45 (15 ): 4320–4323.32735288 38. Bedggood P, Metha A. Direct measurement of pulse wave propagation in capillaries of the human retina. Opt Lett. 2021; 46 (18 ): 4450–4453.34525019 39. Mukaka MM. Statistics corner: a guide to appropriate use of correlation coefficient in medical research. Malawi Med J. 2012; 24 (3 ): 69–71.23638278 40. Burton AC. Physiology and Biophysics of the Circulation: An Introductory Text. Chicago, IL: Year Book Medical Publishers; 1972. 41. Shore AC, Sandeman DD, Tooke JE. Capillary pressure, pulse pressure amplitude, and pressure waveform in healthy volunteers. Am J Physiol. 1995; 268 (1 ): H147–H154.7840258 42. Hudlicka O, Wright AJ, Ziada AM. Angiogenesis in the heart and skeletal muscle. Can J Cardiol. 1986; 2 (2 ): 120–123.2423212 43. Patel-Hett S, D'Amore PA. Signal transduction in vasculogenesis and developmental angiogenesis. Int J Dev Biol. 2011; 55 (4-5 ): 353–363.21732275 44. Pries AR, Secomb TW, Gaehtgens P. Structure and hemodynamics of microvascular networks: heterogeneity and correlations. Am J Physiol. 1995; 269 (5 ): H1713–H1722.7503269 45. Pries AR, Secomb TW. Origins of heterogeneity in tissue perfusion and metabolism. Cardiovasc Res. 2009; 81 (2 ): 328–335.19028725 46. Yu D-Y, Cringle SJ, Paula KY, et al . Retinal capillary perfusion: spatial and temporal heterogeneity. Prog Retin Eye Res. 2019; 70 : 23–54.30769149 47. Jespersen SN, Østergaard L. The roles of cerebral blood flow, capillary transit time heterogeneity, and oxygen tension in brain oxygenation and metabolism. J Cereb Blood Flow Metab. 2012; 32 (2 ): 264–277.22044867 48. Zhong Z, Song H, Chui TYP, Petrig BL, Burns SA. Noninvasive measurements and analysis of blood velocity profiles in human retinal vessels. Invest Ophthalmol Vis Sci. 2011; 52 (7 ): 4151–4157.21467177 49. Palochak CMA, Lee HE, Song J, Geng A, Linsenmeier RA, Burns SA, et al . Retinal blood velocity and flow in early diabetes and diabetic retinopathy using adaptive optics scanning laser ophthalmoscopy. J Clin Med. 2019; 8 (8 ): 1165.31382617 50. Paques M, Tadayoni R, Sercombe R, et al . Structural and hemodynamic analysis of the mouse retinal microcirculation. Invest Ophthalmol Vis Sci. 2003; 44 (11 ): 4960–4967.14578423 51. Seylaz J, Charbonné R, Nanri K, et al . Dynamic in vivo measurement of erythrocyte velocity and flow in capillaries and of microvessel diameter in the rat brain by confocal laser microscopy. J Cereb Blood Flow Metab. 1999; 19 (8 ): 863–870.10458593 52. Cringle SJ, Yu DY, Yu PK, Su EN. Intraretinal oxygen consumption in the rat in vivo. Invest Ophthalmol Vis Sci. 2002; 43 (6 ): 1922–1927.12037000 53. Guevara-Torres A, Williams DR, Schallek JB. Origin of cell contrast in offset aperture adaptive optics ophthalmoscopy. Opt Lett. 2020; 45 (4 ): 840–843.32058484

PMC010xxxxxx/PMC10353744.txt

==== Front Transl Vis Sci Technol Transl Vis Sci Technol TVST Translational Vision Science & Technology 2164-2591 The Association for Research in Vision and Ophthalmology 37440248 10.1167/tvst.12.7.15 TVST-22-5126 Glaucoma Glaucoma Deviated Saccadic Trajectory as a Biometric Signature of Glaucoma Deviated Saccadic Trajectory of Glaucoma Yeon Ji Su 1 Jung Ha Na 1 Kim Jae Young 1 Jung Kyong In 2 3 Park Hae-Young Lopilly 2 3 Park Chan Kee 2 3 Kim Hyo Won 1 Kim Man Soo 1 Kim Yong Chan 2 4 1 Gangnam St. Mary's One Eye Clinic, Seoul, Republic of Korea 2 College of Medicine, The Catholic University of Korea, Seoul, Republic of Korea 3 Department of Ophthalmology, Seoul St. Mary's Hospital, College of Medicine, The Catholic University of Korea, Seoul, Republic of Korea 4 Department of Ophthalmology, Incheon St. Mary's Hospital, College of Medicine, The Catholic University of Korea, Seoul, Republic of Korea * Correspondence: Yong Chan Kim, Department of Ophthalmology, Incheon St. Mary's Hospital, College of Medicine, The Catholic University of Korea, Republic of Korea, 56, Dongsu-ro, Bupyeong-gu, Incheon 21431, Republic of Korea. e-mail: yongchankim@catholic.ac.kr 13 7 2023 7 2023 12 7 1531 5 2023 19 9 2022 Copyright 2023 The Authors 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. Purpose To investigate whether the trajectories of saccadic eye movements (SEMs) significantly differ between glaucoma patients and controls. Methods SEMs were recorded by video-based infrared oculography in 53 patients with glaucoma and 41 age-matched controls. Participants were asked to bilaterally view 24°-horizontal, 14°-vertical, and 20°-diagonal eccentric Goldmann III-sized stimuli. SEMs were evaluated with respect to the saccadic reaction time (SRT), the mean velocity, amplitude, and two novel measures: departure angle (DA) and arrival angle (AA). These parameters were compared between the groups and the associations of SEM parameters with glaucoma parameters and integrated visual field defects were investigated. Results Glaucoma patients exhibited increased mean SRT, DA, and AA values compared with controls for 14°-vertical visual targets (P = 0.05, P < 0.01, and P < 0.01, respectively). The SRT, DA, and AA were significantly associated with the mean and pattern standard deviations of perimetry and with the mean RNFL thickness by OCT (all P < 0.001). Glaucoma was associated with the AA (P = 0.05) and both the SRT (P = 0.01) and DA (P = 0.04) were associated with integrated visual field defects. Conclusions The saccadic trajectories of glaucoma patients depart in an erroneous path and compensate the disparity by deviating the trajectory at arrival. Translational Relevance The initial deviation that we observed (despite continuous exposure to the stimulus) suggests the disoriented spatial perception of glaucoma patients which may be relevant to difficulties encountered daily. saccade spatial cognition glaucoma eye tracking ==== Body pmcIntroduction Glaucoma is a multifactorial, progressive optic neuropathy that leads to gradual and irreversible visual field loss. Glaucoma is a leading cause of acquired blindness worldwide and is characterized by axonal degeneration that affects the afferent visual pathway (i.e., from retinal ganglion cells to the lateral geniculate nucleus and the visual cortex).1–3 Peripheral vision is most susceptible to glaucomatous damage, substantial changes are evident in periphery before any loss of central visual acuity.4 Nonetheless, patients with glaucomatous optic neuropathy (GON) experience difficulties with everyday activities and have an increased risk of falls.5–9 The human visual acuity is highest for images that reach the fovea where the photoreceptor density is greatest and lowest for images that reach peripheral retinal regions.10,11 To maintain adequate vision of the surrounding environment, sufficient perception of peripheral targets and a swift gaze shift are needed to bring the fovea to a position where it can evaluate the area of interest.12 The saccadic system calculates the distance and direction of a peripheral target from the current gaze position, then generates high-velocity movements of both eyes that bring the image onto or near the fovea.13 Therefore, the struggles of daily living may be affected not only by poor perception of peripheral targets, but also because of a miscarried saccadic system that fails to properly shift gaze. The saccadic oculomotor system has been extensively studied because saccadic eye movements (SEMs) can be accurately measured,14–16 the neurons that control saccades are readily accessible to microelectrodes,17–19 and the neural network underlying saccade generation is well-known.12,13 Reliable SEM parameters include the saccadic latency or saccade reaction time (SRT),20 saccadic duration,21 amplitude,22 and peak velocity.23 The SRT is increased in patients with various optic nerve pathologies that affect the retinal ganglion cells necessary to convey visual signals to the saccade-generating network.24–26 These parameters are also used to assess the integrity of the saccade-generating neural network in patients with various brain diseases.27–29 Most recent eye movement studies have used high-speed video-based oculography. Because up to 1000 eye movement coordinates may be registered per second, saccades are often simplified to straight lines that connect the start and end points of an eye movement above a certain velocity and acceleration.30 However, the real trajectory may not always be a straight line, and some of the movement path can be curved or irregular. In particular, it is known that saccades curve away from covertly attended locations and even from visual distractors, and the magnitude of such curvature is associated with the saliency of the distractor.31,32 Based on this speculation, we investigated whether SEMs deviated from the intended paths in subjects with GON. We focused on whether the trajectory changed after commencement of a saccade, which might suggest distorted spatial perception of an intended target. Therefore, the saccade departure angle (DA) and arrival angle (AA) that measures deviation of the path from the intended target were quantified. The angles between the intended target and the actual trajectory throughout an SEM set were measured in 10 different temporal sections. Conventional saccade parameters (e.g., SRT, mean and peak velocities, and amplitude) were also acquired and compared among normal controls and glaucoma patients. Materials and Methods After the study protocol had been approved by the Catholic University Incheon St. Mary's Hospital Research Ethics Board, informed consent was obtained from glaucoma patients and age-matched controls between the ages of 26 and 84 years. Patients with GON (n = 53) were prospectively recruited from the practice of a glaucoma specialist (Y.C.K.) from October 2021 to May 2022. GON was defined by characteristic changes in the optic nerve head and thinning of the retinal nerve fiber layer (RNFL), with corresponding visual field changes determined by white-on-white static automated perimetry (Carl Zeiss Meditec, Jena, Germany). The exclusion criteria were all other causes of secondary glaucoma; any nonglaucomatous eye disease; incisional eye surgery within the previous 1 month; central visual acuity worse than 0.2 logarithm of the minimum angle of resolution; any history of neurological disease, psychiatric condition, or neurological disorder (including cognitive impairment or dementia); and any use of psychotropic medications known to affect saccade velocity.33 Age-matched controls (n = 41) were recruited from among patients who were undergoing follow-up because of high cup-to-disc ratios but otherwise exhibited normal findings during consecutive eye examinations. GON patients and age-matched controls underwent comprehensive clinical eye examinations including Landolt C chart–assisted measurements of logarithm of the minimum angle of resolution best-corrected visual acuity, slit-lamp biomicroscopy, axial length measurement via ocular biometry (IOL Master; Carl Zeiss Meditec), digital color fundus photography (VX-10i camera, Kowa Co., Nagoya, Japan), and optical coherence tomography (DRIOCT Triton; Topcon Corporation, Tokyo, Japan). The study conformed to the Declaration of Helsinki, and all data were anonymized before transfer to a secure computer database at the university. Eye Movement Recordings All glaucoma patients and controls were tested under constant luminance of 300 cd/m2 (LS-150 Luminance Meter, Osaka, Japan). Each participant was seated on a chair with the head stabilized by a chinrest at 60 cm from a light-emitting diode backlit display monitor (59.8 × 33.6 cm; 27 inches; 16:9 aspect ratio) with a resolution of 1920 × 1080 pixels and a refresh rate of 75 Hz (LG 27MK430H, LG Electronics, Seoul, Korea). Binocular eye movements were simultaneously recorded using an EyeLink 1000 Plus device (SR Research Ltd., Mississauga, Ontario, Canada) that recorded eye position coordinates at 1000 Hz. The mean visual angle accuracy exceeded 0.5° over the trackable range of 32° × 25°. The default saccade detection thresholds were velocity greater than 30°/s and acceleration greater than 4000°/s2. Before testing, calibration (13 positions) was performed; the standard was required to be good, as defined by the manufacturer. A drift check was performed between each trial; if drift was substantial, recalibration was conducted automatically. Trial Sequence and Saccade Stimulus Trial sequence is described in Figure 1. (i) Participants were instructed to fixate on a 0.43°-diameter, circular fixation point (equivalent to Goldmann size III, the standard stimulus of the Humphrey Visual Field Analyzer) at the center of the monitor. (ii) After random period of 1500 to 3000 ms, the fixation point disappears and simultaneously a peripheral target appears. Subjects are instructed to perform a single saccade to the target as soon as possible. (iii) After subjects fixated on the target for 5000 ms, the target point disappears and simultaneously the central fixation point reappears at the same location. (iv) The fixation point is presented for random period of 1500 to 3000 ms and another peripheral target in the random order appears with the fixation point disappearing simultaneously. (v) The trials are repeated 16 times at each of 8 locations. Figure 1. Schematic diagram showing the trial sequence. (i) A 0.43°-diameter, circular fixation point appears in the center of the screen and subjects were instructed to fixates on the point. (ii) After random period of 1500 to 3000 ms, the fixation point disappears and simultaneously a peripheral target appears. Subjects are instructed to perform a single saccade to the target as soon as possible. (iii) After subjects fixated on the target for 5000 ms, the target point disappears and simultaneously central fixation point reappears at the same location. (iv) The fixation point is presented for random period of 1500 to 3000 ms and another peripheral target in the random order appears with the fixation point disappearing simultaneously. (v) The trials are repeated 16 times at each of 8 locations. The stimulus had a luminance of 127 cd/m2 (equivalent to 14 dB on the Humphrey Visual Field Analyzer luminance scale) against a background luminance of 10 cd/m2. Figure 2 depicts the location of the saccade stimuli (blue dots) superimposed on the placement of the standard visual field test (black). Eight stimulus locations at which glaucomatous arcuate scotoma frequently occur were designated as stimulus points. The stimulus was a 0.43°-diameter located at 24° of horizontal eccentricity, 14° of vertical eccentricity, and 20° of diagonal eccentricity which corresponded to the placement of the standard automated perimetry (Figure 2). Participants were instructed to look toward the presenting stimulus as rapidly and accurately as possible after presentation. Figure 2. Schematic diagram showing the location of the saccade stimuli (blue dots) superimposed on the placement of the standard visual field test (black). Each black numbered square represents a visual field test stimulus point in the respective left and right eye format. Eight sites at which glaucomatous arcuate scotoma frequently occur were designated as stimulus points. A 0.43°-diameter stimulus at 24° of horizontal eccentricity, 14° of vertical eccentricity, and 20° of diagonal eccentricity appeared for 5000 ms. Integrated Binocular Visual Fields Integrated visual field (IVF) was calculated by choosing the higher sensitivity at two overlapping locations on each monocular Humphrey visual field (HVF) test using the Swedish interactive thresholding algorithm 24-2 (Humphrey Visual Field Analyzer; Carl Zeiss Meditec, Dublin, CA, USA) as described previously.34 The IVF determined whether the participant could identify the stimulus points. A point was labeled the IVF defect (IVFD) if the P value of both pattern standard deviation (PSD) was less than 1%. Each IVF point corresponding with a point of the saccade stimulus was identified (Supplementary Table 1). A reliable visual field was defined as a field with less than 20% fixation loss, as well as false-negative and false-positive rates of less than 15%. Data Analysis Saccadic trajectory analysis was performed using a customized algorithm (iDynamics, Seoul, Korea) that automatically detected the start and end points of all saccades from both eyes, then measured the saccadic trajectories between the onset and the offset. To better define the onset and offset of the saccadic trajectory, it is necessary to extract the hesitating eye movements at the start of the saccade and the wandering eye movements after the saccade ends. Therefore, the algorithm considered the starting direction (within 90°) of the saccade along with the minimum size (>6° amplitude) and speed (>30°/s velocity). The offset of the saccade was defined as the most advanced time within 0.2 seconds after finding the peak velocity with the largest magnitude and velocity. Any trial that lacked data concerning more than one-third of a saccadic progression was excluded. Otherwise, missing data were interpolated. The data from each trial (eight different positions) were converted to single x and y coordinates by affine transform generalization (Equation 1). The x axis was a straight path that connected the start point to the intended target, and the y axis was perpendicular to the x axis. All SEMs were converted into points on these axes, regardless of stimulus position (Equation 2). (1) x'y'=cosθsinθ-sinθcosθxy-xtargetn-1ytargetn-1 (2) θ=tan-1ytargetn-ytargetn-1xtargetn-xtargetn-1 An SEM was divided into 10 equidistant intervals, each of which included k scanpaths; each scanpath position was recorded over 0.001 second. The maximum tangent angles formed by all k scanpaths from the starting points within all sections were calculated (Equation 3). (3) xn=xmin+nxmax-xmin10n=0∼10TangentAnglen+1=maxtan-1yk-yk0xk-xk0xn∼xn+1 The maximum tangent angle of the first interval was designated the DA; the maximum tangent angle of the last interval was designated the AA (Figure 3). The SRT, mean and peak velocities, amplitude, and gain (i.e., the ratio of the amplitude of the saccade to the amplitude of the peripheral visual target) of each saccade were acquired using the default settings of EyeLink DataViewer (SR Research Ltd.). The accuracy of automated saccade detection was verified by experienced investigators (Y.J.S. and Y.C.K.). Figure 3. Measurement of saccadic deviation angle. A saccade was divided into 10 equidistant intervals and the maximum tangent angles formed from the starting points within all sections were calculated. The maximum tangent angle of the first interval was designated the DA and the maximum tangent angle of the last interval was designated the AA. The demographic and comprehensive ophthalmic data of all participants who underwent saccade analyses were compared using the Mann–Whitney U test. Associations between saccadic and glaucoma parameters were calculated using the point-biserial method. We used univariate and multivariate logistic regression analyses to identify factors associated with glaucoma; we calculated adjusted odds ratios with 95% confidence intervals. Five independent variables were tested for the correlation and logistic regression analyses, the adjusted P value to reject the null hypothesis was adjusted to 0.01. A P value of less than 0.05 was considered statistically significant in the Mann–Whitney U test comparison. Demographic and ophthalmic data are presented as medians with interquartile ranges. Statistical analyses were performed and figures were prepared using Python software. Results We included 53 GON patients and 41 age-matched healthy controls with mean ages of 54.60 ± 14.11 years and 56.02 ± 12.36 years, respectively (P = 0.28). The sex ratio, best-corrected visual acuity, and axial length did not significantly differ between the groups. However, differences were apparent in terms of the HVF mean deviation (HVF-MD), HVF-PSD, and the global, superior, and inferior RNFL thicknesses (all P < 0.001) (Table 1). We compared 5746 saccades of the GON group with 3990 saccades of the control group by saccadic direction and eccentricity (Table 2 and Supplementary Figure 1). Although there were various differences according to eccentricity and direction, the difference between the two groups was particularly large in the DA and AA at the 14 degree vertical saccade (both P < 0.001). There was also difference in amplitude and SRT by different eccentricity and direction. Table 1. Demographics and Ocular Clinical Characteristics of Patients With Glaucoma and Healthy Control Groups Variables Control Glaucoma P Value* Age, years 56.02 ± 12.36 54.60 ± 14.11 0.28 Sex 19M, 22F 22M, 31F 0.63† BCVA, logMAR 0.93 ± 0.13 0.90 ± 0.16 0.20 IOP, mm Hg 14.53 ± 3.28 15.12 ± 3.32 0.12 Axial length, mm 24.76 ± 1.31 24.83 ± 1.74 0.57 HVF-MD, dB −1.74 ± 1.94 −7.50 ± 7.52 <0.01 HVF-PSD, dB 2.30 ± 1.49 6.78 ± 3.92 <0.01 Global RNFL-T, µm 99.95 ± 11.18 74.51 ± 15.48 <0.01 Superior RNFL-T, µm 120.67 ± 17.98 91.23 ± 26.43 <0.01 Inferior RNFL-T, µm 128.38 ± 16.18 83.58 ± 25.91 <0.01 Data are presented as mean ± standard deviation and compared using the Mann–Whitney U test. BCVA, best-corrected visual acuity; F, females; IOP, intraocular pressure; logMAR, logarithm of the minimum angle of resolution; M, males; RNFL–T, RNFL thickness. * P < .05 are shown in bold. † χ² test for categorical variables. Table 2. Comparison of Saccadic Characteristics in Glaucoma and Healthy Control Groups by Different Eccentricity and Direction Control Glaucoma P Value* Variables 14˚ Vertical 20˚ Diagonal 24˚ Horizontal 14˚ Vertical 20˚ Diagonal 24˚ Horizontal 14˚ 20˚ 24˚ Amplitude, ° 15.21 ± 7.29 18.96 ± 5.75 23.73 ± 7.08 17.65 ± 9.99 21.04 ± 8.27 25.13 ± 7.83 0.15 0.01 0.04 Average velocity, °/s  126.84 ± 55.55 133.56 ± 55.47 144.06 ± 47.01 126.50 ± 96.95 136.90 ± 72.08 147.20 ± 68.57 0.26 0.63 0.78 SRT, ms 288.24 ± 377.61 268.64 ± 424.78 264.92 ± 367.35 315.59 ± 309.18 303.95 ± 448.79 276.40 ± 363.86 0.05 0.03 0.32 DA, ° 20.11 ± 19.66 20.84 ± 13.25 12.56 ± 13.96 31.03 ± 26.46 21.63 ± 16.04 16.20 ± 18.14 0.00 0.91 0.05 AA, ° 20.60 ± 20.88 18.88 ± 16.41 13.58 ± 12.53 31.10 ± 25.51 19.85 ± 15.72 15.91 ± 18.70 0.00 0.10 0.36 Data are presented as mean ± standard deviation and compared using the Mann–Whitney U test. * P < 0.05 are shown in bold. We tested whether the saccade parameters were associated with the conventional glaucoma parameters of the respective eyes (Table 3). Surprisingly, the HVF-MD and HVF-PSD were significantly associated with saccadic trajectory parameters, including the DA and AA (all P < 0.01). The global RNFL thickness was significantly associated with the DA and AA (P < 0.01 and P = 0.01, respectively). The SRT was associated with the HVF-MD, HVF-PSD, and global RNFL thickness (all P < 0.01). However, the amplitude and average velocity were not significantly associated with most of glaucoma parameters. The extents of association differed according to saccade direction; the saccadic trajectory parameters exhibited stronger associations in the upward and downward directions, whereas the horizontal and diagonal saccades were minimally associated with glaucoma parameters (Supplementary Table 2). Univariate analysis revealed that the DA and AA were significantly associated with glaucoma status (P = 0.020 and P = 0.005, respectively); the AA association remained significant on multivariate analysis (P = 0.048) (Table 4). Table 5 explores whether each saccade involved IVFDs at the corresponding stimulus point. Because tests proceeded simultaneously for both eyes, the presence of an IVFD implied that both eyes exhibited a VFD at the same location. We sought associations between the presence of an IVFD and the trajectory parameters. In contrast with the trajectory parameters associated with glaucoma, only the DA was significantly associated with the presence of an IVFD (P = 0.02) (Table 5). On univariate analysis according to IVFD status, the DA and SRT were significantly associated with defects (P = 0.023 and P = 0.003, respectively). On multivariate analysis, the increases in the SRT and DA remained statistically significant (P = 0.006 and P = 0.043, respectively). Table 3. Associations Between the SEMs Parameters With the Glaucoma Parameters HVF-MD HVF-PSD Superior RNFL-T Inferior RNFL-T Global RNFL-T Variables r P Value* r P Value* r P Value* r P Value* r P Value* Amplitude, ° 0.02 0.56 −0.03 0.41 0.03 0.42 0.01 0.83 0.02 0.60 Average velocity, °/s 0.05 0.22 −0.09 0.02 0.05 0.20 0.07 0.07 0.06 0.13 SRT, ms −0.19 0.00 0.24 0.00 −0.18 0.00 −0.17 0.00 −0.19 0.00 DA, ° −0.19 0.00 0.16 0.00 −0.12 0.00 −0.10 0.01 −0.12 0.00 AA, ° −0.15 0.00 0.10 0.01 −0.09 0.02 −0.12 0.00 −0.09 0.01 RNFL–T, RNFL thickness. * P < 0.01 are shown in bold. Table 4. Saccadic Characteristics Associated With Glaucoma. Univariate Analysis* Multivariate Analysis Variables OR 95% CI P Value† OR 95% CI P Value† Amplitude, ° 1.017 0.980–1.055 0.361 Average velocity, °/s  1.000 0.997–1.002 0.866 SRT, ms 1.000 1.000–1.001 0.100 DA, ° 1.010 1.002–1.018 0.020 1.006 0.996–1.015 0.245 AA, ° 1.012 1.004–1.021 0.005 1.010 1.000–1.019 0.048 CI, confidence interval; OR, odds ratio. * Variables with P < 0.05 in univariate analyses were included in multivariate analyses. † P < 0.05 are shown in bold. Table 5. Saccadic Characteristics Associated With IVFDs Univariate Analysis* Multivariate Analysis Variables OR 95% CI P Value† OR 95% CI P Value† Amplitude, ° 1.028 0.983–1.075 0.227 Average velocity, °/s  0.999 0.995–1.002 0.404 SRT, ms 1.001 1.000–1.001 0.003 1.001 1.000–1.001 0.006 DA, ° 1.011 1.002–1.021 0.023 1.010 1.000–1.020 0.043 AA, ° 1.001 0.990–1.011 0.902 CI, confidence interval; OR, odds ratio. * Variables with P < 0.05 in univariate analyses were included in multivariate analyses. † P < 0.05 are shown in bold. Figure 4A shows the MDs between the intended paths and actual eye movements over the 10 sections. Throughout the trajectory, glaucomatous SEMs exhibited significantly larger deviations from the intended paths in nine of the 10 sections. Figure 4B compares the MDs of glaucoma patients without and with IVFDs. Although the differences in the middle of the saccadic trajectories were not statistically significant, the DA and AA significantly differed (P = 0.02 and P = 0.02, respectively) (Supplementary Table 3). Notably, the DA and AA deviations were opposite in nature; patients without IVFDs exhibited a significantly larger AA compared with patients with IVFDs, whereas patients with IVFDs exhibited a significantly larger DA compared with patients without IVFDs. The glaucoma patients and controls were compared in terms of saccadic directions; no significant differences in amplitude, mean or peak velocity, duration, or gain were apparent. However, in terms of the upward saccades, significant differences in the DA and AA parameters were evident (P = 0.01 and P = 0.05, respectively). These trajectory differences were not statistically significant in either horizontal direction. The diagonal saccades exhibited intermediate results in terms of the trajectory parameters; the AAs of the right-downward saccades significantly differed (P = 0.01) (Figure 5). Figure 4. Mean deviations between the intended paths and actual eye movements over the 10 sections. (A) Throughout the trajectory, glaucomatous SEMs exhibited significantly larger deviations from the intended paths in nine of the 10 sections. (B) Notably, the DA and AA deviations were opposite in nature; patients without IVFDs exhibited a significantly larger AA compared with patients with IVFDs, whereas patients with IVFDs exhibited a significantly larger DA compared with patients without IVFDs. Figure 5. Comparison of saccadic parameters with respective directions. No significant differences in amplitude, mean or peak velocity, duration, or gain were apparent between control (A) and glaucoma (B) group. However, significant differences in the DA and AA parameters were evident in upward and downward saccades. Discussion Our results show that, in a randomized sequence of saccades, SEMs from a GON patient departs in an erroneous path and compensates the disparity by also deviating the trajectory at its arrival compared with age-matched controls. The magnitudes of the deviations were associated with certain glaucoma parameters in terms of HVF and optical coherence tomography. Moreover, our trajectory analysis shows that the glaucomatous saccades have their trajectory significantly deviated from the beginning of its initiation. Our results imply that glaucomatous SEMs without IVFDs actively correct their final trajectories to the intended targets, whereas saccades with IVFDs at the corresponding points exhibit less ability to correct these trajectories. Taken together, our findings suggest that a deviated saccadic trajectory is an inherent characteristic of glaucoma; they may explain the difficulties encountered by glaucoma patients during daily life. It has been suggested that SEMs are generated in the cortical eye fields (primary visual, parietal, frontal, and supplementary) and the subcortical network structures (superior colliculus, thalamus, and striatum).35 Therefore, it can be assumed that malfunctions in these neural networks may cause dysfunctional saccades. Among others, the superior colliculus is presumably the origin of such malfunctions. The superior colliculus is the principal conduit of the output saccadic stream to the oculomotor complex; lesions in the superior colliculus eliminate the stream.36,37 Aizawa et al. found that the injection of muscimol (a GABA receptor agonist) changed saccadic trajectories, such that they became consistently curved and slower, with longer latencies, which is equivalent to our findings regarding glaucomatous SEMs.18,36,38 The anatomical details of direct optic nerve projections to the dorsal midbrain and the superior colliculus have been described as well.19 Thus, a glaucomatous optic nerve may affect SEM generation, such that the saccadic trajectory becomes miscarried. To our knowledge, we are the first to suggest that changes in saccadic trajectory constitute an intrinsic feature of glaucomatous SEMs. Previous studies described reduced saccade velocity,39 hypometric saccade amplitude,23 and delayed SRTs.20,25 However, our data clearly imply that an initial saccade deviation is a hallmark of glaucomatous saccades. This hypothesis is supported by proportional associations between the extent of the deviated angle and the glaucoma parameters in terms of HVF and optical coherence tomography. This hypothesis is also consistent with the spatiotemporal properties of glaucomatous SEMs. Soans et al.40 reported that the increased spatiotemporal properties of vertical SEMs clearly distinguished glaucoma patients from healthy individuals. Their spatiotemporal parameters considered the spatial errors and temporal delays between a stimulus and the SEM trajectory, analogous to the angular deviation measured in our study. Notably, their study showed more prominent upward and downward saccades, consistent with our observations. Furthermore, studies that showed reduced accuracies of glaucomatous SEMs (compared with normal SEMs) may be consistent with our hypothesis; a dysfunctional trajectory may end at an erroneous fixation point.41 Therefore, the deviated trajectories of glaucomatous SEMs that we report may not reflect a specific experimental situation; they may be inherent features of glaucomatous SEMs. Our observations have implications in terms of spatial perception by GON patients. Thus far, functionality in glaucoma patients has usually been evaluated using the HVF. This approach only explores whether a patient recognizes a static stimulus when the eye is stationary. The HVF test does not evaluate how a patient perceives the location of a stimulus. Our data suggest that glaucoma patients not only experience VFDs, but also have disoriented spatial perception. The presence of an angular deviation even at the commencement of saccade generation suggests that the patient is aware of a stimulus, but its perceived location may be inaccurate. This hypothesis is supported by the eye movement perimetry measurements collected when SEM directional biases were imposed during the exploration of stimulus recognition.42 We presented stimuli for 5 seconds, whereas the eye movement perimetry stimulus exposure time in the cited work was 0.2 second. The SRT of a typical SEM is 0.2 second; this is the time before the stimulus disappears. Thus, the SEMs of the eye movement perimetry study proceeded to positions at which the patient remembered that a stimulus had been located, creating a directional bias that revealed perception of the spatial orientation. The initial SEM deviation that we observed (despite continuous exposure to the stimulus) emphasizes the disoriented spatial perception. This theoretical framework is consistent with the proposed paradigm of a decreased spatial cognitive map during explorations of virtual reality,43 which may provide additional explanation for the decreased mobility and poor general performance in glaucoma patients.44,45 According to our paradigm, the deviated trajectory is an intrinsic glaucomatous feature based on disoriented spatial perception. However, it could be hypothesized that the deviation is caused by a wandering gaze, failing to locate the target because of a VFD at the corresponding point. Our comparison of the saccade parameters of glaucoma patients with and without IVFDs suggests that, although the DAs are similar, the AAs significantly differ. The AA increased in patients without IVFDs, whereas the DA increased in patients with IVFDs. These findings suggest that an erroneous departure was actively compensated in patients without IVFDs; patients with IVFDs may have limited capacity to compensate for such a departure. Moreover, an increased SRT was associated with SEMs that involved IVFDs, but not with SEMs that lacked IVFDs, perhaps indicating that SEMs that lacked IVFDs were able to perceive the target, which also indicate the neural capacity to restore the trajectory to an appropriate path. Daga et al.43 recently examined the wayfinding abilities of glaucoma patients in a virtual reality environment. The patients found it more difficult to recognize targets in a virtual room with multiple visual cues compared with a room with fewer such cues. The findings suggested that glaucoma patients’ visual searching was ineffective, which disrupted patients’ ability to build a detailed spatial cognitive map. The authors suggested eye tracking methods when exploring deficient behaviors with respect to wayfinding. These observations are consistent with our hypothesis: glaucoma patients exhibit poor visual field sensitivity and have deviated perception of spatial orientation as well. Extensive visual information must be integrated when recognizing a three-dimensional space; the cognitive map may become distorted by the accumulation of deviated perceptions. Our work had some limitations. First, although the simulated IVF values were in good agreement with the values generated by the binocular Esterman visual field test,34 these values may differ when a bilateral gaze is involved, thereby leading to incorrect VFDs. Additionally, head stabilization by a chinrest also differs from realistic visual conditions, because compensations such as head movements are not allowed. We plan to remove this restriction in future studies. Second, we could not exclude the possible effects of glaucoma medications on SEMs. However, to our knowledge, there have been no suggestions that prostaglandin analogs, alpha agonists, or beta-blockers affect SEMs. Further investigations with additional patients are needed. Third, a cathode ray tube monitor is often used during eye tracking experiments; such monitors exhibit fast reaction and response times, and they permit large viewing angles. A liquid crystal display is slower and the viewing angle is smaller. A light-emitting diode monitor (such as the monitor used in the present study) exhibits a large viewing angle (≤178°) and an adequate response time. Fourth, all stimuli were two-dimensional and not in full color; thus, they did not reflect the dynamic nature of the real world. Additionally, the patients rarely move their eyes by more than 15° in natural conditions. Because we maintained constant saccadic amplitudes in all directions, the saccadic trajectories at various amplitudes should be evaluated in future studies. In conclusion, when binocular saccades were induced, patients with glaucoma exhibited consistently different eye movements, compared with healthy controls. Specifically, the saccadic trajectories of glaucoma patients departed erroneously and the disparities were compensated by deviating at its arrival. These between-group differences were associated with common clinical measures of glaucoma; the DA and AA were associated with changes in HVF-MD and in optical coherence tomography–based measurements of RNFL thickness. The initial changes may not indicate wandering; SEMs in patients without VFDs were actively corrected in terms of the final trajectories, but the SEMs of patients with VFDs exhibited less correction. We present a novel analysis of saccadic trajectory. Our results may be used to analyze the eye movements of glaucoma patients and elucidate the challenges that they experience each day. Supplementary Material Supplement 1 Supplement 2 Acknowledgments Supported by National Research Foundation of Korea (NRF) grant funded by the Ministry of Education (RS-2022-00167024) and the Translational R&D Project through Institute for Bio-Medical convergence, Incheon St. Mary's Hospital, The Catholic University of Korea (IBC-2022M-09). The funding organizations had no role in the design or conduct of this research. YCK is listed as inventors on the Korean patent application “Apparatus and Method for Determining Glaucoma,” which is partially based on the method described in this manuscript. Disclosure: J.S. Yeon, None; H.N. Jung, None; J.Y. Kim, None; K.I. Jung, None; H.-Y.L. Park, None; C.K. Park, None; H.W. Kim, None; M.S. Kim, None; Y.C. Kim, None ==== Refs References 1. Quigley HA. Number of people with glaucoma worldwide. Br J Ophthalmol. 1996; 80 : 389–393.8695555 2. Tham YC, Li X, Wong TY, Quigley HA, Aung T, Cheng CY. Global prevalence of glaucoma and projections of glaucoma burden through 2040: a systematic review and meta-analysis. Ophthalmology. 2014; 121 : 2081–2090.24974815 3. Yucel YH, Zhang Q, Weinreb RN, Kaufman PL, Gupta N. Effects of retinal ganglion cell loss on magno-, parvo-, koniocellular pathways in the lateral geniculate nucleus and visual cortex in glaucoma. Prog Retin Eye Res. 2003; 22 : 465–481.12742392 4. Weinreb RN, Khaw PT. Primary open-angle glaucoma. Lancet. 2004; 363 : 1711–1720.15158634 5. Friedman DS, Freeman E, Munoz B, Jampel HD, West SK. Glaucoma and mobility performance: the Salisbury Eye Evaluation Project. Ophthalmology. 2007; 114 : 2232–2237.17980433 6. Johnson CA, Keltner JL. Incidence of visual field loss in 20,000 eyes and its relationship to driving performance. Arch Ophthalmol. 1983; 101 : 371–375.6830485 7. Viswanathan AC, McNaught AI, Poinoosawmy D, et al . Severity and stability of glaucoma: patient perception compared with objective measurement. Arch Ophthalmol. 1999; 117 : 450–454.10206571 8. Smith ND, Glen FC, Monter VM, Crabb DP. Using eye tracking to assess reading performance in patients with glaucoma: a within-person study. J Ophthalmol. 2014; 2014 : 120528.24883203 9. Patino CM, McKean-Cowdin R, Azen SP, et al . Central and peripheral visual impairment and the risk of falls and falls with injury. Ophthalmology. 2010; 117 : 199–206.e191.20031225 10. Rucci M, Iovin R, Poletti M, Santini F. Miniature eye movements enhance fine spatial detail. Nature. 2007; 447 : 851–854.17568745 11. Ko HK, Poletti M, Rucci M. Microsaccades precisely relocate gaze in a high visual acuity task. Nat Neurosci. 2010; 13 : 1549–1553.21037583 12. Sparks DL. The brainstem control of saccadic eye movements. Nat Rev Neurosci. 2002; 3 : 952–964.12461552 13. Schiller PH, Tehovnik EJ. Neural mechanisms underlying target selection with saccadic eye movements. Prog Brain Res. 2005; 149 : 157–171.16226583 14. Robinson DA. The mechanics of human saccadic eye movement. J Physiol. 1964; 174 : 245–264.14244121 15. Rayner K. Eye movements in reading and information processing: 20 years of research. Psychol Bull. 1998; 124 : 372–422.9849112 16. Cornelissen FW, Peters EM, Palmer J. The Eyelink Toolbox: eye tracking with MATLAB and the Psychophysics Toolbox. Behav Res Methods Instrum Comput. 2002; 34 : 613–617.12564564 17. Moschovakis AK, Scudder CA, Highstein SM. The microscopic anatomy and physiology of the mammalian saccadic system. Prog Neurobiol. 1996; 50 : 133–254.8971981 18. Aizawa H, Wurtz RH. Reversible inactivation of monkey superior colliculus. I. Curvature of saccadic trajectory. J Neurophysiol. 1998; 79 : 2082–2096.9535970 19. Perry VH, Cowey A. Retinal ganglion cells that project to the superior colliculus and pretectum in the macaque monkey. Neuroscience. 1984; 12 : 1125–1137.6483194 20. Kanjee R, Yucel YH, Steinbach MJ, Gonzalez EG, Gupta N. Delayed saccadic eye movements in glaucoma. Eye Brain. 2012; 4 : 63–68.28539782 21. Najjar RP, Sharma S, Drouet M, et al . Disrupted eye movements in preperimetric primary open-angle glaucoma. Invest Ophthalmol Vis Sci. 2017; 58 : 2430–2437.28448671 22. Lee SS, Black AA, Wood JM. Effect of glaucoma on eye movement patterns and laboratory-based hazard detection ability. PloS One. 2017; 12 : e0178876.28570621 23. Lamirel C, Milea D, Cochereau I, Duong MH, Lorenceau J. Impaired saccadic eye movement in primary open-angle glaucoma. J Glaucoma. 2014; 23 : 23–32.22706338 24. Brigell MG, Goodwin JA, Lorance R. Saccadic latency as a measure of afferent visual conduction. Invest Ophthalmol Vis Sci. 1988; 29 : 1331–1338.3417417 25. Thepass G, Lemij HG, Vermeer KA, van der Steen J, Pel JJM. Slowed saccadic reaction times in seemingly normal parts of glaucomatous visual fields. Front Med (Lausanne). 2021; 8 : 679297.34513866 26. Ballae Ganeshrao S, Jaleel A, Madicharla S, et al . Comparison of saccadic eye movements among the high-tension glaucoma, primary angle-closure glaucoma, and normal-tension glaucoma. J Glaucoma. 2021; 30 : e76–e82.33394842 27. Michell AW, Xu Z, Fritz D, et al . Saccadic latency distributions in Parkinson's disease and the effects of L-dopa. Exp Brain Res. 2006; 174 : 7–18.16544135 28. Antoniades CA, Altham PM, Mason SL, Barker RA, Carpenter R. Saccadometry: a new tool for evaluating presymptomatic Huntington patients. Neuroreport. 2007; 18 : 1133–1136.17589313 29. Pearson BC, Armitage KR, Horner CW, Carpenter RH. Saccadometry: the possible application of latency distribution measurement for monitoring concussion. Br J Sports Med. 2007; 41 : 610–612.17496064 30. Nystrom M, Holmqvist K. An adaptive algorithm for fixation, saccade, and glissade detection in eyetracking data. Behav Res Methods. 2010; 42 : 188–204.20160299 31. Dalmaso M, Castelli L, Scatturin P, Galfano G. Trajectories of social vision: eye contact increases saccadic curvature. Visual Cognition. 2017; 25 : 1–3. 32. Doyle M, Walker R. Curved saccade trajectories: voluntary and reflexive saccades curve away from irrelevant distractors. Exp Brain Res. 2001; 139 : 333–344.11545472 33. Reilly JL, Lencer R, Bishop JR, Keedy S, Sweeney JA. Pharmacological treatment effects on eye movement control. Brain Cogn. 2008; 68 : 415–435.19028266 34. Crabb DP, Viswanathan AC, McNaught AI, Poinoosawmy D, Fitzke FW, Hitchings RA. Simulating binocular visual field status in glaucoma. Br J Ophthalmol. 1998; 82 : 1236–1241.9924324 35. McDowell JE, Dyckman KA, Austin BP, Clementz BA. Neurophysiology and neuroanatomy of reflexive and volitional saccades: evidence from studies of humans. Brain Cogn. 2008; 68 : 255–270.18835656 36. McPeek RM, Keller EL. Deficits in saccade target selection after inactivation of superior colliculus. Nat Neurosci. 2004; 7 : 757–763.15195099 37. Mays LE, Sparks DL. Dissociation of visual and saccade-related responses in superior colliculus neurons. J Neurophysiol. 1980; 43 : 207–232.6766178 38. Quaia C, Aizawa H, Optican LM, Wurtz RH. Reversible inactivation of monkey superior colliculus. II. Maps of saccadic deficits. J Neurophysiol. 1998; 79 : 2097–2110.9535971 39. Park SJ, Ko T, Park CK, Kim YC, Choi IY. Deep learning model based on 3D optical coherence tomography images for the automated detection of pathologic myopia. Diagnostics (Basel). 2022; 12 : 742.35328292 40. Soans RS, Grillini A, Saxena R, Renken RJ, Gandhi TK, Cornelissen FW. Eye-movement-based assessment of the perceptual consequences of glaucomatous and neuro-ophthalmological visual field defects. Transl Vis Sci Technol. 2021; 10 : 1. 41. Tatham AJ, Murray IC, McTrusty AD, et al . Speed and accuracy of saccades in patients with glaucoma evaluated using an eye tracking perimeter. BMC Ophthalmol. 2020; 20 : 259.32605609 42. Tatham AJ, McClean P, Murray IC, et al . Development of an age-corrected normative database for saccadic vector optokinetic perimetry (SVOP). J Glaucoma. 2020; 29 : 1106–1114.33264163 43. Daga FB, Macagno E, Stevenson C, et al . Wayfinding and glaucoma: a virtual reality experiment. Invest Ophthalmol Vis Sci. 2017; 58 : 3343–3349.28687845 44. Ramulu P. Glaucoma and disability: which tasks are affected, and at what stage of disease? Curr Opin Ophthalmol. 2009; 20 : 92–98.19240541 45. Turano KA, Rubin GS, Quigley HA. Mobility performance in glaucoma. Invest Ophthalmol Vis Sci. 1999; 40 : 2803–2809.10549639

PMC010xxxxxx/PMC10353746.txt

==== Front Invest Ophthalmol Vis Sci Invest Ophthalmol Vis Sci IOVS Investigative Ophthalmology & Visual Science 0146-0404 1552-5783 The Association for Research in Vision and Ophthalmology 37440262 10.1167/iovs.64.10.13 IOVS-23-37610 Cornea Cornea Obstruction of the Tear Drainage Altered Lacrimal Gland Structure and Function Tear Duct Obstruction Altered Lacrimal Gland Function Xiao Bing 1 Guo Dianlei 1 Liu Ren 1 Tu Mengqian 1 Chen Ziyan 1 Zheng Yingfeng 1 Liu Chunqiao 1 2 Liang Lingyi 1 1 State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Center, Sun Yat-Sen University, Guangdong Provincial Key Laboratory of Ophthalmology and Visual Science, Guangzhou, China 2 Guangdong Province Key Laboratory of Brain Function and Disease, Zhongshan School of Medicine, Sun Yat-Sen University, Guangzhou, China # Correspondence: Lingyi Liang, State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Center, Sun Yat-Sen University, Guangdong Provincial Key Laboratory of Ophthalmology and Visual Science, 7 Jinsui Road, Zhujiang New Town, Tianhe District, Guangzhou 510623, China; lianglingyi@gzzoc.com. Chunqiao Liu, Guangdong Province Key Laboratory of Brain Function and Disease, Zhongshan School of Medicine, Sun Yat-Sen University, #74, Zhongshan No. 2 Road, Guangzhou 510080, People's Republic of China; liuchunq3@mail.sysu.edu.cn. Yingfeng Zheng, State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Center, Sun Yat-Sen University, Guangdong Provincial Key Laboratory of Ophthalmology and Visual Science, 7 Jinsui Road, Zhujiang New Town, Tianhe District, Guangzhou 510623, China; zhengyingfeng@gzzoc.com. * BX and DG contributed equally to the work. 13 7 2023 7 2023 64 10 1319 6 2023 08 5 2023 Copyright 2023 The Authors 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. Purpose Orbital glands and drainage conduits are two distinct entities that constitute the lacrimal apparatus system, the malfunction of which leads to a range of ocular surface disorders. Despite the close functional relationship, how the two parts interact under pathophysiological conditions has not been directly tested. The study aims to investigate the lacrimal gland (LG) structural and functional changes upon the drainage system obstruction, thus, testing their function link. Methods Dacryocystectomy was performed in C57BL/6 mice to create a surgical model for tear duct (TD) obstruction (STDOB). Prickle1 mutant line with congenital nasolacrimal duct dysplasia serves as a genetic model for TD obstruction (GTDOB). Alterations of the LG and the ocular surface in tear duct obstruction mice were examined. Results STDOB and GTDOB mice showed similar ocular surface phenotypes, including epiphora, corneal epithelial defects, and conjunctival goblet cell abnormalities. At the molecular and cellular levels, aberrant secretory vesicle fusion of the LG acinar cells was observed with altered expression and localization of Rab3d, Vamp8, and Snap23, which function in membrane fusion. LG secretion was also altered in that lactoferrin, lipocalin2, and lysozyme expression were increased in both LG and tears. Furthermore, STDOB and GTDOB mice exhibited similar LG transcription profiles. Conclusions Physical obstruction of tear drainage in STDOB or GTDOB mice leads to LG dysfunction, suggesting a long-distance interaction between the tear drainage conduits and the LG. We propose that various components of the lacrimal apparatus should be considered an integral unit in diagnosing and treating ocular surface diseases. tear duct obstruction lacrimal sac removal dacryocystectomy Prickle1 mutant mice lacrimal gland (LG) ocular surface diseases ==== Body pmcThe lacrimal apparatus is a complex system encompassing glands for tear production and sacs and ducts for tear drainage.1 It plays an important role in maintaining tear film stability, corneal transparency, and ocular surface health.2 The lacrimal gland (LG) is the main contributor to the aqueous layer of the tear film, which consists of water, electrolytes, and proteins.3 The drainage or tear duct (TD) system consists of upper and lower puncta, lacrimal canaliculi, lacrimal sacs, and nasolacrimal/tear ducts, providing a route for tears draining into the nasal cavity. Besides draining function, the TD regulates tear dynamics and maintains tear film homeostasis.4–8 Stimulation of the nasal mucosa also leads to the reflex secretion of the LG.9,10 Normal tear quantity and constituents are maintained from the balance of tear secretion and drainage, which are crucial for ocular surface protection. Defective LG secretion is well known to be a major cause of dry eye disease.11 Besides, a plethora of studies indicate that obstruction of the tear drainage also leads to reduced tear production. For example, a complete cessation of tear secretion was reported in rare ectodermal dysplasia of the Acro-Dermato-Ungual-Lacrimal-Tooth syndrome with congenital TD obstruction,12 and reduced tear secretion is observed in the case of lower lid ectropion with obstructed tear outflow13 and in experimental subjects (normal individuals) with unilateral or bilateral punctal occlusion.14,15 Consequently, punctal occlusion (punctual cautery/punctual plug), recommended as an option for treating moderate to severe dry eye disease,16 such as the Sjögren 's syndrome,17,18 Stevens-Johnson syndrome,19 and the post-LASIK subjects, has no improvement in tear secretion, despite the reported improved symptom scores.20 Furthermore, patients with pemphigus of the conjunctiva with blocked punctum also have poor tear production.13 The lacrimal sac is continuous with the nasolacrimal duct. Several ocular diseases, such as chronic dacryocystitis and lacrimal sac tumors, require surgical operations on the lacrimal sac, often causing ocular surface irritations similar to the punctual occlusion or lacrimal gland malfunction. Together with many other observations related above, they imply a functional link between the LG and the TD. However, this link has not been directly tested due to a lack of animal models. A rabbit surgical model was created previously only to investigate structural disorder and dysfunction of the tear duct but not the LG.21 Besides, the rabbit nasolacrimal anatomy differs greatly from the humans in that it has only a single punctum located in the lower eyelid,22 questioning its suitability for studying the human tear duct system. On the other hand, although mice are widely used in modeling human diseases, few studies detail the anatomy of the mouse lacrimal drainage system.23,24 Recently, we described the mouse TD ontogenesis and anatomy, demonstrating its similarity to humans using a Prickle1 mutant mouse line.25 This is, to our knowledge, the first genetic model having incompletely developed TD (GTDOB) by disruption of Prickle1 gene.26–28 In this study, we used two types of mouse models to address the functional link between LG and TD. We first created a surgical mouse model (surgical tear drainage obstruction [STDOB]) with the lacrimal sac removed to study the impact of obstructed tear drainage on LG. We then examined the cellular and structural alterations of LG in both STDOB and GTDOB mice and compared their LG phenotypes in parallel. The results of the two models point to the same conclusion that a long-distance interaction between LG and TD plays an essential role in ocular surface health. Materials and Methods Animals Animal husbandry and experimentation were conducted strictly according to the Association for Research in Vision and Ophthalmology. All experimental procedures were reviewed and approved by the Animal Care and Use Committee (ACUC) at Zhongshan Ophthalmic Center (protocol No. 2021-034). Female wild-type C57BL/6J mice aged 6 to 8 weeks (n = 50) were purchased from the Vital River Laboratory Animal Technology (Beijing, China). Prickle1 mutant mice (the Prickle1a/b group, n = 10, aged 6–8 weeks) were generated by crossbreeding a Prickle1 gene-trap mutant allele (Prickle1a/+) to a straight knockout allele (Prickle1b/+).29 Female adult wild-type C57BL/6 mice were randomly divided into three groups, one receiving bilateral dacryocystectomy (the STDOB group, n = 20), one receiving bilateral sham surgery (the sham group, n = 20), and the other did not undergo any surgery (the wild-type group, n = 10). All mice were euthanized separately at 8 weeks postoperatively, and Prickle1 mutant mice were also euthanized on the same day. Dacryocystectomy All animals were kept at room temperature of 23 ± 2°C with relative humidity of 60 ± 10%, 12 hours of light and 12 hours of darkness, with adequate food and water for 7 days before the lacrimal sac removal surgery. Mice were deeply anesthetized with intraperitoneally injected 1% sodium pentobarbital (50 mg/kg), and 0.5% fast green dye (ab146267) was instilled onto the ocular surface to trace the nasal tear fluid path. The skin of the lacrimal sac area was carefully opened with small straight scissors, and the stained lacrimal sac was found and removed. The skin incision was then closed with a 10-0 nylon suture (the STDOB group). A similar operation was performed for the sham group but with the lacrimal sac left intact. All surgeries were performed by the same surgeon. After surgery, tobramycin eye ointment was applied to both eyes and skin wounds to prevent infection. Measurement of the Tear Flow Tear flow was examined using a phenol red thread (Jingming, Tianjin, China).30 In brief, the thread was applied to the lateral canthus of the eye in the anesthetized mice for 15 seconds. The length of the wet portion was measured in millimeters. Eyes were tested one at a time, first the right eye followed by the left eye. The average tear flow in both eyes was recorded. Corneal Fluorescein Staining Fluorescein staining was used to assess the barrier damage of the corneal epithelium. The staining was performed by instilling 0.25% fluorescein sodium (Jingming, Tianjin, China) onto the cornea and photographed with a digital camera under cobalt blue light. The measurement of fluorescein-stained area was conducted using ImageJ software (version 1.52; National Institutes of Health, Bethesda, MD, USA).31 Percentage of the corneal fluorescein staining (CFS) area was calculated. Gross Morphological and Morphometric Measurement The body weights of mice were recorded 8 weeks after surgery. The extra orbital LGs of each animal were dissected immediately after euthanasia. The LGs were placed on a grid paper sheet, and photographs were taken. The absolute wet weight of each LG was measured using a laboratory precision balance (Sartorius, BSA124S, Germany), and the weight indices were calculated as the LG weight divided by the body mass (LG/body, mg/g).32 The average acinus area measurement on tissue sections was performed on 3 random fields (10 random acini from each field) of each LG at 400× magnification. Five mice (LGs) were measured for each group. The acini area per imaging field was calculated as the acini percent area, three mice (LGs) were measured for each group. For each sample, three random fields were used for analysis. Tissue Histology For Periodic Acid Schiff (PAS) staining, eyeballs were fixed in 10% formalin and embedded in paraffin and sectioned at 5 µm using a paraffin microtome (RM2016, Leica, German). The sections were stained with a PAS staining kit (G1008-20ML; Servicebio, Wuhan, China). Representative images of the conjunctiva were captured with a light microscope (Tissue FAXS Q+, TissueGnostics, Vienna, Austria). Three sections of each nasal, central, and lateral part of an eye from five animals were studied. The number of PAS-positive cells per section was counted.33 For hematoxylin and eosin (H&E) staining, the LGs were fixed in 4% paraformaldehyde (PFA) for 24 hours immediately after dissection. The specimens were then washed and dehydrated in ascending grades of ethanol solutions, and were cleared in xylene for 2 hours after the last wash with pure ethanol. Impregnation and embedding were done first in soft paraffin wax at 59°C for 3 hours and then in hard paraffin at 63°C for 3 hours. The 5  µm sections were prepared every 50  µm for H&E staining. Representative images of the LG were captured with a light microscope (Tissue FAXS Q+, TissueGnostics, Austria). Transmission Electron Microscopy Transmission electron microscopy (TEM) samples were prepared as described previously.34 Briefly, LGs were dissected from the carcass quickly and continued fixation in 2.5% glutaraldehyde solution at 4°C for 24 hours. Sections were prepared and collected on mesh grids, stained with uranyl acetate and lead citrate, and examined under an electron microscope (JEM 1400; JEOL, Tokyo, Japan). All photographs were taken with a bio scan camera (Morada G3; EMSIS, Munster, Germany). Immunohistochemistry As described previously, LG samples were fixed in 4% PFA overnight at 4°C. The tissue blocks were washed, dehydrated, embedded in paraffin, and cut in 5  µm sections. Antigen retrieval was performed by incubation with a 0.01M citrate solution (pH 6.0) for 30 minutes at 98°C. The sections were then blocked with 0.3% BSA and followed by incubation with primary antibodies (Rabbit anti-RAB3d 1:200, Rabbit anti-Vamp8 1:100, and Rabbit anti-Snap23 1:100) and then with Alexa-Fluor 488 goat anti-rabbit secondary antibody (1:1000). Staining was visualized with a confocal Zeiss microscope (LSM-880, Zeiss, Oberkochen, Germany). High-resolution digital images were captured and stored in TIFF format. Antibodies used are rabbit anti-Rab3d polyclonal antibody (Catalog No. 12320-1-AP; Proteintech, Rosemont, IL, USA); rabbit anti-Vamp8 monoclonal antibody (Catalog No. MA5-32502; Invitrogen, Grand Island, NY, USA); rabbit anti-Snap23 polyclonal antibody (Catalog No. DF13314; Affinity Biosciences, Cincinnati, OH, USA); and Alexa-Fluor 488 goat anti-rabbit secondary antibody (Catalog No. 111-545-003; Jackson ImmunoResearch Laboratories, West Grove, PA, USA). Protein Sample Preparation and Western Blotting To prepare tear proteins, pilocarpine (300  µg/kg body weight) was subcutaneously injected to stimulate lacrimal gland secretion.35 Tears were collected from the eyelid margin into Eppendorf tubes using a 0.5  µl micropipette 5 minutes after injection and stored at −80°C. To prepare LG proteins, unstimulated LGs were isolated and proteins were extracted with RIPA buffer (R0278; Sigma-Aldrich, Allentown, PA, USA) and 1 mM phenylmethyl sulfonyl fluoride (329-98-6; Sigma-Aldrich, USA), and a protease inhibitor cocktail (P8340; Sigma-Aldrich, USA). After centrifugation, the supernatant was collected and stored at −80°C. The bicinchoninic acid (BCA) method was used to determine the tear and LG protein concentration (BCA Protein Assay Kit, P0012; Beyotime, Haimen, Jiangsu, China). Western blotting analysis was then performed to determine relative protein expression levels in lacrimal gland tissue and tear fluid. Briefly, proteins were separated by electrophoresis on approximately 4% to 20% sodium dodecyl sulfate-polyacrylamide gels and transferred to polyvinylidene difluoride membrane (PVDF) membrane. After blocking in 5% bovine serum albumin in Tris-buffered saline containing 0.05% Tween-20 (TBST) for 2 hours, membranes were incubated with primary antibodies in TBST at 4°C overnight. After 3 times washing with TBST for 10 minutes each, membranes were incubated with horseradish peroxidase–conjugated secondary antibody (7074S; Cell Signaling Technology, Danvers, MA, USA) for 1 hour. After washing three times with TBST solution, protein signals were developed by enhanced chemiluminescence reagents (ECL; Catalog No. WBKLS0100; Millipore, Chicago, IL, USA). Signal density on membranes was digitalized using ImageJ software. For LG samples, membranes were probed with anti-Gapdh as a loading control. All antibodies for Western blotting were at 1:1000 dilution. Antibodies are rabbit anti-lysozyme (ST50-02; Invitrogen, Grand Island, NY, USA), rabbit anti-Lactoferrin (GXP297214; GenXspan, Shenzhen, China), rabbit anti-lipocalin2 antibody (ab63929; Abcam, Cambridge, MA, USA), and rabbit anti-GAPDH (14C10; Cell Signaling Technology, Danvers, MA, USA). The horseradish peroxidase-conjugated secondary antibody was bought from Cell Signaling Technology (7074S). The rest of the antibodies used for Western blotting were described in the previous section. RNA Sequencing The LGs were dissected from age-matched 8-week-old control, surgical, and Prickle1 mutant mice. Three biological replicates from each group were subjected to RNA sequencing (RNAseq) analysis. Total RNA was extracted using a Trizol reagent kit (15596026, Invitrogen) according to the manufacturer's protocol. RNA quality was assessed on an Agilent 2100 Bioanalyzer (Agilent 2100; Agilent Technologies, Palo Alto, CA, USA) and agarose gel electrophoresis. mRNA was enriched by oligo (dT) beads (NEB #7335; New England Biolabs, Ipswich, MA, USA). The enriched mRNA was fragmented and reversely transcribed into cDNA with random primers (NEB#E 7530; New England Biolabs, Ipswich, MA, USA), followed by second-strand cDNA synthesis using DNA polymerase I (NEB#E 7530; New England Biolabs, Ipswich, MA, USA). The cDNA fragments were purified with a QiaQuick PCR purification kit (Catalog No. 28104; Qiagen, Hilden, Germany) and ligated to Illumina sequencing adapters. The size-selected ligation products from agarose gel electrophoresis were PCR amplified and sequenced using Illumina HiSeq2500 in Gene Denovo Biotechnology Co. (Guangzhou, China). Raw data (raw reads) of fastq format were firstly processed through in-house perl scripts to remove reads containing adapters, ploy-N, and low-quality reads. Q20, Q30, and GC content of the clean data were calculated. Reference genome and gene model annotation files were downloaded from the genome website (http://asia.ensembl.org/Mus_musculus/Info/Index) directly. Index of the reference genome was built using Bowtie version 2.0.6, and paired-end clean reads were aligned to the reference genome as FPKM (fragments per kilobase of transcript per million fragments mapped) using Top Hat version 2.0.9. RNA differential expression analysis was performed by DESeq2 software between two different groups. The genes with a false discovery rate (FDR) below 0.05 and absolute fold change ≥2 were considered differentially expressed genes (DEGs). The data were subjected to Gene Ontology (GO) Enrichment Analysis, which included molecular function, cellular components, and biological process. Pathway Enrichment Analysis based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) database was performed to understand the significantly altered metabolic pathways in STDOB surgical or Prickle1 mutant mice. GO enrichment analysis and pathway analysis were performed with the given criteria of P < 0.05 and FDR <0.05. Real-Time RT-qPCR For measuring gene expression, total RNA was isolated from whole LG with cooled TRIzol reagent (15596026; Invitrogen) and reverse transcribed into cDNA with HiScript III All-in-one RT SuperMix Perfect for qPCR (R333; Vazyme). Real-time quantitative PCR (RT-qPCR) was performed with the ChamQ SYBR Color qPCR Master Mix (Q431; Vazyme) on the Applied Biosystems 7500 system. The PCR program was as follows: 40 cycles of 95°C for 10 seconds and 60°C for 30 seconds after an initial denaturing at 95°C for 30 seconds. The primers for Prickle1 were: 5′-GTATGCTGGCACCCGTCCTG-3′ (forward) and 5′-GCACCGAGGCTTGAGCAGTT-3′ (reverse). The primers for Gapdh were 5′-GGAGAGTGTTTCCTCGTCCC-3′ (forward) and 5′- ATGAAGGGGTCGTTGATGGC-3′ (reverse). The relative expression levels were calculated using the CT method on the default ABI software SDS 2.3 (ThermoFisher Scientific, Waltham, MA, USA), as described previously.36 Statistical Analysis Differences in phenol red thread test values, percentages of CFS area, and goblet cell numbers between two individual groups were analyzed using GraphPad with an unpaired Student t-test to detect statistical powers. Comparisons of LG body weights, average acinus area, percent areas of acini, the relative mRNA levels, and Western blotting signal intensities were conducted with one-way analysis of variance (ANOVA) with Tukey post hoc testing. Differences between the measurements were considered significant if the P value was 0.05 or less. Results Creation of a Surgical Tear Drainage Obstruction Mouse Model by Dacryocystectomy To understand whether the tear drainage system would impact LG function, we performed lacrimal sac removal experiment in mice. We developed a surgical protocol (see Materials and Methods) in which fast green dye was instilled on ocular surface to trace the nasal tear fluid path (Figs. 1A-C). A successful operation will block tear flow, thus, the fast green dye through the nasolacrimal duct. Five out of 20 STDOB mice or sham-operation control mice were randomly picked for tear-outflow examination. None of the STDOB mice showed nasal tear duct staining of fast green, indicating successful lacrimal sac removal. In contrast, all sham operation controls showed green dyes in the tear duct (Figs. 1D, 1E). Figure 1. Schematic illustration of dacryocystectomy to create a surgical tear drainage obstruction (STDOB) mouse model. (A-C) Schematic diagrams of the lacrimal sac removal procedure. (A) Fast green dye was instilled onto the ocular surface to track the tear path through the drainage duct. (B) The lateral nasal skin was carefully opened to expose the lacrimal sac marked by the fast green dye. (C) The lacrimal sac (L.S.) was removed, and the skin was sutured. (D, E) Representative pictures demonstrating dye-filled lacrimal sac of the sham (left) and experiment (right) eyes 8 weeks post operations. White arrows indicate the position of the lacrimal sac. Ocular Surface Abnormalities of the STDOB Mice We first examined whether the obstructed tear drainage in STDOB mice increased tear accumulation on the ocular surface. It was first noticed that abundant white eye discharge was present in the inner canthus of the STDOB but not in the control mice (Figs. 2A, 2B). The tear volume of the STDOB mice measured by the phenol red thread30 also significantly increased in STDOB mice (6.30 ± 0.94 mm for the STDOB group vs. 1.85 ± 0.85 mm for the sham group; P < 0.0001; Fig. 2C). We speculated that the ocular surface directly covered by the tear film might also be altered in STDOB mice. Indeed, when examined by fluorescein staining, the cornea epithelium of STDOB mice also showed larger stained areas than the sham (Figs. 2D, 2E) (10.75 ± 4.55% in the STDOB group vs. 3.41 ± 1.75% in the sham control mice group; P = 0.0002; Fig. 2F). Figure 2. STDOB mice exhibited epiphora, increased corneal fluorescein staining, and increased conjunctival goblet cells. (A) Sham operation: a representative image of the normal ocular surface. (B) Epiphora with white discharge in the inner canthus of the STDOB mice. Arrows point to the inner canthus. (C) Tear flow test using the phenol red thread. (D) Corneal fluorescein staining of the sham-operated mice showing integral surface. (E) Punctate fluorescein staining of the STDOB mouse cornea. (F) Quantifications of the fluorescein-stained corneal area of the STDOB and sham mice. (G) PAS-stained goblet cells in the sham-operated mice had a round shape and uniform size. (H) More conjunctival goblet cells of the STDOB mice stained by the PAS. (I, J) Magnified images of the goblet cells from boxed areas in (G) and (H), respectively. (K) Quantification of the number of goblet cells of the STDOB and sham mice; n = 10 mice for (C) and (F); and n = 5 mice for (K). We next examine whether conjunctival epithelium, another major component of the ocular surface, was also altered in the STDOB mice. By PAS staining, the conjunctival epithelium appeared thickened with increased goblet cells (Figs. 2G-J; 42.47 ± 7.39 cells/section in the STDOB mice group vs. 13.33 ± 3.53 cells/section in the sham controls group; P < 0.0001; Fig. 2K). Additionally, goblet cell secretion revealed by PAS staining appeared remarkably more active in the STDOB than in the sham control mice (see Figs. 2G-J). Thus, the results suggested a systemic alteration of the ocular surface epithelium and secretion upon obstructed tear drainage. Ocular Surface Phenotypes of the STDOB Mice Resemble Those of Prickle1 Mutants We next investigated whether the ocular phenotypes manifested in the STDOB mice were also present in mice with congenital tear drainage obstruction. The only reported genetically engineered mice with tear duct dysplasia are the Prickle1 gene knockouts.27,35 Similar to the STDOB animals, Prickle1 mutant mice showed epiphora with white discharge (Figs. 3A, 3B). The amount of tear secretion measured by the phenol red thread was also increased (7.45 ± 0.49 mm for the Prickle1a/b group vs. 1.9 ± 0.61 mm for the wild type mice group; P < 0.0001; Fig. 3C). Additionally, Prickle1 mutant corneal epithelium showed punctate fluorescein staining (Figs. 3D, 3E), with the mean area of epithelial staining significantly increased (13.17 ± 3.07% for the Prickle1a/b group vs. 1.54 ± 1.63% for the control mice group; P < 0.0001; Fig. 3F). Conjunctiva goblet cells of the Prickle1 mutant mice showed heavily apical staining compared with the controls (Figs. 3G-J). Furthermore, the number of goblet cells in Prickle1 mutant mice was significantly increased as well (40.00 ± 6.78 cells/section for the Prickle1a/b group vs. 9.66 ± 1.79 cells/section for the control mice group; P< 0.0001; Fig. 3K). Thus, Prickle1 mutants fully recapitulate the ocular surface phenotype of STDOB mice, indicating that obstruction of tear duct does cause tear-related ocular surface disorders. Figure 3. STDOB mice showed similar phenotypes to those observed in Prickle 1 mutant mice. (A, D) Normal ocular surface with intact corneal epithelium showed in the wild-type mice. (B, E) Epiphora with white discharge and increased fluorescein stain on the cornea of the Prickle1 mutant mice. The arrow indicates the white discharge in the inner canthus. (C) Tear flow was significantly increased in the Prickle1 mutant mice using the phenol red thread test. (F) Corneal fluorescein staining score revealed more severe staining of the Prickle1 mutant mice. (G) Periodic acid-Schiff (PAS) staining of the conjunctival epithelium of the wild-type mice showed rounded and uniform staining patches. (H) The Prickle1a/b group showed an increase in goblet cells with enhanced staining in the apical areas. (I) and (J), Magnified images from boxed areas in (G) and (H), respectively. (K) The number of goblet cells was significantly increased in the Prickle1a/b group; n = 10 mice for (C) and (F); and n = 5 mice for (K). Tear Components and Lacrimal Gland-Secreted Proteins are Similarly Altered in STDOB and Prickle1a/b Mice The tear fluid is mainly secreted by the LG, playing an important role in the maintenance of ocular health. The observed epiphora and ocular surface abnormalities prompt us to investigate whether major tear proteins were also altered. Lysozyme, tear lipocalin, and lactoferrin of the tear film have diverse roles in the protection of the ocular surface. An examination of the levels of these proteins in the tear fluid by Western blotting revealed increased expression of all three proteins in both STDOB and Prickle1 mutant animals (Fig. 4A). Consistent with the tear fluid, the LG secretion of these proteins was also increased (Fig. 4B). Further quantification of Western blotting showed that the LG lactoferrin, lipocalin, and lysozyme were increased by 1.7-fold (P = 0.0686), 1.7-fold (P = 0.0218), and 1.8-fold (P = 0.0012), respectively, in STDOB mice (Fig. 4C). Prickle1 mutant mice showed similar increasing trends of LG lactoferrin (1.6-fold, P = 0.0113), lipocalin (2.1-fold, P = 0.0744), and lysozyme (2.7-fold, P < 0.0001; see Fig. 4C). Therefore, the LG secretory function is altered upon tear duct obstruction. Figure 4. Alterations of tear component and lacrimal gland-secreted proteins in STDOB and Prickle1 mutant mice. (A, B) Representative images of Western blotting of lactoferrin, lipocalin, and lysozyme in the tear fluid (A) and LG (B). The same amount of proteins (see Methods and materials) were loaded for polyacrylamide gel electrophoresis, with the Gapdh serving as a control for LG protein loading (B). (C), Quantification of Western blotting analysis showed increased lactoferrin, lipocalin, and lysozyme from the LG of the STDOB and Prickle1 mutant mice. Protein levels were normalized to Gapdh, and relative expression was compared between the sham and the treated groups (n = 3 LGs/mice). Aberrant LG Secretory Vesicle Fusion in STDOB and Prickle1 Mutant Mice The altered LG secretory function led us to examine whether the LG size and histology also changes. On stereomicroscope, no obvious differences in LG size or general appearance were observed (Figs. 5A-C). The wet weight of individual LG did not significantly alter when normalized to the body weight (LG mass/body mass: STDOB mice 0.55 ± 0.03 mg/g, Prickle1a/b mice 0.56 ± 0.04 mg/g, and sham mice 0.56 ± 0.03 mg/g; P = 0.89; Fig. 5D). Figure 5. Morphologic alterations of lacrimal glands in the STDOB and Prickle1 mutant mice. (A-C) Stereomicroscopy of the LGs from the sham (A), STDOB (B), and Prickle1 mutant mice (C). (D) Quantification of relative LG weights. The wet LG weights were normalized to the body weights presented as ratio indices (LG/body, mg/g); N =10 LGs. (E-J) H&E staining of the LGs from the sham controls (E, H), STDOB (F, I), and Prickle1 mice (G, J). (H), (I), and (J) are magnified images from the boxed areas of (E), (F), and (G), respectively. (K) Average individual acinus areas. Ten acini from each imaging field are roughly randomly chosen for measuring individual acinus areas. The results are presented as average area/acinus. The n = 5 mice/group. (L), The total average acinar area per imaging field was presented as percentages (n = 3 mice/group). We then performed H&E staining to determine whether pathological changes occurred in the LG among the three groups. LG in all groups showed similar acinus histology (Figs. 5E-J). Additionally, the STDOB (see Figs. 5F, 5I), Prickle1 mutant (see Figs. 5G, 5J), and sham control mice showed similar individual (STDOB 258.4 ± 9.834  µm2; Prickle1a/b 258.0 ± 8.377  µm2; sham 252.7 ± 7.134  µm2; P = 0.1748; Fig. 5K) and total percent (STDOB 78.6 ± 4.643 %; Prickle1a/b 77.12 ± 2.964 %; sham 75.39 ± 3.359 %; P = 0.1579; Fig. 5L) acinar areas. We next investigated whether the ultrastructure of the LG may have been altered corresponding to the secretory function changes. On the TEM of the sham group, LG secretory vesicles had uniform sizes filling in the apical and central areas of the acinar cells with moderate electron density, which were membrane-bound (Figs. 6A, 6B). In contrast, the STDOB and Prickle1 mutant LGs reduced electron density in acinar cells (Figs. 6C-F). Furthermore, the vesicle membrane boundary was blurred, and the fusion of vesicles was noticed (see Figs. 6D, 6F). Thus, despite the relatively normal H&E histology, the LG acinar ultrastructure was altered significantly, consistent with the abnormal tear components and LG secretion. Figure 6. LG ultrastructure revealed by transmission electron microscopy (TEM). (A, B) LG showed uniform electron density of the acinus secretory vesicles in the sham-operated mice. (C-F), Altered electron density of the LG acinar vesicles in the STDOB and Prickle1 mutant groups. (B), (D), (F) Higher magnification of the boxed areas from (A), (C), and (E), respectively. The green arrow in (B) indicates the clear membrane-bound secretory vesicles in the sham mice. The red arrow in (D) (STDOB) and the purple arrow in (F) (Prickle1 mutant) point to the enlarged fused vesicles with lighter electron density and heterogenous membrane curvatures. Altered Expression and Localization of Proteins Related to LG Vesicle Fusion and Secretion To further evaluate how the LG secretory function was altered, we examined protein expression and localization of Rab3d (a vesicle-trafficking GTPase), Vamp8 (a v-SNARE vesicle fusion receptor), and Snap23 (a membrane fusion targeting t-Snare protein), which have been shown to be involved in LG secretion. Immunohistochemistry revealed basally localized Rab3d in the STDOB and Prickle1 mice, contrasting to its normal apical distribution in the sham controls (Fig. 7A). Moreover, Western blotting showed a remarkable decrease of Rab3d expression (0.65-fold, P = 0.025 and 0.69-fold, P = 0.024 for STDOB and Prickle1 mice, respectively; Fig. 7B). Vamp8 and Snap23 were also basally mislocalized (Figs. 7C, 7E), but their expression levels were enhanced (Fig. 7D: Vamp8:1.24-fold for STDOB mice, P = 0.182 and 2.46-fold for Prickle1 mice, P = 0.004; Fig. 7F: Snap23: 1.29-fold for STDOB mice, P = 0.122, and 2.16-fold Prickle1 mice, P = 0.029). Collectively, our findings suggested that apical trafficking or fusion of the tear-containing vesicles might be impaired in the STDOB and Prickle1 mutant mice. Figure 7. Altered expression and localization of proteins relevant to lacrimal gland fusion and secretion. (A) Rab3d immunofluorescence was abundant in the apical lumen side of the acinar cells in the sham LGs (left panels), but uniformly distributed in the STDOB (middle panels) and more basally localized in the Prickle1 mutant acini. The bottom panels are magnified images from the top panels. Apical and basal positions of the acinus lumen are illustrated in the right corner of the first panel. (B) Western blotting demonstrated reduced Rab3d expression in the STDOB and Prickle1 mutant LGs compared with the sham. (C) Vamp8 immunofluorescence showed a similar altered pattern as the Rab3d in the STDOB and Prickle1 mice. (D) Western blotting showed increased LG expression of Vamp8 in the STDOB and Prickle1 mice. (E) Expression of Snap23 in LG acini was also basally shifted and showed increased levels in STDOB and Prickle1 mice compared with the sham (F). The white dotted lines indicate acini boundaries. For Western blotting and quantification, n = 4 LGs/mice. Lacrimal Gland Expression Profiles of STDOB and Prickle1 Mutant Mice The targeted examination of markers relevant to the LG secretion suggested that STDOB mice and the Prickle1 mutant mice shared similar cellular alterations. To further grasp their similarities at a larger scale, we performed RNAseq analysis to examine the LG transcriptomic profiles of the two models. The RNAseq analysis found 682 and 1351 differentially expressed genes (DEGs) in the STDOB and Prickle1 mouse LGs, respectively (Supplementary Figs. 1A, 1B). Three hundred seventy-five altered genes were found in common between the STDOB mice and Prickle1 mice mutant LGs, which was more than half of the STDOB and about one-third of the Prickle1 mutant DEGs (Figs. 8A, 8B). GO analysis demonstrated a remarkable similarity in biological processes, molecular functions, and cellular components between the two models (Supplementary Figs. 1C, 1D). Strikingly, the top 6 terms shared by the total 682 STDOB and 1351 Prickle DEGs are essentially the same in KEGG pathways (Figs. 8C, 8D). We further performed the KEGG enrichment of the common 375 DEGs shared by the two models and found that pathways, including PI3K-Akt, calcium, phospholipase D, and NF-kappa B signaling (Fig. 8E), were all reported to be involved in glandular secretion.37–40 Thus, the STDOB and Prickle1 mutant LGs are generally similar in transcriptomic landscapes. Figure 8. RNA sequencing and KEGG pathway analysis. (A) Differentially expressed genes (DEGs) of LGs from respective STDOB and Prickle1 mutant mice compared with the sham. Three hundred seventy-five common genes were shared by STDOB and Prickle1 mutant groups. (B) A Venn diagram of intersections between up- and downregulated DEGs. (C) Top 6 KEGG pathways enriched for the total 682 STDOB DEGs. (D) Top 6 KEGG pathways enriched for the total 1351 Prickle 1a/b DEGs. (E) The top 6 KEGG pathways enriched for the shared 375 DEGs between STDOB and Prickle1 mutant mice. Prickle1 is not Expressed in the Lacrimal Gland The LG changes of the Prickle1 mutant mice could be directly caused by the loss of Prickle1 in the LG. We, therefore, examined whether Prickle1 is expressed in the lacrimal gland. We first checked the average FPKMs of Prickle1 transcripts in the sham, STDOB, and Prickle1 mice groups, which all appeared very low (1.59, 0.97, and 0.22, respectively; Fig. 9A). We then examined Prickle1 mRNA expression in the LG tissue by RT-qPCR analysis. Compared with the control Gapdh expression, which was detectable at about 20 PCR cycles (Ct20), the Prickle1 expression required more than 35 cycles (Ct35) to be detected in the sham or STDOB LGs, and even more cycles in the Prickle1 mutants (above 40 cycles), as anticipated (Fig. 9B). Last, we took advantage of the Gfp knock-in reporter in the Prickle1 locus to surrogate Prickle1 expression using Prickle1b/+ heterozygous mutants. Immunostained-GFP was found strongly expressed in the hair follicles (Fig. 9C) but undetectable in the LG (Fig. 9D). Thus, the data suggested that obstruction of the tear duct is likely the cause of the abnormalities of LG. Figure 9. Prickle1 is not expressed in the lacrimal gland. (A) The mRNA expression amplitudes of the Prickle1 gene in RNA-seq data from the three experimental groups. FPKM, fragments per kilobase of transcript per million mapped reads. (B) RT-qPCR to determine Prickle1 mRNA levels in the lacrimal gland; CT, cycle threshold. (C) GFP reporter protein was detected by immunohistochemistry in hair follicles (serving as a positive control) but not in the LG acini of the Prickle1b/+ mice. Discussion The tear apparatus consists of many components with diverse yet coordinated functions. Tears are secreted from the LG, distributed by blinking and evaporated from the ocular surface, and drained through the nasolacrimal duct.41 Anatomically, the nasolacrimal duct, conjunctiva, and LG epithelia are continuous and share the same origin from the embryonic conjunctival ectoderm.25,42 Functionally, the lymphoid tissues of the three parts of the tear apparatus are also connected,43 and, as a result, the nasolacrimal system plays an important role in ocular surface innate immune defense.44 Despite the intimate relationships of the three major components of the tear apparatus, the functional connection between either of the two has not been established. In this study, we demonstrated, for the first time, that tear sac/duct impacts LG structure and tear secretion function in both surgical and genetic mouse models with tear duct partially removed. Several previous studies using different animal models demonstrated that ligation of the LG ducts causes reduced tear secretion and LG weights, accompanied by inflammation, increased cell proliferation, apoptosis, and lipid accumulation.30 Our finding that obstruction of the tear duct also led to LG structural and functional changes is striking. Unlike the LG ducts, which physically connect to LG in proximity, tear ducts are separated from LG by a wide-open territory of the ocular surface. Thus, our results hint that a remote signaling system exists between the tear duct and LG to coordinate their physiological activities. It further supports the notion that LG possesses a great remodeling plasticity and capacity illustrated by several studies.45,46 Management of a few ocular surface diseases requires removal or obstruction of the tear drainage components. However, the surgical side effects on the ocular surface have not been systemically evaluated due to a lack of animal models. We created a mouse model with the lacrimal sac surgically removed. Surprisingly, we observed corneal injuries and aberrant secretion of the conjunctival goblet cells accompanied by epiphora phenotypes. The ocular surface phenotypes resemble those seen in Prickle1 mutant mice we previously reported27 and in the current study. These results strongly suggested an inherent relevancy existing between the tear duct and ocular surface. It is intuitive to think that the observed epiphora in our animal models is due to obstructed tear drainage leading to tear accumulation on the ocular surface. However, the tear appearance does not look normal in that white discharges were observed. Therefore, we examined whether LG malfunction is involved. Indeed, examination of LG histology at different levels reveals striking ultrastructural differences by TEM. We further performed immunohistochemistry with molecular markers involved in vesicle secretion, including Rab3d, Vamp8, and Snap23, and found remarkable alterations in their expression and apicobasal distribution in our animal models. Moreover, tear components were also changed by Western blotting. The results suggest that, on the one hand, the epiphora could directly result from the obstruction of the tear duct, but most importantly, LG dysfunction might also be involved. The Rab3d, Vamp8, and Snap23 are crucial for membrane fusion of the secretory vesicles as well as their trafficking. These proteins are usually polarized in their cellular distribution.47–51 Altered distribution pattern of these proteins has been documented as pathological characteristics of exocrine glands in patients with dry eye disease.52,53 The altered expression and subcellular localization of these proteins in acinar cells of our animal models recapitulate these in dry eye disease. For instance, the vesicle-vesicle (instead of vesicle-plasma membrane) fusion of the LG acini in our models is a sign of disruption of the exocytic machinery appeared in the salivary gland of patients with Sjögren's syndrome.54 Similarly, the loss of Rab3d apical localization of the LG was also observed in the mouse model for Sjögren's syndrome55,56 and in the salivary gland of patients with Sjögren's syndrome.52 The loss of apical domains of Vamp8 and Snap23 localization observed in this study (mislocalized or shifted basolaterally) has been reported in the salivary gland acinar cells of Sjögren's syndrome as well.53 Consequently, the lipocalin, lactoferrin, and lysozyme in tears or LG were all altered in our animal models. Altogether, these data indicate that tear duct blockage leads to LG dysfunction, which might also be a suitable dry eye disease model system. Several clinical investigations showed reduced lacrimal fluid drainage was followed by decreased tear production with epiphora.57 For instance, reduced tear production was observed in patients with eversion of the lower punctum or compression of the inferior lacrimal canaliculus.58 Similarly, the dacryocystectomy also leads to a compensatory reduced tear production.13 Although the LG cannot be directly evaluated in these patients, the above results combined with our data suggest that the LGs of these patients may also be impaired, and tear production in our animal models likely decreased rather than overproduced. Several mechanisms could have contributed to the LG dysfunction upon blockage of the tear drainage. First, the excess tears on the ocular surface may send a signal to LG through the conjunctiva openings of lacrimal gland ducts,13 which directly feedbacks tear secretion. Second, a communicative path from the tear duct to LG through regional circulation was interrupted. It has been proposed that the tear components are absorbed into the blood vessels of the cavernous body around the lacrimal sac and nasolacrimal duct, which connect with the extraocular LG.44 Third, neural communications through the “nose (or tear duct)-superior salivatory nucleus-LG” axis59 was impaired leading to LG dysfunctions. Some studies demonstrate higher levels of neuropeptides in the nasal and tear secretions of allergic rhinoconjunctivitis,60,61 which may contribute to this neural axis. Additionally, this neural axis has facilitated intranasal tear neurostimulators to treat dry eye disease in some circumstances.62 Nonetheless, these possible mechanisms require further in-depth investigations in the future. In conclusion, this study offers a new perspective on the importance of the integrity of the nasolacrimal system on the LG function. The long-distance interaction between the nasolacrimal duct and LG may underlie many tear-related ocular surface diseases, including the infamous dry eye disease. As such, surgeries to remove tear duct components should be carefully considered in clinics, and the tear duct may also facilitate the treatment of the dry eye disease. Last, we created both STDOB and genetic mouse models, which could be utilized to investigate a broad range of nasolacrimal-ocular surface diseases. Supplementary Material Supplement 1 Acknowledgments Supported by the National Natural Science Foundation of China (82070922) and the Science and Technology Program of Guangzhou (202201020544). Disclosure: B. Xiao, None; D. Guo, None; R. Liu, None; M. Tu, None; Z. Chen, None; Y. Zheng, None; C. Liu, None; L. Liang, None ==== Refs References 1. Boniuk M. Eyelids, lacrimal apparatus, and conjunctiva. Arch Ophthalmol. 1973; 90 : 239–250.4580976 2. Bron AJ, de Paiva CS, Chauhan SK, et al . TFOS DEWS II pathophysiology report. The Ocular Surf. 2017; 15 : 438–510. 3. Stern ME, Gao J, Siemasko KF, Beuerman RW, Pflugfelder SC. The role of the lacrimal functional unit in the pathophysiology of dry eye. Exp Eye Res. 2004; 78 : 409–416.15106920 4. Paulsen FP, Corfield AP, Hinz M, et al . Characterization of mucins in human lacrimal sac and nasolacrimal duct. Invest Ophthalmol Vis Sci. 2003; 44 : 1807–1813.12714609 5. Jones LT. An anatomical approach to problems of the eyelids and lacrimal apparatus. Arch Ophthalmol. 1961; 66 : 111–124.13790551 6. Hill JC, Bethell W, Smirmaul HJ. Lacrimal drainage − A dynamic evaluation. Part I − mechanics of tear transport. Canadian J Ophthalmol Journal Canadien D'ophtalmologie. 1974; 9 : 411–416. 7. Doane MG. Blinking and the mechanics of the lacrimal drainage system. Ophthalmology. 1981; 88 : 844–851.7322503 8. Thale A, Paulsen F, Rochels R, Tillmann B. Functional anatomy of the human efferent tear ducts: a new theory of tear outflow mechanism. Graefe's Arch Clinic Exp Ophthalmol = Albrecht von Graefes Archiv fur klinische und experimentelle Ophthalmologie. 1998; 236 : 674–678.9782428 9. Boberg-Ans J. Experience in clinical examination of corneal sensitivity; corneal sensitivity and the naso-lacrimal reflex after retrobulbar anaesthesia. The Br J Ophthalmol. 1955; 39 : 705–726.13276583 10. Gupta A, Heigle T, Pflugfelder SC. Nasolacrimal stimulation of aqueous tear production. Cornea. 1997; 16 : 645–648.9395874 11. Dartt DA. Dysfunctional neural regulation of lacrimal gland secretion and its role in the pathogenesis of dry eye syndromes. The Ocular Surf. 2004; 2 : 76–91. 12. Rinne T, Spadoni E, Kjaer KW, et al . Delineation of the ADULT syndrome phenotype due to arginine 298 mutations of the p63 gene. Eur J Human Genetics: EJHG. 2006; 14 : 904–910.16724007 13. Norn MS. Tear secretion in diseased eyes. Keratoconjunctivitis sicca, diseases of the lacrimal system, ectropion, lagophthalmos, conjunctivitis, etc., studied by a new method: lacrimal streak dilution test. Acta Ophthalmologica. 1966; 44 : 25–32.5952955 14. Yen MT, Pflugfelder SC, Feuer WJ. The effect of punctal occlusion on tear production, tear clearance, and ocular surface sensation in normal subjects. Am J Ophthalmol. 2001; 131 : 314–323.11239863 15. Yazici A, Bulbul E, Yazici H, et al . Lacrimal gland volume changes in unilateral primary acquired nasolacrimal obstruction. Invest Ophthalmol Visual Sci. 2015; 56 : 4425–4429.26193918 16. Craig JP, Nelson JD, Azar DT, et al . TFOS DEWS II Report Executive Summary. The Ocular Surf. 2017; 15 : 802–812. 17. Tsifetaki N, Kitsos G, Paschides CA, et al . Oral pilocarpine for the treatment of ocular symptoms in patients with Sjogren's syndrome: A randomised 12 week controlled study. Ann Rheumatic Dis. 2003; 62 : 1204–1207. 18. Latifi G, Banafshe Afshan A, Houshang Beheshtnejad A, et al . Changes in corneal subbasal nerves after punctal occlusion in dry eye disease. Current Eye Res. 2021; 46 : 777–783. 19. Kaido M, Goto E, Dogru M, Tsubota K. Punctal occlusion in the management of chronic Stevens-Johnson syndrome. Ophthalmology. 2004; 111 : 895–900.15121365 20. Yung YH, Toda I, Sakai C, Yoshida A, Tsubota K. Punctal plugs for treatment of post-LASIK dry eye. Japanese J Ophthalmol. 2012; 56 : 208–213. 21. Liu R, Li H, Ai T, Hu W, Luo B, Xiang N. Pathological changes of the nasolacrimal duct in rabbit models of chronic dacryocystitis: correlation with lacrimal endoscopic findings. Graefe's Arch Clinic Exp Ophthalmol = Albrecht von Graefes Archiv fur klinische und experimentelle Ophthalmologie. 2018; 256 : 2103–2112.30187128 22. Rehorek SJ, Holland JR, Johnson JL, et al . Development of the lacrimal apparatus in the rabbit (Oryctolagus cuniculus) and its potential role as an animal model for humans. Anatomy Res Intl. 2011; 2011 : 623186. 23. Lohrberg M, Pabst R, Wilting J. Co-localization of lymphoid aggregates and lymphatic networks in nose- (NALT) and lacrimal duct-associated lymphoid tissue (LDALT) of mice. BMC Immunol. 2018; 19 : 5.29368640 24. Nagatake T, Fukuyama S, Kim DY, et al . Id2-, RORgammat-, and LTbetaR-independent initiation of lymphoid organogenesis in ocular immunity. The J Exp Med. 2009; 206 : 2351–2364.19822644 25. Guo D, Ru J, Mao F, et al . Ontogenesis of the tear drainage system requires Prickle1-driven polarized basement membrane deposition. Development. 2020; 147 : 1–13. 26. Guo D, Yuan Z, Ru J, et al . A spatiotemporal requirement for Prickle 1-mediated PCP signaling in eyelid morphogenesis and homeostasis. Invest Ophthalmol Vis Sci. 2018; 59 : 952–966.29450535 27. Guo D, Li M, Zou B, et al . Ocular surface pathogenesis associated with precocious eyelid opening and necrotic autologous tissue in mouse with disruption of Prickle 1 gene. Exp Eye Res. 2019; 180 : 208–225.30590023 28. Liu C, Lin C, Whitaker DT, et al . Prickle1 is expressed in distinct cell populations of the central nervous system and contributes to neuronal morphogenesis. Human Molec Genetics. 2013; 22 : 2234–2246. 29. Liu C, Lin C, Gao C, May-Simera H, Swaroop A, Li T. Null and hypomorph Prickle1 alleles in mice phenocopy human Robinow syndrome and disrupt signaling downstream of Wnt5a. Biology Open. 2014; 3 : 861–870.25190059 30. He X, Wang S, Sun H, et al . Lacrimal gland microenvironment changes after obstruction of lacrimal gland ducts. Invest Ophthalmol Vis Sci. 2022; 63 : 14. 31. Pellegrini M, Bernabei F, Moscardelli F, et al . Assessment of corneal fluorescein staining in different dry eye subtypes using digital image analysis. Transl Vis Sci Technol. 2019; 8 : 34. 32. de Souza RG, de Paiva CS, Alves MR. Age-related autoimmune changes in lacrimal glands. Immune Network. 2019; 19 : e3.30838158 33. Shirai K, Okada Y, Cheon DJ, et al . Effects of the loss of conjunctival Muc16 on corneal epithelium and stroma in mice. Invest Ophthalmol Vis Sci. 2014; 55 : 3626–3637.24812549 34. Ogawa Y, Yamazaki K, Kuwana M, et al . A significant role of stromal fibroblasts in rapidly progressive dry eye in patients with chronic GVHD. Invest Ophthalmol Vis Sci. 2001; 42 : 111–119.11133855 35. Ru J, Guo D, Fan J, et al . Malformation of tear ducts underlies the epiphora and precocious eyelid opening in Prickle 1 mutant mice: genetic implications for tear duct genesis. Invest Ophthalmol Vis Sci. 2020; 61 : 6. 36. Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods. 2001; 25 : 402–408.11846609 37. Sundermeier T, Matthews G, Brink PR, Walcott B. Calcium dependence of exocytosis in lacrimal gland acinar cells. Am J Physiol Cell Physiol. 2002; 282 : C360–C365.11788347 38. Zoukhri D, Dartt DA. Cholinergic activation of phospholipase D in lacrimal gland acini is independent of protein kinase C and calcium. The Am J Physiol. 1995; 268 : C713–C720.7900776 39. Alves M, Calegari VC, Cunha DA, Saad MJ, Velloso LA, Rocha EM. Increased expression of advanced glycation end-products and their receptor, and activation of nuclear factor kappa-B in lacrimal glands of diabetic rats. Diabetologia. 2005; 48 : 2675–2681.16283249 40. Nakamura H, Kawakami A, Ida H, Koji T, Eguchi K. EGF activates PI3K-Akt and NF-kappaB via distinct pathways in salivary epithelial cells in Sjogren's syndrome. Rheumatol Intl. 2007; 28 : 127–136. 41. Tsubota K. Tear dynamics and dry eye. Prog Retin Eye Res. 1998; 17 : 565–596.9777650 42. Chen Z, Huang J, Liu Y, et al . FGF signaling activates a Sox9-Sox10 pathway for the formation and branching morphogenesis of mouse ocular glands. Development. 2014; 141 : 2691–2701.24924191 43. Knop E, Knop N. The role of eye-associated lymphoid tissue in corneal immune protection. Journal of Anatomy. 2005; 206 : 271–285.15733300 44. Paulsen FP, Schaudig U, Thale AB. Drainage of tears: impact on the ocular surface and lacrimal system. The Ocular Surface. 2003; 1 : 180–191.17075649 45. Lin H, Liu Y, He H, Botsford B, Yiu S. Lacrimal gland repair after short-term obstruction of excretory duct in rabbits. Scientific Rep. 2017; 7 : 8290. 46. Dietrich J, Schlegel C, Roth M, et al . Comparative analysis on the dynamic of lacrimal gland damage and regeneration after Interleukin-1alpha or duct ligation induced dry eye disease in mice. Exp Eye Res. 2018; 172 : 66–77.29605492 47. Pfeffer SR. Rab GTPases: specifying and deciphering organelle identity and function. Trends in Cell Biology. 2001; 11 : 487–491.11719054 48. Sollner TH. Regulated exocytosis and SNARE function (Review). Molecular Membrane Biology. 2003; 20 : 209–220.12893529 49. Schluter OM, Khvotchev M, Jahn R, Sudhof TC. Localization versus function of Rab3 proteins. Evidence for a common regulatory role in controlling fusion. J Biolog Chem. 2002; 277 : 40919–40929. 50. Takai Y, Sasaki T, Matozaki T. Small GTP-binding proteins. Physiological Rev. 2001; 81 : 153–208. 51. Wu K, Jerdeva GV, da Costa SR, Sou E, Schechter JE, Hamm-Alvarez SF. Molecular mechanisms of lacrimal acinar secretory vesicle exocytosis. Experimental Eye Res. 2006; 83 : 84–96. 52. Bahamondes V, Albornoz A, Aguilera S, et al . Changes in Rab3D expression and distribution in the acini of Sjogren's syndrome patients are associated with loss of cell polarity and secretory dysfunction. Arthritis and Rheumatism. 2011; 63 : 3126–3135.21702009 53. Barrera MJ, Sanchez M, Aguilera S, et al . Aberrant localization of fusion receptors involved in regulated exocytosis in salivary glands of Sjogren's syndrome patients is linked to ectopic mucin secretion. J Autoimmunity. 2012; 39 : 83–92.22285554 54. Goicovich E, Molina C, Perez P, et al . Enhanced degradation of proteins of the basal lamina and stroma by matrix metalloproteinases from the salivary glands of Sjogren's syndrome patients: correlation with reduced structural integrity of acini and ducts. Arthritis and Rheumatism. 2003; 48 : 2573–2584.13130477 55. da Costa SR, Wu K, Veigh MM, et al . Male NOD mouse external lacrimal glands exhibit profound changes in the exocytotic pathway early in postnatal development. Exp Eye Res. 2006; 82 : 33–45.16005870 56. Ju Y, Janga SR, Klinngam W, et al . NOD and NOR mice exhibit comparable development of lacrimal gland secretory dysfunction but NOD mice have more severe autoimmune dacryoadenitis. Exp Eye Res. 2018; 176 : 243–251.30201519 57. Zengin N. The effect of dacryocystorhinostomy on tear film flow and stability in patients with chronic dacryocystitis. Acta Ophthalmologica. 1993; 71 : 714–716.8109222 58. Foulds WS. Intra-Canalicular Gelatin Implants in the Treatment of Kerato-Conjunctivitis Sicca. Br J Ophthalmol. 1961; 45 : 625–627.18170714 59. Dartt DA. Neural regulation of lacrimal gland secretory processes: relevance in dry eye diseases. Prog Retin Eye Res. 2009; 28 : 155–177.19376264 60. Meng Y, Lu H, Wang C, et al . Naso-ocular neuropeptide interactions in allergic rhinoconjunctivitis, rhinitis, and conjunctivitis. The World Allergy Organization Journal. 2021; 14 : 100540.34035875 61. Sacchetti M, Micera A, Lambiase A, et al . Tear levels of neuropeptides increase after specific allergen challenge in allergic conjunctivitis. Molecular Vision. 2011; 17 : 47–52.21245958 62. Dieckmann G, Fregni F, Hamrah P. Neurostimulation in dry eye disease-past, present, and future. The Ocular Surface. 2019; 17 : 20–27.30419304

PMC010xxxxxx/PMC10353749.txt

==== Front Transl Vis Sci Technol Transl Vis Sci Technol TVST Translational Vision Science & Technology 2164-2591 The Association for Research in Vision and Ophthalmology 37440249 10.1167/tvst.12.7.14 TVST-23-5636 Review Review A Systematic Review and Meta-Analysis of Applying Deep Learning in the Prediction of the Risk of Cardiovascular Diseases From Retinal Images Retinal Image-Based Cardiovascular Risk Prediction Hu Wenyi 1 2 Yii Fabian S. L. 3 4 Chen Ruiye 1 2 Zhang Xinyu 5 Shang Xianwen 1 Kiburg Katerina 1 Woods Ekaterina 1 Vingrys Algis 1 6 Zhang Lei 7 Zhu Zhuoting 1 He Mingguang 1 2 1 Centre for Eye Research Australia, Royal Victorian Eye and Ear Hospital, East Melbourne, Australia 2 Department of Surgery (Ophthalmology), The University of Melbourne, Melbourne, Australia 3 Centre for Clinical Brain Sciences, Edinburgh Medical School, University of Edinburgh, Edinburgh, UK 4 Curle Ophthalmology Laboratory, Institute for Regeneration and Repair, University of Edinburgh, Edinburgh, UK 5 Shanghai Jiaotong University School of Medicine, Shanghai, China 6 Department of Optometry and Vision Sciences, The University of Melbourne, Melbourne, Australia 7 Central Clinical School, Faculty of Medicine, Nursing and Health Sciences, Monash University, Melbourne, Australia * Correspondence: Mingguang He, Department of Surgery (Ophthalmology), The University of Melbourne, Melbourne, Australia, 32 Gisborne Street, East Melbourne, Victoria 3002, Australia. e-mail: mingguang.he@unimelb.edu.au Zhuoting Zhu, Centre for Eye Research Australia, Royal Victorian Eye and Ear Hospital, 32 Gisborne Street, East Melbourne, Victoria 3002, Australia. e-mail: lisa.zhu@unimelb.edu.au Lei Zhang, Central Clinical School, Faculty of Medicine, Nursing and Health Sciences, Monash University, 37, Rainforest Walk, Clayton, Victoria 3168, Australia. e-mail: lei.zhang1@monash.edu Algis Vingrys, Department of Optometry and Vision Sciences, The University of Melbourne, 800, Swanston Street, Melbourne, Victoria 3053, Australia. e-mail: algis@unimelb.edu.au 13 7 2023 7 2023 12 7 1408 6 2023 24 3 2023 Copyright 2023 The Authors 2023 https://creativecommons.org/licenses/by-nc-nd/4.0/ This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. Purpose The purpose of this study was to perform a systematic review and meta-analysis to synthesize evidence from studies using deep learning (DL) to predict cardiovascular disease (CVD) risk from retinal images. Methods A systematic literature search was performed in MEDLINE, Scopus, and Web of Science up to June 2022. We extracted data pertaining to predicted outcomes, model development, and validation and model performance metrics. Included studies were graded using the Quality Assessment of Diagnostic Accuracies Studies 2 tool. Model performance was pooled across eligible studies using a random-effects meta-analysis model. Results A total of 26 studies were included in the analysis. There were 42 CVD risk-related outcomes predicted from retinal images were identified, including 33 CVD risk factors, 4 cardiac imaging biomarkers, 2 CVD risk scores, the presence of CVD, and incident CVD. Three studies that aimed to predict the development of future CVD events reported an area under the receiver operating curve (AUROC) between 0.68 and 0.81. Models that used retinal images as input data had a pooled mean absolute error of 3.19 years (95% confidence interval [CI] = 2.95–3.43) for age prediction; a pooled AUROC of 0.96 (95% CI = 0.95–0.97) for gender classification; a pooled AUROC of 0.80 (95% CI = 0.73–0.86) for diabetes detection; and a pooled AUROC of 0.86 (95% CI = 0.81–0.92) for the detection of chronic kidney disease. We observed a high level of heterogeneity and variation in study designs. Conclusions Although DL models appear to have reasonably good performance when it comes to predicting CVD risk, further work is necessary to evaluate the real-world applicability and predictive accuracy. Translational Relevance DL-based CVD risk assessment from retinal images holds great promise to be translated to clinical practice as a novel approach for CVD risk assessment, given its simple, quick, and noninvasive nature. artificial intelligence (AI) cardiovascular diseases (CVDs) retinal Imaging ==== Body pmcIntroduction Although cardiovascular diseases (CVDs) are the leading cause of mortality globally, up to 80% of premature CVD events can be prevented by managing modifiable risk factors through lifestyle improvement.1,2 A comprehensive risk assessment considering multiple risk factors is effective in preventing CVD events3 and is therefore considered the first key step for early identification and intervention of high-risk patients.4 Numerous risk assessment tools that integrate various clinical risk factors have been developed in the past, such as the Framingham risk score, Systemic COronary Risk Evaluation (SCORE), and QRISK.5–7 Although these risk-factor-based assessment tools are well-established and even adopted by guidelines, the involvement of blood test results and the limited time for comprehensive data collection and entry in real-world clinics may limit their widespread usage.8 The retina is recognized as a “window” to visualize and assess cardiovascular health noninvasively,9 as it is considered to share similar anatomic structure and physiological function with cardiac vasculature.9 Previous studies have indicated associations between various retinal features and the risk of developing CVD, ranging from retinal vascular geometry/morphology (i.e. vessel caliber, branching angle, tortuosity, and fractal dimension), retinal vascular network patterns, and retinal pathologies (cotton wool spots, arteriovenous nicking, and microaneurysm).10–13 However, exclusively focusing on specific measurements might overlook some implicit information and underestimate the potential of the retina as a whole to inform cardiovascular (CV) health. Deep learning (DL) is a subfield of the artificial intelligence (AI) techniques that focuses on utilizing artificial neural networks with multiple computational layers to learn and extract complex predictive features from high-dimensional data, including medical images.14 Recent studies have developed DL algorithms that could make accurate diagnoses for diseases that are largely dependent on morphology, such as diabetic retinopathy, returning comparable or slightly better performance from AI than human graders.15 With all these resources, it is an emerging area of research to investigate the application of DL to predict the risk of CVD using retinal images. However, distinct datasets and methodologies were adopted in the development and validation of the algorithms, and a variety of CVD-related risk factors or outcomes were used as the prediction outcome in different studies. Recently, Wong et al. have conducted a narrative review on the current scope and future directions of AI on retinal images for CVD prediction.16 Arnould et al. also conducted a review focusing on the application of retinal vascular networks and the development of oculomics in future CVD risk assessment.17 Despite the importance of these studies, very few studies have performed a systematic review to comprehensively evaluate the characteristics of these studies. Therefore, we aim to perform a systematic review and meta-analysis to evaluate the studies that apply DL and retinal images in the prediction of CVD risk, to comprehensively understand the characteristics of these studies in terms of model development and validation, predictors, and CVD-risk-related outcomes and quantify the performances of the DL models in the prediction of CVD risk. Methods We conducted the systematic review following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist.18 This study protocol is registered on the International Prospective Register of Systematic Reviews (PROSPERO) under the ID CRD42022364921. Literature Search We performed a systematic literature search in PubMed, Scopus, and Web of Science on June 30, 2022, using search terms to identify eligible articles with no limitations on the publication year. The keywords used for the search included “artificial intelligence,” “machine learning,” or “deep learning,” combined with “retinal image,” “retinal photo,” “fundus photo,” “fundus image,” and combined “cardiovascular disease,” “cardiovascular risk,” “coronary heart disease,” “myocardial infarction,” or “stroke,” and “risk assessment” or “screening.” The detailed search strategy is shown in the Appendix 1. Titles and abstracts were independently reviewed by two researchers (authors W.H. and R.C.), and all relevant citations were included for full-text analysis. Eligibility Criteria Studies that reported the performance of DL algorithms in predicting CVD risk, including CVD risk factors, CVD risk score, or incident CVD events, and using retinal images as their input were included. Studies that reported DL algorithms using vascular segmentation or extracting specific retinal features from the retinal image to predict CVD risk were excluded from this study as the present review focused on the DL pipeline that took a whole image directly as input, which provides a more holistic approach to risk prediction by understanding different predictive information from all potential retinal features and arguably has greater clinical utility insofar as automated prediction is concerned. Included studies had to be written in English, conducted on human subjects, and report original research. Preprints and conference papers were also included. To find any other potentially relevant research, relevant studies cited by eligible studies were also examined. Data Extraction and Quality Assessment Data extraction and quality assessment were performed by one reviewer (authors W.H.) and underwent double check by another two reviewers (authors X.Z. and F.Y.) independently. Relevant data were extracted from the retrieved articles using a predefined Excel spreadsheet. For all the retrieved articles, the extracted data included the first author, publication year, type of DL model structure, predictive horizon, the study populations, the input(s) for the development of the DL models (retinal images with/without other inputs), the reference standard (the prediction outcome of the DL models), the datasets used, sample sizes, and the number of outcomes for model development and internal validation, the type of internal validation (cross-validation or random split), and external validation if available. We also included the measurement and results used in assessing the diagnostic accuracy or reliability of the DL algorithms in the internal and external validation datasets, and the features highlighted in the retrieved attention maps of the algorithms if available. In addition, more data were extracted for a subset of studies of those that the prediction outcomes were incident CVD cases. The extracted data included basic article information, the definition of CVD events (including the prediction horizon of incident CVD), follow-up periods, predictors (retinal images per se or retina-predicted intermediate traits), sample sizes, and cases of the cohorts, modeling method, and the measurement adopted with results of diagnostic accuracy. Quality assessment was performed during data extraction, using Quality Assessment of Diagnostic Accuracies Studies 2 (QUADAS-2).19 This assessment tool designed for diagnostic accuracy studies systematically evaluated the risk of bias in four domains, including patient selection, index test, reference test, and flow and timing, and concerns of the applicability of the studies regarding the review questions in all the domains except for flow and timing. Data Synthesis and Analysis Data synthesis was performed qualitatively to summarize the characteristics of included studies, including the type of DL model structures used, the CVD risk-related prediction outcomes, the predictors and sample size of studies, the studies that underwent external validation, and the studies that generated attention maps and the features highlighted in different studies for certain CVD risk-related prediction outcomes. Studies with the same predictor (retinal images or retinal images with clinical data), prediction outcome, and measure(s) of diagnostic accuracy or reliability, performed in two or more studies with three or more cohorts were synthesized for meta-analysis. Random-effect models were used to estimate the pooled effect sizes across studies, using precomputed effect sizes and their 95% confidence intervals (CIs) extracted from the articles. If the algorithm was evaluated in both internal validation and external validation in different cohorts, it was considered as a different study in the meta-analysis. I2 was used to evaluate the heterogeneity between studies (25%–50%, 50%–75%, and 75% and over represent low, moderate, and high heterogeneity, respectively). All statistical analyses were performed using Stata version 16 (StataCorp LLC, College Station, TX). Results A total of 859 records were found using the search strategy, and 277 duplicates were removed, resulting in a total of 582 records for screening. Five hundred thirty-eight records were excluded after reviewing the titles and abstracts. Of the remaining articles for full-text screening, 18 studies were excluded because of the adoption of vessel segmentation or feature extraction instead of whole retinal image analysis as the input, using other irrelevant imaging modalities or irrelevant prediction outcomes, being a case study, without sufficient data on the datasets and outcome measures, or being retracted. A total of 26 studies were included in the study for qualitative appraisal.20–45 Three studies were synthesized for the pooled performance of the algorithms used for age prediction, four for gender prediction, three for diabetes detection, and two for chronic kidney disease (CKD) detection. Results from internal and external validation cohorts, if available, of the same study were included for the meta-analysis. The PRISMA flow diagram is shown in Figure 1. The data extracted to demonstrate the characteristics of all the included studies are shown in detail in Appendix 2. The characteristics of a subset of these studies that utilized retinal images or retinal image-predicted intermediate traits to further predict incident CVD events are listed separately in Appendix 3. The quality assessment results are displayed in Figure 2 and details are described in Appendix 4. The main reasons for the risk of bias lie in the adoption of a case-control designs in terms of patient selection, the unclear definition of the reference standards, and the use of inappropriate intervals between the index test (retinal image acquisition) and the reference standard. Figure 1. PRISMA flow diagram. Figure 2. Diagram of quality assessment results of the 26 included studies using QUADAS-2. CVD Risk-Related Outcomes A total of 42 CVD risk-related outcomes were predicted using retinal images in the included studies as some studies predicted more than one outcome. The detailed breakdown of the CVD risk-related outcomes that have been used for prediction from retinal images are depicted in Figure 3. A total of 33 of them were CVD risk factors spanning the following categories: demographic factors and lifestyle, cardiometabolic factors, other blood measures, renal function, and CKD. Four of them are cardiac imaging biomarkers. The other CVD risk-related outcomes included CVD risk scores (10-year ischemic CVD [ICVD] risk score) and 10-year atherosclerotic CVD [ASCVD] risk score), presence of CVD (presence of CVD and presence of peripheral artery disease, and presence of stroke), and incident CVD. Among these, 6 studies predicted the longitudinal CVD risk-related outcomes, including incident type 2 diabetes, incident CKD, and incident CVD. The most common CVD risk-related outcomes were: age (n = 10), gender (n = 6), smoking status (n = 5), diabetes (n = 5), HbA1C (n = 4), body mass index (BMI; n = 4), blood glucose (n = 3), hypertension (n = 3), HDL-cholesterol (n = 3), coronary artery calcium (CAC) score (n = 3), presence of renal impairment or CKD (n = 3), presence of CVD (n = 4), and incident CVD (n = 3). Notably, different definitions of the reference standards may be used in different studies, which can be found in detail in Appendix 2. Figure 3. Summary of CVD risk-related prediction outcomes in the included studies. Model Inputs, Development, and Internal Validation Retinal images as the only input for the development of DL models were investigated in 25 (96.2%) studies, and one study only evaluated the multi-modal DL algorithm with both retinal images and clinical metadata as the input. Three studies experimented with the used of retinal images only and multi-modal input to assess the performance of DL models in the prediction of CVD risk-related outcomes. In the three studies that predicted incident CVD events, Poplin et al. used retinal images per se (and clinical metadata) for prediction,26 whereas Rim et al. and Diaz.Pinto et al. used retina-predicted endophenotypes, including the RetiCAC (i.e. retina-predicted probability of the presence of CAC) with demographic factors and retina-predicted left ventricular mass (LVM)/left ventricular end-diastolic volume (LVEDV) with demographic factors, respectively.22,24 A variety of DL algorithms were used for model development in the studies, including Convolutional Neural Network (CNN) architectures, such as visual geometry graphic (VGG), Inception, ResNet, Inception-ResNet, Xception, EfficientNet, DiaNet, MobileNet, cCondenseNet, DenseNet, NASNet-Large, and one CNN architecture not specified. The CNN architectures were also used in combination with other models. For example, Mueller et al. utilized multiple instance learning (MIL) to process images to reserve high resolution of the input retinal images.29 Diaz Pinto et al. applied a multichannel variational autoencoder (mcVAE), a multimodal AI model, that integrates cardiac magnetic resonance imaging (MRI) scans and retinal images for CVD risk prediction.24 Different terms were used in distinct studies for the development, internal validation, and external validation of the DL model. In this review, development datasets included both training and tuning sets in the original articles. Internal validation in this review was called the internal test in some studies. The sample sizes of the development datasets were reported in terms of retinal images and/or individuals, which ranged between 135 images from 77 participants and 798,866 images from 390,947 participants. Seven studies that predicted a categorical outcome did not specify the number of cases included in the development datasets. Two of them reported the cases in the internal validation set and the number can be estimated given the random split of samples, whereas the other five studies specified the number of cases neither in the development nor internal validation set. There were 18 (69.2%) studies that randomly split the total sample for development and validation, whereas 8 (30.8%) studies used k-fold cross-validation for the development and internal validation of the DL algorithms. The details of the DL model performances of the internal validation can be found in detail in Appendix 2.2. Regarding age prediction from retinal images, the DL algorithms in different studies achieved a mean absolute error (MAE) ranging from 2.74 to 3.55 years.26,31,32,34,42 For gender prediction, the AUROC of the DL algorithms was between 0.704 and 0.978 among all relevant studies.26,28,31,32,40,44 Among the studies that investigated the prediction of smoking status, the models achieved an AUROC that ranged from 0.71 to 0.86.26,28,30,32,44 The DL models had an MAE between 0.61% and 1.39% for the prediction of HbA1c value26,32 and an MAE between 0.652 and 1.06 mmol/l for the prediction of blood glucose level.32,38 Among the studies that predicted the presence of diabetes, the models had an AUROC ranging from 0.731 to 0.923 using retinal images only, which increased to 0.929 by adding clinical data.23,28,33,38,40 The MAE ranged from 8.96 to 11.35 mm Hg for systolic blood pressure prediction, 6.42 to 6.84 mm Hg for diastolic blood pressure, and 3.29 to 4.31 kg/m2 for BMI prediction.26,32 The AUROC for prediction of prevalent CKD was between 0.911 and 0.918 and was up to 0.938 when clinical metadata was added to the model.37,38 As for the prediction of prevalent CVD, the AUROC ranged from 0.499 to 0.700,40,44,45 and the accuracy was reported to be 0.756 when retinal images were used only and increased to 0.783 when dual-energy X-ray absorptiometry (DXA) data was added.39 External Validation Six (23.1%) studies validated the performance of the DL algorithms using a total of nine external validation sets. Sample sizes of the external validation sets ranged from 1054 images from 527 participants to 56,301 participants. The six studies were developed in Korean (n = 2), Chinese (n = 2), multi-ethnic (Malay, Chinese, and Indian, n = 1), and Caucasian (n = 1) populations. Only two of the studies were validated in a population of different ethnicities. Nusinovici et al.25 developed the DL algorithm for the prediction of age in the Korean population and validated the performance in the UK Biobank, which predominantly consisted of Caucasians. The AUROC of predicting the probability of being 65 years and over was 0.968 (95% CI = 0.965–0.970) in the internal validation set, and 0.756 (95% CI = 0.753–0.759) in the external validation set. Yun et al.40 developed an algorithm for the detection of type 2 diabetes in the UK Biobank population, which was validated in a sample composed of Korean patients with diabetes and non-diabetic patients from the UK Biobank. The AUROC in the internal and external validation set was 0.731 (95% CI = 0.707–0.756) and 0.703 (95% CI = 0.691–0.715), respectively. Performance of the AI Algorithm and Meta-Analysis The performance of the DL algorithms was measured in a series of parameters, including the diagnostic accuracy and reliability measures such as sensitivity, specificity, AUROC, accuracy positive predictive value (PPV), negative predictive value (NPV), area under the precision-recall curve (AUPRC), F1 score for binary prediction outcomes, MAE, and the limits of agreement for continuous outcomes. The details of the performance measures were listed by prediction outcome in Appendix 2.2. Meta-analysis was performed for eligible studies that predicted age, gender, diabetes, and the presence of CKD. The detailed results are shown in Appendix 5. CVD Risk Score Prediction Two studies investigate the prediction of CVD risk scores from retinal images only. Ma et al.35 developed and validated the algorithm to predict a 10-year ICVD risk score, which was defined by 7 parameters determined by the Cox regression models, including age, sex, systolic blood pressure, total cholesterol, BMI, smoking status, and diabetes. The AUROC for borderline risk (ICVD risk >5%) was 0.971 (95% CI = 0.967–0.975) and the AUROC for intermediate or high risk (ICVD risk >7.5%) was 0.976 (95% CI = 0.973–0.980) on an individual level. The R2 was 0.876, indicating that 87.6% of the variability in the retina-predicted score is explained by the ICVD risk score. Syed et al.41 performed a study to predict a 10-year ASCVD risk score, which was defined by the pooled cohort equations (PCEs) consisting of age, sex, systolic and diastolic blood pressure, total and high-density lipoprotein cholesterol, diabetes, and smoking status. The MAE was 0.1085 (96% CI = 0.1053, 0.1116) and R2 was 0.5338 (95% CI = 0.5036, 0.5628). Longitudinal Prediction of Incident CVD Events Three studies investigated the prediction of incident CVD events longitudinally (Appendix 3). Poplin et al.26 developed an algorithm using retinal images per se to predict 5-year major adverse CVD events in the UK Biobank study, achieving an AUROC of 0.70 (95% CI = 0.65–0.74). When adding well-established clinical risk factors or SCORE to the model, the AUROCs were 0.73 (95% CI = 0.69-0.77) and 0.72 (95% CI = 0.67–0.76), respectively. Rim et al.22 tested the predictive value of RetiCAC, which was an intermediate trait predicted from retinal images, representing the probability of CAC presence. The combination of RetiCAC with age and gender, or with PCE was modeled to predict incident fatal (and non-fatal) CVD events during a median follow-up period ranging from 4.1 years to 10.3 years in multiple cohorts. The C-statistics ranged between 0.68 (95% CI = 0.58–0.79) (retinal images and age and gender in the CMERC-HI study) and 0.806 (95% CI = 0.790–0.828; retinal images and PCE in the SEED study). Diaz Pinto et al.24 used another endophenotype predicted from the retinal image, namely the retina-predicted LVM and retinal-predicted LVEDV in combination with minimal demographic factors to predict incident myocardial infarction. The mean (± standard deviation) accuracy was 0.74 ± 0.03 and the mean (± standard deviation) AUROC was 0.8 ± 0.02 in the UK Biobank from which the algorithm was developed. The AUROC and accuracy were 0.59 and 0.59 when tested in the age-related eye disease study (AREDS) with all age-related macular degeneration (AMD) images included and 0.70 and 0.68 when all AMD images were excluded. Attention Maps Attention maps were retrieved in 17 (65.4%) studies. For the prediction of age, the features highlighted included the macula, optic disc (including optic nerve head), blood vessels (including vessel arcade at the posterior pole), and regions around the blood vessels.25–27,31,34 Features in gender prediction comprised of optic discs, macula (including the fovea), and vessels.26,31 Specifically, Kim et al. reported that the proximal vascular regions were prominently highlighted in women.31 Blood vessels, perivascular regions, and the fovea were indicated in the attention maps in the prediction of smoking status.26,30 Nonspecific perivascular surroundings were featured in one study predicting HbA1C.26 In the prediction of diabetes, the central retina area between the optic disc and the macula was highlighted in one study.23 In addition, another study found features scattered throughout the whole image that may correspond to diabetic retinopathy (DR) changes in some cases, in patients with diabetes with/without DR.38 Blood vessels and nonspecific features were highlighted in the prediction of systolic blood pressure and diastolic blood pressure, respectively.26 Nonspecific features were found for the prediction of BMI as well.26 Blood vessels, in specific main retinal branches, vessels around the optic disc and vascular arcades, the optic nerve head and its pathologic changes, and retinal pathologies, such as cotton-wool spots were the major features highlighted for the prediction of cardiac imaging biomarkers, including CAC score, branchial-ankle pulse wave velocity (baPWV), and carotid atherosclerosis.21,22,27,43 As for the prediction of renal function impairment or CKD, retinopathy changes, retinal vascular geometry/morphology (including venous caliber, vessel density, vessel branch points, and arterio-venous junctions) and the optic nerve were featured.36–38 Optic disc, vessels, and macula were extracted in the prediction of CVD risk score.41 The central region of the retina, microhemorrhage, vessel arcades, and the optic disc were documented in the studies that predicted the presence of CVD.29,39 Discussion This study was conducted to qualitatively appraise the characteristics of the studies that apply DL to predict CVD risk-related outcomes from retinal images and to quantify the diagnostic accuracy of studies that were designed to predict the same CVD risk-related outcomes. This systematic review identified a variety of CVD risk-related outcomes that were predicted from retinal images and the diagnostic accuracy of commonly predicted CVD risk-related outcomes, such as age, gender, diabetes, and CKD, were highly acceptable in the experimental settings. Nevertheless, the results need to be interpreted with caution given the limited number of studies and the between-study heterogeneities. The application of DL to predict CVD risk from the retinal image is an emerging area of research at the exploratory stage. In terms of prediction outcome of interest, more than 40 CVD risk-related outcomes were summarized in this study. Regarding the performances of the DL algorithms, variations can be found in the performances of the DL algorithms that predicts certain CVD risk-related outcome among different studies. The differences may lie in the variation of sample sizes, different DL algorithms used, and different populations involved. This suggests that even the algorithms achieved a reasonably good performance in its development and validation, it warrants further validation in different datasets to increase the reliability, robustness, and generalizability. In the limited number of studies that are eligible to examine the pooled performance of the algorithms, the results showed that the DL models achieved an area under the curve (AUC) of at least 0.80 in discriminating gender, diabetes status, and CKD. However, although algorithms reported in the study achieved an acceptable performance in predicting single risk factors, it should be borne in mind that some of the risk factors are nonspecific if used alone, and some are intuitive without the need for retinal images and analytical algorithms. In terms of the prediction scheme of the CVD risk-related outcomes, most studies are detecting the CVD risk-related features cross-sectionally, and the prediction of future risk is not defined or further investigated. There could be difficulties in the accurate prediction of future CVD events as the effects of treatment and future comorbid changes will become manifest over a longitudinal period.46 This may explain why the majority of prediction schemes for cross-sectional endophenotypes or CVD risk scores are retinal images only, whereas the prediction of future CVD risk requires the addition of clinical risk factors to improve the performance of the AI algorithms. In terms of study reporting, some studies put more emphasis on algorithm development and did not describe in detail the selected datasets, the definition of the reference tests, the image acquisition protocols, and the flow and timing of data collection, which may potentially introduce biases and limit the validity of the studies from a clinical perspective of view. The performance of the DL algorithms was also reported in a variety of measurements. For some categorical outcomes, only AUC or accuracy was reported but no sensitivity, specificity, PPV, and NPV were displayed, which might limit a holistic interpretation of the algorithms. Further, as different measurements will provide barriers to a pooled analysis and further translation of the technology in terms of establishing clinical cutoff values or criteria.47 This could be addressed in the future by developing and promoting the utilization of standard reporting checklists for these trials, such as Standards for Reporting Diagnostic accuracy studies - Artificial Intelligence (STARD-AI).48 The generalizability of the studies should be taken into consideration. Most studies were developed and validated in the Caucasian and Asian populations, and there is scarce evidence for the other ethnicities that may be at a higher risk of CVD, such as African Americans, aboriginal people in Canada, or indigenous Australians.49–52 The performance of the DL algorithms reported in the studies might consider measures that do not exist or are inappropriate for these populations thereby over- or underestimating their importance so the robustness of such indices needs to be interpreted with caution. Moreover, a primary challenge for ensuring the generalizability of DL models arises from domain shift, referring to the differences in the distributions of the data between the training sets and test sets, such as clinical practical settings. This shift may be due to the distinct imaging protocols or machines provided by vendors or different characteristics of the patient populations.53 In addition, the DL algorithms will perform worse on a new dataset compared with the original one, because of overfitting or data leakage.53 Only a small proportion of studies in the systematic review were externally validated. In addition, for all the studies included in the analysis, retrospective analyses of existing datasets were applied for development and validation. This could be attributable to the challenge that large numbers of images need to recruited for DL algorithm development in cases where prospective data collection are used. This is time-consuming and costly. Finally, as retrospective analysis allows image quality assessment in advance and exempts the consideration of the image acquisition success rate, it can be expected that the application of these prototypes prospectively in real-world clinical settings will face greater challenges. Nevertheless, the application of DL and retinal images in predicting CVD risk shows significant clinical implications. Compared to risk factor-based models, the use of DL and retinal images may create a more efficient and labor-free approach for CVD risk assessment. The traditional parametric prediction models combining multiple risk factors that were considered to be cumbersome in real-world practice.8 The retina-based CVD risk assessment could potentiate efficiency by removing the need for blood tests, and providing extra information on end-organ damage. Retinal photography is one of the most widely adopted and cost-effective imaging modalities used for routine eye care.54 The development of retinal image acquisition technologies, handheld fundus cameras, or smartphone-based fundus photography will further increase the accessibility of the service.55,56 The adoption of DL technology to analyze retinal images has already been tested in a more advanced stage in terms of detecting eye diseases, such as DR, from which the experience can be learnt for the future implementation in terms of CVD risk assessment. Google has conducted a prospective interventional cohort study in Thailand to investigate the performance of AI as a real-time DR screening service in community care settings. For vision-threatening DR, the DL system achieved a comparable accuracy (94.7% vs. 95.3%) and specificity (95.4% vs. 95.5%) to the retinal specialists on-site and higher sensitivity (91.4% vs. 84.8%).57 Based on evidence from prospective studies that prove the diagnostic accuracy of AI application, recent years have also witnessed the approval of two AI screening systems, EyeArt and IDx-DR by the US Food and Drug Administration (FDA) to detect more-than-mild DR (mtmDR).58,59 Taken together, the evidence from DR has shed light on the feasibility and the impact of applying automated AI systems can have on retinal image analysis and disease diagnosis. Although no prospective studies have been performed to validate the real-world performance of the AI systems that predict CVD, we can extrapolate the great potential of implementing it in clinical practice from the DR screening experience. Future studies and collaborations are needed to improve the data source with sufficient sample sizes and upgrade the algorithms to predict CVD events accurately. The constraint of limited sample size could be a major barrier in developing reliable algorithms.60 A couple of studies examined in this review has fewer than 200 labeled images,20,27 potentially due to the high intrinsic cost of obtaining certain sufficient data for various labels, which to some extent prevents the DL models to achieve their full potential. Furthermore, the limited number of longitudinal cohort studies with sufficient incident CVD data might be a reason of the scarcity of evidence on retinal images in the prediction of incident CVD. To date, only one study used retinal images as the sole input for future CVD events, with moderate-to-good discrimination of 0.70. The performance is comparable to well-established clinical risk factor-based models, such as the FRS, PCE, SCORE, and QRISK (AUC = 0.63 to 0.82).61–64 Therefore, future collaborations are needed for the availability of multi-ethnic large-scale longitudinal cohorts with follow-up information on both predictors and outcomes to facilitate the development and validation of algorithms. In addition to the availability of datasets, introducing the concept of open-source algorithms to this field is also a beneficial way to accelerate the translation and customization of the technology as the availability of the codes enables more validation and enhances the performance, facilitates the adjustment of the technology to specific clinical settings, reduces the cost, and enhances benign competition between similar products.65 The clinical implication and integration of this technology into a feasible model of care requires careful consideration. Compared to the diagnosis of eye diseases which directly detected the pathologies from the ocular images, CVD risk prediction is more challenging as it is a multifactorial systemic disease, the feature on the retinal images to inform cardiovascular health can be influenced by the complex interplay of environmental factors and genetic risk factors.66 Therefore, accurate prediction of well-established CVD risk scores might be a solution that could provide extra individualized information than traditional risk factor-based models and simplify the clinical procedure, such as negating invasive tests and complex manual data collection. Multimodal AI models combining retinal images and minimal clinical metadata might be another solution to improve the prediction performance of the algorithms. But the inclusion of clinical data should take into consideration the principle of noninvasive and simple data acquisition procedures to maximize the benefits of using retinal images as the predictor in real practice over traditional risk factor-based models. Strengths and Limitations of the Study This systematic review comprehensively examines the studies using DL to predict CVD risk-related outcomes from retinal images, using structured search terms, literature search, and extracting a wide range of data which enables interpretation of the studies from multiple angles. However, a couple of limitations need to be made aware of. First, there is high heterogeneity in the included studies in terms of CVD risk-related outcomes, methodologies, and measurements of outcomes, which prevented us from performing quantification for all prediction outcomes. Second, studies processing the retinal images with vessel segmentation algorithms to predict CVD risk were excluded in this review for the purpose of keeping the predictors homogeneous to facilitate the interpretation and preventing us from overlooking the value of other retinal features. Nevertheless, retinal vasculature is one of the most significant retinal features in predicting CVD risk, and numerous studies have been conducted on this specific topic, therefore, a review focusing on this topic is worthwhile to be conducted separately. Third, a limited number of studies were eligible for the meta-analysis which increases the uncertainties of the outcome of pooled estimates.67 Finally, as no quality assessment tools are currently available for AI-based diagnostic accuracy studies, we used the QUADAS-2 quality assessment tool of which the interpretation can be varied between studies. Conclusion In conclusion, this systematic review and meta-analysis qualitatively interpret the studies that use DL and retinal images to predict CVD risk-related outcomes. A wide range of the predicted outcomes was investigated but the evidence is scarce on the prediction of incident CVD longitudinally. Future studies are needed to validate and refine the algorithms, especially in large-scale longitudinal cohorts. In addition, prospective studies need to be conducted to prove the applicability of the technology in real-world practice. Supplementary Material Supplement 1 Acknowledgments Supported by Medical Research Future Fund (MRFF; MRFAI000035), NHMRC Investigator Grants (APP1175405 and 2010072) and High-level Talent Flexible Introduction Fund of Guangdong Provincial People's Hospital (KJ012019530). The sponsor or funding organization had no role in the design or conduct of this research. Author Contributions: Study concept and design: M.G.H., W.Y.H., and Z.T.Z. Access to raw data: W.Y.H. and R.Y.C. Acquisition, analysis, or interpretation: W.Y.H., F.Y., and R.Y.C. Drafting of the manuscript: W.Y.H. Critical revision of the manuscript for important intellectual content: M.G.H., A.V., Z.T.Z., X.W.S., K.K., and E.W. Statistical analysis: W.Y.H. and X.W.S. Obtained funding: M.G.H. Administrative, technical, or material support: M.G.H. Study supervision: M.G.H., A.V., Z.T.Z., X.H.Y., X.W.S., K.K., and E.W. Disclosure: W. Hu, None; F.S.L. Yii, None; R. Chen, None; X. Zhang, None; X. Shang, None; K. Kiburg, None; E. Woods, None; A. Vingrys, None; L. Zhang, None; Z. Zhu, None; M. He, None ==== Refs References 1. Stampfer MJ, Hu FB, Manson JE, et al . Primary prevention of coronary heart disease in women through diet and lifestyle. N Engl J Med. 2000; 343 (1 ): 16–22.10882764 2. Chiuve SE, McCullough ML, Sacks FM, Rimm EB. Healthy lifestyle factors in the primary prevention of coronary heart disease among men: benefits among users and nonusers of lipid-lowering and antihypertensive medications. Circulation. 2006; 114 (2 ): 160–167.16818808 3. Manuel DG, Lim J, Tanuseputro P, et al . Revisiting Rose: strategies for reducing coronary heart disease. BMJ. 2006; 332 (7542 ): 659–662.16543339 4. Heart Foundation. Cardiovascular disease (CVD) risk assessment and management. Available at: https://www.heartfoundation.org.au/bundles/for-professionals/cvd-risk-assessment-and-management. Accessed November 16, 2022. 5. D'Agostino RBSr., Vasan RS, Pencina MJ, et al . General cardiovascular risk profile for use in primary care: the Framingham Heart Study. Circulation. 2008; 117 (6 ): 743–753.18212285 6. Group Sw, Collaboration ESCCr. SCORE2 risk prediction algorithms: new models to estimate 10-year risk of cardiovascular disease in Europe. Eur Heart J. 2021; 42 (25 ): 2439–2454.34120177 7. Hippisley-Cox J, Coupland C, Brindle P. Development and validation of QRISK3 risk prediction algorithms to estimate future risk of cardiovascular disease: prospective cohort study. BMJ. 2017; 357 : j2099.28536104 8. Persell SD, Dunne AP, Lloyd-Jones DM, Baker DW. Electronic health record-based cardiac risk assessment and identification of unmet preventive needs. Med Care. 2009; 47 (4 ): 418–424.19238100 9. Flammer J, Konieczka K, Bruno RM, et al . The eye and the heart. Eur Heart J. 2013; 34 (17 ): 1270–1278.23401492 10. Huang L, Aris IM, Teo LLY, et al . Exploring associations between cardiac structure and retinal vascular geometry. J Am Heart Assoc. 2020; 9 (7 ): e014654.32248764 11. McGeechan K, Liew G, Macaskill P, et al . Meta-analysis: retinal vessel caliber and risk for coronary heart disease. Ann Intern Med. 2009; 151 (6 ): 404–413.19755365 12. Seidelmann SB, Claggett B, Bravo PE, et al . Retinal vessel calibers in predicting long-term cardiovascular outcomes: the atherosclerosis risk in communities study. Circulation. 2016; 134 (18 ): 1328–1338.27682886 13. Arnould L, Binquet C, Guenancia C, et al . Association between the retinal vascular network with Singapore “I” Vessel Assessment (SIVA) software, cardiovascular history and risk factors in the elderly: the montrachet study, population-based study. PLoS One. 2018; 13 (4 ): e0194694.29614075 14. LeCun Y, Bengio Y, Hinton G. Deep learning. Nature. 2015; 521 (7553 ): 436–444.26017442 15. Raumviboonsuk P, Krause J, Chotcomwongse P, et al . Deep learning versus human graders for classifying diabetic retinopathy severity in a nationwide screening program. NPJ Digit Med. 2019; 2 : 25.31304372 16. Wong DYL, Lam MC, Ran A, Cheung CY. Artificial intelligence in retinal imaging for cardiovascular disease prediction: current trends and future directions. Curr Opin Ophthalmol. 2022; 33 (5 ): 440–446.35916571 17. Arnould L, Meriaudeau F, Guenancia C, et al . Using artificial intelligence to analyse the retinal vascular network: the future of cardiovascular risk assessment based on oculomics? A Narrative Review. Ophthalmol Ther. 2023; 12 (2 ): 657–674.36562928 18. Page MJ, McKenzie JE, Bossuyt PM, et al . The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021; 372 : n71.33782057 19. Whiting PF, Rutjes AW, Westwood ME, et al . QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med. 2011; 155 (8 ): 529–536.22007046 20. Barriada RG, Simó-Servat O, Planas A, et al . Deep learning of retinal imaging: a useful tool for coronary artery calcium score prediction in diabetic patients. Appl Sci (Switzerland). 2022; 12 (3 ): 1401. 21. Son J, Shin JY, Chun EJ, et al . Predicting high coronary artery calcium score from retinal fundus images with deep learning algorithms. Transl Vis Sci Technol. 2020; 9 (6 ): 28. 22. Rim TH, Lee CJ, Tham YC, et al . Deep-learning-based cardiovascular risk stratification using coronary artery calcium scores predicted from retinal photographs. Lancet Digit Health. 2021; 3 (5 ): e306–e316.33890578 23. Islam MT, Al-Absi HRH, Ruagh EA, Alam T. DiaNet: a deep learning based architecture to diagnose diabetes using retinal images only. IEEE Access. 2021; 9 : 15686–15695. 24. Diaz-Pinto A, Ravikumar N, Attar R, et al . Predicting myocardial infarction through retinal scans and minimal personal information. Nat Mach Intell. 2022; 4 (1 ): 55–61. 25. Nusinovici S, Rim TH, Yu M, et al . Retinal photograph-based deep learning predicts biological age, and stratifies morbidity and mortality risk. Age Ageing. 2022; 51 (4 ): afac065.35363255 26. Poplin R, Varadarajan AV, Blumer K, et al . Prediction of cardiovascular risk factors from retinal fundus photographs via deep learning. Nat Biomed Eng. 2018; 2 (3 ): 158–164.31015713 27. Nagasato D, Tabuchi H, Masumoto H, et al . Prediction of age and brachial-ankle pulse-wave velocity using ultra-wide-field pseudo-color images by deep learning. Sci Rep. 2020; 10 (1 ): 19369.33168888 28. Zhang L, Yuan M, An Z, et al . Prediction of hypertension, hyperglycemia and dyslipidemia from retinal fundus photographs via deep learning: a cross-sectional study of chronic diseases in central China. PLoS One. 2020; 15 (5 ): e0233166.32407418 29. Mueller S, Wintergerst MWM, Falahat P, et al . Multiple instance learning detects peripheral arterial disease from high-resolution color fundus photography. Sci Rep. 2022; 12 (1 ): 1389.35082343 30. Vaghefi E, Yang S, Hill S, et al . Detection of smoking status from retinal images; a convolutional neural network study. Sci Rep. 2019; 9 (1 ): 7180.31073220 31. Kim YD, Noh KJ, Byun SJ, et al . Effects of hypertension, diabetes, and smoking on age and sex prediction from retinal fundus images. Sci Rep. 2020; 10 (1 ): 4623.32165702 32. Gerrits N, Elen B, Craenendonck TV, et al . Age and sex affect deep learning prediction of cardiometabolic risk factors from retinal images. Sci Rep. 2020; 10 : 9432.32523046 33. Heslinga FG, Pluim JPW, Houben AJHM, et al . Direct classification of type 2 diabetes from retinal fundus images in a population-based sample from the Maastricht study. In Medical Imaging 2020: Computer-Aided Diagnosis. 2020; v. 11314 : pp. 383–388. SPIE. 34. Zhu ZT, Shi DL, Guankai P, et al . Retinal age gap as a predictive biomarker for mortality risk. Br J Ophthalmol. 2023; 107.4 : 547–554.35042683 35. Ma Y, Xiong J, Zhu Y, et al . Deep learning algorithm using fundus photographs for 10-year risk assessment of ischemic cardiovascular diseases in China. Science Bull. 2022; 67 : 17–20. 36. Kang EYC, Hsieh YT, Li CH, et al . Deep learning-based detection of early renal function impairment using retinal fundus images: model development and validation. JMIR Medical Informatics. 2020; 8 (11 ): e23472.33139242 37. Sabanayagam C, Xu D, Ting DSW, et al . A deep learning algorithm to detect chronic kidney disease from retinal photographs in community-based populations. The Lancet Digital Health. 2020; 2 (6 ): e295–e302.33328123 38. Zhang K, Liu X, Xu J, et al . Deep-learning models for the detection and incidence prediction of chronic kidney disease and type 2 diabetes from retinal fundus images. Nat Biomed Eng. 2021; 5 (6 ): 533–545.34131321 39. Al-Absi HRH, Islam MT, Refaee MA, et al . Cardiovascular disease diagnosis from DXA scan and retinal images using deep learning. Sensors. 2022; 22 (12 ): 4310.35746092 40. Yun JS, Kim J, Jung SH, Cha SA, Ko SH, Ahn YB, Won HH, Sohn KA, Kim D. A deep learning model for screening type 2 diabetes from retinal photographs. Nutrition, metabolism, and cardiovascular diseases. Nutr Metab Cardiovasc Dis. 2022; 32 (5 ): 1218–1226.35197214 41. Syed MG, Doney A, George G, et al . Are cardiovascular risk scores from genome and retinal image complementary? A deep learning investigation in a diabetic cohort. Ophthalmic Medical Image Analysis. OMIA 2021. Lecture Notes in Computer Science(), vol 12970 . pp. 109–118. 42. Chang J, Shin JY, Ko T, et al . Association of deep learning-based fundus age difference with carotid atherosclerosis and mortality. 2019 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), San Diego: CA, USA, 2019, pp. 1179–1181. 43. Chang J, Ko A, Park SM, et al . Association of cardiovascular mortality and deep learning-funduscopic atherosclerosis score derived from retinal fundus images. Am J Ophthalmol. 2020; 217 : 121–130.32222370 44. Khan NC, Perera C, Dow ER, et al . Predicting systemic health features from retinal fundus images using transfer-learning-based artificial intelligence models. Diagnostics (Basel). 2022; 12 (7 ): 1714.35885619 45. Coronado I, Abdelkhaleq R, Yan J, et al . Towards stroke biomarkers on fundus retinal imaging: a comparison between vasculature embeddings and general purpose convolutional neural networks. Annu Int Conf IEEE Eng Med Biol Soc. 2021: 3873–3876.34892078 46. Reinikainen J, Laatikainen T, Karvanen J, Tolonen H. Lifetime cumulative risk factors predict cardiovascular disease mortality in a 50-year follow-up study in Finland. Int J Epidemiol. 2015; 44 (1 ): 108–116.25501686 47. Park SH, Han K. Methodologic guide for evaluating clinical performance and effect of artificial intelligence technology for medical diagnosis and prediction. Radiology. 2018; 286 (3 ): 800–809.29309734 48. Sounderajah V, Ashrafian H, Golub RM, et al . Developing a reporting guideline for artificial intelligence-centred diagnostic test accuracy studies: the STARD-AI protocol. BMJ Open. 2021; 11 (6 ): e047709. 49. Schmiegelow MD, Hedlin H, Mackey RH, et al . Race and ethnicity, obesity, metabolic health, and risk of cardiovascular disease in postmenopausal women. J Am Heart Assoc. 2015; 4 (5 ): e001695.25994446 50. Hozawa A, Folsom AR, Sharrett AR, Chambless LE. Absolute and attributable risks of cardiovascular disease incidence in relation to optimal and borderline risk factors: comparison of African American with white subjects–Atherosclerosis Risk in Communities Study. Arch Intern Med. 2007; 167 (6 ): 573–579.17389288 51. Anand SS, Yusuf S, Jacobs R, et al . Risk factors, atherosclerosis, and cardiovascular disease among Aboriginal people in Canada: the study of health assessment and risk evaluation in aboriginal peoples (SHARE-AP). Lancet. 2001; 358 (9288 ): 1147–1153.11597669 52. Gardiner FW, Rallah-Baker K, Dos Santos A, et al . Indigenous Australians have a greater prevalence of heart, stroke, and vascular disease, are younger at death, with higher hospitalisation and more aeromedical retrievals from remote regions. EClinicalMedicine. 2021; 42 : 101181.34765955 53. Guan H, Liu M. Domain adaptation for medical image analysis: a survey. IEEE Trans Biomed Eng. 2022; 69 (3 ): 1173–1185.34606445 54. Muftuoglu IK, Gaber R, Bartsch DU, et al . Comparison of conventional color fundus photography and multicolor imaging in choroidal or retinal lesions. Graefes Arch Clin Exp Ophthalmol. 2018; 256 (4 ): 643–649.29492687 55. Fenner BJ, Wong RLM, Lam WC, et al . Advances in retinal imaging and applications in diabetic retinopathy screening: a review. Ophthalmol Ther. 2018; 7 (2 ): 333–346.30415454 56. Das S, Kuht HJ, De Silva I, et al . Feasibility and clinical utility of handheld fundus cameras for retinal imaging. Eye (Lond). 2022; 37 (2 ): 274–279.35022568 57. Ruamviboonsuk P, Tiwari R, Sayres R, et al . Real-time diabetic retinopathy screening by deep learning in a multisite national screening programme: a prospective interventional cohort study. Lancet Digit Health. 2022; 4 (4 ): e235–e244.35272972 58. Ipp E, Liljenquist D, Bode B, et al . Pivotal evaluation of an artificial intelligence system for autonomous detection of referrable and vision-threatening diabetic retinopathy. JAMA Netw Open. 2021; 4 (11 ): e2134254.34779843 59. Abramoff MD, Lavin PT, Birch M, et al . Pivotal trial of an autonomous AI-based diagnostic system for detection of diabetic retinopathy in primary care offices. NPJ Digit Med. 2018; 1 : 39.31304320 60. Kokol P, Kokol M, Zagoranski S. Machine learning on small size samples: a synthetic knowledge synthesis. Sci Prog. 2022; 105 (1 ): 368504211029777.35220816 61. Chia YC, Gray SY, Ching SM, et al . Validation of the Framingham general cardiovascular risk score in a multiethnic Asian population: a retrospective cohort study. BMJ Open. 2015; 5 (5 ): e007324. 62. Qureshi WT, Michos ED, Flueckiger P, et al . Impact of replacing the pooled cohort equation with other cardiovascular disease risk scores on atherosclerotic cardiovascular disease risk assessment (from the Multi-Ethnic Study of Atherosclerosis [MESA]). Am J Cardiol. 2016; 118 (5 ): 691–696.27445216 63. Rabanal KS, Meyer HE, Pylypchuk R, et al . Performance of a Framingham cardiovascular risk model among Indians and Europeans in New Zealand and the role of body mass index and social deprivation. Open Heart. 2018; 5 (2 ): e000821.30018780 64. Siontis GC, Tzoulaki I, Siontis KC, Ioannidis JP. Comparisons of established risk prediction models for cardiovascular disease: systematic review. BMJ. 2012; 344 : e3318.22628003 65. Harish KB, Price WN, Aphinyanaphongs Y. Open-source clinical machine learning models: critical appraisal of feasibility, advantages, and challenges. JMIR Form Res. 2022; 6 (4 ): e33970.35404258 66. Sing CF, Stengard JH, Kardia SL. Genes, environment, and cardiovascular disease. Arterioscler Thromb Vasc Biol. 2003; 23 (7 ): 1190–1196.12730090 67. Goh JX, Hall JA, Rosenthal R. Mini meta-analysis of your own studies: some arguments on why and a primer on how. Social and Personality Psychology Compass. 2016; 10 (10 ): 535–549.

PMC010xxxxxx/PMC10353758.txt

==== Front Age Ageing Age Ageing ageing Age and Ageing 0002-0729 1468-2834 Oxford University Press 10.1093/ageing/afad116 afad116 Research Paper AcademicSubjects/MED00280 ageing/11 Screening instruments to predict adverse outcomes for undifferentiated older adults attending the Emergency Department: Results of SOAED prospective cohort study Leahy Aoife School of Allied Health, Faculty of Education and Health Sciences, Ageing Research Centre, Health Research Institute, University of Limerick, Limerick, Ireland Department of Ageing and Therapeutics, University Hospital Limerick, Limerick, Ireland Corey Gillian School of Allied Health, Faculty of Education and Health Sciences, Ageing Research Centre, Health Research Institute, University of Limerick, Limerick, Ireland ALERT, Department of Emergency Medicine, University Hospital Limerick, Limerick, Ireland Purtill Helen Department of Mathematics & Statistics, University of Limerick, Limerick, Ireland O’Neill Aoife School of Allied Health, Faculty of Education and Health Sciences, Ageing Research Centre, Health Research Institute, University of Limerick, Limerick, Ireland Central Statistics Office, Cork, Ireland Devlin Collette School of Allied Health, Faculty of Education and Health Sciences, Ageing Research Centre, Health Research Institute, University of Limerick, Limerick, Ireland Barry Louise School of Allied Health, Faculty of Education and Health Sciences, Ageing Research Centre, Health Research Institute, University of Limerick, Limerick, Ireland School of Nursing and Midwifery, Faculty of Education and Health Sciences, University of Limerick, Limerick, Ireland Cummins Niamh School of Allied Health, Faculty of Education and Health Sciences, Ageing Research Centre, Health Research Institute, University of Limerick, Limerick, Ireland School of Medicine, University of Limerick, Limerick, Ireland Gabr Ahmed School of Allied Health, Faculty of Education and Health Sciences, Ageing Research Centre, Health Research Institute, University of Limerick, Limerick, Ireland Department of Ageing and Therapeutics, University Hospital Limerick, Limerick, Ireland Mohamed Abdirahman School of Allied Health, Faculty of Education and Health Sciences, Ageing Research Centre, Health Research Institute, University of Limerick, Limerick, Ireland Department of Ageing and Therapeutics, University Hospital Limerick, Limerick, Ireland Shanahan Elaine Department of Ageing and Therapeutics, University Hospital Limerick, Limerick, Ireland Shchetkovsky Denys ALERT, Department of Emergency Medicine, University Hospital Limerick, Limerick, Ireland Ryan Damien ALERT, Department of Emergency Medicine, University Hospital Limerick, Limerick, Ireland School of Medicine, University of Limerick, Limerick, Ireland O’Loughlin Monica Ageing Research Centre, University of Limerick, Limerick, Ireland O'Connor Margaret Department of Ageing and Therapeutics, University Hospital Limerick, Limerick, Ireland Galvin Rose School of Allied Health, Faculty of Education and Health Sciences, Ageing Research Centre, Health Research Institute, University of Limerick, Limerick, Ireland Ageing Research Centre, University of Limerick, Limerick, Ireland Address correspondence to: Aoife Leahy. Tel: 061 301111. Email: aoife.leahy@ul.ie 7 2023 15 7 2023 15 7 2023 52 7 afad11623 10 2022 18 4 2023 © The Author(s) 2023. Published by Oxford University Press on behalf of the British Geriatrics Society. All rights reserved. For permissions, please email: journals.permissions@oup.com 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Abstract Background frailty screening facilitates the stratification of older adults at most risk of adverse events for urgent assessment and subsequent intervention. We assessed the validity of the Identification of Seniors at Risk (ISAR), Clinical Frailty Scale (CFS), Programme on Research for Integrating Services for the Maintenance of Autonomy seven item questionnaire (PRISMA-7) and InterRAI-ED at predicting adverse outcomes at 30 days and 6 months amongst older adults presenting to the Emergency Department (ED). Methods a prospective cohort study of adults ≥65 years who presented to the ED was conducted. The ISAR, CFS, PRISMA-7 and InterRAI-ED were assessed. Blinded follow-up telephone interviews were completed at 30 days and 6 months to assess the incidence of mortality, ED re-attendance, hospital readmission, functional decline and nursing home admission. The sensitivity, specificity, negative predictive value and positive predictive value of the screening tools were calculated using 2 × 2 tables. Results a total of 419 patients were recruited; 47% female with a mean age of 76.9 (Standard deviation = 7.2). The prevalence of frailty varied across the tools (CFS 57% versus InterRAI-ED 70%). At 30 days, the mortality rate was 5.1%, ED re-attendance 18.1%, hospital readmission 14%, functional decline 47.6% and nursing home admission 7.1%. All tools had a high sensitivity and positive predictive value for predicting adverse outcomes. Conclusion older adults who screened positive for frailty were at significantly increased risk of experiencing an adverse outcome at 30 days with the ISAR being the most sensitive tool. We would recommend the implementation of the ISAR in the ED setting to support clinicians in identifying older adults most likely to benefit from specialised geriatric assessment and intervention. screenings instruments adverse outcomes frailty older people Health Research Board 10.13039/100010414 ILP-HSR-2017-014 ==== Body pmcKey Points High prevalence of frailty in older adults presenting to the Emergency Department. Screening tools predict adverse outcomes of mortality, Emergency Department re-attendance, nursing home admission and functional decline. Screening tools may be a method to risk stratify older adults for comprehensive geriatric assessment. Background Older adults presenting to the Emergency Department (ED) comprise of an increasingly complex cohort of patients [1]. Given their higher burden of comorbidities, it is important to provide adequate specialist support. There is an ever-growing population, it is necessary to screen those most at risk of adverse outcomes, who are most likely to benefit from Comprehensive Geriatric Assessment (CGA) that is resource intensive [2, 3]. Screening tools are frequently used in the ED setting to risk stratify patient cohorts. There is great variability in the screening tools used, particularly in relation to frailty screening. In a scoping review by Theou et al., 89 measures were used to indicate frailty, these included 13 established frailty tools and 35 non-frail tools that were used as surrogate markers [4]. The multitude of tools assessed in research makes choice of a single tool for the ED setting difficult. O’Caoimh et al. assessed the diagnostic accuracy of three such tools in the ED environment when compared to the reference standard of CGA, namely the Clinical Frailty Scale (CFS), Identification of Seniors at Risk Tool (ISAR) and the Programme on Research for Integrating Services for the Maintenance of Autonomy seven item questionnaire (PRISMA-7) [5]. This study found that all scales were accurate and reliable at identifying those who were frail. However, the study did not assess their predictive accuracy to identify adverse outcomes [5]. A systematic review by Jorgenson et al. assessed the predictive accuracy of frailty tools in the ED setting. These frailty screening tools predicted those who were at risk of hospitalisation, nursing home admission, mortality and prolonged length of stay over a 6-month period after an initial ED visit but not 30-day readmission [6]. The British Geriatric Society (BGS) recommends frailty assessment for all older adults in primary care and community settings if they are in contact with medical services [7]. Several tools are discussed in this report but no one tool is proposed for widespread use. The ‘Same-Day Acute Frailty Services’ report published by National Health Service (NHS) Improvement, NHS England, the Ambulatory Emergency Care Network and the Acute Frailty Network advocates for the use of the CFS within 30 min of an individual over 65 years presenting to acute services [8]. In the Irish setting, the National Clinical Programme for Integrated Care of Older People highlights the importance of identifying individuals at risk of frailty and completing Comprehensive Geriatric Assessment [9]. This is further expanded on in the Enhanced Community Care Implementation Guidance which mandates frailty screening at the front door ED setting [10]. Similar to the BGS, no one frailty screening tool is recommended for implementation. We aimed to assess the validity of commonly used screening tools including the ISAR, CSF, PRISMA-7 and InterRAI-ED at predicting the adverse outcomes of mortality, ED re-attendance, hospital readmission, functional decline and nursing home placement at 30 days and 6 months amongst older adults who presented to the ED at a University Teaching Hospital in Ireland. Methods Study design and setting This was a prospective cohort study. The STrengthening the Reporting of OBservational studies in Epidemiology standardised reporting guidelines were used to ensure a standardised approach to reporting our findings [11]. The study took place in the ED of a University Teaching Hospital catering for medical and surgical patients in a catchment area of 465,000 people. The protocol for this study is published elsewhere [12]. Population All adults aged ≥65 years who presented to the ED between September 2019 and April 2020 were considered eligible for enrolment if they met the following eligibility criteria: a Manchester Triage Category of 2–5 [13], resident in the catchment area and English speaking. Older adults were informed of the study by the research nurse (RN). Capacity was presumed in all patients, however, in the context of a clinical concern regarding a patient’s capacity to consent, where the patient agreed in principle to participate, the study was discussed with the patient’s next of kin. Written informed consent was obtained from all participants. Patients were not eligible for inclusion if the patients or their caregivers were unable to speak English and therefore unable to consent or provide baseline demographic information. Patients who were acutely unwell were excluded. Recruitment by the RN was 8 am to 5 pm Monday to Friday, therefore patients presenting outside these hours were excluded. Initial assessment Baseline demographic information including age, gender and existing comorbidities (Charlson Comorbidity Score) [14] was collected by the RN. Four frailty screening tools were then administered during this index visit namely the CFS, PRISMA-7, ISAR and InterRAI-ED. The CFS is a pictorial scale rated from 1 to 9 that is based on the patient’s functional status 2 weeks prior to assessment [15]. The PRISMA-7 is a seven-item questionnaire to identify those at risk of frailty [16]. The ISAR is a six-item questionnaire with yes/no answers that is validated to predict mortality, ED re-attendance, hospital readmission, functional decline and nursing home admission [17]. The InterRAI-ED is an app-based screening tool that forms part of the InterRAI Management system [18]. Functional assessment at baseline was documented using the Barthel Index [19]. Outcome assessment Outcomes were assessed by a blinded RN by telephone interview at 30 days and 6 months. Data collected included mortality, ED re-attendance, hospital readmission, subjective change in functional ability noted by the patient, nursing home admission and healthcare utilisation (GP visit, Public Health Nurse visit or health and social care professional review). Hospital management systems (PAS system and MAXIMS system) were used to determine the outcomes of mortality, ED revisit and hospital readmission for those who did not answer the follow-up phone call. The RN tried to contact patients five times at differing times of the day to maximise data capture. Statistical analysis A large sample was required to ensure sufficient observations across the adverse outcome categories. The sample size was calculated using logistic regression estimating a sample size of at least 400 was required for an analysis with six potential predictors as detailed in the study protocol [20]. Descriptive statistics were analysed, using aggregate anonymised participant data linked across baseline, 30 days and 6 months follow-up. Categorical data were described by counts and percentages. Continuous data that approximated a normal distribution were described using means and standard deviations, otherwise the median and interquartile range were presented. The unadjusted relative risk (RR) and corresponding 95% confidence interval (CI) for adverse outcomes (mortality, ED re-admission, hospital readmission, functional decline and nursing home admission) at 30 days and 6 months, were presented at pre-specified cut-offs scores for the four frailty tools. Sensitivity, specificity, negative predictive and positive predictive values and corresponding 95% CIs were calculated to determine the use of the frailty tools to predict adverse outcomes in older adults in the ED. A 5% level of significance was used for all statistical tests. All statistical analysis were undertaken using SPSS Version 24. Patient and public involvement An older adult Patient and Public Involvement (PPI) representative was consulted in relation to the aims of the study and the outcome measures most relevant to older adults. On completion of the first draft of the paper, the results of the study were discussed with the PPI representative and their perspective on meaningful outcome measures in particular were included in further drafts. They reviewed the final draft of the paper and are named as an author of the submitted manuscript. All meetings were conducted over telephone due to COVID restrictions. Results A total of 419 patients were recruited. Figure 1 describes the flow of patients in the study. Table 1 illustrates baseline patient characteristics and measures including frailty scores (ISAR, PRISMA-7, CFS and InterRAI), Barthel Index, and the Charlson co-morbidity score. Females represented 46.7% of the total population and the mean age was 76.9 years [standard deviation (SD) = 7.2 years]. The majority of patients were living rurally (74.2%), with family (65.0%) and were white Irish (95.9%). One third (34%) of patients had visited their GP prior to review in the ED and half of the patients presented by ambulance. The median Barthel score was 18. Overall, the prevalence of frailty in the cohort identified by ISAR, PRISMA-7, CFS and InterRAI-ED was 63.6%, 59%, 57% and 70%, respectively. Figure 1 PRISMA diagram. Table 1 Patient demographics and other characteristics Demographics N (%) Gender Male 223 (53.3) Female 195 (46.7) Age (years) 76.9 (7.2 SD) Location Rural 305 (74.2) Urban 106 (25.8) Ethnicity White Irish 395 (95.9) Other 17 (4.1) Marital Status Married/Partner 203 (49.4) Single 51 (12.4) Separated/Divorced 29 (7.1) Widowed 128 (31.1) Living Status Family 267 (65.0) Alone 125 (30.4) Nursing home 15 (3.6) Other 4 (1.0) Education Primary 178 (43.8) Secondary 162 (39.9) Tertiary 66 (16.3) ED entry mode Ambulance 216 (51.5) Car 191 (45.6) Other 12 (0.03) Referral Source Self/Family 208 (50.7) GP 142 (34.6) Other Healthcare 60 (14.6) Patient Measures Median (IQR) ISAR (0–6) 2 (2) Inter Rai (0–12) 3 (5) PRISMA-7 (0–7) 2 (3) CFS (0–9) 5 (2) Barthel index (0–20) 18 (7) Charlson (0–24) 2 (3) Follow-up at 30 days and 6 months The incidence of adverse outcomes at 30 days and 6 months follow-up are presented in Table 2. Table 2 Patient follow-up Outcome Follow-Up 30-days n (%) 6-months n (%) Mortality 21/415 (5.1%) 37/416 (8.9%) ED re-attendance 75/414 (18.1%) 127/387 (32.8%) Unplanned Hospital Visit 58/414 (14.0%) 108/385 (28.1%) Nursing Home admission 28/392 (7.1%) 22/379 (5.8%) Functional decline 186/391 (47.6%) 193/379 (50.9%) Note different denominators were due to data on outcomes obtained from variable sources, i.e. telephone interview and hospital databases. At 30 days, the mortality rate was 5.1%. ED re-attendance was 18.1%. A total of 14% had an unplanned hospital readmission, 47.6% of the cohort reported functional decline and 7.1% had a nursing home admission. The rates of adverse outcomes at 6 months, were as follows: mortality rate was 8.9%, 32.8% had an ED re-attendance, 28.1% had an unplanned hospital readmission, 50.9% self-reported a decline in functional ability and 5.8% had a nursing home admission. There were low levels of missing data for the outcomes of mortality (<1%) and ED re-attendance at 30 days (<1%). Due to mortality, follow-up data were not available on the other adverse outcomes for some patients (<10%). No statistical adjustment was required given the low levels of missing data. Outcome data were compiled from both telephone interviews and hospital database interrogation explaining the discrepancy in numbers of patients in each group for outcome and at different timepoints. Risk of adverse outcomes at 30 days and 6 months The risk of adverse outcomes at 30 days for each cut-off of the frailty tools is presented in Table 3. Older adults who screen positive for frailty are more likely to experience an adverse outcome. A score of ≥2 on the ISAR and ≥5 on the CFS indicates that an older person is ˃14 times more likely to be dead at 30 days in comparison to their non-frail counterparts. Table 3 Risk of adverse outcomes at 30 days and 6 months Frailty Tool & Cut-off Score Follow-up Period 30-Days 6 Months N (%) RR (95% CI) N (%) RR (95% CI) Mortality ISAR <2 0/151 (0%) 1/152 (0.7%) ≥2 21/264 (8%) NC (1.47, 395.36) 36/264 (13.6%) 20.73 (2.87, 149.67) PRISMA <3 1/170 (0.6%) 2/171 (1.2%) ≥3 20/245 (8.2%) 13.87 (1.88, 102.42) 35/245 (14.3%) 12.21 (2.98, 50.10) CFS <5 1/174 (0.6%) 1/174 (0.6%) ≥5 20/240 (8.3%) 14.5 (1.96, 107.02) 36/204 (15%) 26.25 (3.63, 189.63) Inter RAI <3 1/124 (0.8%) 1/124 (0.8%) ≥3 20/291 (6.9%) 8.52 (.16, 62.80) 36/255 (12.4%) 15.46 (2.14, 111.54) ED Re-attendance ISAR <2 18/154 (11.7%) 35/151(23.2%) ≥2 57/260(21.9%) 1.88 (1.15, 3.06) 92/236 (39%) 1.68 (1.21, 2.34) PRISMA <3 24/172(14%) 45/170 (26.5%) ≥3 51/242 (21.1%) 1.51 (0.97, 2.35) 82/217 (37.8%) 1.43 (1.05, 1.93) CFS <5 25/176 (14.2%) 42/175(24%) ≥5 50/237 (21.1%) 1.49 (0.96, 2.30) 85/212(40.1%) 1.67 (1.22, 2.28) Inter RAI <3 17/126(13.5%) 33/125(26.4%) ≥3 58/288 (20.1%) 1.49 (0.91, 2.46) 94/262 (35.9%) 1.36 (0.97, 1.90) Hospital Readmission ISAR <2 10/153 (6.5%) 27/150 (18%) ≥2 48/261 (18.4%) 2.81 (1.47, 5.40) 81/235 (34.5%) 1.91 (1.30, 2.81) PRISMA <3 13/172 (7.6%) 38/169 (22.5%) ≥3 45/242 (18.6%) 2.46 (1.37, 4.42) 70/216 (32.4%) 1.44 (1.03, 2.02) CFS <5 14/176(8%) 34/174 (19.5%) ≥5 44/237 (18.6%) 2.33 (1.32, 4.12) 74/211 (35.1%) 1.79 (1.26, 2.55) Inter RAI <3 8/126 (6.3%) 25/124 (20.2%) ≥3 50/288 (17.4%) 2.73 (1.34, 5.60) 83/261 (31.8%) 1.58 (1.07, 2.34) Nursing Home Admission ISAR <2 4/149 (2.7%) 5/148 (3.4%) ≥2 24/243 (9.9%) 3.68 (1.30, 10.39) 17/231 (7.4%) 2.18 (0.82, 5.78) PRISMA <3 4/168 (2.4%) 6/166 (3.6%) ≥3 24/224(10.7%) 4.5 (1.59, 12.72) 16/213 (7.5%) 2.08 (0.83, 5.19) CFS <5 5/173 (2.9%) 9/172 (5.2%) ≥5 23/219 (10.5%) 3.63 (1.41, 9.36) 13/207 (6.3%) 1.20 (0.53 2.74) Inter RAI <3 3/123 (2.4%) 5/123 (4.1%) ≥3 25/269 (9.3%) 3.81 (1.172, 12.38) 17/257 (6.6%) 1.61 (0.61 4.27) Functional Decline ISAR <2 58/148 (39.2%) 56/148 (37.8) ≥2 128/243 (52.7%) 1.34 (1.06, 1.70) 137/231 (59.3%) 1.57 (1.24, 1.98) PRISMA <3 64/167 (38.3%) 63/166 (38%) ≥3 122/224 (54.5%) 1.42 (1.13, 1.78) 130/213 (61%) 1.61 (1.29, 2.01) CFS <5 73/172 (42.4%) 68/172(39.5%) ≥5 113/219 (51.6%) 1.22 (0.98, 1.51) 125/207(60.4%) 1.53 (1.23, 1.89) Inter RAI <3 41/122 (33.6%) 42/122 (34.4%) ≥3 145/269 (53.9%) 1.60 (1.22, 2.11) 151/257 (58.8%) 1.71 (1.31, 2.23) Individuals who were frail according to any of the screening tools were more likely to have adverse outcomes at both time points with a similar performance amongst tools. The PRISMA-7 performed best at predicting nursing home admission at 30 days. Those who screened positive for frailty on the ISAR had the highest risk of nursing home admission at 6 months with a relative risk of 2.18 (0.82, 5.78). The risk of functional decline was greater amongst frail older adults. Sensitivity and specificity analysis Table 4 presents the sensitivity and specificity for each frailty tool for the adverse outcomes at 30 days and 6 months. All four frailty screening tools assessed had at least 95% sensitivity for mortality at both 30 days and 6 months. Specificity was much lower. In addition, all screening tools had 82% sensitivity or more for nursing home admission at 30 days. There was high negative predictive value for all tools with the ISAR having the highest negative predictive value for mortality and ED re-attendance. Table 4 Sensitivity, specificity, positive predictive value and negative predictive value analysis 3 month follow-up 6 month follow-up Sensitivity (95% CI) Specificity (95% CI) PPV NPV Sensitivity (95% CI) Specificity (95% CI) PPV NPV Mortality ISAR ≥2 PRISMA ≥3 CFS ≥5 Inter RAI ≥3 1.00 (0.84, 1.00) 0.38 (0.34, 0.43) 0.08 1.00 0.97 (0.86, 1.00) 0.40 (0.35, 0.45) 0.14 0.99 0.95 (0.76, 1.00) 0.43 (0.38, 0.48) 0.08 0.99 0.95 (0.82, 0.99) 0.45 (0.40, 0.50) 0.14 0.99 0.95 (0.76, 1.00) 0.44 (0.39, 0.49) 0.08 0.99 0.97 (0.86, 1.00) 0.46 (0.41, 0.51) 0.15 0.99 0.95 (0.76, 1.00) 0.31 (0.27, 0.36) 0.07 0.99 0.97 (0.86, 1.00) 0.33 (0.28, 0.38) 0.12 0.99 ED Re-attendance ISAR ≥2 PRISMA ≥3 CFS ≥5 Inter RAI ≥3 0.76 (0.65, 0.85) 0.40 (0.35, 0.46) 0.22 0.88 0.72 (0.64, 0.80) 0.45 (0.38, 0.51) 0.39 0.77 0.68 (0.56, 0.78) 0.44 (0.38, 0.49) 0.21 0.86 0.65 (0.56, 0.73) 0.48 (0.42, 0.54) 0.38 0.74 0.67 (0.55, 0.77) 0.45 (0.39, 0.50) 0.21 0.86 0.67 (0.58, 0.75) 0.51 (0.45, 0.57) 0.40 0.76 0.77 (0.66, 0.86) 0.32 (0.27, 0.37) 0.20 0.87 0.74 (0.65, 0.81) 0.35 (0.30, 0.42) 0.36 0.74 Hospital Readmission ISAR ≥2 PRISMA ≥3 CFS ≥5 Inter RAI ≥3 0.83 (0.71, 0.91) 0.40 (0.35, 0.45) 0.18 0.93 0.75 (0.66, 0.83) 0.44 (0.38, 0.50) 0.34 0.82 0.78 (0.65, 0.87) 0.45 (0.39, 0.50) 0.19 0.92 0.65 (0.55, 0.74) 0.47 (0.41, 0.53) 0.32 0.78 0.76 (0.63, 0.86) 0.46 (0.40, 0.51) 0.19 0.92 0.69 (0.59, 0.77) 0.51 (0.44, 0.57) 0.35 0.80 0.86 (0.75, 0.94) 0.33 (0.28, 0.38) 0.17 0.94 0.77 (0.68, 0.84) 0.36 (0.30, 0.42) 0.32 0.80 Nursing Home Admission ISAR ≥2 PRISMA ≥3 CFS ≥5 Inter RAI ≥3 0.86 (0.67, 0.96) 0.40 (0.35, 0.45) 0.10 0.97 0.77 (0.55, 0.92) 0.40 (0.35, 0.45) 0.07 0.97 0.86 (0.67, 0.96) 0.45 (0.40, 0.50) 0.11 0.98 0.73 (0.50, 0.89) 0.45 (0.40, 0.50) 0.08 0.96 0.82 (0.63, 0.94) 0.46 (0.41, 0.51) 0.11 0.97 0.59 (0.36, 0.79) 0.46 (0.40, 0.51) 0.06 0.95 0.89 (0.72, 0.98) 0.33 (0.28, 0.38) 0.09 0.98 0.77 (0.55, 0.92) 0.33 (0.28, 0.38) 0.07 0.96 Functional Decline ISAR ≥2 PRISMA ≥3 CFS ≥5 Inter RAI ≥3 0.69 (0.62, 0.75) 0.44 (0.37, 0.51) 0.53 0.61 0.71 (0.64, 0.77) 0.49 (0.42, 0.57) 0.59 0.62 0.66 (0.58, 0.72) 0.50 (0.43, 0.57) 0.54 0.62 0.67 (0.60, 0.74) 0.55 (0.48, 0.63) 0.61 0.62 0.61 (0.53, 0.68) 0.48 (0.41, 0.55) 0.52 0.58 0.65 (0.58, 0.71) 0.56 (0.48, 0.63) 0.60 0.60 0.78 (0.71, 0.84) 0.40 (0.33, 0.47) 0.54 0.66 0.78 (0.72, 0.84) 0.43 (0.36, 0.50) 0.59 0.66 Discussion In this prospective cohort study, the prevalence of frailty amongst a cohort of older adults who presented to the ED, during 8 am to 5 pm from Monday to Friday, ranged from 57% to 70%. Despite this, the median Barthel score was 18 indicating relative pre-morbid independence. In keeping with the high prevalence of frailty, one in five of the overall cohort represented to ED within 30 days and nearly one in 10 patients died at 6 months. Older adults who screened positive for frailty on any of these tools were more likely to experience all of these adverse outcomes at 30 days and 6 months. The incidence of both mortality and ED re-attendance was higher in the frail group regardless of screening tool administered. The ISAR tool identified a higher risk of mortality, ED re-attendance and hospital readmission at 30 days. All tools studied had a high sensitivity, informing healthcare providers on the low risk of adverse outcomes after discharge in a non-frail group of older adults. The sensitivities of the individual tools differed depending on the outcomes assessed. ISAR was most sensitive for mortality, and second to the InterRAI ED tool for all other outcomes. The CFS and PRISMA-7 had higher specificities for all outcomes (43–55% for PRISMA-7 and 45–56% for CFS). All of the tools had low specificity and positive predictive values for predicting adverse outcomes in frail older adults. However, as a screening tool, it is important to ensure that those who screen negatively are less likely to have an adverse outcome so the sensitivity of the tool is more important than the specificity. All tools have good sensitivity at predicting older adults who are most likely to have adverse outcomes, with the ISAR and the InterRAI ED having the best sensitivity across all outcomes. The negative predictive value was excellent with ISAR performing best at correctly identifying those who did not have a negative outcome. Older patients who present to the ED can commonly be undifferentiated in nature and it is necessary to correctly identify those who are at risk of adverse outcomes. These tools albeit with low specificity can do this with high sensitivity and high negative predictive values. The ISAR, CFS and PRISMA-7 are commonly used in the Irish ED clinical setting. A previous systematic review determining the predictive accuracy of the ISAR showed that it has moderate predictive accuracy for adverse outcomes [20]. The PRISMA-7 has been reviewed in individual cohorts and a systematic review determining its predictive accuracy is currently ongoing [21]. The CFS has been studied extensively in multiple different populations and is predictive of adverse outcomes in hospitalised patients [22], surgical patients [23] and those with COVID-19 [24, 25]. These studies showed that those who screen frail on the CFS had more adverse outcomes. Conversely, InterRAI-ED, an InterRAI management systems domain, is not in routine clinical practice in this setting and there is conflicting evidence of its predictive value. Gretarsdottir et al. reported 100% sensitivity for mortality in the Icelandic population for the Inter-RAI-ED, implementation made feasible with the use of an electronic handheld device [26]. Conversely, Michalski-Monnerat et al., demonstrated that the InterRAI-ED Screener in a Swiss cohort had poor prediction of adverse outcomes including hospital admission (28.8% sensitivity), length of hospital stay (26.3% sensitivity) and 30-day hospital readmission (26.1% sensitivity) [27]. Furthermore, the predictive value for ED re-attendance in an Australian cohort was similarly poor (AUROC 0.55, P = 0.09) [28]. The lack of digitalisation in the majority of Irish ED settings would make the InterRAI ED difficult to implement in clinical practice where handheld devices are not readily available. Furthermore, the lack of integration of IT systems across healthcare settings is also barrier to implementation. Clinical implications Within the ED setting, frailty screening has the potential to identify those at highest risk of future adverse events for whom dedicated intervention such as CGA may be considered. This can be achieved by all tools with high sensitivity. CGA has demonstrated improved outcomes in patients admitted to an acute geriatric ward setting [3]. Whilst the beneficial value of CGA has not been consistently demonstrated in the ED setting, shortly after ED discharge [29] or in community dwelling older adults [30] this may in fact primarily relate to the lack of assessment of CGA effectiveness in a high-risk frail cohort. Methodological issues may have also diluted the benefits such as inadequate implementation of study interventions particularly challenges in timely implementation extending beyond ED discharge. This study identifies a cohort of older adults who are at high risk of adverse outcomes including mortality, functional decline, nursing home admission, ED re-attendance and rehospitalisation. This provides a rational for resource allocation of effective interventions. Due to the competing interests in the ED department, it can be difficult to implement screening strategies. An upcoming qualitative evidence synthesis will explore the barriers and facilitators to screening in the ED. Barry et al. describe the perceived benefits and difficulties regarding screening in a busy ED setting [31]. It is essential to develop strategies to encourage implementation and to support staff to engage with screening programmes. Overall, frailty screening tools provide a robust method to select a high-risk cohort for assessment and intervention, with the ISAR and InterRAI ED most sensitive in this study. In particular, the ISAR tool is quick and easy to employ in the ED triage environment. Strengths and limitations This study has numerous strengths including: the prospective study design; blinded outcome assessment; objective measurement of outcomes measures such as mortality and admission from hospital databases and medical records to avoid recall bias; clinically relevant adverse events; along with robust PPI involvement informing the study from outset to manuscript preparation. There were some limitations noted: recruitment extended across the daytime working week, so this may not be representative of those patients who present outside of normal working hours; acutely unwell patients were excluded, who may be at even greater risk of future adverse outcomes; self-reported functional ability was assessed at 30 days and 6 months, which is subjective and may incur recall bias. The majority of the population lived with their family or a partner, which may have impacted their healthcare utilisation and incidence of functional decline. Furthermore, the population mainly comprised of white Irish and this may affect generalisability of the findings outside of this cohort. Of note, the diagnostic accuracy of the screening tools was not assessed as part of the study but this has recently been investigated by O’Caoimh et al. in a similar cohort [5]. Conclusion The ISAR, PRISMA-7, CFS and InterRAI-ED are frailty screening tools that identify older adults at a high risk of experiencing adverse outcomes at 30 days and 6 months. In particular, the ISAR is a brief, practical frailty screening tool which has demonstrated a high sensitivity for prediction of key adverse outcomes in older adults in the ED setting. This tool is suitable for intervention within the busy ED context and would distinguish a vulnerable cohort who may benefit from targeted resource intensive specialist geriatric assessment and intervention. Declaration of Conflicts of Interest None. Declaration of Sources of Funding Health Research Board (ILP-HSR-2017-014). The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript. Data Availability Statement Anomymised data is stored at Open Science Framework, available at doi 10.17605/OSF.IO/4NG6H. ==== Refs References 1. Thompson  D, Rumley-Buss  M, Conroy  S. Transforming emergency services for frail older people in hospital. Nurs Manag (Harrow)  2015; 22 : 18–9. 2. Ellis  G, Marshall  T, Ritchie  C. Comprehensive geriatric assessment in the emergency department. Clin Interv Aging  2014; 9 : 2033. 10.2147/CIA.S29662.25473275 3. Ellis  G, Gardner  M, Tsiachristas  A  et al.  Comprehensive geriatric assessment for older adults admitted to hospital. Cochrane Database Syst Rev  2017; 2017 : 9. 10.1002/14651858.CD006211.pub3. 4. Theou  O, Campbell  S, Malone  ML, Rockwood  K. Older adults in the emergency department with frailty. Clin Geriatr Med  2018; 34 : 369–86.30031422 5. O'Caoimh  R, Costello  M, Small  C  et al.  Comparison of frailty screening instruments in the emergency department. Int J Environ Res Public Health  2019; 16 : 3626. 10.3390/ijerph16193626. 6. Jørgensen  R, Brabrand  M. Screening of the frail patient in the emergency department: a systematic review. Eur J Intern Med  2017; 45 : 71–3.28986161 7. Turner  G, Clegg  A, Youde  J. Fit for Frailty-Consensus Best Practice Guidance for the Care of Older People Living in Community and Outpatient Settings-a Report from the British Geriatrics Society 2014. London: British Geriatrics Society, 2014. 8. Same Day Frailty Services . Available at  https://www.england.nhs.uk/wp-content/uploads/2021/02/SDEC_guide_frailty_May_2019_update.pdf 9. Making a start in Integrated Care for Older Persons . A Practical Guide to the Local Implementation of Integrated Care Programmes for Older Persons. Health Service Executive. Available at  https://www.icpop.org/_files/ugd/29601c_d8cfeac3dc5343c9945517a4995cd06b.pdf. 10. Enhanced Community Care Implementation Guidance. Health Service Executive. Available at  https://www.icpop.org/_files/ugd/29601c_9cb8a3787eef4e4ebccefbdea164353a.pdf. 11. Von Elm  E, Altman  DG, Egger  M, Pocock  SJ, Gøtzsche  PC, Vandenbroucke  JP. The Strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med  2007; 147 : 573–7.17938396 12. Leahy  A, Corey  G, O'Neill  A  et al.  Screening instruments to predict adverse outcomes for undifferentiated older adults attending the emergency department: protocol for a prospective cohort study. HRB Open Res  2021; 4 : 2. 10.12688/hrbopenres.13131.1. 13. Mackway-Jones  K , ed. Emergency Triage. London: BMJ Publishing, 1997. 14. Charlson  M, Szatrowski  TP, Peterson  J, Gold  J. Validation of a combined comorbidity index. J Clin Epidemiol  1994; 47 : 1245–51.7722560 15. Rockwood  K, Song  X, MacKnight  C  et al.  A global clinical measure of fitness and frailty in elderly people. Cmaj  2005; 173 : 489–95.16129869 16. Raîche  M, Hébert  R, Dubois  MF. PRISMA-7: a case-finding tool to identify older adults with moderate to severe disabilities. Arch Gerontol Geriatr  2008; 47 : 9–18.17723247 17. McCusker  J, Bellavance  F, Cardin  S, Trepanier  S, Verdon  J, Ardman  O. Detection of older people at increased risk of adverse health outcomes after an emergency visit: the ISAR screening tool. J Am Geriatr Soc  1999; 47 : 1229–37.10522957 18. Gray  LC, Peel  NM, Costa  AP  et al.  Profiles of older patients in the emergency department: findings from the interRAI multinational emergency department study. Ann Emerg Med  2013; 62 : 467–74.23809229 19. Mahoney  FI, Barthel  DW. Functional evaluation: the Barthel index: a simple index of independence useful in scoring improvement in the rehabilitation of the chronically ill. Md State Med J  1965; 14 : 61–5. 20. Galvin  R, Gilleit  Y, Wallace  E  et al.  Adverse outcomes in older adults attending emergency departments: a systematic review and meta-analysis of the identification of seniors at risk (ISAR) screening tool. Age Ageing  2017; 46 : 179–86.27989992 21. Higginbotham  O, O'Neill  A, Barry  L  et al.  The diagnostic and predictive accuracy of the PRISMA-7 screening tool for frailty in older adults: a systematic review protocol. HRB Open Res  2020; 3 : 26. 10.12688/hrbopenres.13042.1.34195542 22. Erhag  HF, Guðnadóttir  G, Alfredsson  J  et al.  The association between the clinical frailty scale and adverse health outcomes in older adults in acute clinical settings–a systematic review of the literature. Clin Interv Aging  2023; 18 : 249–61.36843633 23. Aucoin  SD, Hao  M, Sohi  R  et al.  Accuracy and feasibility of clinically applied frailty instruments before surgery: a systematic review and meta-analysis. Anesthesiology  2020; 133 : 78–95.32243326 24. Zhang  X-M, Jiao  J, Cao  J  et al.  Frailty as a predictor of mortality among patients with COVID-19: a systematic review and meta-analysis. BMC Geriatr  2021; 21 : 1–11.33388045 25. Zou  Y, Han  M, Wang  J, Zhao  J, Gan  H, Yang  Y. Predictive value of frailty in the mortality of hospitalized patients with COVID-19: a systematic review and meta-analysis. Ann Transl Med  2022; 10 : 166. 10.21037/atm-22-274.35280387 26. Gretarsdottir  E, Jonsdottir  AB, Sigurthorsdottir  I  et al.  Patients in need of comprehensive geriatric assessment: the utility of the InterRAI emergency department screener. Int Emerg Nurs  2021; 54 : 100943. 10.1016/j.ienj.2020.100943.33370678 27. Michalski-Monnerat  C, Carron  PN, Nguyen  S, Büla  C, Mabire  C. Assessing older Patientsʼ vulnerability in the emergency department: a study ofInterRAI EDScreener accuracy. J Am Geriatr Soc  2020; 68 : 2914–20.32964415 28. Taylor  A, Broadbent  M, Wallis  M, Marsden  E. The predictive validity of the interRAI ED screener for predicting re-presentation within 28 days for older adults at a regional hospital emergency department. Australasian Emerg Care  2019; 22 : 149–55. 29. Jay  S, Whittaker  P, Mcintosh  J, Hadden  N. Can consultant geriatrician led comprehensive geriatric assessment in the emergency department reduce hospital admission rates? A systematic review. Age Ageing  2017; 46 : 366–72.27940568 30. Briggs  R, McDonough  A, Ellis  G, Bennett  K, O'Neill  D, Robinson  D. Comprehensive geriatric assessment for community-dwelling, high-risk, frail, older people. Cochrane Database Syst Rev  2022; 2022 : CD012705. 10.1002/14651858.CD012705.pub2. 31. Barry  L, Galvin  R, Murphy Tighe  S, O'Connor  M, Ryan  D, Meskell  P. The barriers and facilitators to implementing screening in emergency departments: a qualitative evidence synthesis (QES) protocol exploring the experiences of healthcare workers. HRB Open Res  2020; 3 : 50. 10.12688/hrbopenres.13073.2.

PMC010xxxxxx/PMC10353759.txt

==== Front Age Ageing Age Ageing ageing Age and Ageing 0002-0729 1468-2834 Oxford University Press 10.1093/ageing/afad129 afad129 Research Paper AcademicSubjects/MED00280 ageing/10 The use of the World Guidelines for Falls Prevention and Management’s risk stratification algorithm in predicting falls in The Irish Longitudinal Study on Ageing (TILDA) https://orcid.org/0000-0002-1033-5897 Hartley Peter Discipline of Medical Gerontology, School of Medicine, Trinity College Dublin, Dublin, Ireland Department of Public Health and Primary Care, University of Cambridge, Cambridge, UK Department of Physiotherapy, Cambridge University Hospital NHS Foundation Trust, Cambridge, UK Forsyth Faye Department of Public Health and Primary Care, University of Cambridge, Cambridge, UK Rowbotham Scott Department of Physiotherapy, The Queen Elizabeth Hospital King’s Lynn NHS Foundation Trust, King’s Lynn, UK https://orcid.org/0000-0001-9585-2692 Briggs Robert Discipline of Medical Gerontology, School of Medicine, Trinity College Dublin, Dublin, Ireland Mercer’s Institute for Successful Ageing, St James’s Hospital, Dublin, Ireland Kenny Rose Anne Discipline of Medical Gerontology, School of Medicine, Trinity College Dublin, Dublin, Ireland Mercer’s Institute for Successful Ageing, St James’s Hospital, Dublin, Ireland Romero-Ortuno Roman Discipline of Medical Gerontology, School of Medicine, Trinity College Dublin, Dublin, Ireland Mercer’s Institute for Successful Ageing, St James’s Hospital, Dublin, Ireland Global Brain Health Institute, Trinity College Dublin, Dublin, Ireland Address correspondence to: Peter Hartley. Tel.: (+44) 1223 331841. Email: hartleyp@tcd.ie 7 2023 15 7 2023 15 7 2023 52 7 afad1298 3 2023 24 5 2023 © The Author(s) 2023. Published by Oxford University Press on behalf of the British Geriatrics Society. All rights reserved. For permissions, please email: journals.permissions@oup.com 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Abstract Background the aim of this study was to retrospectively operationalise the World Guidelines for Falls Prevention and Management (WGFPM) falls risk stratification algorithm using data from The Irish Longitudinal Study on Ageing (TILDA). We described how easy the algorithm was to operationalise in TILDA and determined its utility in predicting falls in this population. Methods participants aged ≥50 years were stratified as ‘low risk’, ‘intermediate’ or ‘high risk’ as per WGFPM stratification based on their Wave 1 TILDA assessments. Groups were compared for number of falls, number of people who experienced one or more falls and number of people who experienced an injury when falling between Wave 1 and Wave 2 (approximately 2 years). Results 5,882 participants were included in the study; 4,521, 42 and 1,309 were classified as low, intermediate and high risk, respectively, and 10 participants could not be categorised due to missing data. At Wave 2, 17.4%, 43.8% and 40.5% of low-, intermediate- and high-risk groups reported having fallen, and 7.1%, 18.8% and 18.7%, respectively, reported having sustained an injury from falling. Conclusion the implementation of the WGFPM risk assessment algorithm was feasible in TILDA and successfully differentiated those at greater risk of falling. The high number of participants classified in the low-risk group and lack of differences between the intermediate and high-risk groups may be related to the non-clinical nature of the TILDA sample, and further study in other samples is warranted. risk falls stratification longitudinal prediction older people Irish Department of Health 10.13039/100018754 The Healthcare Improvement Studies Institute 10.13039/501100022512 Health Foundation 10.13039/501100000724 Science Foundation Ireland 10.13039/501100001602 18/FRL/6188 ==== Body pmcKey Points We applied the World Guidelines for Falls’ risk stratification algorithm to data of the Irish Longitudinal Study on Ageing. The prevalence of reported falls at follow-up was significantly higher in those at ‘high risk’ than those at ‘low risk’. There may not be a clinically useful differentiation between those categorised at ‘intermediate risk’ and those at ‘high risk’. The algorithm produced a sensitivity score of 39.5% and specificity of 82.0% for predicting one or more falls at follow-up. Background The second largest cause of unintentional injury-related deaths worldwide is due to falls [1], and annually in Ireland, more than 7,000 people aged 65 years and over require hospitalisation due to a fall [2]. In England, fragility fractures alone are estimated to cost health and social care services £1.1 billion annually [3]. Several interventions exist that can reduce the risk of falling, most notably comprehensive geriatric assessment [4] and exercise interventions [5]. However, as with all healthcare, there are limited resources, and one challenge is making sure that people who would benefit from these types of interventions, receive them. Multiple guidelines aiming to reduce falls and improve post-fall management have been published over the last decade. Many have suggested methods or measures to enhance identification of those at high risk of a first fall or recurrent falls. A raft of algorithms or assessment tools exist, with the aim to help busy clinicians rapidly assess and interpret risk in order to inform the acute and long-term management of individuals at risk of falling. However, due to the complex multifactorial nature of falls, there is little consensus on the best approach [6]. The 2022 World Guidelines for Falls Prevention and Management (WGFPM) [7] include a falls risk stratification algorithm. The algorithm stratifies community-dwelling older adults into ‘low risk’, ‘intermediate risk’ and ‘high risk’ of falling. The tool is designed to be used either during opportunistic case finding such as annual health visits or when individuals present to healthcare professionals with a fall or related injury. Having robust tools at hand for all health and social care professionals will ensure they are empowered to proactively and effectively prevent and manage falls to reduce likelihood of occurrence and premature morbidity and mortality, and reduce health and social care costs. The aim of this study was to retrospectively operationalise the WGFPM falls risk stratification algorithm using data from The Irish Longitudinal Study on Ageing (TILDA). In this population, we described how easy the algorithm was to operationalise, and determined its utility in predicting falls. Methods Setting We analysed data from TILDA, a population-based longitudinal study that collects information on the health, economic and social circumstances of community-dwelling adults aged 50 years or over in Ireland. Wave 1 of the study (baseline) took place between October 2009 and July 2011, and subsequent data were collected approximately 2 yearly (Wave 2: February 2012 to March 2013). At each wave, participants completed a computer-assisted personal interview conducted by trained social interviewers in the participants’ own home, and a self-completion questionnaire, which they completed in their own time. Waves 1 and 3 also included a detailed health assessment conducted by trained research nurses at a dedicated health centre or in the participants’ own home. Individuals with cognitive impairment or dementia who could not provide written informed consent to participate in the study were excluded [8]. The full cohort profile has been described elsewhere [9]. Participants For the main analysis, we included TILDA Wave 1 participants aged 50 years or over, who completed the health assessment (at home or at a health assessment centre) and who had self-reported falls information at TILDA Wave 1 (2010). Procedure The WGFPM falls risk stratification algorithm has two points of entry: opportunistic case finding and adults presenting to healthcare with a fall or related injury. For the purpose of this study, the TILDA Wave 1 health assessment was treated as opportunistic case finding, and the entry point of ‘adults presenting to healthcare with a fall or related injury’ was not used. Table 1 shows the variables in the risk stratification algorithm and the variables that these were matched to from the TILDA Wave 1 health assessment. Table 1 Application of the WGFPM risk stratification algorithm to the TILDA dataset WGFPM risk stratification algorithm Variable used for operationalisation with the TILDA data set Assess falls in past 12 months (fall in last 12 months or positive answer to 3KQ) Has fallen in past year? Participants were asked the following:  Have you fallen in the last year? Feels unsteady when standing or walking? Participants were asked the following:  When walking, do you feel: 1. very steady, 2. slightly steady, 3. slightly unsteady, 4. very unsteadya  When standing, do you feel: 1. very steady, 2. slightly steady, 3. slightly unsteady, 4. very unsteadya Worries about falling? Participants were asked the following:  Are you afraid of falling? (yes/no) Assess fall severity (answering ‘yes’ to one criteria is sufficient to satisfy severity criteria) Fall with injuries (severe enough to consult with a physician) If participants reported having fallen, they were asked the following:  Did you injure yourself seriously enough to need medical treatment? ≥2 falls last year If participants reported having fallen, they were asked the following:  How many times have you fallen in the last year? Frailty (previously identified OR positive result on validated instrument, e.g. CFS, FP) Clinical Frailty Scale (CFS) scored using the CFS Decision Tree [10], and operationalised by TILDA [11]. A score of ≥5 considered ‘frail’ Lying on the floor/unable to get up No comparative TILDA variable Loss of consciousness/suspected syncope If participants reported having fallen, they were asked the following:  Have you ever had a blackout or fainted? Assess gait & balance Gait speed ≤0.8 m/s or Timed Up and Go (TUG) > 15 seconds TUG >15 seconds aIf answered ‘slightly unsteady’ or ‘very unsteady’ to either of the two questions, the criteria were considered satisfied. Analysis Data were analysed with R software [12]. Descriptive statistics were presented as mean with standard deviation (± SD), median with interquartile range (IQR) or count with percentage (%). All participants were stratified as low, intermediate or high risk according to the WGFPM risk stratification algorithm at Wave 1. Using data from Wave 2 (approximately 2 years after Wave 1), the groups were compared for number of falls between Wave 1 and Wave 2, number of people who experienced one or more falls between Wave 1 and Wave 2, and number of people who experienced an injury when falling between Wave 1 and Wave 2. To test for differences between groups, the Kruskal–Wallis test was used for the number of falls between Wave 1 and Wave 2. The Chi-squared test or Fisher’s exact test (if ≤10 participants in a group) were used for the categorical outcomes. All post hoc tests used Bonferroni adjustment for multiple comparisons. The sensitivity and specificity of the WGFPM falls risk stratification algorithm in predicting people with one or more falls between Wave 1 and Wave 2 were calculated using the following definitions: – True positives: participants categorised as ‘high risk’ who reported having fallen between Wave 1 and Wave 2; – True negatives: participants categorised as ‘low risk’ who reported no falls between Wave 1 and Wave 2; – False positives: participants categorised as ‘high risk’ who reported no falls between Wave 1 and Wave 2; – False negatives: participants categorised as ‘low risk’ who reported having fallen between Wave 1 and Wave 2. Sensitivity and specificity were calculated either by ignoring the ‘intermediate risk’ group, or by combining the intermediate- and high-risk groups. In addition, sensitivity analysis was conducted by limiting the analysis to those who were 65 years or older at their Wave 1 assessment. Ethics Ethical approval for each wave was obtained from the Faculty of Health Sciences Research Ethics Committee at Trinity College Dublin, Ireland (Wave 1, reference: ‘The Irish Longitudinal Study on Ageing’, date of approval: May 2008; Wave 2, reference: ‘The Irish Longitudinal Study on Ageing’, date of approval: October 2011). All participants provided written informed consent. Results TILDA Wave 1 recruited a total of 8,173 participants aged 50 years or older, of whom 5,891 completed a health assessment (5,031 in a dedicated health centre, 860 in their own home). Of these, 5,882 had self-reported falls information. Figure 1 shows the process of categorising participants as ‘low’ (n = 4,521), ‘intermediate’ (n = 42) and ‘high’ (n = 1,309) risk of falls. Due to missing data, we were unable to categorise 10 participants. A description of the participants in each category is presented in Table 2. Figure 1 Operationalisation of the World Guidelines for Falls Prevention and Management risk stratification algorithm in Wave 1 of The Irish Longitudinal Study on Ageing. * number indicates a positive response/criterion, participants can have multiple positive responses/criteria. Table 2 Sample characteristics by risk category at Wave 1 Variable All participants Low risk Intermediate risk High risk Missing P-value for difference between low-, intermediate- and high-risk groups P-value for difference between groups: L = Low I = Intermediate H = High Number of participants 5,882 4,521 (76.9%) 42 (0.7%) 1,309 (22.3%) 10 (0.2%) Mean age 63.2 (±9.3) 62.1 (±8.7) 77.3 (±8.0) 66.2 (±10.3) 65.6 (±10.4) P < 0.001 L vs. I: P < 0.001 L vs. H: P < 0.001 I vs. H: P < 0.001 Female 3,182 (54.1%) 2,360 (52.2%) 23 (54.8%) 797 (60.9%) 2 (20.0%) P < 0.001 L vs. I: P = 1 L vs. H: P < 0.001 I vs. H: P = 1 Median number of falls in past year 0.0 (0.0–0.0) 0.0 (0.0–0.0) 0.0 (0.0–0.0) 1.0 (0.0–2.0) 0.0 (0.0–0.75) P < 0.001 L vs. I: P = 0.035 L vs. H: p < 0.001 I vs. H: p < 0.001 Mean BMI 28.7 (±5.1) 28.5 (±4.9) 30.6 (±5.7) 29.2 (±5.6) 29.1 (±6.5) P < 0.001 L vs. I: P = 0.039 L vs. H: P = 0.002 I vs. H: P = 0.308 Education up to primary school level 1,540 (26.2%) 1,074 (23.8%) 23 (54.8%) 437 (33.4%) 6 (60.0%) P < 0.001 L vs. I: P < 0.001 L vs. H: P < 0.001 I vs. H: P = 0.005 Education to secondary school level 2,413 (41.0%) 1,902 (42.1%) 14 (33.3%) 493 (37.7%) 4 (40.0%) Education to tertiary/higher level 1,927 (32.8%) 1,543 (34.1%) 5 (11.9%) 379 (29.0%) 0 (0.0%) Median number of chronic conditions 2.0 (1.0–3.0) 1.0 (1.0–2.0) 2.0 (1.0–3.8) 2.0 (1.0–3.0) 2.0 (0.2–3.0) P < 0.001 L vs. I: P < 0.001 L vs. H: P < 0.001 I vs. H: P = 1 Median Clinical Frailty Scale Score 2.0 (2.0–4.0) 2.0 (2.0–3.0) 3.5 (3.0–4.0) 4.0 (2.0–6.0) 4.0 (2.0–4.0) P < 0.001 L vs. I: P < 0.001 L vs. H: P < 0.001 I vs. H: P = 0.68 MI 271 (4.6%) 189 (4.2%) 4 (9.5%) 77 (5.9%) 1 (10.0%) P = 0.009 L vs. I: P = 0.299 L vs. H: P = 0.039 I vs. H: P = 0.938 HF 58 (1.0%) 31 (0.7%) 2 (4.8%) 25 (1.9%) 0 (0.0%) P < 0.001 L vs. I: P = 0.109 L vs. H: P < 0.001 I vs. H: P = 0.611 Stroke or TIA 207 (3.5%) 102 (2.3%) 7 (16.7%) 97 (7.4%) 1 (10.0%) P < 0.001 L vs. I: P < 0.001 L vs. H: P < 0.001 I vs. H: P = 0.112 Cancer 365 (6.2%) 273 (6.0%) 3 (7.1%) 88 (6.7%) 1 (10.0%) P = 0.577 Median number of medications 2.0 (0.0–4.0) 2.0 (0.0–3.0) 4.0 (3.0–7.2) 3.0 (1.0–6.0) 0.5 (0.0–3.8) P < 0.001 L vs. I: P < 0.001 L vs. H: P < 0.001 I vs. H: P = 0.085 Anti-depressants 389 (6.6%) 216 (4.8%) 5 (11.9%) 167 (12.8%) 1 (10.0%) P < 0.001 L vs. I: P = 0.151 L vs. H: P < 0.001 I vs. H: P = 1 Anti-cholinergics 245 (4.2%) 135 (3.0%) 3 (7.1%) 107 (8.2%) 0 (0.0%) P < 0.001 L vs. I: P = 0.361 L vs. H: P < 0.001 I vs. H: P = 1 Anti-hypertensives 2,163 (36.8%) 1,542 (34.1%) 27 (64.3%) 590 (45.1%) 4 (40.0%) P < 0.001 L vs. I: P < 0.001 L vs. H: P < 0.001 I vs. H: P = 0.042 Fear of falling—not afraid 4,521 (76.9%) 3,897 (86.2%) 11 (26.2%) 608 (46.4%) 5 (50.0%) P < 0.001 L vs. I: P < 0.001 L vs. H: P < 0.001 I vs. H: P = 0.160 Fear of falling—somewhat afraid 1,040 (17.7%) 522 (11.5%) 25 (59.5%) 489 (37.4%) 4 (40.0%) Fear of falling—very afraid 318 (5.4%) 100 (2.2%) 6 (14.3%) 211 (16.1%) 1 (10.0%) Mean grip strength (kg) 27.1 (±9.8) 28.0 (±9.8) 20.0 (±6.3) 24.0 (±9.5) 28.2 (±8.9) P < 0.001 L vs. I: P < 0.001 L vs. H: P < 0.001 I vs. H: P = 0.032 Median IADL difficulties 0.0 (0.0–0.0) 0.0 (0.0–0.0) 0.0 (0.0–0.0) 0.0 (0.0–0.0) 0.0 (0.0–0.0) P < 0.001 L vs. I: P = 1 L vs. H: P < 0.001 I vs. H: P = 0.002 Mean TUG time (seconds) 9.2 (±3.8) 8.6 (±1.9) 19.3 (±7.6) 11.0 (±6.5) 8.2 (n = 1) P < 0.001 L vs. I: P < 0.001 L vs. H: P < 0.001 I vs. H: P < 0.001 Median MOCA score 25.0 (23.0–27.0) 26.0 (23.0–28.0) 20.0 (17.0–25.0) 25.0 (22.0–27.0) 22.0 (18.0–23.0) P < 0.001 L vs. I: P < 0.001 L vs. H: P < 0.001 I vs. H: P < 0.001 Data presented as mean and standard deviation (± SD), median and interquartile range (IQR), or count and percentage (%). BMI = body mass index; MI = myocardial infarction; HF = heart failure; TIA = transient ischaemic attack; IADL = instrumental activities of daily living; TUG = Timed Up and Go; MOCA = Montreal Cognitive Assessment. P-values were calculated using Kruskal–Wallis test for continuous outcomes and Chi-squared test or Fisher exact test (if ≤10 participants in a group) for categorical outcomes. All post hoc tests with Bonferroni adjustment. Table 3 presents the differences between participants in the different falls risk categories at Wave 2. All groups had attrition in participants, though this was greatest in the ‘intermediate risk’ group, in part explained by deaths. There were no differences between the intermediate- and high-risk groups in terms of number of falls reported, number of participants who had fallen or number of participants who had sustained an injury from falling. Ignoring the intermediate-risk group, the WGFPM algorithm as applied to TILDA produced a sensitivity score of 39.5% and specificity of 82.0% for predicting one or more falls between Wave 1 and Wave 2 in this cohort. Alternatively, combining participants in the intermediate- and high-risk groups the WGFPM algorithm produced a sensitivity score of 40.2% and specificity of 81.5%. Results for the sensitivity analysis restricting age to ≥65 years at wave 1 are presented in Table 4. Table 3 Differences between risk categories at Wave 2 Variable Low risk Intermediate risk High risk P-value (between groups) P-value (between specific risk groups) Low vs. intermediate Low vs. high Intermediate vs. high Number of participants at Wave 1 4,521 42 1,309 – – – – Missing at Wave 2 342 (7.6%) 10 (23.8%) 131 (10.0%) P < 0.001 P = 0.003 P = 0.017 P = 0.026 Died before Wave 2 55 (1.2%) 5 (11.9%) 50 (3.8%) P < 0.001 P < 0.001 P < 0.001 P = 0.075 Missing for unknown reason 287 (6.3%) 5 (11.9%) 81 (6.2%) P = 0.316 – – – Number of participants at Wave 2 follow-up No age restriction 4,179 32 1,178 – – – – ≥65 years at Wave 1 1,506 29 599 – – – – Median number of falls between Wave 1 and Wave 2 No age restriction 0.0 (0.0–0.0) 0.0 (0.0–1.2) 0.0 (0.0–1.0) P < 0.001 P < 0.001 P < 0.001 P = 1 ≥65 years at Wave 1 0.0 (0.0–0.0) 0.0 (0.0–1.0) 0.0 (0.0–1.0) P < 0.001 P = 0.001 P < 0.001 P = 1 Participants reporting one or more falls between Wave 1 and Wave 2 No age restriction 729 (17.4%) 14 (43.8%) 477 (40.5%) P < 0.001 P < 0.001 P < 0.001 P = 1 ≥65 years at Wave 1 302 (20.1%) 13 (44.8%) 272 (45.4%) P < 0.001 P < 0.001 P = 0.003 P = 1 Participants reporting an injury sustained from a fall between Wave 1 and Wave 2 No age restriction 295 (7.1%) 6 (18.8%) 220 (18.7%) P < 0.001 P = 0.071 P < 0.001 P = 1 ≥65 years at Wave 1 132 (8.8%) 5 (17.2%) 134 (22.4%) P < 0.001 P = 0.52 P < 0.001 P = 1 Data presented as count with percentage (%) or median and interquartile range (IQR). P-values were calculated using Kruskal–Wallis test for continuous outcomes and Chi-squared test or Fisher exact test (if ≤10 participants in a group) for categorical outcomes. All post hoc tests with Bonferroni adjustment. Table 4 Sensitivity and specificity analyses, ignoring or combining the intermediate-risk group Age True positives True negatives Sensitivity Specificity Accuracy Intermediate-risk group removed from analysis No age restriction 477 3,792 39.5% 82.0% 73.2% ≥65 years at Wave 1 272 1,204 47.4% 78.6% 70.1% Combined intermediate- and high-risk groups No age restriction 491 3,792 40.2% 81.5% 72.9% ≥65 years at Wave 1 285 1,204 48.5% 77.8% 69.8% Discussion The aim of this study was to retrospectively operationalise the WGFPM falls risk stratification algorithm using data from TILDA. An increase in incidence of falls and injuries resulting from falls was associated with WGFPM-derived intermediate- and high-risk groups, compared to low risk. The latter implies that the WGFPM risk stratification algorithm successfully identified those at greater risk of falling when using the opportunistic case finding method. However, in this analysis from TILDA, the utility of the intermediate-risk group stratification was less certain, given the small numbers the algorithm identified and that this group did not appear to be significantly different in terms of falls risk from the high-risk group. Of particular note at Wave 2 was the relatively high number of people in the low-risk group who reported having fallen since Wave 1, and a smaller but significant number of participants reported having sustained injuries falling. The WGFPM recommendation for the ‘intermediate risk’ group is for tailored exercises, education on falls risk prevention and reassessment in 1 year’s time. This is contrasted by recommendations of multifactorial risk assessment, individualised tailored interventions and reassessment in 30 to 90 days’ time for those in the high-risk group [7]. In our TILDA-based study, the ‘intermediate risk’ group was very small and not significantly different to the high-risk group in terms of the number of falls or injurious falls reported at Wave 2. From inspection of Table 2, on average the ‘intermediate risk’ participants appeared to be older, have spent less time in formal education, and were slower, weaker and had lower cognitive scores than the high-risk group, which may justify amalgamation with the high-risk group in terms of potential benefit from interventions. Given the retrospective nature of the data available, we can only make limited inferences as to why opportunistic case finding identified such a small proportion of the TILDA cohort as at intermediate risk. We postulate this may be the result of the cut-off point in the Timed Up and Go (TUG) (>15 seconds). Normative TUG data for older adults have been reported as 8.1 seconds (95%: 7.1–9.0) for those aged 60–69 years, 9.2 seconds (95% CI: 8.2–10.2) for 70–79 years and 11.3 seconds (95% CI: 10.0–12.7) for 80–99 years [13]. In the previous TILDA study by Savva et al. [14], the median TUG time for the 7.7% of Wave 1 participants who were aged ≥65 years and had physical frailty was still <15 seconds. A TUG time of more than 15 seconds in the absence of frailty, suspected syncope, or 2 or more falls, or an injury resulting from a fall in the previous year, may be expected to represent a small sample of the population. Despite its suggested benign label, the ‘low risk’ group in TILDA still had a relatively high percentage of participants reporting falls and injuries from falls at Wave 2. Indeed, approximately 60% of people reporting injuries from falls at Wave 2 came from the low-risk group. Conversely, only 40% of the high-risk group went on to have a further fall at Wave 2, which implies that the algorithm resulted in a relatively large proportion of false positives. Ignoring the ‘intermediate risk’ group (no age restriction), the WGFPM algorithm produced a sensitivity score of 39.5% and specificity of 82.0% in this TILDA cohort (assuming ‘high risk’ is interpreted as expected to fall and ‘low risk’ as not expected to fall). These figures are comparable to the sensitivity and specificity reported by Burns et al. [15], 45.3% and 83.4%, respectively, based on asking if participants had fallen in the past year (an affirmative response interpreted as being expected to fall). In our described TILDA cohort, this same ‘fallen in the past year’ question had a sensitivity of 36.7% and specificity of 85.1%, suggesting that the more extensive WGFPM algorithm did not add much accuracy internally. The ability to accurately predict risk of subsequent falls may be affected by the dichotomisation of continuous variables in the WGFPM algorithm (e.g. number of falls, frailty scales, TUG), which is likely to reduce the power of the prediction [16], although there may be benefits to simplifying the risk assessment tool to maximise ease of use. Other risk assessment tools that rely on more advanced computation such as the FRAT-UP [17] or the prediction model by Dormosh et al. [18] provide individualised predicted probabilities or risk of falls on a continuous scale, both of which have been externally validated and show reasonable discriminative power [19, 20]. Dormosh et al. [18] reported sensitivity of 62% and specificity of 70% in their model; however, the feasibility of these methods compared to the approach used by the WGFPM algorithm may be significantly lower where resources are limited. The impact of screening for falls risk is not fully understood and screening should not be necessarily assumed as ‘doing no harm’ [21, 22]. Whether being classified as high risk may cause a fear of falling, a known risk factor of falling, is not clear. Being classified as high risk may increase anxiety and has been shown to be stigmatising, as it may threaten an individual’s identity and autonomy [23–25]. However, it may also lead to lifestyle changes such as avoidance of activities perceived as high risk [26]. Similarly, for some individuals wrongly classified as low risk, this may exacerbate denial of falls risk, which is a known barrier to participation in falls prevention interventions [27]. In other fields, like cancer, where risk stratification to inform eligibility for an intervention is more widely embedded, it would seem that risk stratification prior to being offered an intervention is generally acceptable if people feel it will benefit those who need it most, improve efficiency/resource allocation and reduce false positives [28]. The present study has several limitations. The implementation of the WGFPM risk assessment algorithm was feasible in TILDA, albeit not fully (e.g. the lying on the floor/unable to get up question was not available; gait speed was not available in the full cohort). It should be noted that WGFPM falls risk stratification algorithm also recommended the alternative use of a gait speed cut-off of <0.8 m/s, which we did not assess. Being at high risk of falls and falling are not analogous, and describing people in the high-risk group who did not fall as ‘false positives’ may be disingenuous. Individuals at high risk of falling may have mitigated for their falls risk through lifestyle restrictions and behavioural modifications and may still be considered at high risk of falling. Similarly, ‘false positives’ may be explained by individuals having participated in falls prevention interventions and by Wave 2 no longer being classified as ‘high risk’. While details of the latter are not available, TILDA has observed that falling status is dynamic over time [29]. We did consider various outcomes as a measure of ‘falls risk’; however, we wished to focus on the clinical interpretation of the tool. We felt that any cut-off is relatively arbitrary, and the clinical relevance of distinguishing between different cut-offs (e.g. ‘1 or more falls’, ‘2 or more falls’ or ‘3 or more falls’) was unclear. TILDA Wave 1 household response rate was 62% [30], and as in other longitudinal studies, healthy persons may be over-represented. Our TILDA sample only included community-dwelling individuals, and although at Wave 1 participants living with dementia were excluded, recall bias is possible. The retrospective self-reported nature of falls is liable to recollection and social desirability bias. Previous research has shown that falls tend to be under-reported when relying on verbal prompts and memory, and under-reporting has been attributed to problems recalling falls, stigma attached to falling and differences in how falls are defined and perceived (e.g. what constitutes ‘a fall’) [31, 32]. However, we do not have reasons to believe that this potential bias affected participants classified as low or high risk differently. The WGFPM algorithm is likely targeted at a higher average age than that of our main analysis, which focused on participants aged 50 years or over, and causes of falls may change with age. Yet, the accuracy of the algorithm did not improve when age of participants at Wave 1 was restricted to 65 years or older (Table 4). The follow-up time of approximately 2 years differed to the recommendations of the WGFPM [7] of reassessment within a year for all participants. It is likely that this increased follow-up period exacerbated inaccuracies in recollections of falls. Finally, there were significantly more participants missing at Wave 2 from the high- and intermediate-risk groups than the low-risk group. It is possible that these may represent frailer individuals whose absence may have led to a reduced estimation of the incidence of falls between Wave 1 and Wave 2. Conclusion In TILDA, the WGFPM risk assessment algorithm successfully differentiated those at greater risk of falling when using the opportunistic case finding method. However, the utility of the intermediate-risk group was unclear and may not provide a clinically useful differentiation with the high-risk group. The high number of participants classified in the low-risk group and lack of differences between the intermediate- and high-risk groups may be related to the non-clinical nature of the TILDA sample. Further study in other samples is warranted, including in relation to the WGFPM-recommended actions and interventions. Declaration of Conflicts of Interest R.A.K. was a co-author of the 2022 publication of the World Falls Guidelines (https://doi.org/10.1093/ageing/afac205). Declaration of Sources of Funding TILDA is funded by Atlantic Philanthropies, Irish Life and the Irish Department of Health. P.H. is supported by Homerton College and the Health Foundation’s grant to the University of Cambridge for The Healthcare Improvement Studies Institute (THIS Institute). THIS Institute is supported by the Health Foundation, an independent charity committed to bringing about better health and healthcare for people in the UK. F.F. is funded by the Evelyn Trust. R.R.-O. is funded by a grant from Science Foundation Ireland under grant number 18/FRL/6188. The funders had no role in the conduct of the research or preparation of the article, study design, the collection, analysis and interpretation of data, writing of the report or the decision to submit the paper for publication. Data Availability Statement TILDA provides access to the datasets for research use through anonymised publicly accessible dataset files, and through an on-site Hot Desk Facility. The publicly accessible dataset files are hosted by the Irish Social Science Data Archive based in University College Dublin, and the Interuniversity Consortium for Political and Social Research (ICPSR) based in the University of Michigan. Researchers wishing to access the data must complete a request form, available on either the ISSDA or ICPSR website. ==== Refs References 1. Step Safely: Strategies for Preventing and Managing Falls across the Life-Course. Geneva: World Health Organization, 2021. 2. HSE, NCAOP, DOHC . Strategy to Prevent Falls and Fractures in Ireland’s Ageing Population: Report of the National Steering Group on the Prevention of Falls in Older People and the Prevention and Management of Osteoporosis throughout Life. Dublin, Ireland: Health Service Executive, 2008. Available online:  https://www.hse.ie/eng/services/publications/olderpeople/strategy-to-prevent-falls-and-fractures-in-irelands-ageing-population---full-report.pdf  (accessed 23 October 2022). 3. Great Britain . Office for Health Improvement and Disparities. Falls: Applying all our Health. London: GOV.UK, 2022. 4. Veronese  N, Custodero  C, Demurtas  J  et al.  Comprehensive geriatric assessment in older people: an umbrella review of health outcomes. Age Ageing  2022; 51 : afac104. 10.1093/ageing/afac104. 5. Sherrington  C, Fairhall  NJ, Wallbank  GK  et al.  Exercise for preventing falls in older people living in the community. Cochrane Database Syst Rev  2019; 2019 : CD012424. 10.1002/14651858.CD012424.pub2. 6. Meekes  WM, Korevaar  JC, Leemrijse  CJ, van de  Goor  IA. Practical and validated tool to assess falls risk in the primary care setting: a systematic review. BMJ Open  2021; 11 : e045431. 10.1136/bmjopen-2020-045431. 7. Montero-Odasso  M, van der  Velde  N, Martin  FC  et al.  World guidelines for falls prevention and management for older adults: a global initiative. Age Ageing  2022; 51 : afac205. 10.1093/ageing/afac205. 8. Kearney  PM, Cronin  H, O'Regan  C  et al.  Cohort Profile: The Irish Longitudinal Study on Ageing. Int J Epidemiol  2011; 40 : 877–84.21810894 9. Donoghue  OA, McGarrigle  CA, Foley  M, Fagan  A, Meaney  J, Kenny  RA. Cohort Profile Update: The Irish Longitudinal Study on Ageing (TILDA). Int J Epidemiol  2018; 47 : 1398l. 10.1093/ije/dyy163.30124849 10. Theou  O, Perez-Zepeda  MU, van der  Valk  AM, Searle  SD, Howlett  SE, Rockwood  K. A classification tree to assist with routine scoring of the clinical frailty scale. Age Ageing  2021; 50 : 1406–11.33605412 11. O'Halloran  AM, Hartley  P, Moloney  D, McGarrigle  C, Kenny  RA, Romero-Ortuno  R. Informing patterns of health and social care utilisation in Irish older people according to the clinical frailty scale. HRB Open Res  2021; 4 : 54. 10.12688/hrbopenres.13301.1.34240005 12. R Core Team . R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing, 2020. 13. Bohannon  R . Reference values for the timed up and go test: a descriptive meta-analysis. J Geriatr Phys Ther  2006; 29 : 64–8.16914068 14. Savva  GM, Donoghue  OA, Horgan  F, O'Regan  C, Cronin  H, Kenny  RA. Using timed up-and-go to identify frail members of the older population. J Gerontol A Biol Sci Med Sci  2013; 68 : 441–6.22987796 15. Burns  ER, Lee  R, Hodge  SE, Pineau  VJ, Welch  B, Zhu  M. Validation and comparison of fall screening tools for predicting future falls among older adults. Arch Gerontol Geriatr  2022; 101 : 104713. 10.1016/j.archger.2022.104713.35526339 16. Altman  DG, Royston  P. The cost of dichotomising continuous variables. BMJ  2006; 332 : 1080.16675816 17. Cattelani  L, Palumbo  P, Palmerini  L  et al.  FRAT-up, a web-based fall-risk assessment tool for elderly people living in the community. J Med Internet Res  2015; 17 : e41. 10.2196/jmir.4064.25693419 18. Dormosh  N, Schut  MC, Heymans  MW, van der  Velde  N, Abu-Hanna  A. Development and internal validation of a risk prediction model for falls among older people using primary care electronic health records. J Gerontol A Biol Sci Med Sci  2022; 77 : 1438–45.34637510 19. Palumbo  P, Klenk  J, Cattelani  L  et al.  Predictive performance of a fall risk assessment tool for community-dwelling older people (FRAT-up) in 4 European cohorts. J Am Med Dir Assoc  2016; 17 : 1106–13.27594522 20. Dormosh  N, Heymans  MW, van der  Velde  N  et al.  External validation of a prediction model for falls in older people based on electronic health records in primary care. J Am Med Dir Assoc  2022; 23 : 1691–1697.e3.35963283 21. Gray  JA, Patnick  J, Blanks  RG. Maximising benefit and minimising harm of screening. BMJ  2008; 336 : 480–3.18310003 22. World Health Organisation . WHO Guidelines on Physical Activity and Sedentary Behaviour. Geneva: World Health Organisation, 2020. 23. Anderson  H, Stocker  R, Russell  S  et al.  Identity construction in the very old: a qualitative narrative study. PloS One  2022; 17 : e0279098. 10.1371/journal.pone.0279098.36520876 24. Gardiner  S, Glogowska  M, Stoddart  C, Pendlebury  S, Lasserson  D, Jackson  D. Older people’s experiences of falling and perceived risk of falls in the community: a narrative synthesis of qualitative research. Int J Older People Nurs  2017; 12 : e12151. 10.1111/opn.12151. 25. Hanson  HM, Salmoni  AW, Doyle  PC. Broadening our understanding: approaching falls as a stigmatizing topic for older adults. Disabil Health J  2009; 2 : 36–44.21122741 26. Ellmers  TJ, Wilson  MR, Norris  M, Young  WR. Protective or harmful? A qualitative exploration of older people’s perceptions of worries about falling. Age Ageing  2022; 51 : afac067. 10.1093/ageing/afac067. 27. Yardley  L, Bishop  FL, Beyer  N  et al.  Older people’s views of falls-prevention interventions in six European countries. Gerontologist  2006; 46 : 650–60.17050756 28. Dennison  RA, Boscott  RA, Thomas  R  et al.  A community jury study exploring the public acceptability of using risk stratification to determine eligibility for cancer screening. Health Expect  2022; 25 : 1789–806.35526275 29. Hartley  P, Forsyth  F, O'Halloran  A, Kenny  RA, Romero-Ortuno  R. Eight-year longitudinal falls trajectories and associations with modifiable risk factors: evidence from the Irish Longitudinal Study on Ageing (TILDA). Age Ageing  2023; 52 . 10.1093/ageing/afad037. 30. Donoghue  O, Foley  M, Kenny  RA. Cohort Maintenance Strategies Used by the Irish Longitudinal Study on Ageing (TILDA). Dublin: The Irish Longitudinal Study on Ageing, 2017. Available from:  https://tilda.tcd.ie/publications/reports/pdf/Report_CohortMaintenance.pdf. 31. Hannan  MT, Gagnon  MM, Aneja  J  et al.  Optimizing the tracking of falls in studies of older participants: comparison of quarterly telephone recall with monthly falls calendars in the MOBILIZE Boston study. Am J Epidemiol  2010; 171 : 1031–6.20360242 32. Freiberger  E, de  Vreede  P. Falls recall—limitations of the most used inclusion criteria. Eur Rev Aging Phys Act  2011; 8 : 105–8.

PMC010xxxxxx/PMC10353760.txt

==== Front Age Ageing Age Ageing ageing Age and Ageing 0002-0729 1468-2834 Oxford University Press 10.1093/ageing/afad111 afad111 Qualitative Paper AcademicSubjects/MED00280 ageing/11 ‘Life is about movement—everything that is alive moves’: a mixed methods study to understand barriers and enablers to inpatient mobility from the older patient’s perspective Byrnes Angela Eat Walk Engage Program, Metro North Health, Herston, QLD, Australia Internal Medicine Research Unit, Royal Brisbane and Women’s Hospital, Herston, QLD, Australia Department of Nutrition and Dietetics, Royal Brisbane and Women’s Hospital, Herston, QLD, Australia McRae Prue Eat Walk Engage Program, Metro North Health, Herston, QLD, Australia Internal Medicine Research Unit, Royal Brisbane and Women’s Hospital, Herston, QLD, Australia https://orcid.org/0000-0002-0242-2165 Mudge Alison M Eat Walk Engage Program, Metro North Health, Herston, QLD, Australia Internal Medicine Research Unit, Royal Brisbane and Women’s Hospital, Herston, QLD, Australia Royal Brisbane Clinical Unit, University of Queensland Medical School, Herston, QLD, Australia Address correspondence to: Alison M. Mudge, Eat Walk Engage Program, 6th floor block 7, Royal Brisbane and Women’s Hospital, Butterfield St, Herston, 4029 QLD, Australia. Tel: 61736460854. Email: Alison.Mudge@health.qld.gov.au 7 2023 15 7 2023 15 7 2023 52 7 afad11114 12 2022 25 4 2023 © The Author(s) 2023. Published by Oxford University Press on behalf of the British Geriatrics Society. All rights reserved. For permissions, please email: journals.permissions@oup.com. 2023 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Abstract Background Mobility in hospital is important to maintain independence and prevent complications. Our multi-centre study aimed to measure mobility and identify barriers and enablers to mobility participation from the older patient’s perspective. Methods Mixed methods study including direct observation of adult inpatients on 20 acute care wards in 12 hospitals and semi-structured interviews with adults aged 65 years or older on each of these wards. Interviews were undertaken by trained staff during the inpatient stay. Quantitative data were analysed descriptively. Qualitative data were initially coded deductively using the theoretical domains framework (TDF), with an inductive approach then used to frame belief statements. Results Of 10,178 daytime observations of 503 adult inpatients only 7% of time was spent walking or standing. Two hundred older patient interviews were analysed. Most (85%) patients agreed that mobilising in hospital was very important. Twenty-three belief statements were created across the eight most common TDF domains. Older inpatients recognised mobility benefits and were self-motivated to mobilise in hospital, driven by goals of maintaining or recovering strength and health and returning home. However, they struggled with managing pain, other symptoms and new or pre-existing disability in a rushed, cluttered environment where they did not wish to trouble busy staff. Mobility equipment, meaningful walking destinations and individualised programmes and goals made mobilising easier, but patients also needed permission, encouragement and timely assistance. Conclusion Inpatient mobility was low. Older acute care inpatients frequently faced a physical and/or social environment which did not support their individual capabilities. mobility hospital qualitative research theoretical domains framework physical activity older people Metro North Clinician Researcher Fellowship Queensland Department of Health Clinical Excellence Division ==== Body pmcKey Points Inpatient mobility across 20 wards was generally low but there was substantial variation. Patients were knowledgeable about mobility benefits and mostly self-motivated to maintain mobility. Multiple barriers in the physical and social environment made hospital mobility difficult. Findings highlight the need for locally tailored, patient-informed multicomponent interventions to improve mobility. Introduction Low levels of mobility (ambulation, functional independence activities and exercise) in hospital are common [1, 2]. Low mobility is associated with new hospital-associated disability, which affects around one-third of hospitalised older adults [3, 4]. It is also implicated in other hospital-associated complications including delirium, incontinence, pressure injuries and hospital-acquired pneumonia [5]. Supporting mobility and physical activity in older acute inpatients may reduce hospital-associated disability [6–9]. However, there are many recognised patient, staff and system barriers to mobility [10–12], and implementing programmes to support mobility requires an understanding of these local barriers and a team-based approach to address them [13, 14]. Improvement initiatives often identify barriers through interviewing or surveying health care practitioners, but perspectives of patients and providers may differ [10, 11]. Most qualitative studies of patient-reported barriers have been small single site studies [11, 12, 15]. Eat Walk Engage is a structured ward-based quality improvement programme, which aims to improve mobility as well as nutrition, hydration and meaningful cognitive and social engagement to reduce hospital-associated complications in older acute inpatients [16]. Facilitated practitioner-led improvement is both informed and monitored using observational data and older person interviews, helping practitioners to question their practices by explicitly considering the older patient perspective. A cluster trial [17] demonstrated that the programme significantly reduced delirium, with promising improvements in patient and health service outcomes. Programme expansion to 20 wards in 12 hospitals was funded by the Queensland Government in 2019–20 [18]. To inform implementation of this state-wide programme, baseline data collected at each implementation site included 8 h continuous direct observation of daytime (8 AM to 4 PM) mobility and 10 semi-structured interviews with older patients about mobility importance, barriers and enablers. By pooling and analysing these baseline data across diverse sites, we aimed to provide a comprehensive picture of hospital mobility levels and patient-perceived barriers and enablers to inform improvement strategies tailored to the needs of older inpatients. Methods Study design and research paradigm This study was a mixed methods design embedded in a multi-site hospital quality improvement programme with findings intended to inform practice approaches. Our underpinning philosophical approach was thus pragmatism, employing the method of inquiry and analysis most useful to answering the research question [19]. The study included a cross-sectional observational study of inpatient mobility and a qualitative evaluation of the perceptions and experience of mobility from the perspective of the older patient. Through this mixed methods approach, we aimed to provide an objective estimate of the phenomenon of mobility and privilege the older patient explanatory perspective on why this phenomenon occurs, to inform potential solutions within the acute hospital context. The research was led by three health practitioner researchers with quantitative and qualitative expertise and diverse clinical experience who are part of the state-wide Eat Walk Engage Program team but do not provide facilitation or patient care within study sites. A.B. (dietitian) led initial coding and data summarisation; A.M.M. (physician) and P.M. (physiotherapist) provided subject matter expertise in older patient care and hospital mobility. Setting The study was undertaken in 20 adult wards in 12 publicly funded hospitals in Queensland, Australia (three metropolitan, three outer metropolitan and six regional) as they commenced the Eat Walk Engage Program. Wards included 18 general or specialty medical, one surgical and one geriatric management ward, with an average of 30 beds (range: 19–57). Each ward had funding for a part-time site facilitator (nurse or allied health professional) who supported implementation, including conducting activity mapping and interviews (17). Data were collected between August 2019 and January 2020 prior to commencing quality improvement interventions. Participants Activity mapping was conducted on one weekday in each ward and included all ward inpatients with at least 4 h of observation. Patient interviews were undertaken over a 2–3-week period with a convenience sample of ten older (≥65 years) patients on each ward. Eligible participants (those who had been on the ward for 72 h or more, were not critically ill or dying, and were able to provide verbal consent to participate in an interview) were identified by ward nursing staff. If participants preferred, family members provided support during the interview. Data collection methods and data management Data were collected by Eat Walk Engage site facilitators trained by the state-wide programme team, who have extensive experience using these measures [20–23]. Activity mapping was performed over 8 h (8 AM to 4 PM) using a validated structured observation method [20, 21, 24, 25]. Site facilitators observed all patients within each single- or multi-bedded room over a 2-min period, capturing data about the highest level of mobility using a pre-defined hierarchy (walking, actively wheeling, standing, sitting out of bed, sitting on the bed, sitting up in bed and lying in bed). Observations were repeated in the same fixed order throughout the observation period, providing systematic sampling of daytime activity. Data were entered directly by site facilitators into a customised excel spreadsheet. Semi-structured patient interviews included closed and open-ended questions to elicit patient-reported barriers, enablers and suggestions to inform improvements in team communication and care (see interview tool in Appendix 1). The interview tool was developed iteratively by the Eat Walk Engage Program team in consultation with health practitioners and older patients, and included questions about domains of mobility, nutrition and cognitive engagement [16, 23]. This study analysed responses in the mobility domain only. Interviews were conducted at the bedside and data were recorded in real time by site facilitators using hand-written notes to record patient wording verbatim, which they entered electronically using REDCap [26] (v8.3.2, ©Vanderbilt University, Nashville, United States) hosted at Metro North Hospital and Health Service. Collated response data were exported from REDCap into qualitative data indexing software NVivo v12 for Windows (©QSR International Pty Ltd, Melbourne, Australia) for coding. Data analysis Activity mapping data were summarised overall by calculating the percentage of patient observations across all sites where the patient was observed to be mobile (walking/actively wheeling/standing), sitting out in a chair, or in or on the bed. Percentage of observations where the patient was mobile was displayed for each site to illustrate site variation. We also calculated the overall percentage of patients who were observed to be mobile during any observation over the day. Closed-ended interview questions were analysed descriptively using counts and percentages. Qualitative responses to open-ended questions were analysed using a mixed deductive-inductive approach [27]. As we were interested in factors impacting whether patients performed a specific behaviour (mobilising in hospital), we selected the theoretical domains framework (TDF) as the underpinning theory for analysis. TDF is an empirical framework derived from theories of individual behaviour change and includes 14 domains [28, 29]. These domains align with the major components of the Capability-Opportunity-Motivation-Behaviour (COM-B) system, which theorises that behaviour is a function of an individual’s capability, opportunity and motivation to perform the behaviour [29]. Interview responses were initially coded line-by-line, extracting excerpts aligned with domains from the TDF to create a framework matrix. The relative frequency of excerpts for each domain was described using counts and percentages, with patients the unit of analysis. Data excerpts within domains with a frequency of responses > 10% were analysed inductively to generate proposed belief statements [30] based on recurring patterns observed within patient wording. Illustrative quotations were selected by consensus within the research group. Some concepts were discussed as both a barrier and an enabler; belief statements were framed according to whether participants more frequently identified the concept as a barrier or enabler. Techniques to enhance rigour, trustworthiness and credibility Site facilitators were supported by an audit manual and practiced activity mapping and assigning hierarchical codes with an expert facilitator prior to the full-day measurement. Researcher A.B. cleaned and analysed all activity mapping data. Five patient interviews were dual coded (A.B. and A.M.M.) using TDF domains and discussed in depth prior to A.B. coding all interviews. Four TDF domains with complete indexed data were independently reviewed by two researchers (A.B. and A.M.M.) for coding consistency and analytic memos were discussed to reach consensus on interpretation and application of TDF domains. After all data were assigned within the framework matrix, each domain was reviewed by two of three researchers to create belief statements and select illustrative quotes, with iterative discussion by the whole research team until consensus was achieved. The researchers met regularly throughout data analysis and reporting to discuss assumptions and interpretation, identify researcher implicit biases and iteratively refine data coding, summarisation and interpretation. Ethical considerations Data collection was approved as a service evaluation by the local human research ethics committee. All interview participants provided verbal consent. No individual clinical or identifiable information was recorded during activity mapping or interviews. Results There were 10,178 observations of 503 patients in the activity mapping analysis. Participants were mobile (walking, standing or actively wheeling) for 7% of observations, sitting in a chair for 27% and in or on the bed for 66% of observations. As shown in Figure 1, the percentage mobile varied from 3 to 20% across wards (the outlier geriatric ward data were influenced by three patients with cognitive impairment walking in the hallways on most observations, at times following the observer). Overall, 277/503 (55%) patients had at least one observation where they were mobile. Figure 1 Proportion of time spent mobile during daytime (between 0800 and 1600) by ward, and pooled average across all wards, from activity mapping data. Two hundred patient interviews were analysed. Most participants (n = 176, 88%) answered questions independently with 24 (12%) responses including contributions from a family member or friend. One hundred and sixty-nine participants (85%) rated the importance of mobility in hospital as ‘very important’ and 25 (13%) as ‘somewhat important’, whereas four participants (2%) said it was important, but they were currently unable to participate. Only 2 (1%) rated mobility as ‘not really important’. One hundred and thirteen interviewees (57%) recalled being advised by their health care team that mobility was important. Table 1 shows the relative frequency of each TDF domain in initial deductive coding, grouped by corresponding COM-B component. Within the Motivation component, the most frequent domains were Beliefs about consequences (93% of participants) and Beliefs about capabilities (69%). Beliefs about consequences may have been particularly represented because of the specific interview question: ‘What happens if you don’t stay mobile in hospital?’. Other domains included Intentions (33%), Emotions (22%), Goals (17%), and Social and professional role and identity (15%), whereas Optimism and Reinforcement were rarely indexed. Within the Opportunity component, both Environmental context and resources (88% of participants) and Social influences (49%) were frequently indexed. In contrast, no more than 10% of participants had responses indexed within the Capability component, including Knowledge, Behavioural regulation, Memory, attention and decision processes, and Skills. Narrative summaries of data excerpts within the eight most frequently mentioned domains are provided below. Twenty-three belief statements created from these data are listed in Table 2 with illustrative quotes and are linked in the text below by bracketed reference numbers. Table 1 TDF domain definitions and frequency of interview excerpts, grouped according to COM-B component. COM-B component TDF domain Definition> [29] Frequency of indexed data, n (%) Motivation Beliefs about consequences Acceptance of the truth, reality or validity about outcomes of a behaviour in a given situation 186 (93) Beliefs about capabilities Acceptance of the truth, reality or validity about an ability, talent or facility that a person can put to constructive use 137 (69) Intentions A conscious decision to perform a behaviour or a resolve to act in a certain way 65 (33) Emotion A complex reaction pattern, involving experiential, behavioural and physiological elements, by which the individual attempts to deal with a personally significant matter or event 43 (22) Goals Mental representations of outcomes or end states that an individual wants to achieve 34 (17) Social and professional role and identity A coherent set of behaviours and displayed personal qualities of an individual in a social or work setting 29 (15) Optimism The confidence that things will happen for the best or that desired goals will be attained 4 (2) Reinforcement Increasing the probability of a response by arranging a dependent relationship, or contingency, between the response and a given stimulus 0 (0) Opportunity Environmental context and resources Any circumstance of a person’s situation or environment that discourages or encourages the development of skills and abilities, independence, social competence and adaptive behaviour 175 (88) Social influences Those interpersonal processes that can cause individuals to change their thoughts, feelings or behaviours 97 (49) Capability Behavioural regulation Anything aimed at managing or changing objectively observed or measured actions 19 (10) Knowledge An awareness of the existence of something 19 (10) Skills An ability or proficiency acquired through practice 5 (3) Memory, attention and decision process The ability to retain information, focus selectively on aspects of the environment and choose between two or more alternatives 2 (1) Table 2 Belief statements and illustrative quotes organised by domains from the TDF. TDF domain Belief statement Illustrative quotes Beliefs about consequences 1.1 Use it or lose it You freeze up, if you don’t use it you lose it (61–8) [Walking] is good for your whole metabolism (66–7) In a larger sense, the statement: life is about movement. Everything that is alive moves. If you do not move, you are dead (67–7) 1.2. You go downhill You shrivel up and die (63–16) Sitting or lying all day might give me bedsores (60–5) 1.3. Not being mobile would bring me down If you don’t stay mobile, your muscles shrink, you feel as if there’s no life left, and this makes you lose motivation and affects your mood (61–6) You start thinking about horrible things and the days start to be too long (66–1) 1.4. You won’t get home You don’t go home, do you? (63–15) You won’t be prepared for home and all the walking you’ll need to do (63–9) Beliefs about capabilities 2.1. I have difficulties moving Not being able to see and I am also hard of hearing, so I need someone to help me. I'm not sure where the toilet is in relation to my bed—I get a bit confused (69–15) I have emphysema, so I find I don’t want to go the long distances required in the hallways to get from A to B (65–7) 2.2. Pain holds me back They are also trying to get my medications right to help my gout so I do not have so much pain and I can use my hands properly (65–5) 2.3. I’ve lost my confidence I came into hospital due to a fall so my main challenge is building back my confidence (60–7) . . .fear I will not be able to manage. They may push me into something I cannot do (62–14) Intentions 3.1. You’ve got to push yourself You’ve got to do it if you want to come good, it’s not a choice (64–11) I know I need to keep walking so it is up to me to do so (69–14) 3.2. It depends on how you feel at the time When I was feeling so unwell I didn’t have any willpower to move or any self-motivation (65–15) I don’t really want to a lot of the time (63–31) Emotion 4.1 Moving lifts my spirits Keeping up a positive attitude which makes me more keen to do things (61–3) Being able to at least shuffle around an area gives a sense of happiness and achievement (67–7) 4.2. You can feel trapped You’re like a bird in a cage (62–11) You may as well curl up and die (64–5) You just lie here and lie here and wonder what is going to happen to you (68–13) 4.3. I’m scared to get up I need to make sure I am safe—and [I] don’t feel safe on my feet currently (62–7) If there's no-one to walk with you, you feel intimidated by others in the corridor who are walking quickly, you do not want to hold them up or be in the way (63–4) Goals 5.1. I want to get home The thought of getting out of here and going home makes it easier for me to walk (65–15) And because I have stairs at home, I need to work towards that (68–7) Social and professional role and identity 6.1. I've always been an independent person Mobility gives independence. Independence is the last line of defence for an old person, their whole self-esteem relies on it as you age (67–7) Having a sense of purpose and a positive attitude, I do not allow myself to get bored. I was a nurse, so I know what I need to do (63–4) Environmental context and resources 7.1. I don’t like to bother the busy staff You’ve only got to buzz and they’re here to help (63–29) The physios come every day and help and say what to do (61–9) I’m just left in my chair, staff are too busy to help (63–3) Then the nursing staff need to help me, and they are busy, and I don’t want to take up their time (68–10) 7.2. Having the right equipment helps me Having my waking stick and wheelie walker [makes it easier for me], without them I’d be lost (63–22) IV pole limits me getting to the toilet, it's very difficult to manoeuvre (60–5) 7.3. Small spaces, clutter and bustle make it hard to move around Too many people. It is always crowded. There are no seats I can sit somewhere different or sit somewhere along my walks (65–4) I can go for a wander in hallways, they are nice open flat spaces (60–15) The corridors are so busy there are always staff running around and they do not even look like they are running around for a purpose (65–18) 7.4. I like having somewhere to go It’s nice to walk to the window and look out at the view (69–19) I did find the rose garden outside. . .gives me somewhere to go (68–5) Getting to know where everything was in the ward made it easier to get around (61–1) 7.5. I’m not allowed I'm not allowed to do what I want to do. They tell me not to walk. They make me sit down in the shower (62–1) They're stopping me from doing it. I was showering myself over the weekend, but I'm not allowed today (63–14) I'm on droplet precautions so told not to leave my bed bay (60–9) 7.6. A plan can help motivate me Interventions need to be focused and tailored as you progress throughout your stay, staff need to provide goals and outline expectations, could provide simple exercises to do in spare time (61–6) the physios have provided exercises for me to do (61–2) Social influences 8.1. It helps when staff encourage me The nurses always encourage me too, helps me stay motivated (69–18) When nursing staff are cheerful and proactive it encourages you to do your best (63–8) 8.2. I need permission Make permission to move about more clear (66–8) 8.3. A walking buddy may help Group programs for motivation (61–7) I was able to befriend another patient and walk with them (64–12) Beliefs about consequences Most participants perceived that mobility was important for maintaining and/or recovering their strength, flexibility, mobility, independence and vitality, and that failure to mobilise would result in corresponding losses (Table 2, statements 1.1, 1.2). Many participants used emotive metaphors for the poor outcomes they perceived resulting from poor mobility, e.g. ‘become a vegetable’, ‘shrivel up and die’ or ‘lose the game altogether’, whereas some articulated specific in-hospital complications including falls, pressure injuries, pneumonia, constipation and thrombosis. Some participants linked mobility to mental well-being and identified that poor mobility would lead to boredom, frustration and anxiety (1.3). Many identified a link between their level of mobility and discharge decisions and how well they would cope following discharge (1.4). Beliefs about capabilities Many participants described symptoms such as fatigue, breathlessness, drowsiness and weakness that temporarily reduced their usual capacity to mobilise, or described existing disabilities that made mobilising a challenge, further exacerbated by new symptoms and an unfamiliar environment (2.1). Some participants were aware that they sought reassurance and assistance that they did not usually require, whereas others needed assistance for a long-term disability. Clinical status often changed during their illness trajectory, so that choosing the appropriate timing for mobilisation was important. Many participants specifically mentioned acute or long-standing pain as a major limitation to their capability to mobilise (2.2). Some feared that mobilising would lead to further harm, although a few felt that healthcare staff were too risk averse and imposed unnecessary restrictions on their mobility (2.3). Intentions Many participants described mobility as a personal responsibility and described resolve and self-motivation as central (3.1). However, some identified that maintaining motivation was difficult in the face of illness symptoms (3.2), whereas a few described poor self-motivation and needed encouragement from staff or family. Emotions Some participants described mutually reinforcing relationships between mobility, mood and motivation. Staying mobile was important to maintain a positive outlook, which, in turn, helped to motivate them to work towards their recovery (4.1). Some described that immobility conveyed a sense of confinement leading to frustration, boredom, isolation or depression, and a few identified that immobility made them ruminate and worry (4.2). On the other hand, a few participants were anxious about mobilising, particularly falls or worsening symptoms, or were worried about managing unfamiliar equipment or environment (4.3). Goals Goals identified by participants usually related to getting home (5.1). This included managing their own independence or resuming caring or personal responsibilities. Social and professional role and identity Some participants identified remaining mobile as an important aspect of maintaining their sense of self, providing an important sense of purpose and supporting continuation of responsibilities and enjoyable activities (6.1). Environmental context and resources Most participants expressed gratitude for assistance and encouragement from nursing staff and physiotherapists, but many identified that timely assistance was sometimes compromised by competing staff priorities (7.1). Although many participants appeared stoic and understanding when help was not available, a few described feeling neglected and frustrated. Availability of suitable clinical equipment and mobility aids was an important enabler for mobility and independence, and conversely lack of equipment was recognised as an important barrier by some participants (7.2). The physical ward space could be challenging (e.g. having enough room in bed bays to manoeuvre a mobility device). Corridors often provided the main opportunity for mobilising, but these could be cluttered and busy, without adequate seating to allow for rest (7.3). Participants valued somewhere attractive or meaningful to walk to, such as an outdoor area or café (7.4); however, the presence of such locations and permission to go there were often not communicated to patients. Some participants described restrictions imposed on their mobility by staff, e.g. being confined to their room because of infection control, or restrictions because of concerns about medical stability (e.g. low blood pressure) or falls (7.5). Some participants described the value of instructions and graded plans, such as an exercise programme, suited to their stage of recovery, as well as opportunities for supported activities, such as walking groups or time in the gym (7.6). Social influences Many patients reported that encouragement, reminding and prompting from staff influenced mobilising (8.1). The quality of the interaction with nurses was particularly important. Some patients reported that friendly positive interactions with staff helped to motivate them, whereas other participants described feeling disempowered and uncertain about permission to mobilise, and some appeared to take a passive role to avoid burdening staff who they perceived to be too busy. Clear permission was often reported as an enabler (8.2). Social interaction with other patients on the ward could also encourage and motivate participants to mobilise, with some suggesting structured activities such as walking groups or walking buddies (8.3). Summary Findings are summarised in Figure 2, which demonstrates the wide range of overlapping barriers and enablers and proposes practical strategies to address them. It highlights the need to concurrently address multiple barriers, ideally tailored to local context. For example, creating a walking destination is unlikely to improve mobility if someone has poorly managed pain, does not have permission to leave their bed space or does not feel they can bother busy staff to ask for the assistance or equipment that they require. Figure 2 Belief statements (middle two concentric circles) derived from qualitative interviews aligned with proposed practical strategies for addressing these barriers and enablers. Table 2 aligns these belief statements within the COM-B components and TDF domains and provides illustrative quotes. Discussion Activity mapping revealed limited mobility across acute wards, with two-thirds of daytime patient observations on or in bed and 7% upright and mobile. Only 55% of patients were mobile on at least one observation. These findings confirm previous activity mapping studies reporting low daytime mobility. Only 4% of daytime was spent standing or walking on an Australian general medical ward [24], whereas time spent mobile on medical, vascular surgery and medical oncology wards in another Australian hospital ranged from 6.5 to 11.4% [20]. In the Netherlands, time spent mobile on four specialty medical and surgical wards (cardiology, cardiothoracic surgery, medical oncology and haematology) ranged from 6.6 to 10.2% in one hospital [31], and 5 to 8% on cardiology, cardiothoracic and orthopaedics wards in another [32]. A unique feature of our current study is the large number of hospitals, illustrating substantial variation despite most wards being acute general medical wards with similar populations. Our large interview study also contributes substantial original data to the literature. Two recent systematic reviews have summarised barriers and enablers to mobility reported by adult patients and health providers on acute care wards [11] or medical wards [15]; <400 patient interviews in total were reported in each of these reviews. Our interview findings specifically elucidated older patients’ perceptions of factors that contributed to low mobility (Figure 2). Participants were strongly aware of the importance of mobilising and remaining active in hospital and did not identify mobility knowledge and skills (COM-B Capability component) as barriers to their mobility. They generally had high levels of intrinsic motivation (COM-B Motivation component), driven by goals of maintaining or recovering their strength, health, mood and wellness and returning home. However, they often struggled to mobilise because of pain, other symptoms and new or pre-existing disability, which was harder to manage because of reluctance to burden visibly busy staff. Concerns about safety related to loss of confidence in their abilities also limited some patients’ motivation. Patient mobility was strongly influenced by positive and negative influences within the physical and social environment (COM-B Opportunity component). Availability of mobility equipment, meaningful walking destinations and individualised programmes and goals made mobilising easier, but needed to be supported by clear permission, friendly encouragement and timely assistance to overcome the difficulties imposed by acute and chronic illness symptoms. Uncrowded spaces with rest stops could improve confidence. Our finding that patients perceive mobility to be of central importance to health and recovery is consistent with Pavon et al. [33] who interviewed patients and multidisciplinary healthcare providers. A recent scoping review by Geelen et al. [11] used TDF to synthesise findings from 33 studies of patient-reported barriers and enablers to mobility on acute care wards. Like us, they found few barriers related to domains within the COM-B Capability component and extensive barriers and enablers within the COM-B Opportunity component related to domains of Environmental context and resources (e.g. time, staffing, equipment and culture and routines), and Social influences (e.g. encouragement and assistance from staff, volunteers or other patients). They also found that Goals, Beliefs about consequences and Knowledge domains within the COM-B Motivation component were important enablers. These common findings suggest that additional interventions to educate and motivate inpatients may have limited impact and that attention would be better directed towards modifying the physical and social environment to harness patients’ existing beliefs and motivation. This might include redesign to make spaces more welcoming and improve way-finding [34] and better permission giving, encouragement, reassurance and assistance that recognises patients’ individual and changing symptoms and abilities. This is also supported by findings from Mani et al.’s [15] systematic review on medical wards, which highlighted the importance of symptoms and emotions as barriers, including anxiety about falling or getting lost and not wanting to be a bother to staff. Recent studies underline the interaction between the physical and cultural environment [12, 35, 36], whereby an enabling environment and resources still require staff permission and encouragement to enhance mobility; only 57% of our interview participants recalled direct mobility encouragement from their teams. A major challenge is how to provide this person-centred support in a context where staff are visibly rushed and busy [37], and patients fear that bothering staff will compromise the quality of their care [34], leading them to assume ‘passivity’ that can be misinterpreted as poor motivation [12, 34]. Studies of health practitioners’ perceptions of barriers to inpatient mobility also highlight the central importance of time and competing tasks as barriers, in the context of complex patient needs and a lack of clear roles and responsibilities [12, 15, 38]. These constraints might be addressed by clarifying roles, improving teamwork, reducing competing tasks and/or adding more staff, assistant or volunteer (including family carer) time [39, 40]. These approaches require time, skill and diplomacy to establish trust and question established staff roles and practices, mindful of local culture and resources. Adding more staff or assistants without careful delineation of roles could increase ward busyness and risk duplication, whereas engaging and retaining volunteers is often more complex than anticipated [41, 42]. Strengths of our study include the large sample of hospitalised older adults across multiple hospitals, and use of a theoretical framework aligned with our pragmatic paradigm supporting translation of research findings into potential clinical strategies. We recognise a potential mismatch in that activity mapping was performed for all adults, whereas interviews focused on older adults. However, studies have previously shown that activity levels are similar for older and younger inpatients [20, 43] and concentrating on older adults for interview data is appropriate given the greater potential immobility harm in older inpatients and the focus of our overarching improvement programme. We used activity mapping rather than accelerometry for measuring mobility, although these methods have shown high correlation for estimating time spent upright at the ward level, with activity mapping slightly overestimating upright time [21, 25, 44]. Patient interviews were conducted by multiple facilitators and not audio recorded, which may create bias based on facilitators’ clinical discipline or past experiences. This was mitigated by consistent facilitator training, mentoring and an interview guide. Interviews were relatively brief and structured, which may limit depth, but responses nonetheless yielded rich data across multiple domains. We did not collect details about patient characteristics (e.g. age and frailty), which may limit generalisability, although this is partly mitigated by a large sample across medical and surgical wards in metropolitan and regional hospitals. However, we recognise that our methods under-sampled patients with cognitive impairment, more severe illness and non-English speaking backgrounds who may have identified additional barriers. In conclusion, low mobility is widespread in acute care wards although there was some variation between wards. Older patients perceive complex overlapping barriers and enablers to mobilising. Our findings suggest that simple interventions to inform or educate patients or provide standardised mobility objectives are unlikely to be sufficient to address this complex problem. Interventions must address the physical and social environment of acute care and acknowledge and resolve tensions between workload and person-centred care. Further research may elucidate whether barriers are similar or different in younger patients and in older patient subgroups who are often under-sampled such as those with cognitive impairment. Continuing research to improve inpatient mobility should clearly describe how interventions are informed by the patient perspective and describe implementation in sufficient detail to understand what elements are successfully implemented and how tensions are resolved in practice. Supplementary Material aa-22-2141-File002_afad111 Click here for additional data file. Acknowledgements Preliminary findings were presented in poster format at the 2022 International Association of Geriatrics and Gerontology. Declaration of Conflicts of Interest None. Declaration of Sources of Funding Professor A.M.M. was funded by a Metro North Clinician Researcher Fellowship. The Eat Walk Engage state-wide programme is funded by the Queensland Department of Health Clinical Excellence Division. Data Availability Data are not publicly available but reasonable requests will be considered by the corresponding author. ==== Refs References 1. Brown  C, Redden  D, Flood  K, Allman  R. The underrecognized epidemic of low mobility during hospitalization of older adults. J Am Geriatr Soc  2009; 57 : 1660–5.19682121 2. Baldwin  C, van  Kessel  G, Phillips  A, Johnston  K. Accelerometry shows inpatients with acute medical or surgical conditions spend little time upright and are highly sedentary: systematic review. Phys Ther  2017; 97 : 1044–65.29077906 3. Loyd  C, Markland  AD, Zhang  Y  et al.  Prevalence of hospital-associated disability in older adults: a meta-analysis. J Am Med Dir Assoc  2020; 21 : 455–61.e5.31734122 4. Covinsky  K, Pierlussi  E, Johnston  C. Hospitalization-associated disability. ‘She was probably able to ambulate, but I'm not sure’. JAMA  2011; 306 : 1782–93.22028354 5. Kalisch  BJ, Lee  S, Dabney  BW. Outcomes of inpatient mobilization: a literature review. J Clin Nurs  2014; 23 : 1486–501.24028657 6. Valenzuela  PL, Morales  JS, Castillo-Garcia  A  et al.  Effects of exercise interventions on the functional status of acutely hospitalized older adults: a systematic review and meta-analysis. Ageing Res Rev  2020; 61 : 101076. 10.1016/j.arr.2020.101076.32330558 7. Cohen  Y, Zisberg  A, Chayat  Y  et al.  Walking for better outcomes and recovery: the effect of WALK-FOR in preventing hospital-associated functional decline among older adults. J Gerontol Series A Biol Sci Med Sci  2019; 74 : 1664–70. 8. Reynolds  CD, Brazier  KV, Burgess  EAA  et al.  Effects of unstructured mobility programs in older hospitalized general medicine patients: a systematic review and meta-analysis. J Am Med Dir Assoc  2021; 22 : 2063–73.e6.33434569 9. Smith  TO, Sreekanta  A, Walkeden  S, Penhale  B, Hanson  S. Interventions for reducing hospital-associated deconditioning: a systematic review and meta-analysis. Arch Gerontol Geriatr  2020; 90 : 104176. 10.1016/j.archger.2020.104176.32652367 10. Brown  C, Williams  B, Woodby  L, Davis  L, Allman  R. Barriers to mobility during hospitalization from the perspectives of older patients and their nurses and physicians. J Hosp Med  2007; 2 : 305–13.17935241 11. Geelen  SJG, van  Dijk-Huisman  HC, de  Bie  RA  et al.  Barriers and enablers to physical activity during hospital stay: a scoping review. Syst Rev  2021; 10 : 293. 10.1186/s13643-021-01843-x.34736531 12. Stutzbach  J, Jones  J, Taber  A, Recicar  J, Burke  RE, Stevens-Lapsley  J. Systems approach is needed for in-hospital mobility: a qualitative metasynthesis of patient and clinician perspectives. Arch Phys Med Rehabil  2021; 102 : 984–98.32966808 13. Zisberg  A, Agmon  M, Gur-Yaish  N  et al.  No one size fits all-the development of a theory-driven intervention to increase in-hospital mobility: the ‘WALK-FOR’ study. BMC Geriatr  2018; 18 : 91. 10.1186/s12877-018-0778-3.29653507 14. Moore  JE, Mascarenhas  A, Marquez  C  et al.  Mapping barriers and intervention activities to behaviour change theory for mobilization of vulnerable elders in Ontario (MOVE ON), a multi-site implementation intervention in acute care hospitals. Implement Sci  2014; 9 : 160. 10.1186/s13012-014-0160-6.25928538 15. Mani  H, Mori  C, Mattmann  M  et al.  Barriers and facilitators to mobility of patients hospitalised on an acute medical ward: a systematic review. Age Ageing  2022; 51 : afac159. 10.1093/ageing/afac159. 16. Mudge  A, McRae  P, Cruickshank  M. Eat walk engage: an interdisciplinary collaborative model to improve care of hospitalized elders. Am J Med Qual  2015; 30 : 5–13.24270172 17. Mudge  A, McRae  P, Banks  M  et al.  Effect of a ward-based program on hospital-associated complications and length of stay for older inpatients. The cluster randomized CHERISH trial. JAMA Intern Med  2022; 182 : 274–82.35006265 18. Whiting  E, Scott  I, Hines  L  et al.  A whole of health system approach to improving care of frail older persons. Aust Health Rev  2022; 46 : 629–34. 19. Creswell  JW, Plano Clark  VL. Designing and Planning Mixed Methods Research. 2nd edition. Thousand Oaks: Sage, 2011. 20. Mudge  AM, McRae  P, McHugh  K  et al.  Poor mobility in hospitalized adults of all ages. J Hosp Med  2016; 11 : 289–91.26797978 21. Wu  Y, Smits  EJ, Window  P, Beningfield  A, Johnstone  V, McRae  P. Mobility levels of acute medical patients: is behavioural mapping comparable to accelerometry?  Clin Rehabil  2021; 35 : 595–605.33203223 22. Valkenet  K, McRae  P, Reijneveld  E  et al.  Inpatient physical activity across a large university city hospital: a behavioral mapping study. Physiother Theory Pract  2022. 10.1080/09593985.2022.2112116. 23. Lee-Steere  K, Liddle  J, Mudge  A, Bennett  S, McRae  P, Barrimore  SE. ‘You’ve got to keep moving, keep going’: understanding older patients’ experiences and perceptions of delirium and nonpharmacological delirium prevention strategies in the acute hospital setting. J Clin Nurs  2020; 29 : 2363–77.32220101 24. Kuys  S, Dolecka  U, Guard  A. Activity level of hospital medical inpatients: an observational study. Arch Gerontol Geriatr  2012; 55 : 417–21.22417401 25. Valkenet  K, Bor  P, van  Delft  L, Veenhof  C. Measuring physical activity levels in hospitalized patients: a comparison between behavioural mapping and data from an accelerometer. Clin Rehabil  2019; 33 : 1233–40.30864490 26. Harris  PA, Taylor  R, Minor  BL  et al.  The REDCap consortium: building an international community for software partners. J Biomed Inform  2019; 95 : 103208. 10.1016/j.jbi.2019.103208.31078660 27. Elo  S, Kyngas  H. The qualitative content analysis process. J Adv Nurs  2008; 62 : 107–15.18352969 28. Cane  J, O’Connor  D, Michie  S. Validation of the theoretical domains framework for use in behaviour change and implementation research. Implement Sci  2012; 7 : 37. 10.1186/1748-5908-7-37.22530986 29. Atkins  L, Francis  J, Islam  R  et al.  A guide to using the theoretical domains framework of behaviour change to investigate implementation problems. Implement Sci  2017; 12 : 77. 10.1186/s13012-017-0605-9.28637486 30. Cheung  WJ, Patey  AM, Frank  JR, Mackay  M, Boet  S. Barriers and enablers to direct observation of trainees’ clinical performance: a qualitative study using the theoretical domains framework. Acad Med  2019; 94 : 101–14.30095454 31. van  Delft  L, Bor  P, Valkenet  K, Slooter  AJC, Veenhof  C. The effectiveness of Hospital in Motion: a multidimensional implementation project to improve patients’ movement behaviour during hospitalization. Phys Ther  2020; 100 : 2090–8.32915985 32. Koenders  N, Potkamp-Kloppers  S, Guerts  Y, Akkermans  R, Nijhuis-van der Sanden  M, Hoogeboom  T. Ban Bedcentricity: a multifaceted innovation to reduce sedentary behaviour of patients during the hospital stay. Phys Ther  2021; 101 : pzab054. 10.1093/ptj/pzab054.33564890 33. Pavon  JM, Fish  LJ, Colon-Emeric  CS  et al.  Towards ‘mobility is medicine’: socioecological factors and hospital mobility in older adults. J Am Geriatr Soc  2021; 69 : 1846–55.33755991 34. King  B, Bodden  J, Steege  L, Brown  CJ. Older adults experiences with ambulation during a hospital stay: a qualitative study. Geriatr Nurs  2021; 42 : 225–32.32861430 35. Koenders  N, van  Oorsouw  R, Seeger  J, Nijhuis-van der Sanden  M, van de  Glind  I, Hoogeboom  T. ‘I’m not going to walk, just for the sake of walking…’: a qualitative, phenomenological study on physical activity during hospital stay. Disabil Rehabil  2020; 42 : 78–85.30092714 36. Stefansdottir  N, Pedersen  MM, Tjornhoj-Thomsen  T, Kirk  JW. Older medical patients’ experiences with mobility during hospitalization and the WALK-Copenhagen (WALK-Cph) intervention: a qualitative study in Denmark. Geriatr Nurs  2021; 42 : 46–56.33242706 37. Grealish  L, Chaboyer  W. Older, in hospital and confused—the value of nursing care in preventing falls in older people with cognitive impairment. Int J Nurs Stud  2015; 52 : 1285–7.25726429 38. Geelen  SJG, Giele  B, Engelbert  RHH  et al.  Barriers to and solutions for improving physical activity in adults during hospital stay: a mixed-methods study among healthcare professionals. Disabil Rehabil  2021; 44 : 4004–13.33605171 39. Baczynska  A, Lim  S, Sayer  A, Roberts  H. The use of volunteers to help older medical patients mobilise in hospital: a systematic review. J Clin Nurs  2016; 25 : 3102–12.27477624 40. Snowdon  DA, Storr  B, Davis  A, Taylor  N, Williams  C. The effect of delegation of therapy to allied health assistants on patient and organisational outcomes: a systematic review and meta-analysis. BMC Health Serv Res  2020; 20 : 491. 10.1186/s12913-020-05312-4.32493386 41. Lim  S, Ibrahim  K, Dodds  R  et al.  Physical activity in hospitalised older people: the feasibility and acceptability of a volunteer-led interventions in the SoMoVe study. Age Ageing  2020; 49 : 283–91.31566671 42. Godfrey  M, Green  J, Smith  J  et al.  Process of implementing and delivering the prevention of delirium system of care: a mixed methods preliminary study. BMC Geriatr  2020; 20 : 1. 10.1186/s12877-019-1374-x. 43. Meesters  J, Conjin  D, Vermeulen  HM, Vliet Vielend  TPM. Physical activity during hospitalization: activities and preferences of adults versus older adults. Physiother Theory Pract  2019; 35 : 975–85.29658797 44. Kramer  S, Cumming  T, Churilov  L, Bernhardt  J. Measuring activity levels at an acute stroke ward: comparing observations to a device. Biomed Res Int  2013; 2013 : 1–8.

PMC010xxxxxx/PMC10353840.txt

==== Front Nucleic Acids Res Nucleic Acids Res nar Nucleic Acids Research 0305-1048 1362-4962 Oxford University Press 35609995 10.1093/nar/gkac389 gkac389 AcademicSubjects/SCI00010 Web Server Issue PrankWeb 3: accelerated ligand-binding site predictions for experimental and modelled protein structures Jakubec David Department of Software Engineering, Faculty of Mathematics and Physics, Charles University, Czech Republic Skoda Petr Department of Software Engineering, Faculty of Mathematics and Physics, Charles University, Czech Republic Krivak Radoslav Department of Software Engineering, Faculty of Mathematics and Physics, Charles University, Czech Republic Novotny Marian Department of Cell Biology, Faculty of Science, Charles University, Czech Republic https://orcid.org/0000-0003-4679-0557 Hoksza David Department of Software Engineering, Faculty of Mathematics and Physics, Charles University, Czech Republic To whom correspondence should be addressed. Tel: +420 951 554 227; Email: david.hoksza@matfyz.cuni.cz The authors wish it to be known that, in their opinion, the first two authors should be regarded as Joint First Authors. 05 7 2022 24 5 2022 24 5 2022 50 W1 W593W597 06 5 2022 15 4 2022 25 3 2022 © The Author(s) 2022. Published by Oxford University Press on behalf of Nucleic Acids Research. 2022 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (https://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Abstract Knowledge of protein–ligand binding sites (LBSs) enables research ranging from protein function annotation to structure-based drug design. To this end, we have previously developed a stand-alone tool, P2Rank, and the web server PrankWeb (https://prankweb.cz/) for fast and accurate LBS prediction. Here, we present significant enhancements to PrankWeb. First, a new, more accurate evolutionary conservation estimation pipeline based on the UniRef50 sequence database and the HMMER3 package is introduced. Second, PrankWeb now allows users to enter UniProt ID to carry out LBS predictions in situations where no experimental structure is available by utilizing the AlphaFold model database. Additionally, a range of minor improvements has been implemented. These include the ability to deploy PrankWeb and P2Rank as Docker containers, support for the mmCIF file format, improved public REST API access, or the ability to batch download the LBS predictions for the whole PDB archive and parts of the AlphaFold database. Graphical Abstract Graphical Abstract PrankWeb accepts a protein structure on its input, computes evolutionary conservation, and predicts binding sites which are then mapped onto the structure and can be visually inspected. ELIXIR CZ Research Infrastructure LM2018131 MEYS CR ==== Body pmcINTRODUCTION Interactions of proteins with other molecules drive biological processes at the molecular level. One specific class of such interactions are protein–small molecule (ligand) interactions; identifying the sites and roles of these interactions is crucial for the elucidation of the molecular mechanisms of enzymes, regulation of protein oligomerization, or designing new drugs (e.g., in case drug resistance has occurred) (1,2). In these applications, precise knowledge of the protein’s ligand-binding sites (LBSs) is required. As experimental identification of LBSs is time-consuming and expensive, computational methods have been developed to facilitate LBS identification from the protein three-dimensional (3D) structure. These methods can be broadly categorized as geometric, energetic, evolution-based, and knowledge- or machine learning (ML)-based. Many of the existing methods combine the aforementioned approaches, which is also the case of the P2Rank method (3) developed in our group. P2Rank assigns structural, physico-chemical, and evolutionary features to points on a mesh covering the protein surface and builds an ML model over this representation. The model is used to detect ligandable points, which are then clustered to obtain a list of surface patches corresponding to the predicted LBSs. The approach has achieved state-of-the-art performance and is still on par or outperforming newer deep learning methods (4). The lack of broadly accessible online resources has historically hindered access to the LBS prediction methods. To this end, we have developed PrankWeb (5), an online tool encapsulating the P2Rank approach. PrankWeb has allowed its users to enter a 3D structure as a Protein Data Bank (6) (PDB) file or using a PDB identifier, carried out evolutionary conservation analysis, predicted the LBSs using P2Rank, and enabled visual examination of the results. This paper introduces PrankWeb 3, an improved version of the resource. A limiting aspect of the structure-based LBS prediction approaches is the necessity of having the protein 3D structure determined. Although the number of resolved protein structures keeps increasing, it is still far behind the number of known protein sequences (7). However, recent advances in protein structure prediction, namely the introduction of the AlphaFold 2 method (8) and the AlphaFold Protein Structure Database (AlphaFold DB) (9), have opened the door for the application of structure-based approaches also toward proteins for which only the sequence is known. This development has motivated one of the major improvements in PrankWeb 3: the adoption of the AlphaFold DB, allowing PrankWeb users to enter a UniProt accession number as the input. This change significantly increases the number of proteins to which PrankWeb is applicable (section Predicted structures). Another significant improvement is the replacement of the former evolutionary conservation estimation pipeline with a faster, more consistent version (section Evolutionary conservation calculation pipeline). The last major change has been the refactoring of the PrankWeb application resulting in a modular architecture with strictly separated components. Such architecture enables easy utilization of the application or its parts (such as the conservation calculation pipeline) to advanced users via Docker containers (section Other improvements). A detailed description of the changes follows. EVOLUTIONARY CONSERVATION CALCULATION PIPELINE Evolutionary conservation (EC) has been identified as a powerful indicator of functionally significant regions of protein structures; for this reason, it has been utilized as an optional feature capable of improving the default P2Rank predictions. Previous versions of PrankWeb utilized a series of sequence databases to construct a multiple sequence alignment (MSA) of sequences similar to the given query, and subsequently quantified the EC of its individual columns using Jensen–Shannon divergence (10). This approach possessed two major drawbacks. First, the use of fallback sequence databases for the construction of an MSA of sufficient size resulted in discontinuities in the conservation scores as the number of sequences in the MSA exceeded the threshold. A single P2Rank model was thus unable to account for the different sequence distributions (and, therefore, conservation scores) intrinsic to the individual sequence databases. Second, and more importantly, the previous EC calculation pipeline could take several hours to complete, severely impacting the user’s experience with PrankWeb. Starting with PrankWeb 3, the former EC calculation pipeline has been replaced with a simpler, faster, and more consistent one inspired by the recent Amino Acid Interactions web server v2.0 (11). The new pipeline operates as follows. First, polypeptide chain sequences are extracted from the input file using P2Rank. The phmmer tool from the HMMER software package (http://hmmer.org/) is then used to identify and align similar sequences for each respective query; UniRef50 Release 2021_03 (12) is used as the single target sequence database. Up to 1000 sequences are then randomly selected from each MSA to form the respective sample MSAs; weights are assigned to the individual sequences constituting the sample MSAs using the Gerstein/Sonnhammer/Chothia algorithm (13) implemented in the esl-weight miniapp included with the HMMER software. Finally, per-column information content (i.e. conservation score) and gap character frequency values are calculated using the esl-alistat miniapp, taking the individual sequence weights into account; positions containing the gap character in >50 % of sequences are masked to appear as possessing no conservation at all. The pipeline utilizes a fixed seed value for any random selection, making the output deterministic for a given query. Table 1 shows the runtimes of the new EC calculation pipeline measured on the datasets used for the training, validation, and testing of P2Rank models. It can be seen that for 50% of queries, the EC calculation pipeline (which constitutes most of the time required for PrankWeb predictions) finishes in about 2 min, while nearly all queries finish within 5 min. In comparison, for the previous EC conservation pipeline on the CHEN11 dataset, the median of runtimes was 275 s (4.6 min) while 95th percentile was 854 s (14.2 min). Table 1. The runtimes of the new EC calculation pipeline (in seconds) measured on the datasets used for the training (CHEN11), validation (JOINED), and testing (COACH420 and HOLO4K) of P2Rank models. The computations were performed on a desktop computer running Ubuntu 20.04, HMMER v3.3.2, and using the i7-3770K processor. The numbers in parentheses indicate the number of polypeptide chains in the respective datasets. See the original P2Rank publication (3) for a detailed description of the datasets CHEN11 (251) JOINED (643) COACH420 (420) HOLO4K (8588) Runtime (50th percentile; s) 107 109 108 127 Runtime (95th percentile; s) 139 244 193 324 The adoption of the new EC calculation pipeline necessitated the preparation of a new EC-aware P2Rank model. Table 2 presents the evaluation of all the new P2Rank models prepared for PrankWeb 3, as well as their comparison with the former models; it can be seen that the new Default models exceed the performance of the corresponding old models when evaluated on the representative HOLO4K dataset. Table 2. Identification success rates (in %) measured using the DCA criterion utilizing a 4.0 Å threshold for the distance between the center of the predicted LBS and any ligand atom; only the n or (n + 2), respectively, top-ranking predicted LBSs are considered in the evaluation, where n is the number of ligands in the respective 3D structure. Values for Default (old) and Default + conservation (old) are taken from the original PrankWeb publication (5) and are shown only for comparison, as these models are no longer used. B-factor-free are used with AlphaFold predictions which utilize the B-factor field for confidence scores. Please note that old models were generated by the older version of P2Rank, which used older versions of BioJava and CDK. Using newer versions changed how certain PDB files are parsed, and an upgrade of the CDK library fixed a bug in the algorithm that generates SAS points. This, together with bug fixes in P2Rank itself, causes the scores for the Default (old) and Default models to differ COACH420 HOLO4K Top-n Top-(n + 2) Top-n Top-(n + 2) Default (old) 72.0 78.3 68.6 74.0 Default + conservation (old) 73.2 77.9 72.1 76.7 Default 71.6 76.8 72.7 78.0 Default + conservation 74.3 77.2 74.5 78.4 B-factor-free 71.2 77.5 72.1 77.2 B-factor-free + conservation 74.9 78.5 73.9 77.7 PREDICTED STRUCTURES The AlphaFold DB (9) is a freely and openly accessible resource housing 3D structure models for a selection of biomedically significant proteins predicted using AlphaFold 2 (8). In PrankWeb 3, we have precomputed the P2Rank LBS predictions for two components of the AlphaFold DB—the ‘model organism proteomes’ and ‘Swiss-Prot’—totalling over 800 000 proteins. As the AlphaFold 3D structure models utilize the B-factor fields of the structure files to store the per-residue confidence scores, computing these LBS predictions necessitated the preparation of two additional, B-factor field-agnostic P2Rank models (Table 2); it can be seen that the performance of these on the representative HOLO4K dataset (consisting of experimentally resolved 3D structures) is only marginally worse compared to the models utilizing B-factor as a feature. To show how PrankWeb can be used to predict and visualize binding sites for predicted structures, we chose a protein from the G protein-coupled receptors (GPCR) family. The GPCR family is not only the largest protein family (with over 800 members), but also a family with >160 validated drug targets. GPCRs are membrane proteins and as such have represented a major challenge for structural biology. Advances in cryoEM methodology have brought a revolution in our understanding of intricate differences among GPCR proteins with more than 450 structures of over 80 proteins (14) solved so far, but many proteins indicated in human disease are still without an experimentally solved structure. The availability of high-quality 3D structure models in the AlphaFold DB, however, massively expands the number of proteins that can be investigated with PrankWeb. We used PrankWeb to show predicted binding sites on the AlphaFold model of succinate receptor 1 (uniprot code Q9BXA5), a protein suspected as a major player in the development of kidney hypertension and possibly also metabolic syndrome and thus potential drug target (15) without known experimentally solved 3D structure. The structure submission interface of PrankWeb has been extended to enable fetching predicted structures from the AlphaFold DB via the UniProt accession. After the accession is entered, the structure is downloaded from the AlphaFold DB (if not cached) and binding sites are predicted with P2Rank. Once the results are available, they are visualized in the PrankWeb interface. For AlphaFold predictions, the structure is color-coded by the confidence score. Moreover, PrankWeb enables visualization of only high-confidence regions (pLDDT > 70). The results for the succinate receptor 1 are shown in Figure 1. Figure 1 A displays the best predicted pocket in blue on top. As the experimental structure with, or even without a ligand, is not known, the predicted structure was aligned using PyMOL with the structure of a closely related P2Y12 receptor (PDB ID 4NTJ (16)). The structural alignment (Figure 1B) shows that the best predicted succinate receptor binding pocket is different from ligand binding pocket of P2Y12 receptor as expected due to different properties and size of these ligands, although we can not be completely sure that the predicted binding site is correct as there is no experimentally solved structure of this receptor. This shows that using AlphaFold models for prediction of binding sites provides information that can not be extracted from experimentally solved structures of closely related proteins. Figure 1. P2Rank prediction on an AlphaFold model of human succinate receptor (Q9BXA5). (A) Visualization of the pockets from PrankWeb (available at https://prankweb.cz/analyze?database=v3-alphafold&code=Q9BXA5). The main pocket is in blue on the top of the structure. The structure is colored-coded by AlphaFold confidence (darker being more confident). (B) The predicted succinate receptor structure (in cyan) is aligned with closely related P2Y receptor (in grey, PDB ID 4NTJ) and its ligand (in magenta). The best binding pocket predicted for succinate receptor is shown in blue and is clearly outside of the binding pocket of P2Y receptor (visualized with PyMOL, http://www.pymol.org/pymol). OTHER IMPROVEMENTS Additional updates focus on improving the user experience and usability. The updates range from small quality of life improvements to complete redesign of the PrankWeb architecture. The most noticeable change is in the results visualization page (Figure 2). First, the user can now select a visualization mode for the inspected protein and the predicted binding sites. The modes available are surface, cartoon, and balls and sticks. Second, when a pocket prediction is carried out on a predicted structure, the user can hide low-confident regions, i.e. regions with pLDDT score <70. Finally, the protein surface is colored by conservation score for the experimental structures, and by residue-level confidence scores for the predicted structures. Figure 2. PrankWeb results visualization page. The view shows predicted LBSs on the AlphaFold model of the human striatin-interacting protein (Q5VSL9), available at https://prankweb.cz/analyze?database=v3-alphafold&code=Q5VSL9. Pockets are displayed using surface visualization while the rest of hte structure is shown as cartoon. Different putative pockets are distinguished by color. The parts of the structure which are not part of any pocket are color-coded by tha AlphaFold confidence score, with darker regions being more confident. Finally, the visualization shows only high-confident parts of the structure (pLDDT score > 70) which are connected by dotted lines. Switching between full structure and confident regions only can be controlled by the user. Another addition to the results visualization page is the pocket’s probability score. By default, the pockets are sorted using the P2Rank’s raw pocket score. However, as this value is not bound, it is hard to interpret by a user. To tackle this we added the pocket’s probability score that has a clearly defined maximum value and thus should provide easier interpretation to a user. The pocket probability score is calculated as a monotonous transformation of a raw pocket score to the interval [0,1]. The transformation is calibrated for each model on the HOLO4K dataset in such a way that the probability score represents a ratio of true binding sites among all predicted sites with a comparable raw score. We have also updated the HTTP-based API to v2, indicating breaking changes. The core idea was to shift the API closer to the REST ideas. The change allows users to easily create new prediction tasks for custom structures using POST. GET requests can be used to retrieve prediction status, log, structure or prediction archive. The prediction archive can be also downloaded from the user interface and contains visualizations of the protein in PyMOL, parameters used to run P2rank, prediction log file and information about the predicted pockets in the CSV format. In addition, the archive can contain conservation scores if the user has chosen to use conservation in the prediction. We also added links to the pre-computed predictions described in the section Evolutionary conservation calculation pipeline. Users can thus download all predictions computed for PDB and AlphaFold. For each database, we provide predictions computed with and without the use of conservation. The archive has similar content to the archive for a single prediction, the main difference is in the structure as the archives house multiple predictions. Another modification in PrankWeb 3 is added support for the mmCIF format as the structure definition format. This was necessary as the PDB format has been deprecated due to its limitations. Finally, under the hood, PrankWeb’s architecture has been completely redesigned. The new modular architecture strictly separates web-based user interface, data storage, and an execution component. The execution component is responsible for running the predictions from start to end. Starting with a protein file or UniProt ID, it will compute conservation and produce pocket predictions. Each component corresponds to a Docker image. Combined with docker-compose, it is easy to deploy and update PrankWeb instances. Thanks to the modular architecture, users can deploy only the execution component, using Docker, on their hardware. As a result, it is possible to run predictions on private data without exposing them to third-party servers. Another advantage is that such deployment allows users to run as many predictions as their computation resources allow. On the other hand, we are aware that not every user has the capacity to run the predictions on a large scale database such as PDB and parts of the AlphaFold. DATA AVAILABILITY The PrankWeb web server is publicly available at https://prankweb.cz/. The source codes are available at https://github.com/cusbg/p2rank-framework. ACKNOWLEDGEMENTS Computational resources were supplied by the project ‘e-Infrastruktura CZ’ (e-INFRA CZ LM2018140) supported by the Ministry of Education, Youth and Sports of the Czech Republic. This work was also carried out with the support of the Charles University grant SVV-260588 and the ELIXIR CZ Research Infrastructure (ID LM2018131, MEYS CR), including access to the computational resources. FUNDING Funding for open access charge: ELIXIR CZ Research Infrastructure (ID LM2018131, MEYS CR). Conflict of interest statement. None declared. ==== Refs REFERENCES 1. Konc  J., Lešnik  S., Janežič  D.  Modeling enzyme-ligand binding in drug discovery. J. Cheminform.  2015; 7 :48.26457119 2. Imamura  A., Okada  T., Mase  H., Otani  T., Kobayashi  T., Tamura  M., Kubata  B.K., Inoue  K., Rambo  R.P., Uchiyama  S.  et al .  Allosteric regulation accompanied by oligomeric state changes of Trypanosoma brucei GMP reductase through cystathionine-β-synthase domain. Nat. Commun.  2020; 11 :1837.32296055 3. Krivák  R., Hoksza  D.  P2Rank: machine learning based tool for rapid and accurate prediction of ligand binding sites from protein structure. J. Cheminf.  2018; 10 :39. 4. Mylonas  S.K., Axenopoulos  A., Daras  P.  DeepSurf: a surface-based deep learning approach for the prediction of ligand binding sites on proteins. Bioinformatics. 2021; 37 :1681–1690.33471069 5. Jendele  L., Krivak  R., Skoda  P., Novotny  M., Hoksza  D.  PrankWeb: a web server for ligand binding site prediction and visualization. Nucleic Acids Res.  2019; 47 :W345–W349.31114880 6. Berman  H.M., Westbrook  J., Feng  Z., Gilliland  G., Bhat  T.N., Weissig  H., Shindyalov  I.N., Bourne  P.E.  The protein data bank. Nucleic Acids Res.  2000; 28 :235–242.10592235 7. Mitchell  A.L., Almeida  A., Beracochea  M., Boland  M., Burgin  J., Cochrane  G., Crusoe  M.R., Kale  V., Potter  S.C., Richardson  L.J.  et al .  MGnify: the microbiome analysis resource in 2020. Nucleic Acids Res.  2020; 48 :D570–D578.31696235 8. Jumper  J., Evans  R., Pritzel  A., Green  T., Figurnov  M., Ronneberger  O., Tunyasuvunakool  K., Bates  R., Žídek  A., Potapenko  A.  et al .  Highly accurate protein structure prediction with AlphaFold. Nature. 2021; 596 :583–589.34265844 9. Varadi  M., Anyango  S., Deshpande  M., Nair  S., Natassia  C., Yordanova  G., Yuan  D., Stroe  O., Wood  G., Laydon  A.  et al .  AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models. Nucleic Acids Res.  2022; 50 :D439–D444.34791371 10. Capra  J.A., Singh  M.  Predicting functionally important residues from sequence conservation. Bioinformatics. 2007; 23 :1875–1882.17519246 11. Vymětal  J., Jakubec  D., Galgonek  J., Vondrášek  J.  Amino Acid Interactions (INTAA) web server v2.0: a single service for computation of energetics and conservation in biomolecular 3D structures. Nucleic Acids Res.  2021; 49 :W15–W20.34019656 12. Suzek  B.E., Wang  Y., Huang  H., McGarvey  P.B., Wu  C.H.  UniRef clusters: a comprehensive and scalable alternative for improving sequence similarity searches. Bioinformatics. 2015; 31 :926–932.25398609 13. Gerstein  M., Sonnhammer  E.L., Chothia  C.  Volume changes in protein evolution. J. Mol. Biol.  1994; 236 :1067–1078.8120887 14. Yang  D., Zhou  Q., Labroska  V., Qin  S., Darbalaei  S., Wu  Y., Yuliantie  E., Xie  L., Tao  H., Cheng  J.  et al .  G protein-coupled receptors: structure- and function-based drug discovery. Signal Transduct. Target Ther.  2021; 6 :7.33414387 15. Ariza  A.C., Deen  P.M., Robben  J.H.  The succinate receptor as a novel therapeutic target for oxidative and metabolic stress-related conditions. Front. Endocrinol. (Lausanne). 2012; 3 :22.22649411 16. Zhang  K., Zhang  J., Gao  Z.G., Zhang  D., Zhu  L., Han  G.W., Moss  S.M., Paoletta  S., Kiselev  E., Lu  W.  et al .  Structure of the human P2Y12 receptor in complex with an antithrombotic drug. Nature. 2014; 509 :115–118.24670650

PMC010xxxxxx/PMC10353909.txt

==== Front Eur J Cardiovasc Nurs Eur J Cardiovasc Nurs eurjcn European Journal of Cardiovascular Nursing 1474-5151 1873-1953 Oxford University Press US 36256701 10.1093/eurjcn/zvac098 zvac098 Review Article AcademicSubjects/MED00200 AcademicSubjects/MED00020 AcademicSubjects/MED00600 AcademicSubjects/MED00605 Eurheartj/13 Bed rest duration and complications after transfemoral cardiac catheterization: a network meta-analysis https://orcid.org/0000-0002-1960-2912 Busca Erica Department of Translational Medicine, Università del Piemonte Orientale, Via Solaroli 18, Novara, 28100, Italy https://orcid.org/0000-0002-9114-5816 Airoldi Chiara Department of Translational Medicine, Università del Piemonte Orientale, Via Solaroli 18, Novara, 28100, Italy Bertoncini Fabio Internal Medicine, Ospedale degli Infermi, Ponderano, Via dei Ponderanesi 2, Biella, 13875, Italy Buratti Giulia Internal Medicine, Ospedale degli Infermi, Ponderano, Via dei Ponderanesi 2, Biella, 13875, Italy Casarotto Roberta Emergency Department, Ospedale degli Infermi, Ponderano, Via dei Ponderanesi 2, Biella, 13875, Italy Gaboardi Samanta Emergency Department, Ospedale degli Infermi, Ponderano, Via dei Ponderanesi 2, Biella, 13875, Italy https://orcid.org/0000-0003-2382-0848 Faggiano Fabrizio Department of Translational Medicine, Università del Piemonte Orientale, Via Solaroli 18, Novara, 28100, Italy Epidemiology Centre of Local Health Unit of Vercelli, Largo Giusti 13, Vercelli, 13100, Italy https://orcid.org/0000-0003-4224-8779 Barisone Michela Department of Translational Medicine, Università del Piemonte Orientale, Via Solaroli 18, Novara, 28100, Italy White Ian R Institute of Clinical Trials and Methodology, Faculty of Population Health Sciences, University College London, 90 High Holborn, London WC1V 6LJ, UK https://orcid.org/0000-0002-1634-8330 Allara Elias Department of Translational Medicine, Università del Piemonte Orientale, Via Solaroli 18, Novara, 28100, Italy British Heart Foundation Cardiovascular Epidemiology Unit, Department of Public Health and Primary Care, University of Cambridge, Biomedical Campus, Papworth Road, Trumpington, Cambridge CB2 0BB, UK https://orcid.org/0000-0003-2263-1340 Dal Molin Alberto Department of Translational Medicine, Università del Piemonte Orientale, Via Solaroli 18, Novara, 28100, Italy Corresponding author: Tel: 0321 660567, Email: erica.busca@uniupo.it Lead authors. Senior authors. Conflict of interest: None declared. 7 2023 18 10 2022 18 10 2022 22 5 454462 14 2 2022 07 10 2022 10 10 2022 14 11 2022 © The Author(s) 2022. Published by Oxford University Press on behalf of the European Society of Cardiology. 2022 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (https://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Abstract Aims To assess the effects of bed rest duration on short-term complications following transfemoral catheterization. Methods and results A systematic search was carried out in MEDLINE, Embase, CINAHL, Cochrane Database of Systematic Reviews, Scopus, SciELO and in five registries of grey literature. Randomized controlled trials and quasi-experimental studies comparing different durations of bed rest after transfemoral catheterization were included. Primary outcomes were haematoma and bleeding near the access site. Secondary outcomes were arteriovenous fistula, pseudoaneurysm, back pain, general patient discomfort and urinary discomfort. Study findings were summarized using a network meta-analysis (NMA). Twenty-eight studies and 9217 participants were included (mean age 60.4 years). In NMA, bed rest duration was not consistently associated with either primary outcome, and this was confirmed in sensitivity analyses. There was no evidence of associations with secondary outcomes, except for two effects related to back pain. A bed rest duration of 2–2.9 h was associated with lower risk of back pain [risk ratio (RR) 0.33, 95% confidence interval (CI) 0.17–0.62] and a duration over 12 h with greater risk of back pain (RR 1.94, 95% CI 1.16–3.24), when compared with the 4–5.9 h interval. Post hoc analysis revealed an increased risk of back pain per hour of bed rest (RR 1.08, 95% CI 1.04–1.11). Conclusion A short bed rest was not associated with complications in patients undergoing transfemoral catheterization; the greater the duration of bed rest, the more likely the patients were to experience back pain. Ambulation as early as 2 h after transfemoral catheterization can be safely implemented. Registration PROSPERO: CRD42014014222. Graphical Abstract Graphical Abstract Percutaneous coronary intervention Cardiac catheterization Femoral access Network meta-analysis Systematic review MIUR 10.13039/501100003407 British Heart Foundation Programme 10.13039/501100000274 Medical Research Council Programme UK Medical Research Council 10.13039/501100000265 NIHR Cambridge Biomedical Research Centre 10.13039/501100018956 ==== Body pmcNovelty Early ambulation does not increase the risk of vascular complications. Patients experience more back pain with prolonged bed rest. Ambulation as early as 2 h can be safely implemented. Reducing bed rest duration may optimize patients’ management. Introduction Coronary catheterizations are some of the most frequently performed cardiac procedures.1 Traditional access has been through the femoral artery, which is still performed in more than 500 000 patients each year in Europe and over 400 000 in the USA.2 Although recent trends show an increased utilization of the transradial approach,3 transfemoral access is still common and will likely be used in the future whenever radial access is not applicable.4,5 Unfortunately, transfemoral catheterization can lead to several complications, especially at the access site.6,7 In the past years, vascular closure devices (VCDs) and bed rest were recommended to reduce vascular complications.8 While the effectiveness of VCDs is supported by stronger body of evidence,9 there is more uncertainty regarding the optimal duration of bed rest. Clinical guidelines and consensus documents mention the benefits of early mobilization and the risks of prolonged bed rest, but the recommended duration of bed rest after the interventional procedure is either not specified or inconsistent.8,10–12 Prolonged bed rest may be associated with more discomfort, back pain and voiding problems,13–15 and three reviews suggest that bed rest duration after transfemoral catheterization could be reduced without increasing the rate of vascular complications.16–18 However, previous reviews could only rely on results from comparisons with two treatments at a time and were unable to include more recent studies. Methods We performed a comprehensive network meta-analysis (NMA) review to consider all possible comparisons of bed rest durations on post-intervention complications and provide clinicians with more precise information on the optimal duration of bed rest after transfemoral catheterization. We reported the results consistent with the PRISMA extensions statement for NMA.19 We registered the study in the International Prospective Register of Systematic Reviews (PROSPERO) (CRD42014014222) and published the review’s protocol.20 Ethical approval was not required for this study, and patient involvement was not planned since this was a systematic review based on published primary studies. Search strategy We searched six biomedical databases (MEDLINE, Embase, CINAHL, Cochrane Database of Systematic Reviews, Scopus, and SciELO) until 15 May 2022, without language restrictions. The search terms were a combination of thesaurus-based and free-text terms, and we report them in the Supplementary material online (1. Expanded methods). We explored five sources of grey literature (UpToDate, NHS evidence, Clinicaltrials.gov, WHO International Clinical Trials Registry platform and the ISRCTN registry) and manually extracted studies from the references of previous reviews. Study selection and quality assessment We considered randomized controlled trials (RCTs) and quasi-experimental studies: (i) comparing early with delayed mobilization, (ii) recruiting patients of all ages who underwent diagnostic or therapeutic transfemoral cardiac catheterization and (iii) assessing the effects of bed rest durations in which potential confounders (e.g. postural strategies, catheter size and arterial closure devices) were substantially constant across all study groups. We excluded studies assessing other interventions in addition to bed rest duration. We assessed the methodological quality of the included studies using the Cochrane Effective Practice and Organization of Care (EPOC) Risk of the Bias tool.21 We used the GRADE approach to assess the certainty of the evidence for each primary outcome of interest in each paired comparison for which there is direct evidence. The GRADE system classifies evidence as ‘high’, ‘moderate’, ‘low’ or ‘very low’ certainty. The quality rating start for randomized trial is ‘high’ and may be rated down for limitations concerning risk of bias, inconsistency, indirectness and publication bias. We also used the GRADE approach to assess the certainty in indirect and network (mixed) effect estimates.22,23 Data extraction We extracted information on study characteristics (design, number of patients in each arm, participant age, purpose of procedure, description of intervention, catheter or sheath size, procedure to promote haemostasis) and outcomes. We contacted study authors to complete information not available in the original publication. Categories of bed rest duration are reported in the Supplementary material online (1. Expanded methods). New-onset bleeding and haematoma at the puncture site were our primary outcomes. We extracted information regarding the following secondary outcomes: arteriovenous fistula, pseudoaneurysm, severity of back pain, general patient discomfort and urinary discomfort. Full outcome definitions are included in the Supplementary material online (1. Expanded methods). Data synthesis To maximize utilization of all available data and enable estimation of bed rest duration effects relative to a common control group, we used a random-effects NMA approach. After generating network plots to represent the number of trials and participants for each comparison, we checked key assumptions such as heterogeneity, transitivity and consistency, including exploration of subgroup effects by potential effect modifiers (Supplementary material online, 1. Expanded methods and 2. Expanded results). Although we did not find statistical evidence of between-study heterogeneity, we noted considerable variation in terms of design, patient features and procedures. As such, we conservatively decided to perform random-effects analyses across all outcomes. To further improve power, we also performed post hoc analyses of bed rest duration as a continuous variable. Finally, we conducted sensitivity analyses by including only RCTs and only high-quality RCTs, defined as trials that were not at high risk of bias in any EPOC Risk of Bias domain. We also performed pairwise meta-analysis using all available comparisons. Consistent with the main NMA analysis, we used random-effects models and expressed potential evidence of heterogeneity with the I2 statistic. To assess the potential presence of publication bias, we generated funnel plots (i.e. scatter plots of study effects and their inverted standard errors). We present risk ratios (RRs) and standardized mean differences (SMDs) with their corresponding 95% confidence intervals (CIs) and two-sided P-values. We performed frequentist NMA using Stata 13 with the mvmeta package,24 and pairwise meta-analysis using R v. 3.6.225 with the metafor package v. 2.4.26 Results Study description Search results and study selection We identified 11 700 records from 5 databases and 109 additional papers through sources of grey literature and manuscript references (Figure 1). Based on the assessment of full texts, we found 28 studies that met eligibility criteria and were included in the final review. There was high agreement between the review authors on study selection (Cohen’s Kappa = 0.88). Figure 1 PRISMA flow diagram. Included studies The characteristics of included studies are presented in Supplementary material online, Table S1. Thirteen studies were published before the 2000s27–39 and 15 were published afterwards.13–15,40–51 Twenty-eight studies with 9217 participants compared either bed rest vs. early mobilization or a longer vs. shorter duration of bed rest. All included studies involved an experimental arm where a shorter duration of bed rest was implemented and compared with a longer duration after transfemoral catheterization. The duration of bed rest after catheterization ranged from immediate mobilization directly off the angiographic table to 12 h or longer.14,27–29,35,46 Twenty-four studies had two comparison groups, three studies had three groups37,41,43 and one study had four groups.50 The overall weighted mean age of participants was 60.4 years. Nineteen studies involved patients undergoing diagnostic cardiac catheterization,14,15,27–30,32–34,36–40,42,44,45,47,49 seven studies comprised patients undergoing therapeutic procedures31,35,41,43,46,48,50 and one study both procedures.51 Only one study did not report details about the procedure.13 Catheter and sheath sizes ranged between five and nine French, and 40% of studies used a mean size of six French. Haemostasis was achieved with direct compression manually for 10–20 min in 12 studies,13,27,28,30–32,35,38,44,46–48 with mechanical compression devices in 5 studies36,40,41,43,49 or either.15,34,37,39 In addition, haemostasis was maintained with sandbag in three studies,41,44,47 pressure dressing in seven studies30,31,35,36,40,46,48 or either in five studies.13,14,29,32,42 Risk of bias in included studies We present the summary findings of our quality appraisal in Figure 2 and study specific results in Supplementary material online, Table S2. In general, study quality was good in relation to attrition bias and selective reporting, and poor for the other source of bias considered. Further information is available in Supplementary material online, 2. Expanded results. Figure 2 Risk of bias of included studies. Intervention effects The networks of eligible comparisons for each outcome are available in Figure 3. All comparisons between bed rest durations had at least one trial including a group with bed rest duration falling within the reference category 4–5.9 h. Figure 3 Network plots for all outcomes (A–F) in all included studies. Primary outcomes Twenty-two studies focused on the incidence of bleeding, reporting on 7329 participants and 63 cases. Network meta-analysis (Figure 4B) showed no evidence of association between bed rest duration and bleeding. These findings were confirmed in sensitivity analyses (Supplementary material online, Figures S1B and S2B) and pairwise meta-analyses (Supplementary material online, Figure S3). A post hoc NMA model assuming a linear relationship confirmed lack of association with this outcome (Supplementary material online, Table S3). Figure 4 Random effects network meta-analysis results based on all studies. Twenty-six studies assessed the effect of bed rest duration on the risk of haematoma formation, comprising 9022 participants and 438 cases. There are some suggestions of lower risk of haematoma at shorter durations and higher risk at longer durations, but the finding of one statistically significant result (P = 0.045) out of six tests performed does not suggest evidence of an association, especially since a longer duration showed a lower risk (Figure 4A). No substantial differences from this figure were observed when removing two quasi-experimental studies (Supplementary material online, Figure S1A), when restricting analyses to 12 high-quality RCTs (Supplementary material online, Figure S2A) and in pairwise meta-analyses (Supplementary material online, Figure S4). A post hoc continuous-duration NMA model confirms this and shows no association with haematoma risk [RR 1.00, 95% CI 0.97–1.03] (Supplementary material online, Table S3). We found low heterogeneity for both primary outcomes (Figure 4A and B). Because we noted some differences in the distribution of potential effect modifiers across studies comparing different bed rest durations (Supplementary material online, Table S4), we performed subgroup analyses which revealed no evidence of variation of bed rest duration effects on vascular complications by any of the potential effect modifiers (Supplementary material online, Table S5). Finally, funnel plots were generally symmetrical for both primary outcomes, suggesting that publication bias was unlikely (Supplementary material online, Figure S5). According to the GRADE framework, the certainty of the evidence for the primary outcome is affected by the risk of bias in the included studies and the imprecision of network estimates, which included CIs that include both clinical benefits and possible harms related to bed rest duration. Secondary outcomes Seven studies reported binary back pain in 1832 participants with 247 cases. In NMA, a bed rest duration of 2–2.9 h was associated with lower risk of back pain (RR 0.33, 95% CI 0.17–0.62) and a bed rest of over 12 h with greater risk of back pain (RR 1.94, 95% CI 1.16–3.24), compared with 4–5.9 h (Figure 4E). The post hoc analysis (Supplementary material online, Table S3) supports the hypothesis of an association across durations (RR per 1 h increase in bed rest duration 1.08, 95% CI 1.04–1.11). Pairwise meta-analysis reveals that both studies assessing the >12 vs. 4–5.9 h comparison14,46 have point estimates in the direction of increased risk of back pain, with a pooled effect that is consistent with that generated by NMA and with no evidence of heterogeneity (P = 0.21) (Supplementary material online, Figure S6). Pain intensity measured in two studies42,51 did not differ according to the duration of bed rest (Supplementary material online, Figure S7). General patient discomfort was assessed on a continuous scale by 2 studies in 219 patients. Network meta-analysis (Figure 4F) is limited by the paucity of studies and results are very similar to findings of pairwise meta-analysis (Supplementary material online, Figure S8). There was evidence of greater discomfort among patients allocated to 6–7.9 h bed rest duration compared with 4–5.9 h (SMD 1.06, 95% CI 0.60–1.52, based on one study) but no evidence of association when comparing a bed rest duration >12 vs. 4–5.9 h. Meta-analysis of 2 studies comprising 668 patients and 90 events found no association between rest duration and urinary discomfort (Supplementary material online, Figure S9) when comparing a rest duration >12 vs. 4–5.9 h. Seven studies assessed arteriovenous fistula risk in 2371 participants with 5 cases. There was no evidence of an effect of bed rest duration on such outcome in any analysis (NMA, Figure 4D; post hoc linear NMA, Supplementary material online, Table S2; pairwise meta-analysis, Supplementary material online, Figure S10). Pseudoaneurysm was assessed in 15 studies comprising 7337 participants and 14 cases. Network meta-analysis showed no evidence of association between any bed rest duration and pseudoaneurysm (Figure 4C). This finding was confirmed in a post hoc analysis assuming a linear relationship between bed rest duration and this outcome (Supplementary material online, Table S3), as well as in pairwise meta-analyses (Supplementary material online, Figure S11). Funnel plots were generally symmetrical for all secondary outcomes (Supplementary material online, Figure S12), suggesting little evidence for publication bias. We could not assess if the effects varied by potential effect modifiers for secondary outcomes due to the scarceness of available data. Discussion In our review, the duration of bed rest after coronary catheterization was generally not associated with short-term complications. We also found that short bed rest (2–2.9 h) was associated with lower risk of back pain and long bed rest (>12 h) was associated with higher risk. Back pain is quite common after cardiac catheterization. Lying on supine position for prolonged periods causes cellular ischaemia and pain in the lumbar and the back due to the application of pressure resulting from the position itself. The literature also highlights how changes in patients’ back pain are associated with position change and long bed rest.52,53 The estimates of intervention effect from our study are in line with previous reviews that did not find evidence of difference in the incidence of vascular complications among patients in the categories compared.16–18 In addition to achieving greater precision due to the availability of new studies and the application of NMA, we extend previous published results by adding a new interval of bed rest, 0–1.9 h, which is not associated with risk of haematoma or bleeding. Importantly, our results show low between-study heterogeneity, which is positively surprising considering the high number of studies included and the varying definitions of haematoma and bleeding formation at the puncture site, as well as the varying catheter sizes and haemostasis techniques. Low heterogeneity is however consistent with the findings of previous reviews that found no significant difference in the incidence of vascular complications due to different catheter sizes and the haemostasis technique.9,17 Although we were unable to gather information on the allocation method for some randomized studies and a few additional studies were not randomized, our sensitivity analyses restricted to high-quality RCT confirmed our main results, suggesting that these study characteristics were unlikely to substantially affect the findings of our NMA. Generalizability may be another issue—studies included patients with different mean age and undergoing different procedures (e.g. diagnostic or therapeutic) and the haemostasis technique. However, detailed subgroup analyses showed that these differences are unlikely to modify the effects of bed rest duration, suggesting that these findings may be generalizable to different settings patients and procedures. There are several suggestions for future research in light of the outcomes from this NMA. A particular strength of the network approach is that it can highlight where future comparisons are needed. The connectivity illustrated by the networks suggests that more direct evidence is required on the effects of short bed rest. It is also evident that there is little utility in the continued use of long bed rest. In addition, while a consideration of resources consumption and costs was beyond the scope of this review, it would be useful for future studies to focus on these aspects as well. Conclusions The duration of bed rest after transfemoral catheterization is unlikely to be associated with onset of short-term vascular complications. Ambulation as early as 2 h after transfemoral cardiac catheterization can be safely implemented, if the patient’s physical state allows. A short duration of bed rest will likely result in optimized patient management and reduced risk of complications, therefore lowering in-hospital length of stay and related costs. Findings support the importance of quality nursing care focused on improving patient comfort and early detection of post-procedural complications. Supplementary material Supplementary material is available at European Journal of Cardiovascular Nursing online. Supplementary Material zvac098_Supplementary_Data Click here for additional data file. Acknowledgements We are grateful to Martina Botalla Battistina (Università del Piemonte Orientale, Italy) for contributing to updating search results. Funding This study was funded by the Italian Ministry of Education, University and Research (MIUR) program ‘Departments of Excellence 2018–2022’, AGING Project—Department of Translational Medicine, Università del Piemonte Orientale. E.A. was supported by the EU/EFPIA Innovative Medicines Initiative Joint Undertaking BigData@Heart grant no. 116074 and is currently funded by the British Heart Foundation Programme Grant RG/18/13/33946. I.R.W. was funded by the Medical Research Council MC_UU_12023/21. This work was supported by core funding from the UK Medical Research Council (MR/L003120/1), the British Heart Foundation (RG/13/13/30194; RG/18/13/33946) and the NIHR Cambridge Biomedical Research Centre (BRC-1215–20014)*. This work was also supported by Health Data Research UK, which is funded by the UK Medical Research Council, Engineering and Physical Sciences Research Council, Economic and Social Research Council, Department of Health and Social Care (England), Chief Scientist Office of the Scottish Government Health and Social Care Directorates, Health and Social Care Research and Development Division (Welsh Government), Public Health Agency (Northern Ireland), British Heart Foundation and Wellcome. *The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care. Data availability All data relevant to the study are included in the article or uploaded as supplementary information. ==== Refs References 1 Benjamin EJ , ViraniSS, CallawayCW, ChamberlainAM, ChangAR, ChengS, ChiuveSE, CushmanM, DellingFN, DeoR, De FerrantiSD, FergusonJF, FornageM, GillespieG, IsasiCR, JiménezMC, JordanLC, JuddSE, LacklandD, LichtmanJH, LisabethL, LiuS, LongeneckerCT, LutseyPL, MackeyJS, MatcharDB, MatsushitaK, MussolinoME, NasirK, O'FlahertyM, PalaniappanLP, PandeyA, PandeyDK, ReevesMJ, RitcheyMD, RodriguezCJ, RothGA, RosamondWD, SampsonUKA, SatouGM, ShahSH, SpartanoNL, TirschwellDL, TsaoCW, VoeksJH, WilleyJZ, WilkinsJT, WuJH, AlgerHM, WongSS, MuntnerP, American Heart Association Council on Epidemiology and Prevention Statistics Committee and Stroke Statistics Subcommittee. Heart disease and stroke statistics - 2018 update: a report from the American Heart Association. Circulation 2018;137 :E67–E492.29386200 2 Barbato E , NocM, BaumbachA, DudekD, BuncM, SkalidisE, BanningA, LegutkoJ, WittN, PanM, TilstedHH, NefH, TarantiniG, KazakiewiczD, HuculeciR, CookS, MagdyA, DesmetW, CaylaG, VinereanuD, VoskuilM, GoktekinO, VardasP, TimmisA, HaudeM. Mapping interventional cardiology in Europe: the European Association of Percutaneous Cardiovascular Interventions (EAPCI) Atlas Project. Eur Heart J 2020;41 :2579–2588.32584388 3 Masoudi FA , PonirakisA, de LemosJA, JollisJG, KremersM, MessengerJC, MooreJWM, MoussaI, OetgenWJ, VarosyPD, VincentRN, WeiJ, CurtisJP, RoeMT, SpertusJA . Trends in U.S. Cardiovascular Care: 2016 Report from 4 ACC National Cardiovascular Data Registries. J Am Coll Cardiol 2017;69 :1427–1450.28025065 4 Tanaka Y , MoriyamaN, OchiaiT, TakadaT, TobitaK, ShishidoK, SugitatsuK, YamanakaF, MizunoS, MurakamiM, MatsumiJ, TakahashiS, AkasakaT, SaitoS . Transradial coronary interventions for complex chronic total occlusions. JACC Cardiovasc Interv 2017;10 :235–243.28183464 5 Cahill TJ , ChenM, HayashidaK, LatibA, ModineT, PiazzaN, RedwoodS, SøndergaardL, PrendergastBD . Transcatheter aortic valve implantation: current status and future perspectives. Eur Heart J 2018;39 :2625–2634.29718148 6 Applegate RJ , SacrintyMT, KutcherMA, KahlFR, GandhiSK, SantosRM, LittleWC . Trends in vascular complications after diagnostic cardiac catheterization and percutaneous coronary intervention via the femoral artery, 1998 to 2007. JACC Cardiovasc Interv 2008;1 :317–326.19463320 7 Carrozza J . Complications of diagnostic cardiac catheterization. 2012. Available at: https://www.uptodate.com/contents/complications-of-diagnostic-cardiac-catheterization#! 8 Naidu SS , AronowHD, BoxLC, DuffyPL, KolanskyDM, KupferJM, LatifF, MulukutlaSR, RaoSV, SwaminathanRV, BlankenshipJC. SCAI Expert consensus statement: 2016 best practices in the cardiac catheterization laboratory: (Endorsed by the Cardiological Society of India, and sociedad Latino Americana de Cardiologia intervencionista; affirmation of value by the Canadian Associatio. Catheter Cardiovasc Interv 2016;88 :407–423.27137680 9 Robertson L , AndrasA, ColganF, JacksonR . Vascular closure devices for femoral arterial puncture site haemostasis. Cochrane Database Syst Rev 2016;2016 :CD009541. 10 O’Gara PT , KushnerFG, AscheimDD, CaseyDE Jr, ChungMK, de LemosJA, EttingerSM, FangJC, FesmireFM, FranklinBA, GrangerCB, KrumholzHM, LinderbaumJA, MorrowDA, NewbyLK, OrnatoJP, OuN, RadfordMJ, Tamis-HollandJE, TommasoCL, TracyCM, WooYJ, ZhaoDX, AndersonJL, JacobsAK, HalperinJL, AlbertNM, BrindisRG, CreagerMA, DeMetsD, GuytonRA, HochmanJS, KovacsRJ, KushnerFG, OhmanEM, StevensonWG, YancyCW; American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. 2013 ACCF/AHA guideline for the management of ST-elevation myocardial infarction: a report of the American College of Cardiology Foundation/American Heart Association task force on practice guidelines. J Am Coll Cardiol 2013;127 :e362–e425. 11 Ibánez B , JamesS, AgewallS, AntunesMJ, Bucciarelli-DucciC, BuenoH, CaforioALP, CreaF, GoudevenosJA, HalvorsenS, HindricksG, KastratiA, LenzenMJ, PrescottE, RoffiM, ValgimigliM, VarenhorstC, VranckxP, WidimskýP. 2017 ESC guidelines for the management of acute myocardial infarction in patients presenting with ST-segment elevation. Eur Heart J 2018;39 :119–177.28886621 12 Kimura K , KimuraT, IshiharaM, NakagawaY, NakaoK, MiyauchiK, SakamotoT, TsujitaK, HagiwaraN, MiyazakiS, AkoJ, AraiH, IshiiH, OriguchiH, ShimizuW, TakemuraH, TaharaY, MorinoY, IinoK, ItohT, IwanagaY, UchidaK, EndoH, KongojiK, SakamotoK, ShiomiH, ShimohamaT, SuzukiA, TakahashiJ, TakeuchiI, TanakaA, TamuraT, NakashimaT, NoguchiT, FukamachiD, MizunoT, YamaguchiJ, YodogawaK, KosugeM, KohsakaS, YoshinoH, YasudaS, ShimokawaH, HirayamaA, AkasakaT, HazeK, OgawaH, TsutsuiH, YamazakiT; Japanese Circulation Society Joint Working Group. JCS 2018 guideline on diagnosis and treatment of acute coronary syndrome. Circulation J 2019;83 :1085–1196. 13 Chair SY , ThompsonDR, LiSK. The effect of ambulation after cardiac catheterization on patient outcomes. J Clin Nurs 2007;16 :212–214.17181684 14 Chair SY , YuM, ChoiKC, WongEM, SitJW, IpWJ . Effect of early ambulation after transfemoral cardiac catheterization in Hong Kong: a single-blinded randomized controlled trial. Anadolu Kardiyol Derg 2012;12 :222–230.22366106 15 Matte R , HilárioTS, ReichR, AlitiGB, Rabelo-SilvaER . Reducing bed rest time from five to three hours does not increase complications after cardiac catheterization: the THREE CATH trial. Rev Lat Am Enfermagem 2016;24 :e2796.27463113 16 Mohammady M , AtoofF, SariAA, ZolfaghariM . Bed rest duration after sheath removal following percutaneous coronary interventions: a systematic review and meta-analysis. J Clin Nurs 2014;23 :1476–1485.24028631 17 Mohammady M , HeidariK, Akbari SariA, ZolfaghariM, JananiL . Early ambulation after diagnostic transfemoral catheterisation: a systematic review and meta-analysis. Int J Nurs Stud 2014;51 :39–50.23332719 18 Tongsai S , ThamlikitkulV. The safety of early versus late ambulation in the management of patients after percutaneous coronary interventions: a meta-analysis. Int J Nurs Stud 2012;49 :1084–1090.22521852 19 Hutton B , SalantiG, CaldwellDM, ChaimaniA, SchmidCH, CameronC, IoannidisJP, StrausS, ThorlundK, JansenJP, MulrowC, Catalá-LópezF, GøtzschePC, DickersinK, DickersinI, AltmanDG, MoherD . The PRISMA extension statement for reporting of systematic reviews incorporating network meta-analyses of health care interventions: checklist and explanations. Ann Intern Med 2015;162 :777–784.26030634 20 Dal Molin A , FaggianoF, BertonciniF, BurattiG, BuscaE, CasarottoR, GaboardiS, AllaraE. Bed rest for preventing complications after transfemoral cardiac catheterisation: a protocol of systematic review and network meta-analysis. Syst Rev 2015;4 :47.25903277 21 Higgins JPT, Green SCochrane handbook for systematic reviews of interventions version 5.1.0: The Cochrane Collaboration; 2011. Available from www.training.cochrane.org/handbook 22 Puhan MA , SchünemannHJ, MuradMH, LiT, Brignardello-PetersenR, SinghJA, KesselsAG, GuyattGH, GRADE Working Group. A GRADE Working Group approach for rating the quality of treatment effect estimates from network meta-analysis. BMJ 2014;349 :g5630.25252733 23 Brignardello-Petersen R , BonnerA, AlexanderPE, SiemieniukRA, FurukawaTA, RochwergB, HazlewoodGS, AlhazzaniW, MustafaRA, MuradMH, PuhanMA, SchünemannHJ, GuyattGH, GRADE Working Group. Advances in the GRADE approach to rate the certainty in estimates from a network meta-analysis. J Clin Epidemiol 2018;93 :36–44.29051107 24 White IR . Network meta-analysis. Stata J 2015;15 :951–985. 25 Team RC. R: a language and environment for statistical computing. Vienna. R Foundation for Statistical Computing; 2020. 26 Viechtbauer W . Conducting meta-analyses in R with the metafor. J Stat Softw 2010;36 :1–48. 27 Wong MK , NgH, NgLS, TanKP . Early 4-hour post-angiography ambulation as a feasible alternative to routine 24-hour bedcare. Singapore Med J 1988;29 :63–65.3406772 28 Lau KW , TanA, KohTH, KooCC, QuekN, NgA, JohanA . Early ambulation following diagnostic 7-French cardiac catheterization: a prospective randomized trial. Cathet Cardiovasc Diagn 1993;28 :34–38.8416329 29 Keeling AW , KnightE, TaylorV, NordtLA . Postcardiac catheterization time-in-bed study: enhancing patient comfort through nursing research. Appl Nurs Res 1994;7 :14–7.8203874 30 Barkman A , LunseCP. The effect of early ambulation on patient comfort and delayed bleeding after cardiac angiogram: a pilot study. Heart Lung 1994;23 :112–117.8206767 31 Fowlow B , PriceP, FungT. Ambulation after sheath removal: a comparison of 6 and 8 hours of bedrest after sheath removal in patients following a PTCA procedure. Heart Lung 1995;24 :28–37.7706096 32 Keeling A , TaylorV, NordtLA, PowersE, FisherC . Reducing time in bed after cardiac catheterization (TIBS II). Am J Crit Care 1996;5 :277–281.8811150 33 Baum RA , GanttDS. Safety of decreasing bed rest after coronary angiography. Cathet Cardiovasc Diagn 1996;39 :230–233.8933962 34 Wood RA , LewisBK, HarberDR, KovackPJ, BatesER, StomelRJ . Early ambulation following 6 French diagnostic left heart catheterization: a prospective randomized trial. Cathet Cardiovasc Diagn 1997;42 :8–10.9286529 35 Koch KT , PiekJJ, de WinterRJ, MulderK, DavidGK, LieKI . Early ambulation after coronary angioplasty and stenting with six French guiding catheters and low-dose heparin. Am J Cardiol 1997;80 :1084–1086.9352984 36 Lim R , AndersonH, WaltersMI, KayeGC, NorellMS, CaplinJL . Femoral complications and bed rest duration after coronary arteriography. Am J Cardiol 1997;80 :222–223.9230168 37 Singh N , KuganesanK, GoodeE, RicciAJ . The effect of early ambulation on hematoma formation and vascular complications following 7 French diagnostic cardiac catheterization. Can J Cardiol 1998;14 :1223–1227.9852936 38 Bogart MA , BogartDB, RigdenLB, JungSC, ListonMJ . A prospective randomized trial of early ambulation following 8 French diagnostic cardiac catheterization. Catheter Cardiovasc Interv 1999;47 :175–178.10376499 39 Logemann T , LuetmerP, KaliebeJ, OlsonK, MurdockDK . Two versus six hours of bed rest following left-sided cardiac catheterization and a meta-analysis of early ambulation trials. Am J Cardiol 1999;84 :486–488.10468098 40 Roebuck A , JessopS, TurnerR, CaplinJL . The safety of two-hour versus four-hour bed rest after elective 6-French femoral cardiac catheterization. Coron Health Care 2000;4 :169–173. 41 Vlasic W , AlmondD, MasselD. Reducing bedrest following arterial puncture for coronary interventional procedures–impact on vascular complications: the BAC trial. J Invasive Cardiol 2001;13 :788–792.11731689 42 Wang S-L , RedekerNS, MoreyraAE, DiamondMR . Comparison of comfort and local complications after cardiac catheterization. Clin Nurs Res 2001;10 :29–39.11881749 43 Walker S , JenC, McCoskerF, ClearyS. Comparison of complications in percutaneous coronary intervention patients mobilized at 3, 4, and 6 hours after femoral arterial sheath removal. J Cardiovasc Nurs 2008;23 :407–413.18728513 44 Farmanbar R , ChinikarM, GozalianM, MozhganB, RoushanZ, MoghadamniaM. The effect of post coronary angiography bed-rest time on vascular complications. J Tehran Univ Heart Center 2008;3 :225–228. 45 Ashktorab T , NeishabooryM, PiranfarM, Alavi-MajdH. Effects of bed rest reduction after coronary angiography on local vascular complications and back pain. Adv Nurs Midwifery 2009;18 :34–42. 46 Schiks IEJM , SchoonhovenL, AengevaerenWRM, Nogarede-HoekstraC, van AchterbergT, VerheugtFW. Ambulation after femoral sheath removal in percutaneous coronary intervention: a prospective comparison of early vs. late ambulation. J Clin Nurs 2009;18 :1862–1870.19077015 47 Rocha VS , AlitiG, MoraesMA, RabeloER. Three-hour rest period after cardiac catheterization with a 6 f sheath does not increase complications: a randomized clinical trial. Rev Bras Cardiol Invasiva 2009;17 :512–517. 48 Moeini M , MoradpourF, BabaeiS, RafieianM, KhosraviA . Four hour ambulation after angioplasty is a safe practice method. Iran J Nurs Midwifery Res 2010;15 :109–114.21589772 49 Höglund J , StenestrandU, TödtT, JohanssonI . The effect of early mobilisation for patient undergoing coronary angiography; a pilot study with focus on vascular complications and back pain. Eur J Cardiovasc Nurs 2011;10 :130–136.20620118 50 Larsen EN , HansenCB, ThayssenP, JensenLO . Immediate mobilization after coronary angiography or percutaneous coronary intervention following hemostasis with the AngioSeal vascular closure device (the MOBS study). Eur J Cardiovasc Nurs 2014;13 :466–472.24336239 51 Nørgaard MW , FærchJ, JoshiFR, HøfstenDE, EngstrømT, KelbækH . Is it safe to mobilize patients very early after transfemoral coronary procedures? (SAMOVAR): a randomized clinical trial. J Cardiovasc Nurs 2022;;37 (5):E114–E121.34321432 52 Sarabi HN , FarsiZ, ButlerS, PishgooieAH . Comparison of the effectiveness of position change for patients with pain and vascular complications after transfemoral coronary angiography: a randomized clinical trial. BMC Cardiovasc Disord 2021;21 :114.33632127 53 Mert Boğa S , ÖztekinSD. The effect of position change on vital signs, back pain and vascular complications following percutaneous coronary intervention. J Clin Nurs 2019;28 :1135–1147.30367542

PMC010xxxxxx/PMC10353910.txt

==== Front Adv Med Educ Pract Adv Med Educ Pract amep Advances in Medical Education and Practice 1179-7258 Dove 411323 10.2147/AMEP.S411323 Original Research Knowledge, Attitude, and Practices Toward Tuberculosis Among Health Faculty and Non-Health Faculty Students of Kabul University and Kabul University of Medical Sciences, Kabul, Afghanistan Alimi and Sakhi Alimi and Sakhi Alimi Nagina 1 Sakhi Rohullah 2 1 Epidemiology and Biostatistics, Public Health Faculty, Kabul University of Medical Sciences, Kabul, Afghanistan 2 Environmental and Occupational Health, Public Health Faculty, Kabul University of Medical Sciences, Kabul, Afghanistan Correspondence: Nagina Alimi, Email naginaalimi@gmail.com 13 7 2023 2023 14 753761 13 4 2023 02 7 2023 © 2023 Alimi and Sakhi. 2023 Alimi and Sakhi. https://creativecommons.org/licenses/by-nc/3.0/ This work is published and licensed by Dove Medical Press Limited. The full terms of this license are available at https://www.dovepress.com/terms.php and incorporate the Creative Commons Attribution – Non Commercial (unported, v3.0) License (http://creativecommons.org/licenses/by-nc/3.0/). By accessing the work you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms (https://www.dovepress.com/terms.php). Introduction Afghanistan ranks 24th among the countries with a high TB death rate. The number of TB patients has unfortunately increased by 3% during 2022 compared to 2021. University students are among the high-risk groups for TB. The frequent and high level of person-to-person contact in universities increases the transmission of infectious diseases including TB. This study aimed to evaluate the level of knowledge, attitude, and practices of university students regarding tuberculosis to better understand the situation. Methods A cross-sectional questionnaire-based study was conducted among 415 health and non-health faculty students between October and December 2022. Multi-stage stratified sampling technique was used to collect the data and data were analyzed using SPSSv25. Cross-tabulation and a Chi-Square test were used to identify differences between groups. Results The results of this study showed that 18.1% of health and 2.4% of non-health faculty students had good knowledge about TB. There was a significant difference in the knowledge of health and non-health faculty students (P value<0.01). The level of good attitude of health and non-health faculty students about TB was 26.7% and 14.9%, respectively. Regarding practices, 41.9% of health faculty students and 29.8% of non-health faculty students had good practices about TB. There was a significant difference in the attitude (P value = 0.03) and practices (P value = 0.024) of health and non-health faculty students (health faculty students had better knowledge, attitude, and practices.). Conclusion The knowledge, attitude, and practices of health and non-health faculty students were insufficient about TB. The practice level of health faculty students was poorer than what was expected based on their field. Television and the Internet played a good role in informing students about TB. So, both can be used for transferring good knowledge, positive attitudes, and correct practices about TB to society. Keywords tuberculosis students Kabul University Kabul University of Medical Sciences knowledge attitude practice ==== Body pmcIntroduction Tuberculosis (TB) is a severe bacterial infection of the lungs that spread through the air when a person speaks, coughs, or sneezes.1 TB is the leading cause of death worldwide from an infectious agent;2 the thirteenth leading cause of death and the second deadliest infectious agent after COVID-19, surpassing HIV.3 Ending the TB epidemic by 2030 is one of the health goals of the United Nations Sustainable Development Goals (SDGs).3 There is a significant gap between the global burden of TB and the perceived importance of addressing this burden, even in areas where TB is endemic.4 Each year, approximately 10 million people are diagnosed with TB, despite the fact that TB is a preventable and treatable disease. And 1.5 million people die from TB each year; making it a major infectious disease.3 In 2021, an estimated 10.6 million cases of TB were reported, and approximately 1.6 million people died from TB that year.3,5 Most cases of TB occur in low- and middle-income countries; however, TB exists worldwide. TB mostly affects adults in their most productive years. More than 95% of cases and deaths occur in developing countries.6 According to the Global TB Report, Afghanistan is one of the countries with the highest incidence of TB in the Eastern Mediterranean Region7 and is among the 24 countries with high TB death rates worldwide.8 In 2020, TB killed nearly 10,000 people in Afghanistan, which was three times higher than COVID-19 deaths.9 The incidence of TB in 2022 has increased by 3% compared to 2021 in the country.10 The World Health Organization believes that cultural barriers, lack of knowledge and proper attitudes about TB, and security problems continue to pose serious challenges to reducing this deadly infectious disease in Afghanistan.9 Studies have shown that awareness and knowledge about TB and access to its services in high TB burden environments is insufficient.11 On average, one TB patient can infect 10–15 people.12 Timely screening and medications for TB management are essential to reduce the likelihood of infection. Drug-resistant TB is a global challenge that is growing, resulting from incorrect use and mismanagement of TB drugs due to a lack of proper knowledge about this disease.6 Each year, approximately half a million people become resistant to multiple drugs for TB, which is expensive and complicated to treat and has potentially life-threatening side effects.4 Every year on March 24, world TB Day is commemorated to raise public awareness about TB and its devastating health, social and economic consequences and to raise efforts to end the global TB epidemic.13 No study has evaluated knowledge, attitudes, and practices regarding TB among university students in Afghanistan. This study aimed to provide an overview of the current TB knowledge, attitude and practices in two major governmental universities in Afghanistan to better understand the situation and to identify potential areas for improvement in TB prevention and control efforts. The findings of this study will provide valuable insights into the current state of knowledge, attitude, and practices among students and can be used to develop effective TB control strategies in the country. Materials and Methods Study Design, Setting, and Period This cross-sectional study measured health and non-health faculty students’ knowledge, attitude, and practices toward tuberculosis in two large and famous governmental universities of Afghanistan; Kabul University and Kabul University of Medical Sciences. This study was conducted between October and December 2022. Sample Size and Sampling Method In 2022, Kabul University and Kabul University of Medical Sciences had 20,171 students. The number of samples was calculated using a confidence interval of 95%, with an absolute error of 5% and a population proportion of 0.5 through Cochran’s formula and Epi Info 7.2.5 software. The estimated sample size was 415 along with a 10% non-response bias. We used a multi-stage stratified sampling technique in this study. The universities were first divided into two sub-groups (strata), and then each university was also divided into subgroups by faculty. Proportionate share for females and males and each stratum was considered. Then, in each stratum (faculty), first, classes were randomly selected, and then within the selected class, samples (students) were also randomly selected. Ethics The proposal of this study was accepted and confirmed by Public Health Faculty Institutional Review Board. Participation in this study was voluntary and written informed consent to participation was obtained from all participants before participation. All data were collected and kept anonymously without identification. University policy, Afghani cultures and costume were considered completely. We maintain the welfare of research participants by doing no physical or mental harm to them. Data Source and Measurement Based on a review of similar studies, we developed a 35-item questionnaire divided into five sections to achieve the study objectives.1,14,15 The questionnaire was reviewed for content validity by expert researchers and their comments were considered. The questionnaire consisted of 5 demographic questions, 19 knowledge questions, 7 attitude questions, 3 practice questions and one question about source of information of TB. The questionnaire was translated into local languages by a professional and a pilot study of 60 participants was done prior to conducting the main research to confirm the practicability and clarity of questions. Students’ overall knowledge, attitude, and practice were categorized according to Bloom’s cutoff point into good, moderate, and poor based on the total percentage score as follows: good for scores 80% and above, moderate for scores 60–79%, and poor for scores below 60%. For the knowledge questions, participants were required to answer “Yes”, “No”, or “Do not know”. Each correct answer had a score of 1, while the answer “Do not know” had a score of 0. The maximum score for knowledge was 19, for attitude the score ranged from 4 to 16, while for practice questions, the score ranged from 1 to 3. And for the source of information, we found a percentage of each answer. Statistical Analysis For statistical analysis of the data collected, we used descriptive statistics: mean, standard deviation, frequency and percentage. Descriptive analysis such as frequency and percentage was used for categorical variables and for describing the demographic characteristics of the participants. Differences in participants’ knowledge, attitude and practices according to participants’ characteristics were analyzed using Cross-tabulation and Chi-Square test. Significance level was set at p-value <0.05. Results Students’ Sociodemographic Characteristics Of the 415 questionnaires, 400 questionnaires returned. Students’ sociodemographic characteristics are presented in Table 1. Of all the participants, 105 (26.3%) were health faculty students, whereas 295 (73.8%) were non-health faculty students (Table 1). Participants age in this study ranged from (17–28) years with a mean of 21.32 and with a standard deviation of 1.85.Table 1 Students’ Sociodemographic Characteristics Characteristics Number (Percentage) Sex Female 199 (49.8%) Male 201 (50.2%) Faculty Health- faculties 105 (26.3%) Non-health faculties 295 (73.7%) Residence City 338 (84.5%) Village 62 (15.5%) Year of education 1st year students 124 (31%) 2nd year students 95 (23.8%) 3rd year students 88 (22%) 4th year students 93 (23.25%) Students’ Knowledge About TB The results of this study showed that 18.1% of health faculty and 2.4% of non-health faculty students had good knowledge, 63.8% of health and 39.7% of non-health faculty students had moderate knowledge and 18.1% of health and 58% of non-health faculty students had overall poor knowledge of Tuberculosis. As might be expected, knowledge about TB among health faculty students was better than that among non-health faculty students. Additional data on students’ knowledge are presented in Table 2.Table 2 Students’ Knowledge About TB Variables Health Faculty Students Non-Health Faculty Students Yes n (%) No n (%) Do not Know n (%) Yes n (%) No n (%) Do not Know n (%) 1. Have you ever heard of an illness called TB? 104 (99%) 1 (1%) 0 290 (98.3%) 3 (1%) 2 (0.7%) 2. TB can occur anywhere in the body. 55 (52.4%) 41 (39%) 9 (8.6%) 109 (36.9%) 91 (30.8%) 95 32.2% 3. What causes Tuberculosis? a. Cold Air 50 (47.5%) 31 (29.5%) 24 (22.9%) 178 (60.3%) 46 (15.6%) 71 (24.1%) b. Smoking 69 (65.7%) 20 (19%) 16 (15.2%) 180 (61%) 36 (12.2%) 79 (26.8%) c. Dust 82 (78.1%) 10 (9.5%) 13 (12.3%) 215 (72.9%) 16 (5.4%) 64 (21.7%) 4. Can TB transmit through…? a. Cough/breathe 99 (94.3%) 4 (3.8%) 2 (1.9%) 235 (79.6%) 29 (9.8%) 31 (10.5%) b. Sexual contact 36 (34.3%) 49 (46.7%) 20 (19%) 119 (40.3%) 76 (25.8%) 100 (33.9%) 5. What are Symptoms of Tuberculosis? a. Cough for two weeks 91 (86.7%) 4 (3.8%) 10 (9.5%) 249 (84.4%) 11 (3.7%) 35 (11.9%) b. Chest pain 98 (93.35) 3 (2.9%) 4 (3.8%) 234 (79.3%) 12 (4.1%) 49 (16.6%) c. Weight and appetite loss 83 (79%) 12 (11.4%) 10 (9.5%) 209 (70.8%) 19 (6.4%) 67 (22.7%) d. Night sweating 73 (69.5%) 11 (10.5%) 21 (20%) 148 (50.2%) 18 (6.1%) 129 (43.7%) 6. Is TB preventable disease? 100 (95.2%) 3 (2.9%) 2 (1.9%) 257 (87.1%) 11 (3.7%) 27 (9.2%) 7. Can TB be prevented by covering mouth while coughing? 89 (84.8%) 6 (5.7%) 10 (9.5%) 209 (70.8%) 36 (12.2%) 45 (17%) 8. Can TB be Cured? 100 (95.2%) 3 (2.9%) 2 (1.9%) 256 (86.8%) 15 (5.1%) 24 (8.1%) 9. Can TB be cured by herbal remedies? 33 (31.4%) 20 (19%) 52 (49.5%) 102 (34.6%) 57 (19.3%) 136 (46.1%) 10. Can TB be cured by home rest without treatment? 8 (7.6%) 81 (77.1%) 16 (15.2%) 42 (14.2%) 184 (62.4%) 69 (23.4%) 11. TB can only be treated if there are obvious symptoms. 48 (45.7%) 49 (46.7%) 8 (7.6%) 129 (43.7%) 88 (29.9%) 78 (26.4%) 12. Is TB treatment difficult and if anti TB drugs are not taken regularly, it can lead to drug resistance? 57 (54.3%) 19 (18.1%) 29 (27.6%) 109 (36.9%) 53 (18%) 133 (45.1%) 13. Is TB treatment free in Afghanistan? 53 (50.5%) 30 (28.6%) 22 (21%) 82 (27.8%) 108 (36.6%) 105 (35.6%) Abbreviations: n, number of students; %, percentage. According to the results of this study, 99% of health faculty students and 98.3% of non-health faculty students had heard of TB. 65.7% of health faculty students and 61% of non-health faculty students confirmed that smoking causes TB. 47.5% of health faculty students and 60.3% of non-health faculty students believed that cold weather causes TB. 86.7% of health and 84.4% of non-health faculty students were informed that coughs for two weeks is a sign of TB. 95.2% of health and 86.8% of non-health faculty students confirmed that TB can be cured. 19 % of health and 19.3% of non-health faculty students knew that TB cannot be cured by herbal remedies. Only 50.5% of health faculty students and 27.8% of non-health faculty students knew that TB treatment is free in the country (Table 2). Students’ Attitude of TB The results of this study showed that 26.7% of health faculty students and 14.9% of non-health faculty students had good attitude, 60% of health and 61.4% of non-health faculty students had moderate attitude and 12.4% of health and 23.7% of non-health faculty students had overall poor attitude of TB. Additional data on students’ attitudes are presented in Table 3.Table 3 Students’ Attitude About TB Variables Health Faculty Students Non-Health Faculty Students Yes n (%) No n (%) Do Not Know n (%) Yes n (%) No n (%) Do Not Know n (%) Would you be willing to work with someone previously treated for TB? 72 (68.6%) 23 (21.9%) 10 (9.5%) 169 (57.3%) 87 (29.5%) 39 (13.2%) Would you want a family member’s TB to be kept secret? 62 (59%) 29 (27.6%) 14 (13.3%) 181 (61.4%) 89 (30.2%) 25 (8.5%) Do you think you can get TB? 63 (60%) 20 (19%) 22 (21%) 158 (53.6%) 70 (23.7%) 67 (22.7%) What would be your reaction if you found out that you have TB? Answers Fear Hopelessness Shame Sadness Acceptance* Health Faculties 8 (7.6%) 4 (3.8%) 4 (3.8%) 2 (1.9%) 87 (82.9%) Non-Health Faculties 53 (18%) 13 (4.4%) 19 (6.4%) 24 (8.1%) 186 (63.1%) I am interested in finding out more about TB. Answers SA, n(%) A, n(%) DA, n(%) SD, n(%) Health Faculties 11 (10.5%) 83 (79%) 10 (9.5%) 1 (1%) Non-Health Faculties 42 (14.2%) 215 (72.9%) 24 (8.1%) 14 (4.7%) I think education about TB is very much needed. Answers SA, n(%) A, n(%) DA, n(%) SD, n(%) Health Faculties 14 (13.3%) 66 (62.9%) 22 (21%) 3 (2.9%) Non-Health Faculties 30 (10.2%) 128 (43.4%) 109 (36.9%) 28 (9.5%) I would encourage those with TB around me to obtain treatment. Answers SA, n(%) A, n(%) DA, n(%) SD, n(%) Health Faculties 24 (22.9%) 70 (66.7%) 11 (10.5%) 0 Non-Health Faculties 40 (13.6%) 198 (67.1%) 38 (12.9%) 19 (6.4%) Note: *Acceptance and seeking health care. Abbreviations: SA, Strongly agree; A, Agree; DA, Disagree; SD, Strongly disagree; n, number of students; %, percentage. The results of this study showed that 21.9% of health faculty students and 29.5% of non-health faculty students were unwilling to work with someone who was previously treated for TB. Sixty percent of health faculty students and 53.6% of non-health faculty students confirmed that they could also become infected with TB. Fifty-nine percent of health faculty students and 61.4% of non-health faculty students wanted a family member’s TB to be kept secret. 13.3% of health faculty students and 10.2% of non-health faculty students strongly agreed with the statement that “education about TB is very much needed” (Table 3). Students’ Practices of TB The results of this study showed that 41.9% of health faculty and 29.8% of non-health faculty students had good practices and 58.1% of health and 70.2% of non-health faculty students had overall poor practices of tuberculosis. You can see additional data on students’ practices in Table 4.Table 4 Students’ Practices About TB What would you do if you thought you had symptoms of TB? Answers Go to a health facility Go to a pharmacy Go to traditional healers self-treatment Health Faculties 67 (63.8%) 8 (7.6%) 6 (5.7%) 24 (22.9%) Non-Health Faculties 173 (58.6%) 80 (27.1%) 21 (7.1%) 21 (7.1%) If you had symptoms of TB, at what point would you seek medical help? Answers When treatment on my own does not work When TB symptoms last for two or more weeks As soon as I realize TB symptoms Don’t know Health Faculties 14 (13.3%) 36 (34.3%) 50 (47.6%) 5 (4.8%) Non-Health Faculties 69 (23.4%) 101 (34.2%) 102 (34.6%) 23 (7.8%) If you would not go to the health facility, what is the reason? Answers I didn’t refuse to go to the hospital Not sure where to go Cost of the services Do not trust health care workers Can’t leave my work Health Faculties 62 (59%) 3 (2.9%) 18 (17.1%) 13 (12.4%) 9 (8.6%) Non-Health Faculties 148 (50.2%) 15 (5.1%) 82 (27.8%) 39 (13.2%) 11 (3.7%) The results of this study showed that 7.6% of health faculty and 27.1% of non-health faculty students went to the pharmacy if they had symptoms of TB. 22.9% of health faculty students confirmed that they will do self-treatment in case of having TB symptoms. 27.8% of non-health faculty students cited cost of the services as the reason for not seeking medical care. 34.3% of health faculty students and 34.2% of non-health faculty students confirmed that they would seek medical help if symptoms of TB lasted more than two weeks (Table 4). Differences in Students’ Knowledge, Attitude, and Practices Toward TB According to Students’ Characteristics According to the results of this study, health faculty students demonstrated greater knowledge, better attitudes, and practices toward TB than non-health faculty students (p < 0.00, p = 0.003, and p = 0.024 respectively). In this study, there were no significant differences in students’ knowledge, attitude, and practices according to sociodemographic factors (sex and year of education). Additional data on differences in students’ knowledge, attitude, and practices toward TB according to their characteristics are presented in Table 5.Table 5 Differences in Students’ Knowledge, Attitude and Practices Toward TB According to Students’ Characteristics Variables Categories Knowledge Attitude Practices Level of Good Knowledge *P value Level of Good Attitude P value Level of Good Practices P value Faculty Health faculties 18.1% <0.01 27.6% 0.03 41.9% 0.024 Non-health faculties 2.4% 14.9% 29.8% Sex Female 5.5% 0.731 19.6% 0.722 31.2% 0.435 Male 7.5% 16.9% 34.8% Year of education 1st year Students 3.22% 0.07 16.12% 0.871 36.2% 0.561 2nd year Students 4.21% 16.8% 27.3% 3rd year Students 6.81% 22.7% 32.95% 4th year Students 12.9% 18.27% 34.4% Notes: All analyses were by chi-square-test and cross-tabulation, *Significance level was set at p-value < 0.05. Students’ Source of Information About TB 13.2% of the participants in this study said that they obtained information about TB from healthcare workers, 25.5% from family and friends, 17.13% from Internet and social media, 21.22% from television, 3.32% from radio, 2.80% from posters, and 16.60% of the participants mentioned other sources. Discussion The results of this study indicated insufficient knowledge of tuberculosis among health and non-health faculty students, with only 18.1% of health faculty students and 2.4% of non-health faculty students having good knowledge of tuberculosis. In this study, the knowledge of health faculty students was higher than non-health faculty students and a significant relationship between knowledge of tuberculosis and faculty was found (P < 0.01). There was no difference in knowledge based on year of education and gender. The results of this study are in complete contrast to a study conducted in Kabul in 2022 among hospital outpatients, where 87.7% of the participants had good knowledge. This difference may be due to differences in the study population, questionnaire, and measurement methods, indicating that knowledge level among students is very low and more studies are needed on this topic among different study populations in the country.6 50.5% of health faculty students and 27.8% of non-health faculty students in this study knew that treatment of tuberculosis is free in the country which is very low in a developing country like Afghanistan and not knowing that TB treatment is free in many cases can delay seeking health care and ultimately lead to drug-resistant TB. Therefore, raising awareness about the fact that TB treatment is free in the country is necessary. In a study which was conducted among health and non-health faculty students in Indonesia in 2021, the average knowledge score among health and non-health faculty students was respectively, 7.03 ± 2.36 and 4.98 ± 2.20 out of 11. However, in this study, the average knowledge score in health and non-health faculty students was 12.05 ± 2.56 and 9.87 ± 2.61 out of 19, which is almost similar.16 Also, in the study conducted in Indonesia and another study in Malaysia,17 the knowledge of health faculty students was higher than non-health faculty students which was also observed in this study. In Indonesia, female students had good knowledge than male students, but no difference based on gender was observed in this study. In a study in Ethiopia, 35.7% of non-health faculty students had good knowledge about tuberculosis, but in this study, only 2.4% of non-health faculty students had good knowledge, showing the difference.18 In a study in Iran, most health faculty students had moderate to good knowledge, which is contrary to the present study; this difference may be because the participants of the Iran study was only final year students.19 The results of this study regarding attitudes showed that most participants had moderate attitude towards tuberculosis. Health faculty students in this study had a better attitude than non-health faculty students, and this relationship was statistically significant (P = 0.03). In a study conducted in Iran, it was found that most health faculty students had moderate to good attitude towards TB, while in this study, 26.7% of students had a good attitude, and the most had moderate attitude. This similarity may be due to the proximity and cultural similarities of the two countries.20 In another study conducted in China, it was found that 89.64% of non-health faculty students had a positive attitude towards tuberculosis, in another study in Iraq, 76% of medical students had positive attitude while in the present study, only 14.9% of non-health faculty students and 26.7% of health faculty students had a positive attitude, indicating a large difference.21,22 Regarding the practice of tuberculosis, the results of this study showed that 41.9% of health faculty students and 29.8% of non-health faculty students had good practices, while 58.1% of health faculty students and 70.2% of non-health faculty students had poor practices of TB. These results indicate that the level of good practice in both health and non-health faculty students is inadequate. There was a significant difference in practice based on faculty (p = 0.024); health faculty students had better practices compared to non-health faculty students. Still practices of health faculty students are considered weak based on their field of study and regular studying and trainings in medical subjects. There was no significant difference in practices based on year of education and gender. A study conducted in Iran showed that the practice of health faculty students regarding TB was weak, which is similar to the results of this study where 58.1% of health faculty students had poor practices of tuberculosis.19 In another study conducted in Saudi Arabia, 59.4% of the participants had poor practices of TB, which is similar to the results of this study despite having differences in study population.15 In the study in Saudi Arabia, 61.4% stated that they would go to a doctor if they had TB symptoms, but in this study, 60% stated that they would go to a health center.15 In a study in Malaysia among health and non-health faculty students, most of the students said that their source of information was the Internet, while in this study 17.13% said that their source of information was the Internet and social media.17 In this study television was a significant source of information for TB, so mass media and television programs can be used to inform people and increase knowledge of TB in the country. The findings of this study suggest that there is a need for increased public awareness campaigns to educate individuals about TB and its transmission, symptoms and treatment. This could be achieved through various means such as mass media, community outreach programs, and health education sessions. According to the results and insufficient knowledge, attitude, and practices of TB in students, TB education can be integrated into the curriculum of health and non-health faculty students. This can help students to develop a better understanding of TB and its impact on public health. Conclusion According to the results of this study, the knowledge about tuberculosis in health and non-health faculty students was insufficient, but the knowledge of health faculty students was more than that of non-health faculty students (P < 0.01). Most of the health faculty students had moderate knowledge, but the majority of non-health faculty students had poor knowledge about TB. Regarding the attitude, most of the health and non-health faculty students had moderate attitude. There was also a significant difference in the level of attitude of health and non-health faculty students (P = 0.03). Regarding the practice, most of the health faculty students had poorer practices than what was expected based on their field, there was a significant difference in the practices between health and non-health faculty students (P = 0.024). In conclusion, the students in this study did not have adequate knowledge, attitude and practices about TB. In this study, television, Internet and social media had made a significant contribution in transferring of knowledge about TB, which indicates that it is possible to transfer better knowledge to the society with the help of these tools and cause their attitude and practices to improve. In short, improving the knowledge, attitude and practice level of TB among students (the educated class of society) is crucial for reducing the prevalence of TB in the country. By developing educational materials, conducting awareness campaigns, integrating TB education into the curriculum, community outreach programs, social media, Internet and television, we can raise awareness about TB and prevent its spread. Disclosure The authors report no conflicts of interest in this work. ==== Refs References 1. NHS. Tuberculosis (TB); 2022. Available from: https://www.nhs.uk/conditions/tuberculosis-tb/. Accessed July 3, 2023. 2. Datiko DG, Hate D, Jerene D, Suarez P. Knowledge, attitudes and practices related to TB among the general population of Ethiopia: Findings from a national cross-sectional survey. PLoS One. 2019;14 (10 ):1–16 doi:10.1371/journal.pone.0224196. 3. World Health Organization. Tubercolusis; April 21, 2023. Available from: https://www.who.int/news-room/fact-sheets/detail/tuberculosis. Accessed July 3, 2023. 4. ECDC. Tuberculosis remains one of the deadliest infectious diseases worldwide, warns new report; 2022. Available from: https://www.ecdc.europa.eu/en/news-events/tuberculosis-remains-one-deadliest-infectious-diseases-worldwide-warns-new-report. Accessed March 24, 2022. 5. World Health Organization. Global Tuberculosis Report 2022, TB disease burden, TB mortality; 2022. Available from: https://www.who.int/teams/global-tuberculosis-programme/tb-reports/global-tuberculosis-report-2022/tb-disease-burden/2-2-tb-mortality. Accessed July 3, 2023. 6. Yasir Essar M, Jan Rezayee K, Ahmad S, et al. Knowledge attitude and practices toward tuberculosis among hospital outpatients in Kabul, Afghanistan. Frontiers. 2022;10 (933005 ):1–10, 11 7 doi:10.3389/fpubh.2022.933005. 7. EMRO. Tuberculosis, Programmes, Afghanistan; 2021. Available from: https://www.emro.who.int/afg/programmes/stop-tuberculosis-stb.html. Accessed July 3, 2023. 8. World Life Expectancy. Tuberculosis; 2023. Available from: http://www.worldlifeexpectancy.com/cause-of-death/tuberculosis/by-country/. Accessed July 3, 2023. 9. Anadolu Agency. More Afghans Killed by tuberculosis than COVID 19 in 2020; 2021. Available from: https://www.aa.com.tr/en/asia-pacific/more-afghans-killed-by-tubercolusis-than-covid-in-2020/2187078. Accessed July 3, 2023. 10. The Kabul Times. Increasing number of tuberculosis patients in Afghanistan; 2023. Available from: http;//thekabultimes.com/increasing-number-of-tuberculosis-patients-in-Afghanistan/thekabultimes.com. Accessed July 3, 2023. 11. Reliefweb. Drug-resistant tuberculosis in Afghanistan: We must continue to put people at the center of treatment; 2022. Available from: https://reliefweb.int/report/afghanistan/drug-resistant-tuberculosis-afghanistan-we-must-continue-put-people-center. Accessed July 3, 2023. 12. WHO. Tuberculosis; 2022. Available from: http://www.who.int/news-room/fact-sheets/detail/tuberculosis. Accessed July 3, 2023. 13. PAHO. World Tuberculosis Day, 24 March 2022; 2022. Available from: https://www.paho.org/en/campaigns/world-tuberculosis-day-2022. Accessed July 3, 2023. 14. Regassa Luba T, Tang S, Qiaoyan L, Afewerki Gebremedhin S, Kisasi MD, Feng Z. Knowledge, attitude and associated factors towards tuberculosis in Lesotho: a population based study. BMC Infect Dis. 2014;19 (96 ):1–10 doi:10.1186/s12879-019-3688-x. 15. Almalki ME, Almalki FSA, Alasmari R, et al. A cross sectional study of tuberculosis knowledge, attitude and practice among the general population in the western region of Saudi Arabia. Cureus. 2022;14 (10 ):1–11 doi:10.7759/cureus.29987. 16. Melyani Puspitasari I, Kurnia Sinuraya R, Nurhaqiqi Aminudin A, Rahmi Kamilah R. Knowledge, attitudes, and preventive behavior toward tuberculosis in university students in Indonesia. Infect Drug Resist. 2022;15 :4720–4733 doi:10.2147/IDR.S365852. 17. Izham MN, Rahman NA. Knowledge, attitude and practices related to tuberculosis among students in a public university in East Coast Malaysia. Adv Human Biol. 2022;12 (2 ):190–197. doi:10.4103/aihb.aihb_25_22 18. Mekonnen A, Collins JM, Klinkenberg E, et al. Tuberculosis knowledge and attitude among non-health science university students needs attention: a cross sectional study in three Ethiopian universities. BMC Public Health. 2020;20 (631 ):1–9. doi:10.1186/s12889-020-08788-1 31898494 19. Behnaz F. Assessment of knowledge, attitudes and practices regarding tuberculosis among final year students in Yazd, central Iran. J Epidemiol Glob Health. 2013;09 (003 ):81–85 doi:10.1016/j.jegh.2013.09.003. 20. Yusuf L. Recent studies on knowledge, attitude, and practice toward tuberculosis among university students. J App Pharm Sci. 2021;11 (08 ):178–183 doi:10.7324/JAPS.2021.110823. 21. Zuheir Abbas A, Jasim Mohammad S. Assessment of knowledge and attitudes of medical students on tuberculosis. Iraqi Postgrad Med J. 2022;21 (3 ):36O–367 doi:10.1186/s12889-020-08788-1. 22. Du G, Li C, Liu Y, et al. Study on influencing factors of knowledge, attitudes and practice about tuberculosis among freshman in Jiangsu, China: a cross sectional study. Infect Drug Resist. 2021;2022 :1235–1245.

PMC010xxxxxx/PMC10353911.txt

==== Front J Korean Med Sci J Korean Med Sci JKMS Journal of Korean Medical Science 1011-8934 1598-6357 The Korean Academy of Medical Sciences 10.3346/jkms.2023.38.e216 Original Article Gastroenterology & Hepatology Efficacy of Antiviral Prophylaxis up to 6 or 12 Months From Completion of Rituximab in Resolved Hepatitis B Patients: A Multicenter, Randomized Study https://orcid.org/0000-0003-2425-7698 Jang Heejoon 1* https://orcid.org/0000-0001-8888-7977 Yu Su Jong 2* https://orcid.org/0000-0002-1922-7155 Lee Hong Ghi 3 https://orcid.org/0000-0001-6145-4426 Kim Tae Min 4 https://orcid.org/0000-0002-3193-9745 Lee Yun Bin 2 https://orcid.org/0000-0002-2677-3189 Cho Eun Ju 2 https://orcid.org/0000-0002-0315-2080 Lee Jeong-Hoon 2 https://orcid.org/0000-0002-9128-3610 Yoon Jung-Hwan 2 https://orcid.org/0000-0001-9141-7773 Kim Yoon Jun 2 1 Department of Internal Medicine, Seoul Metropolitan Government-Seoul National University Boramae Medical Center, Seoul, Korea. 2 Department of Internal Medicine and Liver Research Institute, Seoul National University College of Medicine, Seoul, Korea. 3 Division of Hematology-Oncology, Department of Internal Medicine, Konkuk University Medical Center, Seoul, Korea. 4 Department of Internal Medicine, Seoul National University Hospital, Seoul, Korea. Address for Correspondence: Yoon Jun Kim, MD, PhD. Department of Internal Medicine and Liver Research Institute, Seoul National University College of Medicine, 101 Daehak-ro, Jongno-gu, Seoul 03080, Korea. yoonjun@snu.ac.kr *Heejoon Jang and Su Jong Yu contributed equally to this work. 17 7 2023 13 6 2023 38 28 e21619 10 2022 22 3 2023 © 2023 The Korean Academy of Medical Sciences. 2023 The Korean Academy of Medical Sciences https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Background Rituximab occasionally induces reactivation of hepatitis B virus (HBV) in patients with resolved HBV, at times with fatal consequences. The optimal duration of prophylactic antiviral therapy in this situation is unclear. We aimed to investigate the difference in HBV reactivation according to the duration of prophylactic tenofovir disoproxil fumarate (TDF) in patients with resolved HBV and receiving rituximab. Methods A multicenter, randomized, open-label, prospective study was conducted in hepatitis B surface antigen-negative and anti-HBc-positive non-Hodgkin’s lymphoma patients treated with rituximab-based chemotherapy. A total of 90 patients were randomized and received prophylactic TDF from the initiation of rituximab until 6 months (the 6-month group) or 12 months (the 12-month group) after the completion of rituximab. The primary outcome was the difference in HBV reactivation and the secondary outcomes were the difference in hepatitis flare and adverse events between the two groups. Results In an intention to treat (ITT) analysis, HBV reactivation occurred in 1 of 43 patients (2.3%; 95% confidence interval [CI], 0.41–12%) at a median of 13.3 months in the 6-month group and 2 of 41 patients (4.9%; 95% CI, 1.4–16%) at a median of 13.7 months in the 12-month group. In a per protocol (PP) analysis, HBV reactivation occurred in 1 of 18 patients (5.6%; 95% CI, 0.99–26%) at 13.3 months in the 6-month group and 1 of 13 patients (7.7%; 95% CI, 1.4–33%) at 9.7 months in the 12-month group. The cumulative incidence of HBV reactivation was not significantly different between the two groups in ITT and PP analyses (P = 0.502 and 0.795, respectively). The occurrence of adverse events was not significantly different between the two groups in ITT (9.3% in the 6-month group, 22.0% in the 12-month group, P = 0.193) and PP analyses (5.6% in the 6-month group, 7.7% in the 12-month group, P > 0.999). Conclusion Prophylactic TDF up to 6 months after completion of rituximab-based chemotherapy is sufficient in terms of the efficacy and safety of reducing HBV reactivation in patients with resolved HBV. Trial Registration ClinicalTrials.gov Identifier: NCT02585947 Graphical Abstract Hepatitis B Resolved Reactivation Tenofovir Prophylaxis Gilead Sciences https://doi.org/10.13039/100005564 0620153240 ==== Body pmcINTRODUCTION Hepatitis B virus (HBV) reactivation is a major complication in patients who have a history of HBV infection and receive immunosuppressive therapy or chemotherapy. HBV reactivation has a variety of clinical manifestations from asymptomatic to life-threatening liver failure caused by fulminant hepatitis.1 Patients with chronic hepatitis B (CHB) who receive immunosuppressive therapy or chemotherapy are at risk for HBV reactivation.2 Patients with resolved HBV, which is characterized by negative hepatitis B surface antigen (HBsAg) and positive immunoglobulin G subclass of antibody to hepatitis B core antigen (anti-HBc) on blood tests, may have a low level of HBV genome in their hepatocytes.34 Therefore, patients with resolved HBV may also experience HBV reactivation if their immune function decreases due to immunosuppressive therapy or chemotherapy.567 Since the activity of HBV is determined by an interaction between viral replication and the host immune response, the possibility of HBV reactivation varies depending on the type of immunosuppressive therapy or chemotherapy.8 Among patients using immunosuppressive therapy or chemotherapy, the risk of HBV reactivation is high in those on B lymphocyte-depleting agents such as rituximab.91011 Rituximab is a monoclonal antibody against cluster of differentiation 20 (CD20) on the surface of the B lymphocytes. B lymphocyte count decreases by 90% within 3 days of rituximab administration and recovers after 6 to 12 months.12 Rituximab is the basis of combination chemotherapy in the treatment of non-Hodgkin lymphoma.13 A recent meta-analysis showed that non-Hodgkin lymphoma patients treated with rituximab had a higher risk of HBV reactivation than those treated with anticancer drugs other than rituximab.14 Previous studies have suggested that HBsAg-positive patients (i.e., patients with CHB) should receive prophylactic antiviral therapy to reduce the risk of HBV reactivation when being treated with rituximab.15161718192021 Patients who have achieved seroclearance of HBsAg (HBsAg-negative) are less likely to develop HBV reactivation than HBsAg-positive patients; for HBsAg-negative and anti-HBc-positive patients (i.e., patients with resolved HBV), some guidelines recommend prophylactic antiviral therapy for 6 months or 12 months, while some guidelines recommend close monitoring rather than prophylactic antiviral therapy.22232425 Prolonged antiviral therapy leads to a variety of side effects and a decrease in adherence, consequently lowering its benefits compared to its risks.262728 However, there has been insufficient evidence for the optimal duration of prophylactic antiviral treatment when rituximab is administrated in HBsAg-negative and anti-HBc-positive patients. We aimed to compare the efficacy and safety of tenofovir disoproxil fumarate (TDF) in preventing HBV reactivation between 6-month and 12-month duration of prophylactic treatment in HBsAg-negative and anti-HBc-positive non-Hodgkin’s lymphoma patients treated with rituximab-based chemotherapy. METHODS Study design This study was designed as a multicenter, randomized, open-label, prospective study. Participants in the study were randomized at a 1:1 ratio and assigned to the 6-month group or the 12-month group by the Medical Research Collaborating Center (MRCC) at Seoul National University Hospital, a center independent from researchers. For the generation of randomization codes and the operation of randomization procedures, Interactive Web Response System was used, which was operated by MRCC. Patients and interventions Patients aged 18 to 80 years who were diagnosed with CD20 positive non-Hodgkin's lymphoma and who received rituximab-based chemotherapy were enrolled in this study. Participants had Eastern Cooperative Oncology Group performance status between 0 and 2, were HBsAg-negative, anti-HBc-positive, and anti-hepatitis C virus antibody (anti-HCV) negative, and had a serum creatinine level of < 2 mg/dL, hemoglobin ≥ 9 g/dL, absolute neutrophil counts ≥ 1,500 /μL, platelet counts ≥ 75,000 /μL (unless abnormalities were due to bone marrow involvement by lymphoma), a serum alanine aminotransferase (ALT) level of < 80 IU/mL, and a serum total bilirubin level of < 3.0 mg/dL (unless abnormalities were due to liver involvement by lymphoma or tumor lysis syndrome). Patients with Child-Pugh class C, autoimmune hepatitis, Wilson’s disease, other chronic liver diseases, galactose intolerance, Lapp lactase deficiency, glucose-galactose malabsorption, pregnancy, lactation, or hypersensitivity to TDF were excluded. The 6-month group received oral TDF 300 mg (Gilead, Foster City, CA, USA) daily during the period of rituximab treatment and for an additional 6 months after the completion of rituximab treatment. The 12-month group received oral TDF 300 mg daily during the period of rituximab treatment and for an additional 12 months after the completion of rituximab treatment. During the planning of the study, there was limited evidence concerning the efficacy of prophylactic TDF in patients with resolved HBV infection and treated with rituximab. A clinical trial reported that rituximab has an immunosuppressive effect that lasts for up to six months after treatment, at which point the immune system slowly recovers over the next three to six months.29 One study found that resolved HBV patients treated with rituximab had a risk of HBV reactivation of about 23% in the first 6 months and up to 19% thereafter.10 Another study reported that resolved HBV patients treated with rituximab did not experience HBV reactivation while maintaining prophylactic TDF.30 Based on the previous studies, we assumed that the HBV reactivation rate in the 6-month group is 9%, approximately half of 19%, because the group is at risk of HBV reactivation since prophylactic TDF is discontinued during the period when immune function is slowly regaining but not yet fully recovered. On the other hand, we hypothesized that the HBV reactivation rate in the 12-month group is 0.01% since prophylactic TDF is used adequately during the period of rituximab-induced immunosuppression. To test the null hypothesis by power of 80% and type I error of 0.05 on both sides, assuming that the reactivation rate of HBV is 9% for the 6-month group and 0.01% for the 12-month group, each group required 41 participants.3132 In consideration of the 10% dropout rate, a total of 90 participants were needed. Assessments and outcomes Blood tests including ALT, albumin, total bilirubin, creatinine, platelet, HBsAg, anti-hepatitis B surface antibody (anti-HBs), hepatitis B e antigen (HBeAg), anti-HCV, anti-human immunodeficiency virus antigen (anti-HIV), and quantitative HBV DNA were performed at the screening visit. Liver function tests and HBV DNA tests were performed 4 times during chemotherapy: the first day of chemotherapy, 4 and 12 weeks from the start of chemotherapy, and the last day of chemotherapy. After the completion of rituximab-based chemotherapy, liver function tests and HBV DNA tests were performed 6 times: every 12 weeks up to 72 weeks from the end of chemotherapy. HBsAg and anti-HBs were tested every 24 weeks after the discontinuation of TDF. If hepatitis flare occurred, liver function tests were performed every week regardless of the chemotherapy schedule. The primary outcome was the development of HBV reactivation. HBV reactivation was defined as HBV DNA rising by more than 2,000 IU/mL compared to baseline and/or HBsAg seroreversion to HBsAg positive during the period of participation in this study.2333 The secondary outcomes were the development of hepatitis flare and adverse events. Hepatitis flare was defined as ALT of ≥ 100 IU/L with HBV reactivation. The outcomes were analyzed by modified intention to treat (ITT) and per protocol (PP) approaches. The ITT population was composed of patients who received at least one dose of rituximab and TDF. The PP population was composed of patients who completed 72 weeks of follow-up after the completion of chemotherapy with rituximab. Statistical analyses Continuous data were described as median with interquartile range. Categorical data were described as the frequency with percentage. Student’s t-test or Mann-Whitney U test was performed on continuous variables. The χ2 test or Fisher’s exact test was performed on categorical variables. The cumulative incidence of HBV DNA reactivation and hepatitis flare were compared between the 6- and 12-month groups using the Kaplan-Meier model and log-rank test. The rates of HBV reactivation and 95% confidence intervals (CI) were calculated in the 6-month group and 12-month group. In addition, it was also examined whether the groups differed in terms of the HBV reactivation rates and 95% CIs.34 The odds ratio and 95% CI were calculated for the 12-month group compared to the 6-month group. The primary outcome was further analyzed using the missing equals failure method.35 All statistical analyses were done by a two-sided test. A P value lower than 0.05 was considered statistically significant. R version 3.6.3 (http://www.r-project.org/; R Foundation, Vienna, Austria) was used for all statistical analyses. Ethics statement This study was approved by the Institutional Review Boards of Seoul National Hospital (H-1506-020-678) and Konkuk University Medical Center (KUH1010712). The study was conducted in compliance with the ethical principles in the Declaration of Helsinki (revised in 2013) and registered at ClinicalTrials.gov (NCT02585947). All participants provided written informed consent prior to enrolling in the study. RESULTS Baseline characteristics A total of 90 non-Hodgkin lymphoma patients from November 2015 to October 2017 were enrolled and randomized. After randomization, three patients decided to receive chemotherapy without rituximab, and three other patients withdrew consent before starting prophylactic TDF treatment. Consequently, 43 patients were assigned to the 6-month group and 41 patients were assigned to the 12-month group for ITT analysis (Fig. 1). During the study period, 33 patients were lost to follow-up, 16 patients withdrew consent, and 4 patients died. Finally, PP analysis was performed with 18 patients in the 6-month group and 13 patients in the 12-month group (Fig. 1). Fig. 1 Consolidated Standards of Reporting Trials diagram. ITT = intention to treat, PP = per protocol. HBV DNA was undetectable, HBeAg was negative, and anti-HIV was negative at baseline in all patients included in the ITT or PP analysis. There were no significant differences in age, sex, anti-HBs positivity, anti-HBs titer, ALT, albumin, total bilirubin, creatinine, platelet count, fibrosis-4 score at baseline, type of lymphoma, cycles of chemotherapy, cumulative dose of rituximab, and regimen of chemotherapy between the 6-month group and the 12-month group in ITT analysis (all P > 0.05; Table 1). There was no significant difference in all baseline characteristics except age (median 62.5 years in the 6-month group, median 57.0 months in the 12-month group; P = 0.021; all other P > 0.05; Supplementary Table 1) in PP analysis. Table 1 Baseline characteristics in the population for intention to treat analysis Variables Total (N = 84) Duration of prophylactic tenofovira P value 6-mon group (n = 43) 12-mon group (n = 41) Age, yr 64.5 (57.0–71.2) 64.0 (57.5–72.0) 65.0 (57.0–71.0) 0.588 Male 39 (46.4) 20 (46.5) 19 (46.3) > 0.990 Anti-HBs, positive 70 (84.3) 35 (81.4) 35 (87.5) 0.644 Anti-HBs titer, mIU/mL 54.9 (18.0–179.2) 51.5 (17.7–141.7) 58.2 (18.5–247.3) 0.505 ALT, IU/L 19.0 (14.8–26.2) 18.0 (13.0–23.5) 20.0 (15.0–28.0) 0.202 Albumin, g/dL 4.0 (3.6–4.2) 4.0 (3.8–4.3) 3.9 (3.6–4.2) 0.487 Total bilirubin, mg/dL 0.6 (0.4–0.8) 0.7 (0.4–0.8) 0.6 (0.4–0.7) 0.284 Creatinine, mg/dL 0.8 (0.7–0.9) 0.8 (0.7–0.9) 0.8 (0.7–0.9) 0.792 Platelet count, × 103/μL 257 (205–306) 254 (197–293) 261 (213–308) 0.758 FIB-4 score 1.2 (0.9–1.9) 1.2 (0.9–2.1) 1.2 (0.9–1.7) 0.968 Type of lymphoma 0.783 Diffuse large B cell lymphoma 73 (86.9) 38 (88.4) 35 (85.4) Follicular lymphoma 6 (7.1) 2 (4.7) 4 (9.8) Mantle-cell lymphoma 2 (2.4) 1 (2.3) 1 (2.4) Small lymphocytic lymphoma 3 (3.6) 2 (4.7) 1 (2.4) Regimen of chemotherapy 0.432 (SC) R-CHOP 4 (4.8) 3 (7.0) 1 (2.4) (SC) R-CVP 6 (7.1) 3 (7.0) 3 (7.3) (SC) R-miniCHOP 2 (2.4) 0 (0.0) 2 (4.9) FCR 3 (3.6) 2 (4.7) 1 (2.4) R-CHOP 41 (48.8) 18 (41.9) 23 (56.1) R-CVP 8 (9.5) 5 (11.6) 3 (7.3) R-EPOCH 1 (1.2) 1 (2.3) 0 (0.0) R-hyperCVAD 2 (2.4) 1 (2.3) 1 (2.4) R-HyperCVAD/MA 1 (1.2) 0 (0.0) 1 (2.4) R-ICE 1 (1.2) 0 (0.0) 1 (2.4) R-miniCHOP 3 (3.6) 3 (7.0) 0 (0.0) R-MVP 13 (15.5) 8 (18.6) 5 (12.2) Data are expressed as number (%) or median with interquartile range. Anti-HBs = anti-hepatitis B surface antibody, ALT = alanine transaminase, FIB-4 = fibrosis-4, SC = subcutaneous, R-CHOP = rituximab with cyclophosphamide, doxorubicin, vincristine, and prednisone, R-CVP = rituximab with cyclophosphamide, vincristine and prednisolone, R-miniCHOP = rituximab and reduced dose cyclophosphamide, doxorubicin, vincristine, and prednisone, FCR = fludarabine, cyclophosphamide and rituximab, R-EPOCH = rituximab, etoposide phosphate, prednisone, vincristine sulfate, cyclophosphamide and doxorubicin hydrochloride, R-hyperCVAD = rituximab plus fractionated cyclophosphamide, vincristine, doxorubicin and dexamethasone, R-hyperCVAD/MA = rituximab plus fractionated cyclophosphamide, vincristine sulfate, doxorubicin and dexamethasone alternating with high-dose methotrexate and cytarabine, R-ICE = rituximab, ifosfamide, carboplatin and etoposide, R-MVP = rituximab, methotrexate, vincristine, procarbazine. aReceived prophylactic tenofovir disoproxil fumarate 300 mg daily from the start of rituximab for 6 months (6-month group) or 12 months (12-month group) after completion of rituximab. Primary outcome In ITT analysis, HBV reactivation occurred in 1 of 43 patients (2.3%; 95% CI, 0.41–12%) in the 6-month group and 2 of 41 patients (4.9%; 95% CI, 1.4–16%) in the 12-month group during the study period (P = 0.966; Table 2). The difference in HBV reactivation rate between the groups was 2.6% (95% CI, −5.4–11%). The time from the initial administration of rituximab to HBV reactivation was 13.3 months in the 6-month group and a median of 13.7 months in the 12-month group (P > 0.999; Table 2). There was no significant difference in the cumulative incidence of HBV reactivation between the two groups (log-rank test P = 0.502; Fig. 2A). HBsAg seroreversion occurred at 17.6 months of enrollment in 1 patient in the 12-month group who self-discontinued prophylactic TDF and changed to entecavir 1 month after starting the drug. Elevations of HBV DNA levels above 2,000 IU/mL were observed at 9.7 months from enrollment in 1 patient in the 12-month group who missed 9 days of prophylactic TDF administration and at 13.3 months from enrollment in 1 patient in the 6-month group who completed prophylactic TDF administration. In both patients described above, AST and ALT levels were normal, and HBV DNA was not detected at the next visit. Compared to the 6-month group, the 12-month group had an odds ratio of 2.0 and a 95% CI of 0.16 to 65 in HBV reactivation. Additionally, there was no difference in HBV reactivation between the 6-month group and the 12-month group in ITT using the missing equals failure method. (P = 0.692; Supplementary Table 2). Table 2 HBV reactivation according to the duration of prophylactic tenofovir administration after rituximab-based chemotherapy Variables Total Duration of prophylactic tenofovira P value 6-mon group 12-mon group Intention to treat analysis No. of patients 84 43 41 HBV reactivation 3 (3.6, 1.2–10) 1 (2.3, 0.41–12) 2 (4.9, 1.4–16) 0.966 Time to HBV reactivation, months 13.3 (11.5–15.5) 13.3 (13.3–13.3) 13.7 (11.7–15.7) > 0.999 Per protocol analysis No. of patients 31 18 13 HBV reactivation 2 (6.5, 1.8–21) 1 (5.6, 0.99–26) 1 (7.7, 1.4–33) > 0.999 Time to HBV reactivation, months 11.5 (10.6–12.4) 13.3 (13.3–13.3) 9.7 (9.7–9.7) 0.317 Data with parentheses are expressed as number (%, 95% confidence interval) or median (interquartile range). HBV reactivation was defined as HBV DNA rising by more than 2,000 IU/mL compared to baseline and/or HBsAg seroreversion to HBsAg positive during the period of participation in this study. HBV = hepatitis B virus, HBsAg = hepatitis B surface antigen. aReceived prophylactic tenofovir disoproxil fumarate 300 mg daily from the start of rituximab up to 6 months (6-month group) or 12 months (12-month group) after completion of rituximab. Fig. 2 Cumulative incidence of hepatitis B virus reactivation according to the duration of prophylactic tenofovir administration after rituximab-based chemotherapy in (A) intention to treat analysis and (B) per protocol analysis. In PP analysis, HBV reactivation occurred in 1 of 18 patients (5.6%; 95% CI, 0.99–26%) in the 6-month group and 1 of 13 patients (7.7%; 95% CI, 1.4–33%) in the 12-month group during the follow-up (P > 0.999; Table 2). The difference in HBV reactivation rate between the groups was 2.1% (95% CI, −16–20%). The time from the initial administration of rituximab to HBV reactivation was 13.3 months in the 6-month group and 9.7 months in the 12-month group (P = 0.317; Table 2). There was no significant difference in the cumulative incidence of HBV reactivation between the two groups (log-rank test P = 0.795; Fig. 2B). The 12-month group had an odds ratio of 1.4 and a 95% CI of 0.03 to 58 in terms of HBV reactivation compared to the 6-month group. Secondary outcomes Neither the ITT analysis nor the PP analysis group experienced hepatitis flare during the participation period. In ITT analysis, the frequency of adverse events was not significantly different between the two groups (4 of 43, 9.3% in the 6-month group; 9 of 41, 22.0% in the 12-month group; P = 0.193; Supplementary Table 3). There were no significant differences in severity and type of adverse events between the two groups (all P > 0.05; Supplementary Table 3). In PP analysis, the number of adverse events was not significantly different between the two groups (1 of 18, 5.6% in the 6-month group; 1 of 13, 7.7% in the 12-month group; P > 0.999; Supplementary Table 4). There were no significant differences in severity and type of adverse events between the two groups (all P > 0.05; Supplementary Table 4). DISCUSSION This study was designed as a multicenter, randomized, open-label, prospective study to investigate the optimal duration of TDF to prevent HBV reactivation in patients with HBsAg-negative and anti-HBc-positive non-Hodgkin’s lymphoma using rituximab-based chemotherapy. Based on the report that 6 to 12 months is needed for the recovery of B cell function after administration of rituximab and the guidelines that recommend 6 or 12 months of prophylactic antiviral therapy in patients with resolved or occult HBV undergoing chemotherapy,1223 we compared 6 months to 12 months of prophylactic antiviral therapy to determine the optimal treatment duration. The 6-month group and the 12-month group received TDF during the same period as the rituximab treatment, but the durations of additional prophylactic TDF after completion of rituximab treatment were 6 and 12 months, respectively. This study found no significant difference in HBV reactivation and adverse events between the 6-month group and the 12-month group. HBV reactivation has been variously defined, such as elevation of HBV DNA or reappearance of HBV DNA, HBsAg, HBeAg, or anti-HBc IgM in previous studies.910333637 In a randomized controlled trial evaluating the efficacy of prophylactic entecavir in patients with resolved HBV and lymphoma using rituximab, Huang et al.33 defined HBV reactivation as when the HBV viral load is increased by 2,000 IU/mL for two consecutive times. In a randomized prospective investigation evaluating the efficacy of prophylactic TDF in anti-HBc-positive hematologic malignancy patients using rituximab-based chemotherapy, Buti et al.38 defined HBV reactivation as HBsAg detection, HBV DNA detection, or HBV DNA level increased by 1 log10 IU/mL or more compared to baseline. The definition of HBV reactivation in a recent guideline of the American Association for the Study of Liver Diseases (AASLD) is loss of immune control against HBV when receiving immunosuppressive treatment in anti-HBc-positive and HBsAg-positive or -negative patients, which means either the HBV DNA level rises above the basal level or seroreversion of HBsAg (transition from negative to positive).23 We defined HBV reactivation as an increase in HBV DNA levels of more than 2,000 IU/mL and/or HBsAg seroreversion compared to baseline, which was in line with the definitions in the literature.2333 Various strategies to prevent HBV reactivation have been proposed in HBsAg-negative and anti-HBc-positive patients receiving rituximab. AASLD recommends that HBsAg-negative and anti-HBc-positive patients undergo anti-HBV prophylaxis during immunosuppression treatment and for up to 12 months after completion of immunosuppression when receiving anti-CD20 antibodies such as rituximab.23 The European Association for the Study of the Liver recommends that HBsAg-negative and anti-HBc-positive patients receiving rituximab undergo HBV prophylaxis for at least 18 months after discontinuation of rituximab.22 The Asian Pacific Association for the Study of the Liver noted that while HBsAg-negative and anti-HBc-positive patients receiving rituximab are at risk for HBV reactivation, further study is needed on the comparison of prophylactic antiviral agents and HBV DNA monitoring.24 Some reports in HBsAg-negative and anti-HBc-positive lymphoma patients receiving rituximab showed that the strategy of close monitoring without prophylactic antiviral agents and on-demand antiviral administration was also successful.103339 At least 3 months of prophylactic antiviral treatment after completion of rituximab is required to reduce HBV reactivation in resolved HBV patients receiving rituximab.333840 Huang et al.33 reported that HBV reactivation was significantly lower in the group treated with prophylactic entecavir for up to 3 months from completion of the chemotherapy compared to the control group in CD20+ lymphoma patients with resolved HBV. In addition, the effect was maintained up to 18 months; the cumulative HBV reactivation rates were 4.3% in the prophylactic entecavir group and 25.9% in the control group at 18 months from completion of chemotherapy. Buti et al.38 reported that HBV reactivation at 18 months was 0% (0 out of 33 patients) when prophylactic TDF was administered for 18 months from the start of chemotherapy in resolved HBV patients with hematologic cancer treated with rituximab, whereas HBV reactivation at 18 months in the close monitoring group was 10.7% (3 out of 28 patients). Another study by Kusumoto et al.40 in HBsAg-negative and anti-HBc-positive non-Hodgkin B-cell lymphoma patients receiving rituximab or obinutuzumab showed that the HBV reactivation rate was 10.8% (25 out of 232 patients) without prophylactic antiviral drugs and 2.1% (2 out of 94 patients) with prophylactic antiviral drugs, all of which were prophylactic lamivudine, while patients on prophylactic entecavir or tenofovir did not develop HBV reactivation. Antiviral drugs with a high genetic barrier may effectively prevent HBV reactivation for a period shorter than 12 months in resolved HBV patients receiving rituximab. Entecavir, an antiviral drug with a high genetic barrier, for only 3 months of prophylaxis maintained the effect of preventing HBV reactivation for up to 18 months in resolved HBV patients treated with rituximab.33 TDF, another antiviral drug with a high genetic barrier, did not differ in the effectiveness and safety of preventing HBV reactivation between the 6-month group and the 12-month group in our study. Furthermore, none of the patients with resolved HBV using rituximab developed HBV-related hepatitis, even though they used prophylactic TDF for up to 6 months after completion of chemotherapy in our study. Since prolonged prophylactic antiviral treatment may result in decreased adherence and increased side effects (e.g., decrease in renal function and bone mineral density) and costs, evaluation of the appropriate administration period is crucial.26272841 Our results suggested that prophylactic TDF up to 6 months after completion of chemotherapy is adequately effective in reducing HBV reactivation in resolved HBV patients using rituximab. Recently, novel biomarkers have been proposed for hepatitis B patients, and some of these biomarkers, such as Hepatitis B core-related antigen(HBcrAg), may be useful in evaluating the prognosis of resolved HBV patients. When HBV infects hepatocytes, the HBV genome forms covalently closed circular DNA (cccDNA) from relaxed circular DNA or integrates into the genome of the host hepatocyte.4243 Even though HBsAg clearance is achieved by the host’s immune response, some of the HBV genomes that have formed cccDNA or integrated into the host genome may remain.34 There are some reports that HBV DNA remains in the host's liver for years to decades and leads to HBV reactivation if the host’s immune system is compromised in serologically resolved HBV patients.32244 The level of HBcrAg is related to the level of intrahepatic cccDNA. Therefore, in resolved HBV patients for whom HBV DNA cannot be measured in serum, the level of HBcrAg may indicate the burden of HBV.45 Furthermore, high levels of HBcrAg have been associated with a greater risk of hepatocellular carcinoma.46 A recent study investigated the risk of hepatocellular carcinoma by assessing where HBV DNA is integrated into the host genome.47 To our best knowledge, this is the first multicenter, randomized, open-label, prospective clinical trial to investigate the difference in HBV reactivation according to the duration of prophylactic TDF in HBsAg-negative and anti-HBc-positive patients using rituximab-based chemotherapy. Previous studies in resolved HBV patients using rituximab investigated the difference in HBV reactivation by focusing on the use vs. nonuse of prophylactic antiviral agents, but our study is the first randomized controlled clinical trial to focus on the difference in the effect of HBV reactivation according to the duration of prophylactic antiviral drug use. There are several limitations in this study. First, a large proportion of enrolled patients in this study were lost to follow-up and so the power to see statistical differences in outcomes was lost. Poor adherence to antiviral prophylaxis may be explained by the patients not being aware that they are at risk for hepatitis. Patients’ chemotherapy schedules, disease progression, and experience of adverse effects to medication may also be factors in discontinuing treatment. Greater non-adherence in patients on long-term prophylactic therapy underscores the result in our study that short-term prophylactic antiviral treatment was sufficient in terms of efficacy and safety. Second, since this study was designed as a superiority test, the interpretation of the results is limited. To validate the results of this study, a non-inferiority-designed study is needed to compare the efficacy of 6- and 12-month use of prophylactic TDF in patients with resolved HBV using rituximab. In conclusion, prophylactic TDF up to 6 months after completion of rituximab-based chemotherapy is sufficient in terms of the efficacy and safety of reducing HBV reactivation in HBsAg-negative and anti-HBc-positive non-Hodgkin lymphoma patients treated with rituximab-based chemotherapy. ACKNOWLEDGMENTS We would like to acknowledge The Liver Meeting 2022, The Liver Week 2022, and Seoul International Digestive Disease Symposium 2023 for providing the opportunity to present this research as a poster and oral presentation. The feedback and discussions during the conference greatly contributed to the development of this manuscript. SUPPLEMENTARY MATERIALS Supplementary Table 1 Baseline characteristics in the population for per protocol analysis Supplementary Table 2 HBV reactivation according to the duration of prophylactic tenofovir administration after rituximab-based chemotherapy using the missing equals failure method Supplementary Table 3 AEs according to the duration of prophylactic tenofovir administration after rituximab-based chemotherapy in the population for intention to treat analysis Supplementary Table 4 Adverse events according to the duration of prophylactic tenofovir administration after rituximab-based chemotherapy in the population for per protocol analysis Supplementary Data 1 Funding: This work was supported by funding from Gilead Sciences (grant number 0620153240). Disclosure: Su Jong Yu received compensation for lectures from Bayer HealthCare Pharmaceuticals. Jeong-Hoon Lee received compensation for lectures from GreenCross Cell, Daewoong Pharmaceuticals and Gilead Korea; Jung-Hwan Yoon received a research grant from Bayer HealthCare Pharmaceuticals, Bukwang Pharmaceuticals and Daewoong Pharmaceuticals; and Yoon Jun Kim received research grants from BTG, AstraZeneca, Gilead Sciences, Daewoong, Bayer, Samjin, and Yuhan Pharmaceuticals. Data Sharing Statement: Data sharing statement is provided in Supplementary Data 1. Author Contributions: Conceptualization: Kim YJ. Data curation: Jang H, Yu SJ, Lee HG, Kim TM, Lee YB, Cho EJ, Lee JH, Yoon JH, Kim YJ. Formal analysis: Jang H. Funding acquisition: Kim YJ. Investigation: Jang H. Methodology: Kim YJ. Project administration: Kim YJ. Supervision: Kim YJ. Writing - original draft: Jang H, Yu SJ. Writing - review & editing: Jang H, Yu SJ, Lee HG, Kim TM, Lee YB, Cho EJ, Lee JH, Yoon JH, Kim YJ. ==== Refs 1 Lau JY Lai CL Lin HJ Lok AS Liang RH Wu PC Fatal reactivation of chronic hepatitis B virus infection following withdrawal of chemotherapy in lymphoma patients Q J Med 1989 73 270 911 917 2629023 2 Lok AS Liang RH Chiu EK Wong KL Chan TK Todd D Reactivation of hepatitis B virus replication in patients receiving cytotoxic therapy. Report of a prospective study Gastroenterology 1991 100 1 182 188 1983820 3 Mason AL Xu L Guo L Kuhns M Perrillo RP Molecular basis for persistent hepatitis B virus infection in the liver after clearance of serum hepatitis B surface antigen Hepatology 1998 27 6 1736 1742 9620351 4 Murakami Y Minami M Daimon Y Okanoue T Hepatitis B virus DNA in liver, serum, and peripheral blood mononuclear cells after the clearance of serum hepatitis B virus surface antigen J Med Virol 2004 72 2 203 214 14695661 5 Yeo W Chan PK Zhong S Ho WM Steinberg JL Tam JS Frequency of hepatitis B virus reactivation in cancer patients undergoing cytotoxic chemotherapy: a prospective study of 626 patients with identification of risk factors J Med Virol 2000 62 3 299 307 11055239 6 Conjeevaram HS Lok AS Occult hepatitis B virus infection: a hidden menace? Hepatology 2001 34 1 204 206 11431752 7 Hoofnagle JH Reactivation of hepatitis B Hepatology 2009 49 5 Suppl S156 S165 19399803 8 Loomba R Liang TJ Hepatitis B reactivation associated with immune suppressive and biological modifier therapies: current concepts, management strategies, and future directions Gastroenterology 2017 152 6 1297 1309 28219691 9 Hsu C Tsou HH Lin SJ Wang MC Yao M Hwang WL Chemotherapy-induced hepatitis B reactivation in lymphoma patients with resolved HBV infection: a prospective study Hepatology 2014 59 6 2092 2100 24002804 10 Seto WK Chan TS Hwang YY Wong DK Fung J Liu KS Hepatitis B reactivation in patients with previous hepatitis B virus exposure undergoing rituximab-containing chemotherapy for lymphoma: a prospective study J Clin Oncol 2014 32 33 3736 3743 25287829 11 Pan J Yao T Cheng H Zhu Y Wang Y B lymphocyte-mediated humoral immunity in the pathogenesis of chronic hepatitis B infection Liver Res 2020 4 3 124 128 12 Onrust SV Lamb HM Balfour JA Rituximab Drugs 1999 58 1 79 88 10439931 13 Bowzyk Al-Naeeb A Ajithkumar T Behan S Hodson DJ Non-Hodgkin lymphoma BMJ 2018 362 k3204 30135071 14 Dong HJ Ni LN Sheng GF Song HL Xu JZ Ling Y Risk of hepatitis B virus (HBV) reactivation in non-Hodgkin lymphoma patients receiving rituximab-chemotherapy: a meta-analysis J Clin Virol 2013 57 3 209 214 23562041 15 Tsutsumi Y Kanamori H Mori A Tanaka J Asaka M Imamura M Reactivation of hepatitis B virus with rituximab Expert Opin Drug Saf 2005 4 3 599 608 15934864 16 Lok AS McMahon BJ Chronic hepatitis B Hepatology 2007 45 2 507 539 17256718 17 Lai CL Chien RN Leung NW Chang TT Guan R Tai DI A one-year trial of lamivudine for chronic hepatitis B N Engl J Med 1998 339 2 61 68 9654535 18 Andreone P Caraceni P Grazi GL Belli L Milandri GL Ercolani G Lamivudine treatment for acute hepatitis B after liver transplantation J Hepatol 1998 29 6 985 989 9875646 19 Yeo W Chan PK Ho WM Zee B Lam KC Lei KI Lamivudine for the prevention of hepatitis B virus reactivation in hepatitis B s-antigen seropositive cancer patients undergoing cytotoxic chemotherapy J Clin Oncol 2004 22 5 927 934 14990649 20 Kohrt HE Ouyang DL Keeffe EB Systematic review: lamivudine prophylaxis for chemotherapy-induced reactivation of chronic hepatitis B virus infection Aliment Pharmacol Ther 2006 24 7 1003 1016 16984494 21 Li YH He YF Jiang WQ Wang FH Lin XB Zhang L Lamivudine prophylaxis reduces the incidence and severity of hepatitis in hepatitis B virus carriers who receive chemotherapy for lymphoma Cancer 2006 106 6 1320 1325 16470607 22 European Association for the Study of the Liver EASL 2017 clinical practice guidelines on the management of hepatitis B virus infection J Hepatol 2017 67 2 370 398 28427875 23 Terrault NA Lok AS McMahon BJ Chang KM Hwang JP Jonas MM Update on prevention, diagnosis, and treatment of chronic hepatitis B: AASLD 2018 hepatitis B guidance Hepatology 2018 67 4 1560 1599 29405329 24 Sarin SK Kumar M Lau GK Abbas Z Chan HL Chen CJ Asian-Pacific clinical practice guidelines on the management of hepatitis B: a 2015 update Hepatol Int 2016 10 1 1 98 25 Yim HJ Kim JH Park JY Yoon EL Park H Kwon JH Comparison of clinical practice guidelines for the management of chronic hepatitis B: When to start, when to change, and when to stop Clin Mol Hepatol 2020 26 4 411 429 32854458 26 Xu K Liu LM Farazi PA Wang H Rochling FA Watanabe-Galloway S Adherence and perceived barriers to oral antiviral therapy for chronic hepatitis B Glob Health Action 2018 11 1 1433987 29447614 27 Lee SB Jeong J Park JH Jung SW Jeong ID Bang SJ Low-level viremia and cirrhotic complications in patients with chronic hepatitis B according to adherence to entecavir Clin Mol Hepatol 2020 26 3 364 375 32466635 28 Kayaaslan B Guner R Adverse effects of oral antiviral therapy in chronic hepatitis B World J Hepatol 2017 9 5 227 241 28261380 29 McLaughlin P Grillo-López AJ Link BK Levy R Czuczman MS Williams ME Rituximab chimeric anti-CD20 monoclonal antibody therapy for relapsed indolent lymphoma: half of patients respond to a four-dose treatment program J Clin Oncol 1998 16 8 2825 2833 9704735 30 Koskinas JS Deutsch M Adamidi S Skondra M Tampaki M Alexopoulou A The role of tenofovir in preventing and treating hepatitis B virus (HBV) reactivation in immunosuppressed patients. A real life experience from a tertiary center Eur J Intern Med 2014 25 8 768 771 25037900 31 Casagrande JT Pike MC Smith PG An improved approximate formula for calculating sample sizes for comparing two binomial distributions Biometrics 1978 34 3 483 486 719125 32 Chow SC Wang H Shao J Sample Size Calculations in Clinical Research Boca Raton, FL, USA CRC Press 2007 33 Huang YH Hsiao LT Hong YC Chiou TJ Yu YB Gau JP Randomized controlled trial of entecavir prophylaxis for rituximab-associated hepatitis B virus reactivation in patients with lymphoma and resolved hepatitis B J Clin Oncol 2013 31 22 2765 2772 23775967 34 Simon R Confidence intervals for reporting results of clinical trials Ann Intern Med 1986 105 3 429 435 3740683 35 Papp KA Fonjallaz P Casset-Semanaz F Krueger JG Wittkowski KM Analytical approaches to reporting long-term clinical trial data Curr Med Res Opin 2008 24 7 2001 2008 18534049 36 Shibolet O Ilan Y Gillis S Hubert A Shouval D Safadi R Lamivudine therapy for prevention of immunosuppressive-induced hepatitis B virus reactivation in hepatitis B surface antigen carriers Blood 2002 100 2 391 396 12091327 37 Yeo W Chan TC Leung NW Lam WY Mo FK Chu MT Hepatitis B virus reactivation in lymphoma patients with prior resolved hepatitis B undergoing anticancer therapy with or without rituximab J Clin Oncol 2009 27 4 605 611 19075267 38 Buti M Manzano ML Morillas RM García-Retortillo M Martín L Prieto M Randomized prospective study evaluating tenofovir disoproxil fumarate prophylaxis against hepatitis B virus reactivation in anti-HBc-positive patients with rituximab-based regimens to treat hematologic malignancies: the Preblin study PLoS One 2017 12 9 e0184550 28898281 39 Masarone M De Renzo A La Mura V Sasso FC Romano M Signoriello G Management of the HBV reactivation in isolated HBcAb positive patients affected with non Hodgkin lymphoma BMC Gastroenterol 2014 14 1 31 24533834 40 Kusumoto S Arcaini L Hong X Jin J Kim WS Kwong YL Risk of HBV reactivation in patients with B-cell lymphomas receiving obinutuzumab or rituximab immunochemotherapy Blood 2019 133 2 137 146 30341058 41 Buti M Riveiro-Barciela M Esteban R Long-term safety and efficacy of nucleo(t)side analogue therapy in hepatitis B Liver Int 2018 38 Suppl 1 84 89 42 Tuttleman JS Pourcel C Summers J Formation of the pool of covalently closed circular viral DNA in hepadnavirus-infected cells Cell 1986 47 3 451 460 3768961 43 Yang W Summers J Integration of hepadnavirus DNA in infected liver: evidence for a linear precursor J Virol 1999 73 12 9710 9717 10559280 44 Rehermann B Ferrari C Pasquinelli C Chisari FV The hepatitis B virus persists for decades after patients’ recovery from acute viral hepatitis despite active maintenance of a cytotoxic T-lymphocyte response Nat Med 1996 2 10 1104 1108 8837608 45 Inoue T Tanaka Y Novel biomarkers for the management of chronic hepatitis B Clin Mol Hepatol 2020 26 3 261 279 32536045 46 Wu JW Kao JH Tseng TC Three heads are better than two: hepatitis B core-related antigen as a new predictor of hepatitis B virus-related hepatocellular carcinoma Clin Mol Hepatol 2021 27 4 524 534 33618507 47 Jang JW Kim JS Kim HS Tak KY Nam H Sung PS Persistence of intrahepatic hepatitis B virus DNA integration in patients developing hepatocellular carcinoma after hepatitis B surface antigen seroclearance Clin Mol Hepatol 2021 27 1 207 218 33317255

PMC010xxxxxx/PMC10353912.txt

==== Front J Korean Med Sci J Korean Med Sci JKMS Journal of Korean Medical Science 1011-8934 1598-6357 The Korean Academy of Medical Sciences 10.3346/jkms.2023.38.e223 Brief Communication Infectious Diseases, Microbiology & Parasitology The Impact of Entry Restrictions on the Spread of Severe Acute Respiratory Syndrome Coronavirus Variants Between 2021 and 2022 https://orcid.org/0000-0003-1855-9649 Hong Jinwook 12* https://orcid.org/0000-0002-6898-9340 Park Ae Kyung 3* https://orcid.org/0000-0003-4968-5031 Radnaabaatar Munkhzul 1* https://orcid.org/0000-0001-6784-8004 Kim Eun-Jin 3 https://orcid.org/0000-0002-4478-3794 Kim Dong Wook 45 https://orcid.org/0000-0002-4856-3668 Jung Jaehun 12 1 Artificial Intelligence and Big-Data Convergence Center, Gil Medical Center, Gachon University College of Medicine, Incheon, Korea. 2 Department of Preventive Medicine, Gachon University College of Medicine, Incheon, Korea. 3 Division of Emerging Infectious Diseases, Bureau of Infectious Disease Diagnosis Control, Korea Disease Control and Prevention Agency, Cheongju, Korea. 4 Department of Information and Statistics, Research Institute of Natural Science, Gyeongsang National University, Jinju, Korea. 5 Department of Bio & Medical Bigdata (BK21 Plus), Gyeongsang National University, Jinju, Korea. Address for Correspondence: Jaehun Jung, MD, PhD. Department of Preventive Medicine, Gachon University College of Medicine, 38-13 Dokjeom-ro 3-beon-gil, Namdong-gu, Incheon 21565, Republic of Korea. eastside1st@gmail.com Address for Correspondence: Dong Wook Kim, PhD. Department of Information and Statistics, Research Institute of Natural Science, Gyeongsang National University, 501 Jinju-daero, Jinju 52858, Republic of Korea. kimdw2269@gmail.com *Jinwook Hong, Ae Kyung Park, and Munkhzul Radnaabaatar contributed equally to this work as first authors. 17 7 2023 21 6 2023 38 28 e22306 12 2022 25 5 2023 © 2023 The Korean Academy of Medical Sciences. 2023 The Korean Academy of Medical Sciences https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. To contain the surge of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the South Korean government has implemented non-pharmacological interventions as well as border restrictions. The efficacy of entry restrictions should be evaluated to facilitate their preparation for new variants of SARS-CoV-2. This study explored the impact of border policy changes on overseas entrants and local cases of SARS-CoV-2 variants. Data from the Korea Disease Control and Prevention Agency randomly collected between April 11, 2021 and August 20, 2022 were evaluated using the Granger causality model. The results showed that the outbreak gap of delta variants between international and domestic cases was 10 weeks, while that of omicron variants was approximately 2 weeks, meaning that the quarantine policy helped contain delta variants rather than more transmissible variants. It is recommended that countries implement quarantine policies based on particular purposes accounting for the specific features of different variants to avoid potential negative impacts on the economy. Graphical Abstract SARS-CoV-2 Overseas Entrants Border Policy Variants Ministry of Science and ICT, South Korea https://doi.org/10.13039/501100014188 National Research Foundation of Korea https://doi.org/10.13039/501100003725 NRF-2021R1A5A2030333 Ministry of Health and Welfare https://doi.org/10.13039/501100003625 HG22C0094 ==== Body pmcFollowing the rapid spread of the novel infectious disease caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in China, the World Health Organization declared a global pandemic in March 2020.1 Governments have implemented various strategies to control the virus globally.2 However, globalization and human mobility have accelerated the spread of coronavirus disease 2019 (COVID-19) within a short period, which encourages strengthened border control and lockdowns.3 The South Korean government has conducted non-pharmacological interventions, including large-scale testing, contact tracing, mandatory self-isolation, and border control.4 Although the Korean government used non-pharmacological interventions, SARS-CoV-2 still spread among non-close contacts and evolved into other variants, including delta and omicron.5 Specifically, the omicron variant caused a worldwide resurgence of COVID-19 compared with the delta variant, regardless of vaccination status.6 Therefore, the Korean government has set quarantine guidelines for overseas arrivals in line with the COVID-19 pandemic. Since February 4, 2020, a new entry procedure was introduced for all passengers entering South Korea from China and was gradually extended to include all international arrivals. Starting from April 1, 2020, a 14-day quarantine was required for all travelers entering Korea from the day after arrival.78 Subsequently, strict border measures were altered due to the administration of vaccines and spread of novel SARS-CoV-2 variants. For instance, entrants who were fully vaccinated in Korea were excluded from quarantine starting on May 5, 2021.9 Thereafter, the mandated 14-day quarantine was reduced to 10 days on November 1, 202110 and to 7 days on February 4, 2022.11 Since June 8, 2022, quarantine has not been required anymore, regardless of vaccination status. Since September 3, 2022, no passengers have been required to submit a negative polymerase chain reaction (PCR) or rapid antigen test result before arrival in Korea.12 To prepare for future pandemics, the efficacy of entry restrictions should be evaluated, particularly regarding new variants of SARS-CoV-2. This study explored the impact of border policy changes on overseas entrants and local cases of SARS-CoV-2 variants. As shown in Fig. 1, since the outbreak of COVID-19 in January 2020, the Korea Disease Control and Prevention Agency has been continuously monitoring the genotype and variants of SARS-CoV-2 from the randomly collected nasopharyngeal- and oropharyngeal samples from patients with SARS-CoV-2 infection among the number of new weekly confirmed cases.13 The variant was identified using whole-genome sequencing, targeted sequencing of the spike protein, and a genotyping PCR test; subsequently, the results were categorized, and a rule-based decision algorithm detected the variants and mutations. We applied the Granger causality approach to elucidate the associations of SARS-CoV-2 variants, including delta and omicron, between overseas entrants and national outbreaks. The Granger causality approach is a statistical hypothesis test for determining whether a single time series is a valuable tool for forecasting others.14 However, because of the autoregressive nature of the time series, we first applied it to test the stationarity of the time-varying variables. Therefore, if the model did not satisfy the stationary condition, a different technique before the Granger causality test was to be used. After pre-testing satisfied the stationary condition, we found that the entire time series data were the non-stationary. Thus, we used the differencing technique before the Granger causality test. The results of the pre-testing are shown in Supplementary Data 1. To assess the associations between overseas entrants and national outbreaks of SARS-CoV-2 variants using this definition, we built the Granger causality models as follows: Yt = ∑j=1pΦjYt−j + ∑j=0pΘjXt−j+εt where Yt is the dependent variable at time t and Φj denotes an (N × N) matrix of autoregressive coefficients for j = 1,2,…, p. εt is a vector with Ω (N × N) symmetric definite matrix as the autoregressive zero-mean white-noise error terms. Xt = (X1t,…,Xwt)′ is a W-dimensional exogenous time series vector, and Θj denotes an (N × N) matrix of coefficients. The null hypothesis that a given variable, Y1t, has been reported to Granger-cause Y2t; however, Y2t does not Granger-cause Y1t, tested by calculating the F-statistic. If the F-statistic is significant, it is concluded that the time j lagged by the Y1t variable Granger-causes the Y2t variable. Specifically, the future value of Y2t depends on the present value of Y1t. Herein, we used the SARS-CoV-2 variants as explanatory variables in the Granger causality model. The sensitivity results are shown in Supplementary Data 2. All statistical analyses were performed using the Granger test function from the “lmtest” package in R Statistical Software (version 4.2.0; R and R Studio Foundation for Statistical Computing, Vienna, Austria). Table 1 shows the randomly selected weekly statistics of the cumulative Delta and Omicron rates in South Korea between April 16, 2021, and August 3, 2022. Tables 2 and 3 shows the results of the Granger causality tests. Panel A classifies the null hypothesis based on the quarantine period and shows that the delta virus from overseas entrants (DEO) and national outbreaks when 14 days of quarantine were mandated existed in a one-way causal direction that examines the new confirmed cases of DEO predicting the new confirmed cases of delta virus in national outbreaks (DEN) in the future. We conducted a Granger causality test on 10 different lags. We found that the value of DEO is valuable for forecasting the future value of DEN. However, when the Granger causality test was performed in reverse, the results showed that DEN’s value does not forecast DEO’s future value. Furthermore, Table 2 demonstrates the association between the omicron virus from overseas entrants (OMO) and national outbreaks when 7, 10 days, and no days of quarantine were mandated. In 7 or 10 days of quarantine, there was a bidirectional association between overseas entrants and national outbreaks. There was a one-way association that examined the value of OMO to predict the future value of omicron virus in national outbreaks (OMN) using two different lags (P = 0.008), meaning that the value of OMO is helpful for forecasting the future value of OMN. Table 3 shows the results of the omicron sub-variants based on quarantine. Some omicron sub-variants, including BA1.1 and BA2.75, were significant in forecasting the future value of national outbreaks. However, since the observation period of omicron sub-variants was only 6 weeks, it was insufficient to examine the Granger causality test. The models may include a low power issue, resulting in increased variance for estimation. Fig. 1A shows a time series of the delta virus, including overseas entrants and national outbreaks, as the quarantine time mandated changed from 14 to 10 days. Fig. 1B illustrates a time series of the Omicron virus, including sub-variants, among overseas entrants and national outbreaks as the quarantine time mandated changed from 10 to 7 days and from 7 to no days. Fig. 2A shows the results of the Granger causality test for the delta virus between overseas entrants and national outbreaks with 14 days of quarantine. The pattern in overseas entrants approximately happened in national outbreaks after 10 lag weeks. Overseas entrants peaked at approximately 18 weeks, and national outbreaks peaked at approximately 28 weeks. The dotted blue line indicates the gap between overseas entrants and national outbreaks. Thus, the overseas entrants’ value of the past 10 weeks can be used to predict the future value of national outbreaks. Fig. 2B illustrates the results of the Granger causality test for the omicron virus between overseas entrants and national outbreaks after the quarantine was lifted. The pattern in overseas entrants approximately happened in national outbreaks after 2 lag weeks. Overseas entrants and national outbreaks peaked at approximately 32 and 34 weeks, respectively. Therefore, the value of overseas entrants in the past 2 weeks can be used to predict the future value of national outbreaks. According to our findings, the outbreak gap between international and domestic cases was estimated to be 10 and 2 weeks for delta and omicron variants, respectively, meaning that the quarantine policy successfully delayed the risk of local transmission from incoming travelers for delta variants. Quarantine duration is typically set to prevent post-quarantine transmission with or without testing.15 However, travel quarantine does not fully contain local infections without severe domestic controls before a sufficient vaccination rate.16 Particularly, the quarantine policy during delta variants may be effective due to vaccine-induced herd immunity,17 lower remission rates than omicron variants,18 and adherence to social distancing schemes.8 Conversely, the rate of local cases was not different from that of imported cases during the surge of omicron variants. Regarding the lifted quarantine policy during omicron variants, our findings suggest that border restrictions cannot play an important role in preventing imported cases of new variants. The reduced outbreak gap for the omicron variant could have been due to the reduced effectiveness of vaccines against the omicron variant compared to the Delta variant,19 increased transmission rates, which are noted by many countries,16 or increased mobility,20 since the government announced a post-pandemic strategy with gradual mitigation of social distancing in November 2021. This study had some limitations. First, the Granger causality approach does not directly evaluate the relationship between variables; depending on the lag lengths of variables, results may be changed or true causality may not be accurately measured. Thus, the Granger causality approach is close to the association between variables. Second, since this study focused on delta and omicron virus era for the impact of entry restriction policy, it excluded the 1st through the 3rd wave of COVID-19 in South Korea, which cannot determine the potential super-spreading of SARS-CoV-2. However, despite these limitations, we evaluated the impact of the quarantine policy using a time-series analysis and provided the effectiveness of the quarantine policy decreases over time. In conclusion, the quarantine policy for delta variants helped contain the virus but may not be effective for more transmissible variants. Therefore, it is recommended that countries implement quarantine policies based on particular purposes accounting for the specific features of different variants to avoid potential negative impacts on the economy and distortion of global mobility. Ethics statement We used open data sources that are publicly available. Thus, approval from the Institutional Review Board and participant consent were not required. All procedures were carried out according to the tenets of the Declaration of Helsinki. SUPPLEMENTARY MATERIALS Supplementary Data 1 Pre-testing of stationarity Supplementary Data 2 Sensitivity analysis Fig. 1 Time series of the SARS-CoV-2 variant. (A) Delta. (B) Omicron. SARS-CoV-2 = severe acute respiratory syndrome coronavirus 2. Fig. 2 Granger causality test in the SARS-CoV-2 variant. (A) Delta. (B) Omicron. SARS-CoV-2 = severe acute respiratory syndrome coronavirus 2. Table 1 Summary weekly statistics for the delta and omicron virus rates, including variants in South Korea from April 16, 2021, to August 3, 2022 Gender and age group Cumulative delta rates, % (95% CI) Cumulative omicron rates, % (95% CI) Omicron variants (%) BA.1 (95% CI) BA1.1 (95% CI) BA.2 (95% CI) BA.2.3 (95% CI) BA.2.12.1 (95% CI) BA.4 (95% CI) BA.5 (95% CI) BA.2.75 (95% CI) Overseas entrants Male 66.1 (66.0–66.2) 56.1 (56.0–56.1) 62.9 (62.8–62.9) 56.1 (56.1–56.2) 65.8 (65.7–65.8) 65.9 (65.8–65.9) 62.5 (62.4–62.5) 54.3 (54.2–54.3) 52.5 (52.4–52.5) 73.3 (73.2–73.3) Female 33.9 (33.8–33.9) 43.9 (43.8–43.9) 37.1 (37.0–37.1) 43.9 (43.8–43.9) 34.2 (34.1–34.2) 34.1 (34.0–34.1) 37.5 (37.4–37.5) 45.7 (45.6–45.7) 47.5 (47.4–47.5) 26.7 (26.7–26.8) Over 60 8.1 (8.1–8.2) 11.3 (11.3–11.4) 5.7 (5.7–5.8) 9.9 (9.9–10.0) 8.4 (8.3–8.4) 15.2 (15.2–15.3) 10.2 (10.1–10.2) 13.9 (13.9–14.0) 12.4 (12.3–12.4) 6.7 (6.7–6.8) Under 60 91.9 (91.8–91.9) 88.7 (88.6–88.7) 94.3 (94.2–94.3) 90.1 (90.0–90.1) 91.6 (91.5–91.6) 84.8 (84.7–84.8) 89.8 (89.7–89.8) 86.1 (86.0–86.1) 87.6 (87.5–87.6) 93.3 (93.2–93.3) National outbreak Male 52.4 (52.4–52.5) 50.0 (50.0–50.1) 45.8 (45.7–45.8) 47.0 (47.0–47.1) 48.4 (48.4–48.5) 49.1 (49.0–49.1) 52.2 (52.2–52.3) 51.0 (50.9–51.0) 57.3 (57.2–57.3) 43.8 (43.7–43.8) Female 47.6 (47.5–47.6) 50.0 (50.0–50.1) 54.2 (54.2–54.3) 53.0 (52.9–53.0) 51.6 (51.5–51.6) 50.9 (50.8–50.9) 47.8 (47.8–47.9) 49.0 (48.9–49.0) 42.7 (42.6–42.7) 56.3 (56.2–56.3) Over 60 21.4 (21.3–21.4) 22.4 (22.4–22.5) 9.1 (9.0–9.1) 15.2 (15.1–15.2) 25.1 (25.0–25.1) 28.0 (27.9–28.0) 16.1 (16.0–16.1) 15.4 (15.4–15.5) 20.4 (20.4–20.5) 18.8 (18.8–18.9) Under 60 78.6 (78.5–78.6) 77.6 (77.5–77.6) 90.9 (90.8–90.9) 84.8 (84.8–84.9) 74.9 (74.8–74.9) 72.0 (71.9–72.0) 83.9 (83.8–83.9) 84.6 (84.5–84.6) 79.6 (79.5–79.6) 81.3 (81.2–81.3) CI = confidence interval. Table 2 Granger causality test results for delta and omicron variants Null hypothesis Quarantine Lags, wk 1 2 3 4 10 F-statistic P value F-statistic P value F-statistic P value F-statistic P value F-statistic P value DEO does not Granger-cause DEN 14 days 1.397 0.246 0.613 0.549 1.047 0.389 1.088 0.386 8.584 0.014* DEN does not Granger-cause DEO 14 days 0.093 0.762 0.108 0.898 0.637 0.598 0.610 0.660 0.530 0.816 DEO does not Granger-cause DEN 10 days 1.487 0.231 1.163 0.325 1.761 0.525 1.589 0.281 2.248 < 0.001** DEN does not Granger-cause DEO 10 days 0.101 0.752 0.063 0.939 1.019 0.398 1.721 0.174 6.995 < 0.001** OMO does not Granger-cause OMN 7 or 10 days 4.314 0.050* 6.978 0.010** 2.646 0.113 4.829 0.044* - - OMN does not Granger-cause OMO 7 or 10 days 13.851 0.002** 5.173 0.024* 2.338 0.142 2.517 0.150 - - OMO does not Granger-cause OMN Released 0.682 0.433 14.989 0.008** 14.499 0.065 - - - - OMN does not Granger-cause OMO Released 0.157 0.703 1.803 0.257 4.919 0.174 - - - - DEO = delta virus from overseas entrants, DEN = delta virus national outbreaks, OMO = omicron virus from overseas entrants, OMN = omicron virus in national outbreaks. *Indicates significance at P < 0.05; **Indicates significance at P < 0.01. Table 3 Granger causality test results for omicron sub-variants Null hypothesis: Quarantine Lags, wk 1 2 3 4 10 F-statistic P value F-statistic P value F-statistic P value F-statistic P value F-statistic P value BA1O does not Granger-cause BA1N 7 or 10 days 1.475 0.243 36.505 < 0.001** 10.136 0.003** 35.162 < 0.001** - - BA1N does not Granger-cause BA1O 7 or 10 days 6.331 0.024* 1.013 0.392 1.382 0.310 20.333 0.001** - - BA1.1O does not Granger-cause BA1.1N 7 or 10 days 9.273 0.008** 9.420 0.003** 11.707 0.002** 2.070 0.203 - - BA1.1N does not Granger-cause BA1.1O 7 or 10 days 3.740 0.072 1.834 0.202 2.352 0.140 0.723 0.607 - - BA2O does not Granger-cause BA2N 7 or 10 days 0.005 0.946 3.612 0.059 1.772 0.222 6.072 0.026* - - BA2N does not Granger-cause BA2O 7 or 10 days 0.101 0.755 0.580 0.575 0.406 0.752 0.189 0.936 - - BA2.3O does not Granger-cause BA2.3N 7 or 10 days 0.135 0.718 0.044 0.957 19.401 < 0.001** - - - - BA2.3N does not Granger-cause BA2.3O 7 or 10 days 3.244 0.092 4.916 0.028* 26.773 < 0.001** - - - - BA2.12.1O does not Granger-cause BA2.12.1N Released 0.275 0.615 0.361 0.714 0.440 0.749 - - - - BA2.12.1N does not Granger-cause BA2.12.1O Released 0.314 0.591 0.156 0.859 0.144 0.925 - - - - BA4O does not Granger-cause BA4N Released 13.425 0.006** 6.223 0.044* 0.799 0.598 - - - - BA4N does not Granger-cause BA4O Released 5.891 0.041* 5.719 0.050* 2.103 0.338 - - - - BA5O does not Granger-cause BA5N Released 0.767 0.407 56.658 < 0.001** 15.197 0.063 - - - - BA5N does not Granger-cause BA5O Released 0.007 0.937 2.283 0.198 82.084 0.012* - - - - BA2.75O does not Granger-cause BA2.75N Released 163.823 < 0.001** 52.584 < 0.001** - - - - - - BA2.75N does not Granger-cause BA2.75O Released 3.913 0.083 0.592 0.588 - - - - - - BA1O = omicron sub-variants (BA.1) from overseas entrants, BA1N = omicron sub-variants (BA.1) in national outbreaks, BA1.1O = omicron sub-variants (BA.1.1) from overseas entrants, BA1.1N = omicron sub-variants (BA.1.1) in national outbreaks, BA2O = omicron sub-variants (BA.2) from overseas entrants, BA2N = omicron sub-variants (BA.2) in national outbreaks, BA2.3O = omicron sub-variants (BA.2.3) from overseas entrants, BA2.3N = omicron sub-variants (BA.2.3) in national outbreaks, BA2.12.1O = omicron sub-variants (BA.2.12.1) from overseas entrants, BA2.12.1N = omicron sub-variants (BA.2.12.1) in national outbreaks, BA4O = omicron sub-variants (BA.4) from overseas entrants, BA4N = omicron sub-variants (BA.4) in national outbreaks, BA5O = omicron sub-variants (BA.5) from overseas entrants, BA5N = omicron sub-variants (BA.5) in national outbreaks, BA2.75O = omicron sub-variants (BA.2.75) from overseas entrants, BA2.75N = omicron sub-variants (BA.2.75) in national outbreaks. *Indicates significance at P < 0.05; **Indicates significance at P < 0.01. Funding: This study was supported in part by the MSIT (Ministry of Science and ICT), Korea, under the “AI Convergence New Infectious Disease Response System” supervised by the NIPA (National IT Industry Promotion Agency), and National Research Foundation of Korea grant funded by the Korean government (NRF- 2021R1C1C101177411 and NRF-2021R1A5A2030333). It was also supported by a grant of the project for Infectious Disease Medical Safety, funded by the Ministry of Health & Welfare, Republic of Korea (grant number: HG22C0094). The funder of the study had no role in the study design, data collection, data analysis, data interpretation, or writing of the report. Disclosure: The authors have no potential conflicts of interest to disclose. Data Availability Statement: Data is available from the Korea Disease Control and Prevention Agency. Author Contributions: Conceptualization: Kim DW, Jung J. Data curation: Hong J, Park AK, Kim EJ. Formal analysis: Hong J. Methodology: Hong J, Park AK, Jung J. Writing - original draft: Hong J, Park AK, Radnaabaatar M. Writing - review & editing: all authors. ==== Refs 1 Lee Y Kim M Oh K Kang E Rhie YJ Lee J Comparison of initial presentation of pediatric diabetes before and during the coronavirus disease 2019 pandemic era J Korean Med Sci 2022 37 22 e176 35668686 2 Jo Y Shrestha S Radnaabaatar M Park H Jung J Optimal social distancing policy for COVID-19 control in Korea: a model-based analysis J Korean Med Sci 2022 37 23 e189 35698839 3 Lee H Kim Y Kim E Lee S Risk assessment of importation and local transmission of COVID-19 in South Korea: statistical modeling approach JMIR Public Health Surveill 2021 7 6 e26784 33819165 4 Jung J Preparing for the coronavirus disease (COVID-19) vaccination: evidence, plans, and implications J Korean Med Sci 2021 36 7 e59 33619920 5 Jung J Lee J Kim E Namgung S Kim Y Yun M Frequent occurrence of SARS-CoV-2 transmission among non-close contacts exposed to COVID-19 patients J Korean Med Sci 2021 36 33 e233 34427062 6 Kim YC Kim B Son NH Heo N Nam Y Shin A Vaccine effect on household transmission of omicron and delta SARS-CoV-2 variants J Korean Med Sci 2023 38 1 e9 36593690 7 Korea Disease Control and Prevention Agency Press release. (3.30) Regular briefing of central disaster and safety countermeasure headquarters on COVID-19 Updated March 31, 2020 Accessed November 30, 2022 https://ncov.kdca.go.kr/en/tcmBoardView.do?brdId=12&brdGubun=125&dataGubun=&ncvContSeq=353828&contSeq=353828&board_id=1365&gubun= 8 Chen H Shi L Zhang Y Wang X Sun G A cross-country core strategy comparison in China, Japan, Singapore and South Korea during the early COVID-19 pandemic Global Health 2021 17 1 22 33618688 9 Embassy of the Republic of Korea in the USA New Entry/Quarantine Procedures for Individuals who Completed COVID-19 Vaccination in Korea Updated 2021 Accessed Accessed June 12, 2023 https://overseas.mofa.go.kr/us-en/brd/m_4500/view.do?seq=761013&srchFr=&srchTo=&srchWord=&srchTp=&multi_itm_seq=0&itm_seq_1=0&itm_seq_2=0&company_cd=&company_nm=&page=1 10 Korea Disease Control and Prevention Agency Guidance on shortening the quarantine and manual surveillance period for contacts and overseas arrivals Updated November 1, 2021 Accessed November 30, 2022 http://nqs.kdca.go.kr/nqs/quaStation/busan.do?gubun=notice&fromMainYn=Y&ctx=PS1&contentid=221854 11 Korea Disease Control and Prevention Agency Omicron mutation-related changes in quarantine measures for overseas arrivals Updated February 3, 2022 Accessed November 30, 2022 http://nqs.kdca.go.kr/nqs/quaStation/gimhae.do?gubun=notice&fromMainYn=Y&ctx=BB1&contentid=226274 12 Korea Disease Control and Prevention Agency Notice of suspension of pre-entry inspection for overseas arrivals Updated Aug 31, 2022 Accessed December 2, 2022 https://www.kdca.go.kr/board.es?mid=a20504000000&bid=0014&nPage=2 13 Kim IH Park AK Kim JM Kim HM Lee NJ Woo SH COVID-19 variant surveillance in the Republic of Korea Public Health Wkly Rep 2021 14 16 922 929 14 Granger CW Investigating causal relations by econometric models and cross-spectral methods Econometrica 1969 37 3 424 438 15 Wells CR Pandey A Fitzpatrick MC Crystal WS Singer BH Moghadas SM Quarantine and testing strategies to ameliorate transmission due to travel during the COVID-19 pandemic: a modelling study Lancet Reg Health Eur 2022 14 100304 35036981 16 Kucharski AJ Jit M Logan JG Cotten M Clifford S Quilty BJ Travel measures in the SARS-CoV-2 variant era need clear objectives Lancet 2022 399 10333 1367 1369 35247312 17 Kwon SL Oh J COVID-19 vaccination program in South Korea: a long journey toward a new normal Health Policy Technol 2022 11 2 100601 35127400 18 Jalali N Brustad HK Frigessi A MacDonald EA Meijerink H Feruglio SL Increased household transmission and immune escape of the SARS-CoV-2 omicron compared to delta variants Nat Commun 2022 13 1 5706 36175424 19 Andrews N Stowe J Kirsebom F Toffa S Rickeard T Gallagher E COVID-19 vaccine effectiveness against the omicron (B.1.1.529) variant N Engl J Med 2022 386 16 1532 1546 35249272 20 Central Disease Control Headquarters Press release. South Korea announces the roadmap for gradual return to normal (10.29) Updated October 29, 2021 Accessed December 1, 2022 https://ncov.kdca.go.kr/en/tcmBoardView.do?brdId=12&brdGubun=125&dataGubun=&ncvContSeq=368308&contSeq=368308&board_id=1365&gubun=

PMC010xxxxxx/PMC10353913.txt

==== Front J Korean Med Sci J Korean Med Sci JKMS Journal of Korean Medical Science 1011-8934 1598-6357 The Korean Academy of Medical Sciences 10.3346/jkms.2023.38.e218 Original Article Global Health A Three-Year Longitudinal Study of Risk Factors for Suicidality in North Korean Defectors https://orcid.org/0000-0001-8599-0231 Lee Hyerin 1* https://orcid.org/0000-0002-1628-9617 An Ji Hyun 1* https://orcid.org/0000-0001-8632-4978 Chang Hyein 2 https://orcid.org/0000-0001-7703-0505 Jun Jin Yong 3 https://orcid.org/0000-0001-5384-2605 Hong Jin Pyo 1 1 Department of Psychiatry, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea. 2 Department of Psychology, Sungkyunkwan University, Seoul, Korea. 3 Department of Psychiatry, National Center for Mental Health, Seoul, Korea. Address for Correspondence: Jin Pyo Hong, MD, PhD. Department of Psychiatry, Samsung Medical Center, Sungkyunkwan University School of Medicine, 81 Irwon-ro, Gangnam-gu, Seoul 06351, Korea. suhurhong@gmail.com *Hyerin Lee and Ji Hyun An contributed equally to this article. 17 7 2023 16 6 2023 38 28 e21831 10 2022 26 4 2023 © 2023 The Korean Academy of Medical Sciences. 2023 The Korean Academy of Medical Sciences https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Background This longitudinal study examined risk factors for future suicidality among North Korean defectors (NKDs) living in South Korea. Methods The subjects were 300 NKDs registered with a regional adaptation center (the Hana Center) in South Korea. Face-to-face interviews were conducted using the North Korean version of the World Health Organization’s Composite International Diagnostic Interview to diagnose mental disorders according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition. Subjects were also asked about sociodemographic and clinical factors at baseline. At follow-up after three years, the NKDs (n = 172 respondents) were asked to participate in an online survey, responding to self-questionnaires about suicidality. Logistic regression analyses were used to explore associations between baseline variables and future suicidality among NKDs. Results Thirty (17.4%) of the 172 survey respondents reported suicidality at follow-up. The presence of health problems over the past year, any prior suicidality at baseline, a higher score on a trauma-related scale, and a lower score on a resilience scale at baseline were associated with greater odds of suicidality at follow-up after adjusting for age, sex, and educational level. Of all mental disorder categories, major depressive disorder, dysthymia, agoraphobia, and social phobia were also associated with significantly increased odds of suicidality at follow-up after adjusting for age, sex, educational level, and prior suicidality at baseline. Conclusion Resilience, a previous history of suicidality, and the presence of lifetime depressive disorder and anxiety disorder should be given consideration in mental health support and suicide prevention in NKDs. Graphical Abstract Suicide Depression Resilience North Korean Defectors Mental Health Ministry of Health and Welfare https://doi.org/10.13039/501100003625 HM15C1072 HL19C0018 ==== Body pmcINTRODUCTION The number of persons defecting from North Korea has increased considerably since the late 90s, and as of 2019, the total number of North Korean defectors (NKDs) residing in South Korea is estimated at about 33,523.1 Like other refugees, NKDs are often exposed to psychological trauma over the course of their residence in North Korea and during their escape.2 Even after they have settled in South Korea, NKDs often experience high levels of acculturative stress in adapting to an unfamiliar culture.3 Given this series of stressful events, NKDs are highly vulnerable to mental health problems. A number of studies report that NKDs exhibit high prevalence and severity of psychiatric symptoms such as depression, anxiety, and somatization. Prevalence of major psychiatric disorders including major depressive disorder and posttraumatic stress disorder (PTSD) in NKDs is much higher than in the general Korean population.456 Suicidality rates among NKDs have been reported as being higher than those of the general South Korean population; in one study, 31.3% experienced lifetime suicidal ideation, suicidal plan, or suicide attempt.7 According to the 2019 NKDs’ social survey8 conducted by the Korea Hana Foundation, 12.4% of respondents reported suicidal ideation over the past year, this rate more than double that of South Koreans (5.1%).9 Previous studies of other refugees have also reported higher prevalence of suicidal thoughts and behaviors among refugees than in the general population.101112 Considering the high rate of suicidal ideation among NKDs, addressing post-defection adjustment problems, including suicidality, is a priority. A previous study of suicidal ideation among youth NKDs reported that lower levels of familial cohesion and higher levels of emotional suppression were associated with suicidal ideation.13 However, this study was limited to subjects aged 13 to 27 years and did not investigate the relationship between suicidal thoughts and mental disorders. Our previous cross-sectional study of suicidality among NKDs investigated factors related to lifetime suicidal thoughts and behaviors,7 finding that female sex; the presence of health problems in the past year; and the presence of mental disorders including agoraphobia listed in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) were associated with greater odds of lifetime suicidal thoughts and behaviors. However, to the best of our knowledge, no prior studies have investigated future suicidality among NKDs living in South Korea and the wide range of factors contributing to such. The purpose of this longitudinal study was therefore to examine associations of previously recorded sociodemographic characteristics, clinical factors (e.g., depressive symptoms, resilience, trauma-related symptoms, and loneliness) and mental disorders with future suicidality among NKDs in South Korea. METHODS Subjects and study design This prospective cohort study included a total of 300 NKDs selected from eight regional resettlement centers (Hana Center) across the country. The subjects were all eligible residents aged 18 to 70 years who had arrived in South Korea within the last three years before the first assessment. All subjects who voluntarily agreed to participate in the research were fully informed about the purpose and research procedure in advance. The first assessments were conducted through face-to-face interviews between June and October 2016. The interviewers included psychology, nursing, and social work graduate students in mental health-related departments who had received training in World Health Organization—recommended Composite International Diagnostic Interview (CIDI) administration. Subjects were also asked to complete questionnaires including sociodemographic variables and self-rating scales for clinical factors at the baseline interview. Follow-up evaluations were conducted through an online survey three years later. Of the original 300 subjects, 172 NKDs responded to a questionnaire on suicidality at follow-up. Each subject’s data were collected using an anonymized identification number to protect personal information. Measures Assessment of lifetime DSM-IV disorders This study administered the North Korean version of the CIDI version 2.1 (NK-CIDI 2.1) to each participant to assess lifetime DSM-IV disorders at the baseline interview. The CIDI (World Health Organization, 1990) is a fully structured diagnostic interview designed to confirm psychiatric diagnoses using the definitions and criteria of the DSM-IV.14 The NK-CIDI tool was developed by Lee et al.,15 who modified the Korean language version of CIDI to suit the North Korean sociocultural background. Satisfactory reliability and validity of the new NK-CIDI 2.1 tool was confirmed among NKDs. Lifetime DSM-IV disorders included major depressive disorder, dysthymia, PTSD, generalized anxiety disorder, panic disorder, agoraphobia, social phobia, specific phobia, alcohol use disorder (both alcohol dependence and abuse), and nicotine use disorder. Assessment of suicidality (suicidal thoughts and behaviors) The NK-CIDI module on suicide was used to assess lifetime suicidality at the baseline interview. The presence of suicidality was defined on a lifetime basis, with subjects answering “yes” to at least one of three questions: “Have you ever seriously thought of dying by suicide?” “Have you ever made a plan for suicide?” or “Have you ever attempted suicide?” classified as having suicidality. During the three-year follow-up assessment, subjects were asked the three following questions and those who answered “yes” to at least were regarded as having experienced suicidality over the past year: “Have you ever seriously thought of dying by suicide in the past year?” “Have you ever made a plan for suicide in the past year?” and “Have you ever attempted suicide in the past year?” Assessment of psychiatric clinical factors In addition to being asked about basic sociodemographic variables, subjects were also asked about their mental health status including depressive symptoms, trauma-related symptoms, loneliness, and resilience at the baseline interview. Depressive symptoms were assessed with the 20-item North Korean version of the Center for Epidemiological Studies–Depression Scale (CES-D-NK), which has previously been validated with Cronbach’s alpha coefficients 0.91 for male and 0.93 for female.16 The CES-D total score ranges from zero to 60 points, with higher scores indicating greater depressive symptoms. Trauma-related symptoms were assessed with the 22-item Impact of Event Scale Revised–North Korea Scale (IES-R-NK), with a total score that ranges from zero to 88 points. The IES-R-NK has previously been validated (Cronbach’s alpha coefficients 0.95 for male and 0.94 for female).17 Also, we measured loneliness at baseline evaluation with the 10-item University of California, Los Angeles (UCLA) loneliness scale (Cronbach’s alpha coefficients 0.93).18 Subjects were asked to rate how often they felt the way described by the items on a scale ranging from one (never) to four (often) points, with higher scores reflecting greater loneliness. Finally, the 25-item Connor–Davidson Resilience Scale (CD-RISC) was administered to assess resilience.1920 The CD-RISC is a reliable and valid instrument for measuring resilience (Cronbach’s alpha coefficients 0.89) and has been administered and validated in several different populations including a community sample, psychiatric outpatients, and patients with PTSD and generalized anxiety disorder. Each item is rated on a five-point scale (0–4 points), with a possible total score ranging from zero to 100 points. Higher scores indicate greater resilience. Statistical analyses The χ2 test and logistic regression analyses were conducted to examine baseline characteristic differences between those who participated in follow-up assessment and those who did not respond to the follow-up survey invitation. Baseline sociodemographic and clinical characteristics of those with and without suicidality over the follow-up period of three years were summarized using descriptive statistics. We used logistic regression analyses to examine associations between baseline variables and suicidality at follow-up, adjusting for the potentially confounding effects of sex, age, and educational status. Next, to identify which variables were independent predictors for future suicidality, baseline variables showing a significant association in prior logistic analyses, and potential confounding factors including sex, age, educational status, and severity of depressive symptoms, were entered into multivariable logistic analyses. Logistic regression analyses were carried out to examine the association between lifetime DSM-IV psychiatric disorders and future suicidality. First, we regressed suicidality at follow-up on each baseline DSM-IV disorder (model 1). Then, confounding variables were sequentially entered into models: model 2 was adjusted for sex, age, and educational status, and model 3 was additionally adjusted for lifetime suicidality as assessed at baseline evaluation. Although age, sex, or educational level were not significant variables in this study, they are typically adjusted for and analyzed together in previous studies regarding psychiatric epidemiology studies for NKDs and other refugees.2122 Multicollinearity among univariate variables was not observed in the multivariable models. A Hosmer-Lemeshow goodness of fit test indicated a P value of 0.52. Some of the baseline variables had missing data. Missing cases were excluded in statistical analyses. All statistical analyses were performed with the Statistical Package for the Social Sciences version 25.0 software program (IBM Corp., Armonk, NY, USA) with a statistical significance cutoff set at an alpha level of 0.05. Ethics statement This study was approved by the Institutional Review Board of Samsung Medical Center (SMC 2015-05-042-002). Informed consent was obtained from all subjects prior to the interview. We have obtained permissions for all mental health scales (NK-CIDI, CES-D-NK, IES-NK, Korean version of CD-RISC) used in this study, and we have confirmed that no authorization is needed to use the UCLA loneliness score for research purposes. RESULTS In the three-year follow-up study, 172 of the original 300 subjects responded to an online survey. Subjects participating in follow-up were more often married (38.4% vs. 22.8%; P = 0.028), more frequently living with someone else (34.3% vs. 21.9%; P = 0.019), had more lifetime suicidality (36.6% vs. 23.4%; P = 0.015), had more PTSD diagnoses (19.2% vs. 10.2%; P = 0.032), had less alcohol abuse diagnoses (7.6% vs. 15.6%; P = 0.027), and had more lifetime DSM-IV disorders (62.2% vs. 48.4%; P = 0.017) at baseline than not participating in follow-up. No differences between groups were found for any of the other sociodemographic and clinical characteristics assessed at baseline. At baseline, the mean age of the study sample (n = 172) was 38.8 years (standard deviation [SD], 11.9 years), 80.2% of the population was female, and the mean years of education of the respondents was 11.2 years (SD, 2.5 years). Sixty-three respondents (36.6%) reported lifetime suicidality at baseline. Mean CES-D-NK score was 14.1 points (SD, 13.2 points), mean IES-NK score was 29.8 points (SD, 20.2 points), mean CD-RISC was 73.5 points (SD, 19.0 points), and mean UCLA loneliness scale score was 17.9 points (SD, 5.4 points). Table 1 summarizes the characteristics of subjects with future suicidality vs. those without suicidality at follow-up. Thirty (17.4%) of the 172 surveyed subjects reported suicidality at follow-up. In both unadjusted and adjusted analyses, significant evidence for association of future suicidality with presence of health problems over the past year (adjusted odds ratio [OR], 3.82; 95% confidence interval [CI], 1.22–11.97; P = 0.021), lifetime suicidality at baseline (OR, 6.42; 95% CI, 2.60–15.86; P < 0.001), higher trauma-related scale (IES-NK) score (OR, 1.03; 95% CI, 1.01–1.05; P = 0.007), and lower resilience scale (CD-RISC) score (OR, 0.97; 95% CI, 0.95–0.99; P = 0.009) was found. There was also weak evidence of association between female sex (OR, 3.85; 95% CI, 0.86–17.14; P = 0.077) and future suicidality. Table 1 Odds ratios of baseline sociodemographic and clinical factors among survey respondents with and without suicidalitya at three years of follow-up Variables With suicidality at follow-up Without suicidality at follow-up Unadjusted OR (95% CI) P value aORb (95% CI) P value Sex Male 2 (6.7) 32 (22.5) 1 (ref.) 1 (ref.) Female 28 (93.3) 110 (77.5) 4.07 (0.92–18.03) 0.064 3.85 (0.86–17.14) 0.077 Age, years 37.8 ± 11.6 40.2 ± 11.9 0.98 (0.95–1.02) 0.313 0.99 (0.95–1.02) 0.449 Education, years 10.9 ± 3.4 11.2 ± 2.3 0.96 (0.82–1.12) 0.567 0.98 (0.83–1.15) 0.779 Marital status Married 9 (30.0) 57 (40.1) 1 (ref.) 1 (ref.) Unmarried 10 (33.3) 49 (34.5) 1.29 (0.49–3.44) 0.607 0.78 (0.26–2.30) 0.645 Divorced/Separated/Widowed 11 (36.7) 36 (25.4) 1.94 (0.73–5.13) 0.184 2.11 (0.79–6.13) 0.168 Living arrangement Alone 23 (76.7) 90 (63.4) 1.90 (0.76–4.73) 0.168 1.61 (0.64–4.09) 0.314 With someone else 7 (23.3) 52 (36.6) 1 (ref.) 1 (ref.) Occupational status Yes 6 (20.0) 32 (22.5) 1 (ref.) 1 (ref.) No 24 (80.0) 110 (77.5) 1.16 (0.44–3.09) 0.761 1.08 (0.40–2.94) 0.888 Economic status (monthly income) ≤ $820c 17 (60.7) 89 (63.6) 1 (ref.) 1 (ref.) > $820c 11 (39.3) 51 (36.4) 1.13 (0.49–2.60) 0.775 1.36 (0.57–3.21) 0.486 Religion Yes 16 (53.3) 82 (58.2) 1 (ref.) 1 (ref.) No 14 (46.7) 59 (41.8) 1.22 (0.55–2.68) 0.628 1.49 (0.66–3.36) 0.342 Age of defection, years 30.1 ± 12.0 33.5 ± 13.2 0.98 (0.95–1.01) 0.198 0.99 (0.92–1.05) 0.645 Period of stay in transit country, mon 66.4 ± 77.4 55.2 ± 69.9 1.00 (1.00–1.01) 0.435 1.00 (1.00–1.01) 0.598 Forced repatriation experience Yes 4 (13.3) 20 (14.1) 0.94 (0.30–2.98) 0.914 1.07 (0.33–3.51) 0.909 No 26 (86.7) 122 (85.9) 1 (ref.) 1 (ref.) Health problems over the past year Yes 26 (86.7) 93 (65.5) 3.43 (1.13–10.37) 0.029 3.82 (1.22–11.97) 0.021 No 4 (13.3) 49 (34.5) 1 (ref.) 1 (ref.) Lifetime suicidalitya Yes 22 (73.3) 41 (28.9) 6.77 (2.79–16.45) 0.000 6.42 (2.60–15.86) 0.000 No 8 (26.7) 101 (71.1) 1 (ref.) 1 (ref.) CES-D-NK 17.6 ± 14.4 13.3 ± 12.8 1.02 (1.00–1.05) 0.112 1.02 (0.99–1.05) 0.222 IES-R-NK 39.3 ± 18.9 27.7 ± 19.9 1.03 (1.01–1.05) 0.007 1.03 (1.01–1.05) 0.007 UCLA loneliness 18.9 ± 6.3 17.6 ± 5.1 1.04 (0.97–1.12) 0.240 1.05 (0.98–1.13) 0.180 CD-RISC 63.4 ± 18.7 75.3 ± 18.5 0.97 (0.95–0.99) 0.007 0.97 (0.95–0.99) 0.009 Values are presented as number (%) or mean ± standard deviation. OR = odds ratio, aOR = adjusted odds ratio, CI = confidence interval, CES-D-NK = North Korean version of the Center for Epidemiological Studies–Depression Scale, IES-R-NK = Impact of Event Scale Revised–North Korea Scale, CD-RISC = Connor–Davidson Resilience Scale. aSuicidality: either suicidal ideation suicidal plan or suicide attempt; bAdjusted for sex, age, and educational level; cAbout one million Korean Won. Table 2 shows the results of multivariate logistic regression analyses of potential baseline sociodemographic and clinical characteristics. Lifetime suicidality at baseline (OR, 6.37; 95% CI, 2.03–19.97; P = 0.001) and a lower resilience scale (CD-RISC) score (OR, 0.96; 95% CI, 0.93–0.99; P = 0.011) were significantly associated with future suicidal thoughts and behaviors. There was weak evidence of association with female sex (OR, 9.61; 95% CI, 0.96–96.58; P = 0.055) and presence of health problems over the past year (OR, 4.22; 95% CI, 0.79–22.49; P = 0.092). Table 2 Sociodemographic and clinical variables associated with future suicidalitya (n = 147) Variables aOR (95% CI) P value Female sex 9.61 (0.96–96.58) 0.055 Age, years 0.97 (0.92–1.02) 0.176 Education, years 0.88 (0.71–1.09) 0.242 Health problems over the past year 4.22 (0.79–22.49) 0.092 Lifetime suicidalitya 6.37 (2.03–19.97) 0.001 CES-D-NK 0.97 (0.93–1.02) 0.294 IES-R-NK 1.02 (0.99–1.05) 0.275 CD-RISC 0.96 (0.93–0.99) 0.011 aOR = adjusted odds ratio, CI = confidence interval, CES-D-NK = North Korean version of the Center for Epidemiological Studies–Depression Scale, IES-R-NK = Impact of Event Scale Revised–North Korea Scale, CD-RISC = Connor–Davidson Resilience Scale. aSuicidality: either suicidal ideation suicidal plan or suicide attempt. Table 3 shows associations between each baseline DSM-IV disorder and suicidality at follow-up assessment. Lifetime major depressive disorder (OR, 1.31; 95% CI, 1.04–1.64; P = 0.023), dysthymia (OR, 1.52; 95% CI, 1.07–2.15; P = 0.018), agoraphobia (OR, 1.63; 95% CI, 1.01–2.64; P = 0.046), and social phobia (OR, 1.43; 95% CI, 1.04–1.97; P = 0.030) were significantly associated with suicidality at follow-up, even after adjusting for age, sex, educational level, and lifetime suicidality at baseline (model 3). Although lifetime PTSD was positively associated with future suicidality in model 1, the association disappeared following adjustment for potential confounding variables including lifetime suicidality (OR, 1.12; 95% CI, 0.88–1.43; P = 0.360). There was significant association between specific phobia and future suicidality in model 1 and model 2, yet the association disappeared after adjusting for lifetime suicidality (OR, 1.11; 95% CI, 0.88–1.41; P = 0.365). Of all measured mental disorder categories, subjects with lifetime agoraphobia had the highest odds of future suicidality (OR, 1.63; 95% CI, 1.01–2.64; P = 0.046). Table 3 ORs for each DSM-IV/CIDI mental disorder among respondents with vs. without suicidalitya at the time of follow-up (n = 172) Disorders Model 1b Model 2c Model 3d OR (95% CI) P value aOR (95% CI) P value aOR (95% CI) P value Major depressive disorder 1.46 (1.19–1.79) 0.000 1.44 (1.16–1.79) 0.001 1.31 (1.04–1.64) 0.023 Dysthymia 1.48 (1.11–1.99) 0.009 1.54 (1.13–2.11) 0.007 1.52 (1.07–2.15) 0.018 Posttraumatic stress disorder 1.27 (1.02–1.58) 0.034 1.23 (0.98–1.54) 0.073 1.12 (0.88–1.43) 0.360 Generalized anxiety disorder 1.05 (0.80–1.37) 0.716 1.04 (0.78–1.37) 0.802 0.96 (0.71–1.29) 0.766 Panic disorder 1.27 (0.92–1.74) 0.145 1.25 (0.90–1.74) 0.182 1.17 (0.82–1.67) 0.394 Agoraphobia 1.81 (1.17–2.81) 0.008 1.82 (1.15–2.87) 0.010 1.63 (1.01–2.64) 0.046 Social phobia 1.62 (1.21–2.18) 0.001 1.55 (1.15–2.08) 0.004 1.43 (1.04–1.97) 0.030 Specific phobia 1.34 (1.09–1.65) 0.005 1.31 (1.06–1.61) 0.012 1.11 (0.88–1.41) 0.365 Alcohol dependence 0.99 (0.71–1.37) 0.927 1.08 (0.77–1.53) 0.645 0.92 (0.64–1.32) 0.641 Alcohol abuse 1.10 (0.78–1.54) 0.580 1.24 (0.85–1.80) 0.257 1.15 (0.77–1.71) 0.493 Nicotine use disorder 0.94 (0.64–1.38) 0.746 1.28 (0.77–2.14) 0.346 1.11 (0.66–1.85) 0.699 Any DSM-IV disorders 1.41 (1.09–1.82) 0.009 1.45 (1.12–1.88) 0.005 1.27 (0.97–1.68) 0.087 DSM-IV = Diagnostic Statistical Manual of Mental Disorders, Fourth Edition, CIDI = Composite International Diagnostic Interview, OR = odds ratio, aOR = adjusted odds ratio, CI = confidence interval. aSuicidality: either suicidal ideation suicidal plan or suicide attempt; bUnivariate logistic regression analyses for each DSM-IV/CIDI mental disorder; cMultivariate logistic regression analyses for each DSM-IV/CIDI mental disorder, adjusted for age, sex, and educational level; dMultivariate logistic regression analyses for each DSM-IV/CIDI mental disorder, adjusted for age, sex, educational level, and lifetime suicidality. DISCUSSION To our knowledge, this is the first longitudinal cohort study investigating risk and protective factors associated with future suicidality in a sample composed of NKDs aged 18 to 70 years having arrived in South Korea within the last three years. The rate of suicidality among respondents was higher than in the South Korean general population. The key finding is that the presence of health problems over the past year, lifetime suicidality, higher trauma-related scale (IES-NK) score, and lower resilience scale (CD-RISC) score at initial assessment were significantly associated with future suicidality among NKDs. Lifetime suicidality and a lower resilience scale score were significantly associated with future suicidality even after multiple regression. Also, lifetime mental disorders including major depressive disorder, dysthymia, agoraphobia, and social phobia at initial assessment were significantly associated with higher suicidality at follow-up, even after adjusting for age, sex, educational level, and lifetime suicidality. Due to the control of the government, it is difficult to conduct actual research on the free culture and emotional differences of North Koreans, so most of the studies have been conducted indirectly through defectors. North Korean culture places less emphasis on individual unconsciousness and more on sharing and distribution, with a strong focus on survival and present-oriented thinking, resulting in a culture that suppresses emotions and lacks information about mental health.23 This may be related to the fact that North Koreans commonly express depression, anxiety, and stress through nonspecific somatic symptoms such as headaches, insomnia, and digestive problems.4 Furthermore, since suicide is considered a crime of betraying the nation in North Korea, its incidence rate is not well known. It is also possible that the suicide rate among NKDs may increase after their escape from the socialist regime. The results of the study provide additional validation for the societal concerns of increased suicidality among NKDs living in South Korea. In the present study, 17.4% of respondents reported experience of suicidal thoughts or behaviors over the past year, this rates higher than that recorded among the South Korean general population (5.1%).9 This finding is consistent with previous studies of refugees suggesting that suicidality among refugees is higher than in the general population.101112 The rate of suicidality at follow-up assessment observed in this study was also similar to that found in a 2019 social survey assessing NKDs, where 12.4% of NKDs reported suicidal ideation in the past year. In this longitudinal study, low resilience at initial assessment predicted greater odds of suicidality at follow-up, controlling for several confounding factors including suicidality at initial assessment. Resilience is defined by Conner and Davidson as “[a series of] qualities that enable one to thrive in the face of adversity.” One’s ability to cope with stressful internal and external life events is influenced by both successful and unsuccessful adaptations to previous disruptions.19 Resilience has been presented as a protective factor among refugees for successful adaptation to a new culture, healthy functioning over time, withstanding adversities including traumatic events, and sustaining well-being.2425 For NKDs, resilience has been shown to mitigate the effects of depression and suicidal ideation. One study revealed that depressive symptoms and the resilience level of NKDs are inversely related, and that the resilience score measured by CD-RISC was significantly lower for women, for subjects having resettled more than one year ago, and for unmarried individuals.26 The relationship between suicidal ideation and resilience in youth NKDs was previously investigated in a cross-sectional study, which showed that subjects with suicidal ideation exhibited significantly lower resilience than those without.13 This supports our finding that the suicidality and resilience profiles of NKDs are inversely related across time. It is noteworthy that resilience may be a protective factor mitigating risk of suicidality. Considering that resilience can be modifiable and can improve with pharmacotherapeutic and psychotherapeutic intervention,19 our findings have important implications for clinical care and for guiding future research efforts to increase resilience among NKDs with high suicidality. A previous history of suicidality (e.g., suicidal ideation, suicidal plan, and suicidal attempt) is a known risk factor for suicide. In our study, lifetime suicidality was significantly associated with future suicidality among NKDs even after controlling for possible confounding factors. Although we did not investigate the individual impacts of suicidal ideation, suicidal plan, and suicidal attempt, our results imply that a previous history of suicidality may be an important suicide risk factor among NKDs. Other factors related to future suicidality included the presence of health problems over the past year and a higher trauma-related scale score measured using the IES-NK. In our study, both of these factors were significantly associated with future suicidality after adjusting for age, sex and educational level; however, the association was not apparent in multiple regression analysis. One refugee study showed that poor physical health is associated with increased levels of thwarted belongingness and perceived burdensomeness, which are two main factors underpinning interpersonal theory of suicide hypotheses.27 Our previous cross-sectional study also reported an association between presence of health problems and lifetime suicidality.7 Poor state of health may contribute to increased suicidal behavior via several aspects such as psychological stress and poor access to social support. Meanwhile, trauma exposure among refugees is extremely high and about 49.3% of NKDs reported having experienced or witnessed traumatic events such as the death or arrest of family members and/or suffered physical abuse from acquaintances.2829 There is some evidence that exposure to chronic fear and repeated trauma is associated with increased risk of suicide among refugees.27 In this study, it is possible that the effects of the presence of health problems and trauma on future suicidality were overshadowed in multivariate analysis by the influence of other factors such as previous suicide and resilience. Further research and replication will be necessary to discern the causal relationship of health problems and trauma experience to future suicidality among NKDs. Mental disorders are known to be associated with an elevated risk of suicide in the general population. For refugees, inconsistent findings have been reported regarding the association between mental disorders and suicidality. One study showed that rates for suicide attempt in individuals with mental disorders were lower in refugees within a Swedish-born reference group,30 while other studies found that patients with mental disorders such as depression and PTSD showed greater frequency of suicidal thoughts and behaviors.731 In our study, the presence of lifetime mental disorders diagnosed according to the DSM-IV was associated with significantly increased odds of future suicidality among NKDs. Agoraphobia was most strongly associated with future suicidality, consistent with a previous cross-sectional study.7 Other depression and anxiety disorders including major depressive disorder, dysthymia, and social phobia also showed significant association with future suicidality, even after controlling for lifetime history of suicide. However, the association between PTSD and future suicidality disappeared after adjusting for confounding factors. This result was inconsistent with those of previous cross-sectional refugee studies reporting possible association between PTSD and suicide.31 Considering the high coexistence rate of depression and anxiety disorders in PTSD patients and the greater risk of suicide in the presence of these coexisting disorders,31 in the long term, depression and phobia may be more likely than PTSD to cause impairment in one’s functioning and to affect future suicidality. Further research is needed to investigate the relationship between PTSD and future suicidality in the NKD population. Interestingly, social and economic factors such as age, marital status, and employment status were not significantly associated with depression and suicide among NKDs with a relatively short period of settlement but showed an increasing relationship as the settlement period lengthened.2132 The participants in this study had a relatively short settlement period and did not experience significant difficulties in adapting to South Korean culture, possibly due to the provision of early settlement support policies. Therefore, further in-depth studies are needed to examine the temporal relationship between stress experiences and suicide risk during the settlement process among NKDs. This study has several limitations to consider. First, this longitudinal follow-up study included a relatively small sample size, and 128 out of 300 subjects did not respond to the follow-up assessment. The small sample size and rarity of suicide attempts in our sample precluded meaningful investigation of correlates of suicidality in those who have attempted suicide as compared with those who only had suicidal ideation or a suicidal plan. Previous studies on NKDs also employed the same methodology, and as this population is considered a special group, the sample cohort in studies related to this topic typically ranges from 150 to 300 individuals.467 Further longitudinal research is needed to explore whether there are differences in the risk factors between those who only think about suicide and those who have attempted suicide. Second, it is also possible that serious suicide cases, such as complete suicide or suicide attempts with severe sequelae, have not been tracked. However, the assessment of suicidality is the standard of clinical care and a proxy for assessing the risk of suicide. Third, our study did not include other possible factors that may be associated with future suicidality such as occupation before defection, whether family members defected together, and postmigration challenges. Fourth, there may be a possibility of recall bias because lifelong history of suicidality and psychiatric diagnoses were self-reported. Finally, although the high correlation between bipolar disorder and suicide is well-known in the general population, none of the subjects in the present study was diagnosed with bipolar disorder. It is possible that bipolar disorder-related functional decline deters persons with the disorder from even attempting the escape from North Korea. Thus, in this study, the relationship between suicide and bipolar disorder among NKDs may have been underestimated. Despite these limitations, this study is the first longitudinal follow-up study to date investigating associations of sociodemographic and mental disorder risk factors with future suicidality among NKDs within three years of entry into South Korea. It is noteworthy that our study diagnosed DSM-IV mental disorders using a fully structured diagnostic interview tool (NK-CIDI 2.1) and evaluated mental disorders as possible predictors of future suicidality in this high-risk group. The study can be considered an important step toward the development of targeted suicide prevention programs for NKDs and other immigrants and refugees. In conclusion, lower resilience and previous history of suicidality were strongly associated with future suicidality among NKDs. Also, NKDs with a lifetime history of mental disorders including major depressive disorder, dysthymia, agoraphobia, and social phobia are at high risk for future suicide. Further research should focus on tracking changes in suicide risk factors at each post-defection adaptation stage, and on the development of preventive therapeutic interventions, especially those geared toward suicide prevention. Funding: This work was supported by the Korea Healthcare Technology R&D project, Ministry of Health and Welfare, Republic of Korea (HM15C1072, HL19C0018). Disclosure: The authors have no potential conflicts of interest to disclose. Data Availability Statement: The data supporting the findings of the present study are available from the corresponding author upon reasonable request with permission from the Korean National Center for Mental Health. Author Contributions: Conceptualization: An JH, Lee H, Chang HI, Jun JY, Hong JP. Data curation: Jun JY. Formal analysis: An JH, Lee H. Investigation: An JH. Methodology: Lee H, Chang HI, Jun JY. Supervision: Hong JP. Writing - original draft: An JH, Lee H. Writing - review & editing: Chang HI, Jun JY, Hong JP. ==== Refs 1 Ministry of Unification (KR) North Korean Defectors Government Policy Seoul, Korea Ministry of Unification, Department of Settlement Support 2019 2 Jeon WT Yu SE Cho YA Eom JS Traumatic experiences and mental health of North Korean refugees in South Korea Psychiatry Investig 2008 5 4 213 220 3 Jeon WT Issues and problems of adaptation of North Korean defectors to South Korean society: an in-depth interview study with 32 defectors Yonsei Med J 2000 41 3 362 371 10957891 4 Kim HH Lee YJ Kim HK Kim JE Kim SJ Bae SM Prevalence and correlates of psychiatric symptoms in North Korean defectors Psychiatry Investig 2011 8 3 179 185 5 Lee KE An JH Kim DE Moon CS Hong JP Clinical characteristics of post-traumatic stress disorder among North Korean defectors Anxiety Mood 2018 14 2 80 87 6 Lee KE Moon CS An JH Lee HC Kim DE Park S Prevalence of DSM-IV major psychiatric disorders among North Korean defectors in South Korea Psychiatry Investig 2020 17 6 541 546 7 An JH Lee KE Lee HC Kim HS Jun JY Chang HI Prevalence and correlates of suicidal thoughts and behaviors among North Korean defectors Psychiatry Investig 2018 15 5 445 451 8 Korea Hana Foundation Settlement Survey of North Korean Refugees in South Korea Seoul, Korea Korea Hana Foundation 2019 9 Statistics Korea Social Survey Daejeon, Korea Statistics Korea 2018 10 Hagaman AK Sivilli TI Ao T Blanton C Ellis H Lopes Cardozo B An investigation into suicides among bhutanese refugees resettled in the united states between 2008 and 2011 J Immigr Minor Health 2016 18 4 819 827 26758579 11 Akinyemi OO Atilola O Soyannwo T Suicidal ideation: are refugees more at risk compared to host population? Findings from a preliminary assessment in a refugee community in Nigeria Asian J Psychiatr 2015 18 81 85 26412050 12 Goosen S Kunst AE Stronks K van Oostrum IE Uitenbroek DG Kerkhof AJ Suicide death and hospital-treated suicidal behaviour in asylum seekers in the Netherlands: a national registry-based study BMC Public Health 2011 11 1 484 21693002 13 Park S Rim SJ Jun JY Related factors of suicidal ideation among North Korean refugee youth in South Korea Int J Environ Res Public Health 2018 15 8 1694 30096867 14 Kessler RC Ustün TB The World Mental Health (WMH) survey initiative version of the World Health Organization (WHO) Composite International Diagnostic Interview (CIDI) Int J Methods Psychiatr Res 2004 13 2 93 121 15297906 15 Lee T Ahn MH Jun JY Han JM Lee SH Hahm BJ Development of North Korean version of the Composite International Diagnostic Interview J Korean Neuropsychiatr Assoc 2015 54 2 228 235 16 Park SJ Lee SH Jun JY Lee T Han JM Ahn MH Reliability and validity of the North Korean version of Center for Epidemiologic Studies Depression Scale (CES-D-NK) Anxiety Mood 2015 11 1 54 60 17 Won SD Lee SH Hong JP Jun JY Han JM Shin MN A study on reliability and validity of the Impact of Event Scale-Revised-North Korea (IES-R-NK) J Korean Neuropsychiatr Assoc 2015 54 1 97 104 18 Russell DW UCLA loneliness scale (version 3): reliability, validity, and factor structure J Pers Assess 1996 66 1 20 40 8576833 19 Connor KM Davidson JR Development of a new resilience scale: the Connor-Davidson Resilience Scale (CD-RISC) Depress Anxiety 2003 18 2 76 82 12964174 20 Baek HS Lee KU Joo EJ Lee MY Choi KS Reliability and validity of the Korean version of the Connor-Davidson Resilience Scale Psychiatry Investig 2010 7 2 109 115 21 Lee Y Lee M Park S Mental health status of North Korean refugees in South Korea and risk and protective factors: a 10-year review of the literature Eur J Psychotraumatol 2017 8 sup2 1369833 29038687 22 Bogic M Ajdukovic D Bremner S Franciskovic T Galeazzi GM Kucukalic A Factors associated with mental disorders in long-settled war refugees: refugees from the former Yugoslavia in Germany, Italy and the UK Br J Psychiatry 2012 200 3 216 223 22282430 23 Kim SJ Park YS Lee H Park SM Current situation of psychiatry in North Korean: from the viewpoint of North Korean medical doctors Korean J Psychosom Med 2012 20 1 32 39 24 Panter-Brick C Hadfield K Dajani R Eggerman M Ager A Ungar M Resilience in context: a brief and culturally grounded measure for Syrian refugee and Jordanian host-community adolescents Child Dev 2018 89 5 1803 1820 28617937 25 Marley C Mauki B Resilience and protective factors among refugee children post-migration to high-income countries: a systematic review Eur J Public Health 2019 29 4 706 713 30380016 26 An JH Lee KE Lee HC Jun JY Kim HS Hong JP The relationship between depressive symptoms and psychological resilience among North Korean defectors J Korean Assoc Soc Psychiatry 2017 22 2 67 74 27 Ellis BH Lankau EW Ao T Benson MA Miller AB Shetty S Understanding Bhutanese refugee suicide through the interpersonal-psychological theory of suicidal behavior Am J Orthopsychiatry 2015 85 1 43 55 25642653 28 Steel Z Chey T Silove D Marnane C Bryant RA van Ommeren M Association of torture and other potentially traumatic events with mental health outcomes among populations exposed to mass conflict and displacement: a systematic review and meta-analysis JAMA 2009 302 5 537 549 19654388 29 Nam B Kim JY DeVylder JE Song A Family functioning, resilience, and depression among North Korean refugees Psychiatry Res 2016 245 451 457 27620328 30 Björkenstam E Helgesson M Amin R Mittendorfer-Rutz E Mental disorders, suicide attempt and suicide: differences in the association in refugees compared with Swedish-born individuals Br J Psychiatry 2020 217 6 679 685 31608856 31 Ferrada-Noli M Asberg M Ormstad K Lundin T Sundbom E Suicidal behavior after severe trauma. Part 1: PTSD diagnoses, psychiatric comorbidity, and assessments of suicidal behavior J Trauma Stress 1998 11 1 103 112 9479679 32 Cho YA Jeun WT Yu JJ Um JS Predictors of depression among North Korean defectors: a 3-year follow-up study Korean J Couns Psychother 2005 17 2 467 484

PMC010xxxxxx/PMC10353914.txt

==== Front J Korean Med Sci J Korean Med Sci JKMS Journal of Korean Medical Science 1011-8934 1598-6357 The Korean Academy of Medical Sciences 10.3346/jkms.2023.38.e217 Original Article Infectious Diseases, Microbiology & Parasitology Clinical Characteristics and Risk Factors for Mortality in Critical COVID-19 Patients Aged 50 Years or Younger During Omicron Wave in Korea: Comparison With Patients Older Than 50 Years of Age https://orcid.org/0000-0002-7571-8485 Shi Hye Jin 1* https://orcid.org/0000-0001-8378-8937 Yang Jinyoung 2* https://orcid.org/0000-0003-2744-1159 Eom Joong Sik 1 https://orcid.org/0000-0002-9490-6609 Ko Jae-Hoon 2 https://orcid.org/0000-0002-7464-9780 Peck Kyong Ran 2 https://orcid.org/0000-0002-8463-6297 Kim Uh Jin 3 https://orcid.org/0000-0002-1577-678X Jung Sook In 3 https://orcid.org/0000-0002-1603-0963 Kim Seulki 4 https://orcid.org/0000-0002-2032-9538 Seok Hyeri 5 https://orcid.org/0000-0003-0062-342X Hyun Miri 6 https://orcid.org/0000-0002-9125-7156 Kim Hyun Ah 6 https://orcid.org/0000-0002-2197-642X Kim Bomi 7 https://orcid.org/0000-0003-4741-4406 Joo Eun-Jeong 7 https://orcid.org/0000-0001-8702-1987 Cheong Hae Suk 7 https://orcid.org/0000-0003-4676-2257 Jun Cheon Hoo 8 https://orcid.org/0000-0003-3625-3328 Wi Yu Mi 8 https://orcid.org/0000-0002-0694-1579 Kim Jungok 9 https://orcid.org/0000-0003-3518-966X Kym Sungmin 9 https://orcid.org/0000-0001-7939-9744 Lim Seungjin 410 https://orcid.org/0000-0002-2644-3606 Park Yoonseon 1 1 Division of Infectious Diseases, Department of Internal Medicine, Gil Medical Center, Gachon University College of Medicine, Incheon, Korea. 2 Division of Infectious Diseases, Department of Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea. 3 Department of Infectious Diseases, Chonnam National University Medical School, Gwangju, Korea. 4 Division of Infectious Diseases, Department of Internal Medicine, Pusan National University Yangsan Hospital, Yangsan, Korea. 5 Division of Infectious Diseases, Department of Medicine, Korea University Ansan Hospital, Korea University College of Medicine, Ansan, Korea. 6 Department of Infectious Diseases, Keimyung University Dongsan Hospital, Keimyung University School of Medicine, Daegu, Korea. 7 Division of Infectious Diseases, Department of Medicine, Kangbuk Samsung Hospital, Sungkyunkwan University School of Medicine, Seoul, Korea. 8 Division of Infectious Diseases, Department of Medicine, Changwon Samsung Hospital, Sungkyunkwan University School of Medicine, Changwon, Korea. 9 Division of Infectious Diseases, Chungnam National University Sejong Hospital, Sejong, Korea. 10 Department of Internal Medicine, School of Medicine, Pusan National University, Busan, Korea. Address for Correspondence: Yoonseon Park, MD, PhD. Division of Infectious Diseases, Department of Internal Medicine, Gil Medical Center, Gachon University School of Medicine, 21 Namdong-daero 774-beon-gil, Namdong-gu, Incheon 21565, Republic of Korea. yoonseony@gilhospital.com Address for Correspondence: Seungjin Lim, MD. Division of Infectious Disease, Department of Internal Medicine, Pusan National University Yangsan Hospital, 20 Geumo-ro, Mulgeum-eup, Yangsan 50612, Republic of Korea. babopm@naver.com *Hye Jin Shi and Jinyoung Yang contributed equally as first-authors to this work. 17 7 2023 14 6 2023 38 28 e21701 12 2022 22 3 2023 © 2023 The Korean Academy of Medical Sciences. 2023 The Korean Academy of Medical Sciences https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Background The coronavirus disease 2019 (COVID-19) pandemic has caused the death of thousands of patients worldwide. Although age is known to be a risk factor for morbidity and mortality in COVID-19 patients, critical illness or death is occurring even in the younger age group as the epidemic spreads. In early 2022, omicron became the dominant variant of the COVID-19 virus in South Korea, and the epidemic proceeded on a large scale. Accordingly, this study aimed to determine whether young adults (aged ≤ 50 years) with critical COVID-19 infection during the omicron period had different characteristics from older patients and to determine the risk factors for mortality in this specific age group. Methods We evaluated 213 critical adult patients (high flow nasal cannula or higher respiratory support) hospitalized for polymerase chain reaction-confirmed COVID-19 in nine hospitals in South Korea between February 1, 2022 and April 30, 2022. Demographic characteristics, including body mass index (BMI) and vaccination status; underlying diseases; clinical features and laboratory findings; clinical course; treatment received; and outcomes were collected from electronic medical records (EMRs) and analyzed according to age and mortality. Results Overall, 71 critically ill patients aged ≤ 50 years were enrolled, and 142 critically ill patients aged over 50 years were selected through 1:2 matching based on the date of diagnosis. The most frequent underlying diseases among those aged ≤ 50 years were diabetes and hypertension, and all 14 patients who died had either a BMI ≥ 25 kg/m2 or an underlying disease. The total case fatality rate among severe patients (S-CFR) was 31.0%, and the S-CFR differed according to age and was higher than that during the delta period. The S-CFR was 19.7% for those aged ≤ 50 years, 36.6% for those aged > 50 years, and 38.1% for those aged ≥ 65 years. In multivariate analysis, age (odds ratio [OR], 1.084; 95% confidence interval [CI], 1.043–1.127), initial low-density lipoprotein > 600 IU/L (OR, 4.782; 95% CI, 1.584–14.434), initial C-reactive protein > 8 mg/dL (OR, 2.940; 95% CI, 1.042–8.293), highest aspartate aminotransferase > 200 IU/L (OR, 12.931; 95% CI, 1.691–98.908), and mechanical ventilation implementation (OR, 3.671; 95% CI, 1.294–10.420) were significant independent predictors of mortality in critical COVID-19 patients during the omicron wave. A similar pattern was shown when analyzing the data by age group, but most had no statistical significance owing to the small number of deaths in the young critical group. Although the vaccination completion rate of all the patients (31.0%) was higher than that in the delta wave period (13.6%), it was still lower than that of the general population. Further, only 15 (21.1%) critically ill patients aged ≤ 50 years were fully vaccinated. Overall, the severity of hospitalized critical patients was significantly higher than that in the delta period, indicating that it was difficult to find common risk factors in the two periods only with a simple comparison. Conclusion Overall, the S-CFR of critically ill COVID-19 patients in the omicron period was higher than that in the delta period, especially in those aged ≤ 50 years. All of the patients who died had an underlying disease or obesity. In the same population, the vaccination rate was very low compared to that in the delta wave, indicating that non-vaccination significantly affected the progression to critical illness. Notably, there was a lack of prescription for Paxlovid for these patients although they satisfied the prescription criteria. Early diagnosis and active initial treatment was necessary, along with the proven methods of vaccination and personal hygiene. Further studies are needed to determine how each variant affects critically ill patients. Graphical Abstract COVID-19 Risk Factors Young Critical Mortality Omicron Variant Gachon University https://doi.org/10.13039/501100002631 GCU-2022-202209660001 Organization of academic resource management system for clinical trial of infectious disease HE20C002201139820037 ==== Body pmcINTRODUCTION Since the outbreak of the coronavirus disease 2019 (COVID-19) pandemic, several variants of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) have emerged. Among these variants, omicron (B.1.1.529) was first identified in November 2021 and had been subsequently declared as a variant of concern (VOC) by the World Health Organization (WHO).1 In South Korea, the first patient carrying the omicron variant was identified on December 1st, 2021,2 initiating a wide spread of the omicron variant (January 30, 2022 to April 24, 2022),3 which followed the preceding delta epidemic (July 7, 2021 to January 29, 2021). Early national and international reports suggested that the clinical course of patients infected with the omicron variant was better than that of patients infected with the delta variant or variants of the other epidemics4; however, there were also reports suggesting the contrary.567 The SARS-CoV-2 omicron epidemic in South Korea was massive compared to that in other countries possibly owing to immune evasion8 or the previous epidemics,9 and there were reports of many severe patients. Patients with severe COVID-19 infections require respiratory support, such as a high-flow nasal cannula (HFNC) or mechanical ventilation (MV),10 and the case fatality rate (CFR) in those patients varies depending on the size and speed of the outbreak. In a previous study, age > 65 years and underlying diseases were found to be risk factors for severe COVID-19 infection and death.10111213 Studies on young patients, although relatively limited,141516 reported that not only underlying diseases but also obesity15 and non-vaccination171819 had emerged as risk factors of complications in this particular population. A study conducted in South Korea during the fourth epidemic after the delta variant20 showed that for critical COVID-19 patients aged ≤ 50 years,21 age, elevated creatinine (Cr), decreased platelet, MV, continuous renal replacement therapy (CRRT), central line-associated bloodstream infection (CLABSI), obesity, and lack of vaccination were risk factors for mortality. It was suggested that in South Korea, there might be specific differences between the omicron variant epidemic (fifth epidemic) and the previous epidemics with respect to the characteristics and severity of the variant driving the epidemic, epidemic magnitude and spread, and vaccination completion rate at that period. Therefore, this study aimed to analyze the clinical characteristics and risk factors of death in young critical COVID-19 patients in South Korea during the omicron epidemic in comparison to those of patients older than 50 years and to also comparing this period to the delta variant period based on previous data. METHODS Study population and data sources This study enrolled all critical patients hospitalized for COVID-19 in nine hospitals in South Korea between February 1, 2022 and April 30, 2022. The inclusion criterion was age ≥ 19 years, and the exclusion criteria were pediatric cases, re-confirmed cases, and cases confirmed after death. Each patient aged ≤ 50 years was matched to two patients aged > 50 years based on the diagnosis date. Matching in the control group consisted of the nearest date of diagnosis. If there were multiple control groups on the same diagnosis date, one oldest and one lowest age was selected for the correction of age variables. These patients were then evaluated and compared with respect to their clinical characteristics and prognosis. All patients were diagnosed with COVID-19 using SARS-CoV-2 real-time reverse transcription polymerase chain reaction performed using nasopharyngeal swabs or sputum specimens or by expert rapid antigen testing according to national guidelines. Electronic medical records were reviewed to collect demographic characteristics, including BMI; underlying diseases; clinical features and laboratory findings on the day of admission; clinical course; treatment received; and outcomes. The Health Insurance Review and Assessment system was used to identify information about the status of vaccination, route of infection, living environment before hospitalization, and variant contracted. The patients were followed until death or discharge, whichever occurred first. Study outcomes and definitions Critically ill patients were defined as those who received HFNC or higher respiratory support during hospitalization, as defined by the Korea Disease Control and Prevention Agency (KDCA).22 The completion of vaccination was defined as completion of the second dose of Janssen or the third dose of Moderna/Pfizer/AstraZeneca taking into consideration the timing of the investigation. Breakthrough infection was defined as infection 2 weeks after completion of vaccination or 14–90 days after the second dose of the vaccine. The outcome measures were all-cause hospital mortality and MV or extracorporeal membrane oxygenation (ECMO). The severity of the clinical course was assessed based on the highest respiratory support required during the hospitalization period. These were classified as HFNC, MV, and ECMO. Since the study population comprised critically ill patients, patients maintained on nasal cannula or on room air were not included. Non-invasive positive pressure ventilation was not performed for the patients in this study. The outbreak was defined as the simultaneous occurrence of multiple confirmed cases in a nursing home/long-term care hospital, workplace, or religious or group facility. Hospital-acquired infection (HAI) was defined as confirmed cases during admission to acute care hospitals or long-term care facilities for other unrelated illnesses, except for outbreaks. Statistical analysis Continuous variables were expressed as the mean ± standard deviation (SD), while categorical variables were expressed as the frequency and percentage. Continuous variables were compared using the Student’s t-test or Mann-Whitney U test, as appropriate, while categorical variables were compared using the χ2 test or Fisher’s exact test. The all-cause mortality rate was compared between the two groups using Kaplan-Meier curves. Significant variables (P < 0.050) were analyzed using multivariate analysis (logistic regression model), and significant results were presented as the odds ratio (OR) and 95% confidence interval (CI). A logistic regression model was used to control for confounding variables. Statistically significant variables in the univariate analyses were then used in the multivariate analysis in addition to the variables of clinical significance. We performed a collinearity analysis on the results that were significant in the univariate analysis. A multivariate analysis was performed on the test results that emerged as independent variables in the collinearity analysis. A collinearity test was also performed to exclude the possibility of cross-influences. Risks factors for mortality were reported with their ORs and 95% CIs. When the median and interquartile ranges (IQRs) were reported, the median was assumed to reflect the mean and the IQR was assumed to be 1.35 SD. All P-values were reported to three decimal places. P values of < 0.050 were considered statistically significant. All statistical analyses were performed using IBM SPSS for Windows version 24.0 (IBM Corp., Armonk, NY, USA). Ethics statement The present study protocol was reviewed and approved by the Institutional Review Board of Gil Medical Center (approval No. GCIRB2021-462). This study adhered to the principles embodied in the Declaration of Helsinki. The requirement for written informed consent was waived because de-identified data were collected retrospectively. RESULTS Comparison of clinical characteristics between age groups A total of 213 patients were evaluated. Among them, 71 patients were aged ≤ 50 years, while 142 patients were aged > 50 years (110 of 142 were aged 65 years or older) (Table 1). The mean patient age was 62.44 ± 19.10 years. The patients aged ≤ 50 years had a mean ± SD age of 39.51 ± 8.92 years, while patients aged > 50 years had a mean ± SD age of 73.91 ± 10.56 years. Women accounted for 49.3% in the group of patients aged ≤ 50 years and for 42.3% of patients aged > 50 years, corresponding to nearly half in each group. Patients aged ≤ 50 years had fewer outbreak-related associations than those aged > 50 years (2.8% vs. 14.8%), but the difference was not significant (Table 1). Of the entire population, only 31.0% were fully vaccinated, and only 15 of 71 patients aged ≤ 50 years had complete vaccination, significantly lower than the complete vaccination rate in patients aged > 50 years (21.1% vs. 35.9%, P = 0.019).Compared with death group (29.3%), survivor group have a higher vaccine completion rate(34.8%), but no stastically significant.(Table 2). Table 1 Patient characteristics Characteristics Total (N = 213) Age > 50 years group (n = 142) Age ≤ 50 years group (n = 71) P value Female sex 95 (44.5) 60 (42.3) 35 (49.3) 0.204 Age, yr 62.44 ± 19.10 73.91 ± 10.56 39.51 ± 8.92 < 0.001*** BMI ≥ 25 kg/m2 72 (34.1) 37 (26.4) 35 (49.3) 0.001** Mode of infection 0.057 Outbreak related 23 (10.8) 21 (14.8) 2 (2.8) No obvious exposure 155 (72.8) 100 (69) 57 (80.3) Full vaccination 66 (31.0) 51 (35.9) 15 (21.1) 0.019* Breakthrough infection 81 (38.0) 62 (43.7) 19 (26.8) 0.017* Underlying disease Any underlying disease 186 (87.3) 134 (94.4) 52 (73.2) < 0.001*** Diabetes mellitus 79 (37.1) 59 (41.5) 20 (28.2) 0.039* Hypertension 88 (41.3) 72 (50.7) 16 (22.5) < 0.001*** Chronic lung disease 35 (16.4) 30 (21.1) 5 (7.0) 0.010* Heart disease 42 (19.7) 36 (25.4) 6 (8.5) 0.003** ESRD or CKD 27 (12.7) 18 (12.7) 9 (12.7) 1.000 Chronic neurologic disease 61 (28.6) 53 (37.3) 8 (11.3) < 0.001*** Solid tumor 38 (17.8) 31 (21.8) 7 (9.9) 0.037* Taking immunosuppressant 21 (10.0) 10 (7.1) 11 (15.7) 0.055 Clinical manifestation on admission Lowest oxygen saturation (%) 89.7809 ± 8.60 88.90 ± 9.05 91.68 ± 6.18 0.001** Initial chest infiltration 195 (91.5) 133 (93.7) 62 (87.3) 0.125 Lowest O2 saturation < 90 79 (37.1) 50 (35.2) 29 (40.8) 0.454 Vasopressor use 42 (19.7) 26 (18.3) 16 (22.5) 0.470 Initial SBP < 90 mmHg 48 (22.6) 33 (23.25) 15 (21.4) 0.862 Initial respiratory rate ≥ 22 breaths/min 143 (67.5) 97 (68.3) 46 (65.7) 0.756 Altered mentality 55 (25.8) 39 (27.5) 16 (22.5) 0.508 Initial laboratory findings WBC > 12,000 or 4,000 mm3 79 (37.1) 50 (35.2) 29 (40.8) 0.454 Platelet count < 100 109/L 27 (12.7) 13 (9.2) 14 (19.7) 0.047* Creatinine > 1.5 mg/dL 64 (30.0) 47 (33.1) 17 (23.9) 0.205 LDH > 600 IU/L 47 (30.9) 27 (26.0) 20 (41.7) 0.060 Laboratory finding (Hosp) Highest WBC (109/L) 17.14 ± 8.95 15.97 ± 12.74 0.436 > 12.000 135 (63.4) 97 (68.3) 38 (53.5) 0.049* Lowest lymphocyte count < 900 109/L 158 (74.2) 110 (77.5) 48 (67.6) 0.137 Lowest platelet count < 100 109/L 77 (36.2) 50 (35.2) 27 (38.0) 0.763 Lowest hemoglobin count < 8 g/dL 77 (36.2) 56 (39.4) 21 (29.6) 0.176 Highest ALT > 200 IU/L 35 (16.4) 18 (12.7) 17 (2.9) 0.049* Highest LDH IU/L > 600 IU/L 67 (39.4) 44 (37.9) 23 (42.6) 0.615 Highest Cr > 1.5 mg/dL 89 (41.8) 62 (43.7) 27 (38.0) 0.464 Highest CRP > 8 mg/dL 149 (70.0) 106 (74.6) 43 (60.6) 0.040* Highest procalcitonin > 0.5 mg/dL 109 (59.9) 72 (60.0) 37 (59.7) 1.000 Treatment Remdesivir 185 (86.9) 128 (90.1) 57 (80.3) 0.054 Baricitinib 13 (6.1) 5 (3.5) 8 (11.3) 0.035 Paxlovid 7 (3.3) 6 (4.2) 1 (1.4) 0.429 Othersa 7 (3.3) 5 (3.5) 2 (2.8) 1.000 MV 90 (42.3) 59 (41.5) 31 (43.7) 0.771 CRRT 29 (13.6) 16 (11.3) 13 (18.3) 0.203 ECMO 7 (3.3) 2 (1.4) 5 (7.0) 0.043* Complication Blood stream infection 16 (7.5) 9 (6.3) 7 (9.9) 0.411 Fungal 11 (5.2) 8 (5.6) 3 (4.2) 0.755 Bacterial pneumonia 54 (25.4) 39 (27.5) 15 (21.1) 0.404 Death 66 (31.0) 52 (36.6) 14 (19.7) 0.012* Mean hospital day of death 15.61 ± 16.64 16.75 ± 18.29 11.36 ± 6.94 0.285 Values are presented as mean ± standard deviation or number (%). BMI= body mass index, ESRD = end-stage renal disease, CKD = chronic kidney disease, O2 = oxygen, SBP = systolic blood pressure, WBC = white blood cell, ALT = alanine aminotransferase, LDH = lactate dehydrogenase, Cr = creatinine, CRP = C-reactive protein, MV= mechanical ventilation, CRRT = continuous renal replacement therapy, ECMO = extracorporeal membrane oxygenation. aOther treatments: antibiotics, prone position, immunoglobulins, etc. *P < 0.05, **P < 0.01, ***P < 0.001. Table 2 Predictors of mortality Risk factors Univariate analysis Multivariate analysis Survivor group (n = 147) Death group (n = 66) P OR 95% CI Age, yr 59.54 ± 19.22 68.89 ± 17.281 < 0.001*** 1.084 1.043–1.127 BMI ≥ 25 kg/m2 58 (39.7) 14 (21.5) 0.007** Fully vaccinated 43 (29.3) 23 (34.8) 0.427 Breakthrough infection 49 (33.3) 32 (48.5) 0.026* Underlying disease Any 126 (85.7) 60 (90.9) 0.205 Diabetes mellitus 56 (38.1) 23 (34.8) 0.384 Malignancy 22 (15.0) 16 (24.2) 0.077 Chronic liver disease 1 (0.6) 4 (6.1) 0.009 Taking immune suppressant 14 (9.7) 7 (10.6) 0.503 Initial manifestation Vasopressor use 23 (15.6) 19 (28.8) 0.022* Initial respiratory rate ≥ 22 breaths/min 105 (71.4) 38 (58.5) 0.046* Altered mentality 32 (21.8) 23 (34.8) 0.034* Initial laboratory findings BUN, mg/dL 26.31 ± 23.82 36.24 ± 24.16 0.006** LDH > 600 IU/L 22 (21.2) 25 (52.1) < 0.001*** 4.782 1.584–14.434 CRP, mg/dL 9.97 ± 9.79 14.16 ± 10.10 0.005** > 8 mg/dL 81 (55.1) 54 (81.8) < 0.001*** 2.940 1.042–8.293 Laboratory finding (Hosp) Highest WBC, 109/L 13.85 ± 5.97 23.20 ± 14.44 < 0.001*** > 12.000 81 (55.1) 54 (81.8) < 0.001*** Lowest platelet count < 100 109/L 40 (27.2) 37 (56.1) < 0.001*** Lowest hemoglobin count < 8 g/dL 43 (29.3) 34 (51.5) 0.002** Highest ALT > 200 IU/L 12 (8.2) 23 (34.8) < 0.001*** Highest AST > 200 IU/L 10 (6.8) 24 (36.4) < 0.001*** 12.931 1.691–98.908 Highest TB > 1.2 IU/L 31 (21.1) 31 (47.0) < 0.001*** Highest LDH > 600 IU/L 31 (26.5) 36 (67.9) < 0.001*** Highest Cr > 1.5 mg/dL 44 (29.9) 45 (68.2) < 0.001*** Highest CRP > 8 mg/dL 93 (63.3) 56 (84.8) 0.001** Highest procalcitonin, mg/dL 6.14 ± 14.51 15.67 ± 39.60 0.019* > 0.5 mg/dL 67 (53.2) 42 (75) 0.004** MV 47 (32.0) 43 (65.2) < 0.001*** 3.671 1.294–10.420 ECMO 2 (1.4) 5 (7.6) 0.031* CRRT 10 (6.8) 19 (28.8) < 0.001*** Complication Blood stream infection 8 (5.4) 7 (12.1) 0.080 Bacterial pneumonia 30 (20.4) 24 (36.4) 0.012* Other complication 24 (16.6) 21 (31.8) 0.011* Values are presented as mean ± standard deviation or number (%). OR = odds ratio, CI = confidential interval, BMI= body mass index, BUN= blood urea nitrogen, LDH = lactate dehydrogenase, CRP = C=reactive protein, WBC = white blood cell, ALT = alanine aminotransferase, AST = aspartate aminotransferase, TB = total bilirubin, Cr = creatinine, CRP = C-reactive protein, MV= mechanical ventilation, ECMO = extracorporeal membrane oxygenation, CRRT = continuous renal replacement therapy. *P < 0.05, **P < 0.01, ***P < 0.001. The ≤ 50 years group included a significantly higher percentage of patients with a BMI > 25 kg/m2 than did the > 50 years group (49.3% vs. 26.4%, P < 0.001) (Table 1); however, overall, there were fewer patients with a BMI of ≥ 25 kg/m2 during this omicron epidemic period than that during the delta epidemic period (Supplementary Table 1). The number of patients aged ≤ 50 years and had underlying diseases was also significantly lesser in the omicron epidemic period than in the delta epidemic period (73.2% vs. 94.4%, P < 0.001) (Table 1), but the number of underlying diseases in the patient was nearly three times higher than that in the delta epidemic period (Supplementary Table 1). Only 6 (8.4%) of the 71 critical patients aged ≤ 50 years had no risk factors (BMI ≥ 25 kg/m2 or underlying diseases). The most frequent underlying diseases in those aged ≤ 50 years were diabetes mellitus (28.2%) and hypertension (22.5%) followed by immunosuppression (15.7%). Patients aged ≤ 50 years had higher oxygen saturation (91.68 ± 6.18 vs. 88.90 ± 9.05, P = 0.001) and had more frequent thrombocytopenia (19.7% vs. 9.2%, P = 0.047) than those aged >50 years. There was no significant difference between the two groups in the usage of MV and CRRT during the treatment course. However, ECMO was used significantly more in patients aged ≤ 50 years, which was attributed to the reduced usage of futile ECMO in elderly patients (7% vs. 1.4%; P = 0.043). The mean time to death in patients aged ≤ 50 years was 11.36 ± 6.94 days, shorter than that in patients aged > 50 years (16.75 ± 18.29 days) (Table 3). Table 3 Risk factors for mortality in the young (age < 50 years) critical patients (N = 71) Risk factors Survivor group (n = 57) Death group (n = 14) OR (95% CI) P Female sex 27 (47.4%) 8 (7.1%) 0.675 (0.208–2.195) 0.563 Age, yr 38.75 ± 8.83 42.57 ± 8.96 0.153 BMI ≥ 25 kg/m2 30 (52.6%) 5 (35.7%) 0.737 (0.457–1.187) 0.372 Any underlying or BMI ≥ 25 kg/m2 51 (89.5%) 14 (100.0%) 1.117 (1.022–1.221) 0.591 Breakthrough infection 13 (22.8%) 6 (42.9%) 2.538 (0.745–8.651) 0.178 Underlying disease Any 40 (70.2%) 12 (85.7%) 2.550 (0.514–12.642) 0.324 DM 16 (28.1%) 4 (28.6%) 1.025 (0.281–3.744) 1.000 Hypertension 14 (24.6%) 2 (14.3%) 0.880 (0.679–1.142) 0.501 Heart disease 5 (8.8%) 1 (7.1%) 0.982 (0.832–1.160) 1.000 ESRD or CKD 7 (12.3%) 2 (14.3%) 1.023 (0.809–1.294) 1.000 Chronic neurologic disease 7 (12.3%) 1 (7.1%) 0.945 (0.793–1.125) 1.000 Solid tumor 4 (7.0%) 3 (21.4%) 1.183 (0.892–1.570) 1.000 Initial vasopressor use 9 (15.8%) 7 (50.0%) 1.684 (0.986–2.878) 0.010* RR over 22 breath/min 41 (71.9%) 5 (38.5%) 0.456 (0.251–0.829) 0.048* Initial laboratory results Platelet count < 100 109/L 9 (15.8%) 5 (35.7%) 1.310 (0.879–1.967) 0.132 Creatinine > 1.5 U/L 13 (22.8%) 4 (28.6%) 1.081 (0.754–1.549) 0.730 LDH > 600 U/L 15 (26.3%) 5 (35.7%) 1.385 (0.640–2.995) 0.460 CRP > 8 mg/L 24 (42.1%) 6 (42.9%) 1.158 (0.613–2.216) 0.751 Laboratory findings (during treatment) Lowest neutrophil count < 1 109/L 5 (8.8%) 5 (35.7%) 1.501 (1.013–2.225) 0.003** Lowest platelet count < 100 109/L 16 (28.1%) 11 (78.6%) 3.357 (1.215–9.272) 0.001** Highest ALT > 200 IU/L 10 (17.5%) 7 (50.0%) 1.649 (0.964–2.822) 0.031* Highest AST > 200 IU/L 6 (10.5%) 10 (71.4%) 3.132 (1.361–7.203) < 0.001*** Highest LDH > 600 U/L 16 (35.6%) 7 (77.8%) 2.900 (0.838–10.035) 0.028* Highest creatinine > 1.5 U/L 15 (26.3%) 12 (85.7%) 5.158 (1.416–18.783) < 0.001*** Highest procalcitonin >0.5 mg/dL 27 (52.9%) 10 (90.9%) 5.176 (0.781–34.309) 0.038* MV 20 (5.1%) 11 (78.6%) 3.029 (1.091–8.409) 0.006** ECMO 2 (3.5%) 3 (21.4%) 1.228 (0.90–1.622) 0.049* CRRT 4 (7.0%) 9 (64.3%) 2.604 (1.285–5.726) < 0.001*** Complication Blood stream infection 6 (10.5%) 1 (7.1%) 0.964 (0.813–1.143) 1.000 Fungal infection 2 (3.5%) 1 (7.1%) 1.039 (0.891–1.212) 0.488 Bacterial pneumonia 11 (19.3%) 4 (28.6%) 1.130 (0.792–1.611) 0.475 OR = odds ratio, CI = confidential interval, BMI = body mass index, DM = diabetes mellitus, ESRD = end-stage renal disease, CKD = chronic kidney disease, RR = respiratory rate, LDH = lactate dehydrogenase, CRP = C=reactive protein, ALT = alanine aminotransferase, AST = aspartate aminotransferase, LDH = lactate dehydrogenase, MV= mechanical ventilation, ECMO = extracorporeal membrane oxygenation, CRRT = continuous renal replacement therapy. *P < 0.05, **P < 0.01, ***P < 0.001. Risk factors for mortality in the whole population The total case fatality rate among severe patients (S-CFR) was 31.0%, higher than the 21.0% in the delta period (Supplementary Table 1). The fatality rate increased with age, from 19.7% in those aged ≤ 50 years to 36.6% in those aged > 50 years and 38.2% in those aged ≥ 65 years. Particularly, the mortality rate in patients aged ≤ 50 years was higher during the omicron epidemic than during the delta epidemic (Fig. 1, Supplementary Table 1). With respect to the survival rate by age, the mortality rate significantly increased as the age increased in the groups aged 50 years or younger (19.7%), 50–65 years (36.6%), and 65 years or older (38.2%) (P = 0.032) (Fig. 1). In MV patients, the mortality rate in those aged ≤ 50 years was 32%, lower than the mortality rate in those aged ≥ 65 years (46.2%). In ECMO patients, the mortality rate of those aged ≤ 50 years was close to 50% (Fig. 1). Fig. 1 Severe case fatality rate by the highest respiratory support and age group. MV = mechanical ventilation, ECMO = extracorporeal membrane oxygenation. Table 2 shows the analysis results of predictive factors of mortality in all critical patients (N = 213) during the omicron wave period. The frequency of vasopressor use (28.8% vs. 15.6%, P = 0.022) and decreased consciousness (34.8% vs. 21.8%, P = 0.034) were significantly higher in the death group than in the survivor group. This finding along with the significant differences in initial laboratory findings indicate that the severity of disease at the time of initial hospitalization was higher in the death group than in the survivor group (Table 2). In the univariate analysis, age, breakthrough infection, initial vasopressor use, initial respiratory rate ≥ 22 breaths/min, altered mental status, initial blood urea nitrogen (BUN) >20 mg/dL, initial lactate dehydrogenase (LDH) > 600 IU/L, initial C-reactive protein (CRP) >8 mg/dL, poor laboratory findings during treatment (thrombocytopenia, low-density lipoprotein [LDL] elevation, creatinine elevation, and CRP elevation), application of MV, ECMO, or CRRT, and occurrence of bacterial pneumonia all showed significant differences between the death and survivor groups. The results of the multivariate analysis showed that five factors, age (OR, 1.084; 95% CI, 1.043–1.127), initial LDL > 600 IU/L (OR, 4.782; 95% CI, 1.584–1434), initial CRP > 8 mg/dL (OR, 2.940; 95% CI, 1.042–8.293), highest AST > 200 IU/L (OR, 12.931; 95% CI, 1.691–98.908), and application of MV (OR, 3.671; 95% CI, 1.294–10.420), were independent predictors of mortality in critical COVID-19 patients (Table 2). Initial saturation, infiltration, and treatment drugs (steroid, baricitinib, tocilizumab, Paxlovid, Lagevrio, remdesivir) had no significant effect on mortality (data not shown). The vaccination completion rate in all critical patients was only 31%, and there was no difference between the nonsurvivor and survivor groups (34.8% vs. 29.3%, P = 0.427) (Table 2). Risk factors for mortality by age group A subgroup analysis was performed in critical patients aged ≤ 50 years based on the characteristics evaluated in all critical COVID-19 patients (Table 3). Of the 71 patients, 14 patients died, corresponding to a significantly higher mortality rate of 19.7% during the omicron epidemic than the 5.6% rate during the delta epidemic (Supplementary Table 2). The proportion of patients with a BMI ≥ 25 kg/m2 was lower in the death group than in the survival group (35.7% vs. 52.6%). This indicated that the survival group included a greater number of obese patients, but the difference was not statistically significant (P = 0.372). All 14 patients aged ≤ 50 years (100%) who died had an underlying disease or had a BMI ≥ 25 kg/m2. Only two patients who died had no underlying disease, and 70% of the survivors also had underlying diseases. The importance of BMI decreased relatively as the severity of hospitalized critical patients increased (Table 3, Supplementary Table 3). There was no significant difference between the death and survivor groups with respect to the initial laboratory results, and most of the laboratory tests during treatment showed significantly worse values in patients aged ≤ 50 years (Table 3). MV, ECMO, and CRRT were all significantly more common in critical patients aged ≤ 50 years (Table 3). In elderly critical patients aged > 50 years, 92.3% of patients in the death group had underlying disease. Further, similar to the entire critical patient group, the risk of death was high according to age; state of consciousness at admission; and the initial laboratory results such as platelet count, BUN, Cr, LDH, and CRP (Supplementary Tables 3 and 4). In patients who were older than 50 years, both MV (61.5% vs. 30.0%, P < 0.001) and CRRT (19.2% vs. 6.7%, P = 0.029) were used significantly more often in the death group. ECMO was not performed in any of the patients in the survivor group and in only two patients in the death group (3.8%), reflecting a lower ECMO implementation rate in the elderly than that during the delta wave (Supplementary Tables 2 and 4). MV, ECMO, and CRRT were significantly more frequently performed in the death group, both in patients aged ≤ 50 years and > 50 years. Patients who had leukocytosis, lymphopenia, thrombocytopenia, AST/alanine aminotransferase levels 5 times higher than the upper limit of normal, LDH > 600 IU/L, Cr > 1.5 mg/dL, and procalcitonin > 0.5 during treatment had a significantly higher mortality rate (Table 3, Supplementary Table 4), while there was no difference in mortality rate according to chest radiography results and treatment at admission (data not shown). The completion rate of vaccination in the patients aged ≤ 50 years was only 21.3%, which corresponded to only one-third of the national average at 87%. There was no difference in the vaccination rate between the survivor and death groups, possibly because of the low vaccination completion rate in both groups. General characteristics of deceased patients younger than 50 years In this study, 14 patients aged ≤ 50 years (19.7%) died; among them, 12 patients had underlying diseases, and 9 patients had a BMI of ≥ 25 kg/m2 (Supplementary Table 5). Overall, more than half of the patients had underlying diseases and had two or more underlying diseases. At the time of admission, 12 patients had findings of bilateral chest infiltration, and 7 patients (50%) had decreased consciousness or use of vasopressors from the time of admission. One patient developed complications of CLABSI and hospital-acquired pneumonia, and the mean time to death was 11.35 days. Although the national vaccination rate was over 80% during the study period, 50% of the patients aged ≤ 50 years who died were not vaccinated (never vaccinated, n = 7), and only three patients, who were all immunocompromised (having a tumor or taking immunosuppressants), received the third vaccine dose. Comparison of clinical characteristics between periods Data of severe patients with COVID-19 hospitalized during the delta period (between July 1, 2021 and November 30, 2021) were compared to those of severe patients with COVID-19 hospitalized during the omicron period. The overall mortality rate increased from 21.0% during the delta period to 30.6% during the omicron period. Particularly, the mortality rate in those aged ≥ 65 years was not significantly different between the delta and omicron periods (38.1% vs. 38.2%). In contrast, the mortality rate of in those aged ≤ 50 years was more than three times higher in the omicron period than in the delta period (19.7% vs. 5.6%, P = 0.003) (Supplementary Table 2). With respect to the route of infection, the rate of unknown infection route was significantly higher during the omicron period than during the delta period (72.8% vs. 33.9%, P < 0.001). Although the vaccination completion rate increased from 13.6% in the delta period to 31.0% (P < 0.001), the overall vaccination completion rate was still only 31.0% (Supplementary Table 1). When examining the underlying diseases and risk factors, the percentage of patients with a BMI of 25 kg/m2 was lower in the omicron period than in the delta period (34.1% vs. 53.4%, P < 0.001), but the frequency of the underlying disease was higher than in the delta period (87.3% vs. 60.9%, P < 0.001). With respect to underlying disease, there were significantly more patients with any underlying disease or BMI ≥ 25 kg/m2 in the omicron period than in the delta period (93.9% vs. 82.4%, P < 0.001). Particular, all patients aged ≤ 50 years who died during the omicron period either had any underlying disease or had a BMI ≥ 25 kg/m2. The rate of initial vasopressor usage or altered mental status was higher in the omicron period than in the delta period regardless of age, and both MV and CRRT usage rates were significantly higher in the omicron period (P < 0.001) (Supplementary Table 1). With respect to treatment, Paxlovid and Lagevrio were newly used in the omicron period, and the frequency of other treatments (plasma treatment, antibiotics, and conservative treatment) significantly decreased from 22.7% in the delta period to 3.3% in the omicron period as a standard treatment was established (P < 0.001) (Supplementary Table 1). In contrast to the higher mortality in the omicron period, the rate of complications such as CLABSI and fungal infection did not differ significantly between the omicron and the delta periods. However, the incidence of bacterial pneumonia was approximately 10% higher in the omicron period (21.0% vs. 31.0%, P = 0.008). The period from the time of diagnosis to death was shorter by 5 days in the omicron period than in the delta period (16.68 ± 16.758 days vs. 22.51 ± 17.137 days, P = 0.032) (Supplementary Table 1). For patients aged ≤ 50 years, the percentage of those with BMI ≥ 25 kg/m2 significantly decreased from 69.7% in the delta period to 49.3% in the omicron period (P = 0.004), but there was no significant difference in the percentage of patients with BMI ≥ 30 kg/m2 between the two periods. Meanwhile, the percentage of patients who had any underlying disease or BMI ≥ 25 kg/m2 was higher by more than 10% in the omicron period than in the delta period (77.5% vs. 91.5%, P = 0.013) (Supplementary Table 2). The proportion of patients with underlying diseases more than doubled to 73.2%, and the number of underlying diseases in each patient also increased. DISCUSSION A previous study21 conducted in early 2022 analyzed the fatality rate and mortality risk factors in critical COVID-19 patients aged ≤ 50 years who received HFNC or higher respiratory therapy in the delta period, and we attempted to identify the specific characteristics of critical patients aged ≤ 50 years with the same condition in the omicron period. The overall S-CFR reported in the current study was 30.6%, much higher than the overall mortality rate of 1.54% based on the WHO data.1 Further, it was higher by more than 10% compared to the 21% mortality rate reported by a study conducted during the delta period in patients with similar characteristics and conditions. By age, the CFR for those aged ≤ 50 years increased from 5.6% to 19.7%, which was steeper than that in those aged > 50 years (from 29.3% to 36.6%). The CFR in the current study differs from those in epidemiological reports of the overall COVID-19 omicron pandemic. This is because the CFR analyzed in the current study is the S-CFR, which is the rate of death in severely ill patients. One of the important findings of this study is that unlike the overall population, severely ill patients had high S-CFR, and additional research for other variants are needed. Among the critical patients in the omicron period, in addition to age, initial LDL >600 IU/L, initial CRP > 8 mg/dL, highest AST > 200 IU/L, and MV implementation were significant independent risk factors of mortality, consistent with previous studies.212324 Meanwhile, BMI, underlying disease, and vaccination status were not significant factors, possibly owing to the higher prevalence of BMI > 25 kg/m2 and underlying disease in the entire critical patient group than in the general patient population and to the small proportion of fully vaccinated patients. In a study on adults aged ≥ 65 years in South Korea,24 HAI, diabetes mellitus, chronic lung disease, chronic neurologic disease, hypoxia, altered mental status, and CRP > 8.0 mg/dL were reported as risk factors for mortality in COVID-19 patients. Unlike data from the delta wave, no significance was observed for underlying diseases in both univariate and multivariate analyses in this study. This result is thought to be because underlying disease is a common risk factor for severity and death. Although the fatality rate was high in the patients with underlying diseases, it did not show any significance as the rate of underlying diseases increased in all critical patients. In contrast to previous findings of underlying diseases such as nosocomial infection24 and hypertension being a risk factor, the current study found no significance for underlying diseases. This is believed to be possibly owing to the higher frequency of underlying disease among all patients in the omicron period than in the delta period. The current study showed that the frequency of underlying disease in critical patients was significantly higher in the omicron period than in the delta period (87.3% vs. 60.9%, P < 0.001). The KDCA report3 showed that 16,118,490 cases were confirmed during the omicron outbreak (fifth period), which corresponded to 95.2% of the cumulative confirmed cases. The daily average number of confirmed cases was 187,424, with a minimum of 17,075 and a maximum of 621,177. This corresponded to 60 times the daily average number of confirmed cases with a total of 649,534 confirmed cases and 3,137.8 daily average confirmed cases during the delta period (fourth period). As mentioned previously, reports have shown that it is difficult to find consistent mortality predictors except for age in large-scale studies in the real world, and the actual prediction rates of several known risk factors are even not high.2526 Moreover, further studies are needed to explore whether the underlying disease status, age, or the severity of condition at the time of admission predicts morbidity and mortality in a large-scale epidemic. This will be helpful in efficient bed management in the future. With limited resources (hospitals, health care workers), it is crucial to predict the risk of aggravation or mortality of infectious disease because it is a key component of hospital operation. Although 73.2% of critical patients aged ≤ 50 years during the omicron wave had an underlying disease, Paxlovid and Lagevrio were not prescribed. This was despite their proven effectiveness in preventing exacerbations and their recommendation to be prescribed in patients with COVID-19 with underlying medical diseases. Paxlovid was initially allowed only for limited patients; these included patients aged ≥ 65 years, immunosuppressed patients aged ≥ 12 years, and patients with underlying diseases aged ≥ 50 years or older. After February 21, 2022, the indication for Paxlovid was expanded to senior citizens aged ≥ 60 years, patients with underlying diseases aged ≥ 40 years, and immunocompromised patients aged ≥ 12 years. However, the prescription rate was still very low. This may be proof that there are many cases of late detection in young people who have already progressed to severe disease at the time of diagnosis. Further, it can also be reflection of the low rates of drug prescription for COVID-19 in the general public. The March report of the KDCA27 showed that Paxlovid, which was introduced on January 13, 2022, was prescribed for only 47,000 cases nationwide over a 2-month period, supporting the above finding. Currently used treatment drugs for COVID-19 have already been proven to be effective.272829 It is considered that early pharmacological intervention in patients with underlying diseases and risk factors (e.g., obesity), along with an active initial diagnosis, will markedly reduce the severity and mortality related to COVID-19 infection. A previous study reported that compared to critical patients aged > 50 years, critical patients aged ≤ 50 years had a higher BMI and a lower fatality rate because of a lower severity of their condition at admission. The current study also found that the average BMI of patients in this age group was higher than that of the general population. Further, for patients in this age group, those critical patients during the omicron period had worse S-CFR at the time of hospitalization than those infected in the delta period. Regarding obesity, 49.3% of all critical patients aged ≤ 50 years were obese (BMI ≥ 25 kg/m2), which is higher than the average in adults. In a study evaluating Korean pediatric COVID-19 patients,30 8 of the 39,000 pediatric patients had severe disease, and their mean BMI was 29.3 kg/m2, higher than the national average. Similar trends were observed in our study. Overseas studies1431 also support the hypothesis that obesity is associated with exacerbation of COVID-19 and mortality. One study explored the mechanism by which obesity is associated with the critical severity of COVID-1932 and found that the cytokine storm caused by interleukin-6 secretion by adipocytes increased the risk of complications and led to hospitalization and intensive care unit admission. A recent report33 also showed that 38% of the adult population in 2020 had a BMI ≥ 25 kg/m2; accordingly, proper COVID-19 management in the increasing obese population should be emphasized. In the current study, among the 14 patients aged ≤ 50 years who died, 5 patients had a BMI of ≥ 25 kg/m2 (35.7%) (Supplementary Table 5), which was not significantly different from the average BMI of adults. However, when the underlying diseases were also considered, all of these 14 patients who died had an underlying disease or BMI ≥ 25 kg/m2. These results can be explained by the relatively low evaluation of BMI in critically ill patients who died as the epidemic increased and the number of critically ill patients increased. Furthermore, some studies abroad have reported that omicron has a lower correlation with BMI than delta,7 which requires further investigation. Examining Tables 1-3 and Supplementary Table 1, the laboratory test results or the severity of condition at the time of hospitalization increased significantly in the omicron period from the delta period, and many patients with severe conditions were hospitalized. We hypothesized that this was because of the difference in the epidemic size rather than the effect of the omicron variant itself. As previously mentioned, the total number of confirmed cases during the omicron epidemic was 16,118,490, with an average of 187,424.3 daily confirmed cases, and this was 60 times higher than that during the delta epidemic. The number of beds for severe cases differed only by 500 beds from the maximum estimate of 2,360 beds in the delta period to the maximum estimate of 2,825 (private + public) in the omicron period. As a result, the severity of severe hospitalized patients seems to have differed significantly between the periods. There have been reports that the epidemic during the omicron period was more large scale in South Korea than in other countries because the confirmation rate in previous epidemics was not high.89 Accordingly, it is thought to have shown different characteristics compared to those of previous epidemics. Hospitals have limited capacity for increasing the number of beds, especially for critically ill patients, owing to limited human resources and facility equipment. Thus, efficient management of severely ill patients using limited beds is more emphasized. In addition, it will be of great help if further studies are conducted for each variant to determine which of the various risk factors are significant in large-scale epidemic situations. The disease severity, vaccine effectiveness, and mortality rate of the omicron variant infection have not yet been standardized, and findings have been conflicting. Most studies showed that the infection caused by the omicron variant has lower severity and mortality rates than those caused by the delta or alpha variant.3435 However, there is also evidence that omicron variant has low severity but not mortality.36 Although some studies reported that for the omicron variant, the probability of critical aggravation was 0.23%,37 the infectivity was 13 times higher. 38 Moreover, the overall number of critical cases did not seem to decrease significantly, considering the high reinfection rate and immune evasion. In addition, there are still unknown effects of the omicron variant in the real world because there are reports of increasing deterioration of other diseases, 39 in addition to the mortality rate being not low or rather high in specific patient groups,40 unvaccinated groups,618 or partially vaccinated groups.1941 Studies from other countries540 have also reported that it is difficult to predict the severity, vaccine effectiveness, and mortality rate of the omicron variant infection. The main reasons why it is challenging to predict the effect of COVID-19 variants like omicron are related to the acquisition of immunity from vaccination and previous infections and the need to analyze different risk factors in specific groups. Studies have shown high mortality related to the omicron variant in certain patient groups,40 including the hospitalized, solid organ transplantation, chronic dialysis, and hematologic malignancy patients. In this study, only Korean critical patients who needed hospitalization were evaluated, and although there was a difference by age, the mortality rate was as high as 19.7% even for those aged 50 years or younger. These results highlight that with the omicron variant, disease severity at the time of diagnosis was worse in certain groups of patients, and further studies should explore this in other critical patients. Some studies also that showed high mortality in the unvaccinated618 or partially vaccinated1941 patients, and consistent results were found in this study. The mortality differs by 3 to 23 times between unvaccinated and vaccinated individuals, but several studies in Europe,41 South America,19 Africa,6 and Asia18 consistently reported a low mortality rate with the omicron variant owing to the effect of vaccination and immunity. Although the vaccination rate of critical patients during the omicron period was significantly higher than that during the delta period, it was still only 31.0%, which corresponded to less than 50% the national data. In addition, only 15 of the 71 critical patients aged ≤ 50 had completed their vaccination, accounting for only 20%. In particular, only 3 of the 14 patients aged ≤ 50 years who died were vaccinated, showing that the vaccination completion rate was relatively low for all critically ill patients. In this study, which recruited critical patients, although there was no significant difference in the vaccination completion rate between the survivor and death groups, the results highlighted a low vaccination rate in the overall population of critically ill patients. This study has several limitations. First, there were many cases in which the variant type was not identified, and these cases could not be analyzed according to the variant. However, considering the KCDA report42 that more than 90% of all patients during the omicron period had omicron and sub-variants, it was thought there would be no significant bias even if all patients were considered to have an omicron variant. Second, as pointed out in the previous study, we only included severely ill inpatients, which may not be representative of all COVID-19 patients. It is challenging to directly compare the fatality rate before and after the omicron period. Death and survival were compared among critically ill patients, and thus, it was possible that factors found to be significant in other studies (e.g., vaccination and obesity) were found to be insignificant in the current study. While we recognize the limitations of our study design, it is important to note that the title of our study specifically focuses on severely ill patients, which is consistent with the study population we included. Despite these limitations, our study provides important insights into the impact of the Omicron variant on critically ill COVID-19 patients aged ≤ 50 years. Finally, there may be outcome variables that did not derive a significant outcome value, as there were only 14 deaths among those aged ≤ 50 years. This is believed to be the result of restricting the study population to a specific age range and to specific time periods of the epidemic. Nevertheless, the 14 deaths in patients aged ≤ 50 is significant considering their age and the fact that they are critical patients. We tried to compensate for this by analyzing characteristics of 14 deaths (Supplementary Table 5). In addition, we compared the delta wave group and the omicron wave group by recruiting the same patient population (i.e., those who had similar characteristics). Compared to other studies, the current study included a larger number of critical patients aged ≤ 50 years, and data were compared between periods of variant infection. The findings provide important baseline evidence for determining the risk factors of critical disease in the future. Furthermore, because this study was a multicenter study that conducted an additional comparison between epidemic periods, the results have important implications in developing the risk factor criteria for critical patients with COVID-19 omicron infection. In conclusion, critically ill COVID-19 patients aged ≤ 50 years infected during the omicron period, more than 70% patients have underlying disease. Further, disease severity at the time of hospitalization was worse than that during the delta period owing to the increased epidemic size and the limited number of beds. The effect of obesity on disease severity was relatively low, and the S-CFR reached 20% even for those aged ≤ 50 years. Compared with general population, those with severe infection in the overall population, the age ≤ 50 years group, and the age > 50 years group had significantly lower vaccination rates. All the patients aged ≤ 50 years who died had an underlying disease or had a BMI of ≥ 25 kg/m2. Notably, there was a lack of prescription for Paxlovid for these patients although they satisfied the prescription criteria. Although the severity of the omicron variant epidemic was expected to be relatively low, the total number of critical patients increased as the epidemic size increased. In addition, there was a possibility of progression to critical illness or death in young adult patients infected with the omicron variant. Early diagnosis and active initial treatment was necessary, along with the proven methods of vaccination and personal hygiene. Further studies are needed to explore the difference in mortality among SARS-CoV-2 variants, especially in critical patients. SUPPLEMENTARY MATERIALS Supplementary Table 1 Clinicodemographic patient characteristics by variant period Supplementary Table 2 Clinicodemographic characteristics by variant period in patients aged ≤ 50 years Supplementary Table 3 Predictors of mortality (univariate and multivariate analysis) Supplementary Table 4 Risk factors for mortality in patients aged >50 years (N = 142, univariate analysis) Supplementary Table 5 Clinical characteristics of patients aged ≤ 50 years who died Funding: This work was supported by the Gachon University research fund of 2022 (GCU-2022-202209660001). This work was supported by Organization of academic resource management system for clinical trial of infectious disease (HE20C002201139820037). Disclosure: The authors have no potential conflicts of interest to disclose. Author Contributions: Conceptualization: Shi HJ, Yang J, Jung SI, Kim S, Seok H, Kim B. Data curation: Yang J. Formal analysis: Kym S. Funding acquisition: Kim B, Wi YM, Kym S. Investigation: Shi HJ, Ko JH, Wi YM, Lim S. Methodology: Peck KR, Kim UJ, Kim HA, Wi YM, Lim S. Project administration: Jung SI, Kim HA. Resources: Kim S, Seok H. Software: Seok H, Hyun M, Cheong HS, Jeon CH. Supervision: Joo EJ, Cheong HS, Jeon CH, Kim J, Park Y. Validation: Hyun M, Joo EJ, Jeon CH, Kim J, Park Y. Visualization: Shi HJ. Writing - original draft: Shi HJ. Writing - review & editing: Shi HJ, Eom JS, Peck KR, Park Y. ==== Refs 1 World Health Organization Weekly Epidemiological Update on COVID-19 - 24 August 2022 Geneva, Switzerland World Health Organization 2022 2 Korea Disease Control and Prevention Agency Variants of COVID-19 Updated 2021 Accessed August 30, 2022 https://ncv.kdca.go.kr/hcp/page.do?mid=060102 3 Korea Disease Control and Prevention Agency Analysis of the COVID-19 Outbreak During the Category 1 Statutory Infectious Disease Designation Period Cheongju, Korea Korea Disease Control and Prevention Agency 2022 4 Kim MK Lee B Choi YY Um J Lee KS Sung HK Clinical characteristics of 40 patients infected with the SARS-CoV-2 omicron variant in Korea J Korean Med Sci 2022 37 3 e31 35040299 5 Bhattacharyya RP Hanage WP Challenges in inferring intrinsic severity of the SARS-CoV-2 omicron variant N Engl J Med 2022 386 7 e14 35108465 6 Ribeiro Xavier C Sachetto Oliveira R da Fonseca Vieira V Lobosco M Weber Dos Santos R Characterisation of Omicron Variant during COVID-19 Pandemic and the Impact of Vaccination, Transmission Rate, Mortality, and Reinfection in South Africa, Germany, and Brazil BioTech (Basel) 2022 11 2 12 35822785 7 Bouzid D Visseaux B Kassasseya C Daoud A Fémy F Hermand C Comparison of patients infected with delta versus omicron COVID-19 variants presenting to Paris emergency departments : a retrospective cohort study Ann Intern Med 2022 175 6 831 837 35286147 8 Shim E Choi W Kwon D Kim T Song Y Transmission potential of the omicron variant of severe acute respiratory syndrome coronavirus 2 in South Korea, 25 November 2021-8 January 2022 Open Forum Infect Dis 2022 9 7 ofac248 35855956 9 Lee JJ Choe YJ Jeong H Kim M Kim S Yoo H Importation and transmission of SARS-CoV-2 B.1.1.529 (omicron) variant of concern in Korea, November 2021 J Korean Med Sci 2021 36 50 e346 34962117 10 Wu C Chen X Cai Y Xia J Zhou X Xu S Risk factors associated with acute respiratory distress syndrome and death in patients with coronavirus disease 2019 pneumonia in Wuhan, China JAMA Intern Med 2020 180 7 934 943 32167524 11 Jang JG Hur J Choi EY Hong KS Lee W Ahn JH Prognostic factors for severe coronavirus disease 2019 in Daegu, Korea J Korean Med Sci 2020 35 23 e209 32537954 12 Kang SJ Jung SI Age-related morbidity and mortality among patients with COVID-19 Infect Chemother 2020 52 2 154 164 32537961 13 Heald-Sargent T Muller WJ Zheng X Rippe J Patel AB Kociolek LK Age-related differences in nasopharyngeal severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) levels in patients with mild to moderate coronavirus disease 2019 (COVID-19) JAMA Pediatr 2020 174 9 902 903 32745201 14 Murillo-Zamora E Aguilar-Sollano F Delgado-Enciso I Hernandez-Suarez CM Predictors of laboratory-positive COVID-19 in children and teenagers Public Health 2020 189 153 157 33246302 15 Steinberg E Wright E Kushner B In young adults with COVID-19, obesity is associated with adverse outcomes West J Emerg Med 2020 21 4 752 755 32726235 16 Lu Y Huang Z Wang M Tang K Wang S Gao P Clinical characteristics and predictors of mortality in young adults with severe COVID-19: a retrospective observational study Ann Clin Microbiol Antimicrob 2021 20 1 3 33407543 17 Forchette L Sebastian W Liu T A comprehensive review of COVID-19 virology, vaccines, variants, and therapeutics Curr Med Sci 2021 41 6 1037 1051 34241776 18 Smith DJ Hakim AJ Leung GM Xu W Schluter WW Novak RT COVID-19 mortality and vaccine coverage - Hong Kong Special Administrative Region, China, January 6, 2022-March 21, 2022 MMWR Morb Mortal Wkly Rep 2022 71 15 545 548 35421076 19 Martins-Filho PR de Souza Araújo AA Quintans-Júnior LJ Soares BD Barboza WS Cavalcante TF Dynamics of hospitalizations and in-hospital deaths from COVID-19 in northeast Brazil: a retrospective analysis based on the circulation of SARS-CoV-2 variants and vaccination coverage Epidemiol Health 2022 44 e2022036 35413166 20 Singanayagam A Hakki S Dunning J Madon KJ Crone MA Koycheva A Community transmission and viral load kinetics of the SARS-CoV-2 delta (B.1.617.2) variant in vaccinated and unvaccinated individuals in the UK: a prospective, longitudinal, cohort study Lancet Infect Dis 2022 22 2 183 195 34756186 21 Shi HJ Nham E Kim B Joo EJ Cheong HS Hong SH Clinical characteristics and risk factors for mortality in critical coronavirus disease 2019 patients 50 years of age or younger during the delta wave: comparison with patients > 50 years in Korea J Korean Med Sci 2022 37 22 e175 35668685 22 Korea Disease Control and Prevention Agency COVID-19 domestic outbreaks Accessed February 03, 2022 https://www.kdca.go.kr/board/boardApi.es?mid=a20507050000&bid=0080&api_code=20220214 23 Liu Z Liu J Ye L Yu K Luo Z Liang C Predictors of mortality for hospitalized young adults aged less than 60 years old with severe COVID-19: a retrospective study J Thorac Dis 2021 13 6 3628 3642 34277055 24 Lee JY Kim HA Huh K Hyun M Rhee JY Jang S Risk factors for mortality and respiratory support in elderly patients hospitalized with COVID-19 in Korea J Korean Med Sci 2020 35 23 e223 32537957 25 Capuzzo M Amaral AC Liu VX Assess COVID-19 prognosis … but be aware of your instrument’s accuracy! Intensive Care Med 2021 47 12 1472 1474 34608529 26 He F Page JH Weinberg KR Mishra A The development and validation of simplified machine learning algorithms to predict prognosis of hospitalized patients with COVID-19: multicenter, retrospective study J Med Internet Res 2022 24 1 e31549 34951865 27 Jeon HW [COVID-19] Paxlovid was introduced in Korea about two month ago... but still perscribed only 47000 patients Aju Business Daily 2022 03 16 28 Drożdżal S Rosik J Lechowicz K Machaj F Szostak B Przybyciński J An update on drugs with therapeutic potential for SARS-CoV-2 (COVID-19) treatment Drug Resist Updat 2021 59 100794 34991982 29 Wen W Chen C Tang J Wang C Zhou M Cheng Y Efficacy and safety of three new oral antiviral treatment (molnupiravir, fluvoxamine and Paxlovid) for COVID-19: a meta-analysis Ann Med 2022 54 1 516 523 35118917 30 Lee H Choi S Park JY Jo DS Choi UY Lee H Analysis of critical COVID-19 cases among children in Korea J Korean Med Sci 2022 37 1 e13 34981683 31 Bienvenu LA Noonan J Wang X Peter K Higher mortality of COVID-19 in males: sex differences in immune response and cardiovascular comorbidities Cardiovasc Res 2020 116 14 2197 2206 33063089 32 Lim S Shin SM Nam GE Jung CH Koo BK Proper management of people with obesity during the COVID-19 pandemic J Obes Metab Syndr 2020 29 2 84 98 32544885 33 Jung E Four out of 10 people in the country are obese... COVID-19 obesity rate ‘Biggest Ever’ Korean Financial 2022 03 15 34 Ward IL Bermingham C Ayoubkhani D Gethings OJ Pouwels KB Yates T Risk of covid-19 related deaths for SARS-CoV-2 omicron (B.1.1.529) compared with delta (B.1.617.2): retrospective cohort study BMJ 2022 378 e070695 35918098 35 Nyberg T Ferguson NM Nash SG Webster HH Flaxman S Andrews N Comparative analysis of the risks of hospitalisation and death associated with SARS-CoV-2 omicron (B.1.1.529) and delta (B.1.617.2) variants in England: a cohort study Lancet 2022 399 10332 1303 1312 35305296 36 Lauring AS Tenforde MW Chappell JD Gaglani M Ginde AA McNeal T Clinical severity of, and effectiveness of mRNA vaccines against, covid-19 from omicron, delta, and alpha SARS-CoV-2 variants in the United States: prospective observational study BMJ 2022 376 e069761 35264324 37 Chen X Yan X Sun K Zheng N Sun R Zhou J Estimation of disease burden and clinical severity of COVID-19 caused by omicron BA.2 in Shanghai, February-June 2022 Emerg Microbes Infect 2022 11 1 2800 2807 36205530 38 Aleem A Akbar Samad AB Vaqar S Emerging Variants of SARS-CoV-2 And Novel Therapeutics Against Coronavirus (COVID-19) Treasure Island, FL, USA StatPearls 2023 39 Ahmad Malik J Ahmed S Shinde M Almermesh MH Alghamdi S Hussain A The impact of COVID-19 on comorbidities: a review of recent updates for combating it Saudi J Biol Sci 2022 29 5 3586 3599 35165505 40 Anjan S Khatri A Viotti JB Cheung T Garcia LA Simkins J Is the Omicron variant truly less virulent in solid organ transplant recipients? Transpl Infect Dis 2022 24 6 e13923 35915957 41 Pinato DJ Aguilar-Company J Ferrante D Hanbury G Bower M Salazar R Outcomes of the SARS-CoV-2 omicron (B.1.1.529) variant outbreak among vaccinated and unvaccinated patients with cancer in Europe: results from the retrospective, multicentre, OnCovid registry study Lancet Oncol 2022 23 7 865 875 35660139 42 Korea Disease Control and Prevention Agency Domestic COVID-19 Variant Virus Detection Rate Cheongju, Korea Korea Disease Control and Prevention Agency 2022

PMC010xxxxxx/PMC10353915.txt

==== Front J Korean Med Sci J Korean Med Sci JKMS Journal of Korean Medical Science 1011-8934 1598-6357 The Korean Academy of Medical Sciences 10.3346/jkms.2023.38.e210 Original Article Preventive & Social Medicine Risk Factors for the Occurrence and Severity of Vertebral Fractures in Inflammatory Bowel Disease Patients: A Nationwide Population-Based Cohort Study https://orcid.org/0000-0002-5532-0103 Choi Arum 1* https://orcid.org/0000-0001-9075-2027 Jung Sung Hoon 2* https://orcid.org/0000-0001-9730-9845 Kim Sukil 1 https://orcid.org/0000-0003-4321-9611 Lee Jun-Seok 3 1 Department of Preventive Medicine and Public Health, College of Medicine, The Catholic University of Korea, Seoul, Korea. 2 Department of Internal Medicine, Eunpyeong St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul, Korea. 3 Department of Orthopedic Surgery, Eunpyeong St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul, Korea. Address for Correspondence: Jun-Seok Lee, MD, PhD. Department of Orthopedic Surgery, Eunpyeong St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, 1021 Tongil-ro, Eunpyeong-gu, Seoul 03312, Korea. junband@naver.com *Arum Choi and Sung Hoon Jung contributed equally to this study as co-first authors. 17 7 2023 08 6 2023 38 28 e21019 10 2022 25 4 2023 © 2023 The Korean Academy of Medical Sciences. 2023 The Korean Academy of Medical Sciences https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Background The risk of vertebral fractures is increased in inflammatory bowel disease (IBD) patients. However, whether the severity of vertebral fractures differs between IBD patients and the general population, or between patients with Crohn’s disease (CD) and ulcerative colitis (UC), is unknown. Methods We investigated risk factors associated with the occurrence and severity of vertebral fractures in IBD patients using The National Healthcare Insurance Service (NHIS) database. We defined the patients who underwent vertebroplasty or kyphoplasty after being diagnosed with a vertebral fracture as having a severe vertebral fracture than those with only diagnosis codes. Results From 2008 to 2018, there were 33,778 patients with IBD (24,370 UC patients and 9,408 CD patients) and 101,265 patients in the reference population. The incidence rate ratio of vertebral fractures in the IBD patients was 1.27 per 1,000 person-years (95% confidence interval [CI], 1.26–1.27). The risk of vertebral fracture was higher in CD and UC patients than in the matched reference group (hazard ratio [HR], 1.59; 95% CI, 1.31–1.92; P < 0.001 and HR, 1.26; 95% CI, 1.14–1.41; P < 0.001, respectively). In a multivariate analysis, the occurrence of vertebral fracture was associated with CD (HR, 1.31; 95% CI, 1.08–1.59; P = 0.006), older age (CD: HR, 1.09; 95% CI, 1.08–1.09; P < 0.001 and UC: HR, 1.09; 95% CI, 1.08–1.09; P < 0.001), female sex (CD: HR, 1.81; 95% CI, 1.63–2.01; P < 0.001 and UC: HR, 2.02; 95% CI, 1.83–2.22; P < 0.001), high Charlson Comorbidity Index (CCI) score (CD: HR, 1.42; 95% CI, 1.23–1.63; P < 0.001 and UC: HR, 1.46; 95% CI, 1.29–1.65, P < 0.001), and long-term steroid use (CD: HR, 3.71; 95% CI, 2.84–3.37; P < 0.001 and UC: HR, 3.88; 95% CI, 3.07–4.91; P < 0.001). The severity of vertebral fractures was associated with IBD (CD: HR, 1.82; 95% CI, 1.17–2.83; P = 0.008 and UC: HR, 1.49; 95% CI, 1.17–1.89; P < 0.001) and older age (HR, 1.06; 95% CI, 1.05–1.07; P < 0.001). Conclusion Vertebral fractures occur frequently and more severely in IBD patients, particularly those with CD. Therefore, we suggest monitoring of bone density, regular vitamin D supply, and reducing the use of corticosteroids to prevent vertebral fractures in IBD patients who are older, female, or have comorbidities. Graphical Abstract Vertebral Fracture Inflammatory Bowel Disease Crohn’s Disease Ulcerative Colitis Steroid Incidence Korea Korea Health Industry Development Institute https://doi.org/10.13039/501100003710 HI19C1298 ==== Body pmcINTRODUCTION Vertebral compression fracture is the most common fragility fracture and its incidence is increasing worldwide.1 These fractures cause persistent pain and deterioration in the quality of daily life, and increase medical expenditures, causing pain and increasing social and economic burdens.23 Although vertebral compression fractures can result from osteoporosis caused by old age or menopause, they can also be triggered by secondary osteoporosis caused by chronic disease or medications.45 Considering the serious public health burden, investigation of vertebral compression fractures induced by chronic disease or medications is important.5 Inflammatory bowel disease (IBD), which includes Crohn’s disease (CD) and ulcerative colitis (UC), is a chronic relapsing inflammatory disease of the gastrointestinal tract.6 Chronic intestinal inflammation causes malabsorption, leading to deficiencies in iron, zinc, calcium, and vitamin D.7 Moreover, the medications used to treat IBD, such as corticosteroids and immunomodulators, can adversely affect bone metabolism.8 In addition, the aging of the IBD population and the increased disease duration may affect the risk of comorbidities such as vertebral fractures.9 There is a correlation between IBD and the risk of vertebral compression fractures.101112 The risk of vertebral fractures is increased in IBD patients, particularly in CD patients, and risk factors include age, sex, corticosteroids, and comorbidities.10111213 However, whether the severity of vertebral compression fractures differs between the IBD patients and healthy controls, or between CD and UC groups, was unclear. We investigated risk factors and the severity of vertebral compression fractures in IBD patients using the National Healthcare Insurance Service (NHIS) database of South Korea. METHODS Data source The NHIS database was used for this population-based cohort study. We used the diagnostic codes of the International Classification of Diseases, 10th revision (ICD-10) code and the V code in the rare intractable diseases (RID) database. RID registration in South Korea requires clinical, endoscopic, and pathologic findings diagnostic of IBD. Study population IBD patients were defined as those diagnosed with ICD-10 and RID registration system (V code). CD patients were defined as those with ICD-10 code K50 and V code V130 and UC patients as ICD-10 code K51 and V code V131. After a washout period of 1 year (2007), 34,158 patients with IBD aged 18–79 years were included between 2008 and 2018. We excluded patients with a history of vertebral fractures before the index date. The sex- and 10-year age matched non-IBD reference population was selected from the NHIS database during the same study period at a matching ratio of 1:3. Outcomes Demographic features (age, sex), comorbidities, and use of UC medications were analyzed. The Charlson Comorbidity Index (CCI) was used to assess the severity of underlying comorbidities. Based on the index date, the CCI was calculated by observing comorbidities over the past year and assigned weighted scores of 0, 1, 2, and ≥ 3. We collected data on IBD medications, including 5-aminosalicylic acid (5-ASA), immunomodulators (azathioprine, mercaptopurine, cyclosporine, tacrolimus, and methotrexate), steroids, and biologics (infliximab, adalimumab, golimumab, vedolizumab). Vertebral fractures were defined as ICD-10 codes M48.4, M48.5, M49.5, M80.88, S22.0, S22.1, S32.0, S32.7 and T08.x and procedure codes as N0471, N0472, N0473, N0474, and N0630. Vertebral compression fractures are treated conservatively—bed rest, analgesics, brace, and physical therapy.9 However, in severe cases such as kyphosis due to excessive vertebral compression or intractable pain, vertebral augmentation such as vertebroplasty or kyphoplasty can be used.9 Therefore, patients who have undergone procedures such as vertebroplasty or kyphoplasty for vertebral compression fractures have a greater fracture severity than those treated conservatively. We defined the patients who underwent vertebroplasty or kyphoplasty after being diagnosed with a vertebral fracture, cases with both a diagnosis code and a procedure code, as having a severe vertebral fracture than the patients with only diagnosis codes. When extracting data, we included the patients with vertebral fractures that occurred after the diagnosis of IBD and clarified the precedent relationship between IBD, vertebral fractures, and procedures performed for treatment after vertebral fractures occurred. Statistical analysis We performed a descriptive statistical analysis of the patients’ characteristics. The χ2 test was used for binary or categorical variables and the Mann-Whitney U test for continuous variables. We assessed the cumulative vertebral fracture incidence between IBD patients and the general population and performed Cox regression analyses to determine whether the duration of steroid use was associated with the occurrence and severity of vertebral fractures in IBD patients and the general population. Statistical analysis was performed using R version 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria) and SAS version 7.1 (SAS Institute Inc., Cary, NC, USA). Ethics statement This study was approved by the Institutional Review Board of Eunpyeong St. Mary’s Hospital, The Catholic University of Korea (IRB approval number PC19ZNSE0128, November 7, 2019). No informed consent was required from patients due to the nature of public data from NHIS. RESULTS Demographic characteristics From 2008 to 2018, there were 33,778 patients with IBD (24,370 UC patients and 9,408 CD patients) and 101,265 patients in the reference population. The median age at IBD diagnosis was 39 years, and patients were followed up for a median of 5.0 years. The CD patients were younger and more likely to be male than the UC patients. The characteristics of the subjects are listed in Table 1. Table 1 Baseline characteristics of the study population Variables General population (N = 101,265) IBD (n = 33,778) CD (n = 9,408) UC (n = 24,370) P valuea Age at diagnosis, yr 39.0 (27.0–52.0) 39.0 (27.0–52.0) 28.0 (21.0–41.0) 43.0 (31.0–54.0) < 0.001 Sex, male 62,369 (61.6) 20,822 (61.6) 6,495 (69.0) 14,327 (58.8) < 0.001 Vertebral fractures 1,413 (1.4) 559 (1.7) 115 (1.2) 444 (1.8) < 0.001 Severity < 0.001 Diagnostic code only 1,167 (1.2) 436 (1.3) 93 (1.0) 343 (1.4) Operation 246 (0.2) 123 (0.4) 22 (0.2) 101 (0.4) CCI group < 0.001 0 60,432 (59.7) 14,738 (43.6) 3,899 (41.4) 10,839 (44.5) 1 21,791 (21.5) 9,783 (29.0) 2,910 (30.9) 6,873 (28.2) 2 9,460 (9.3) 4,956 (14.7) 1,494 (16.0) 3,462 (14.2) 3+ 9,582 (9.5) 4,301 (12.7) 1,105 (11.7) 3,196 (13.1) Use of medication 5-ASA 27,593 (81.7) 8,081 (85.9) 19,512 (80.1) < 0.001 Immune suppressors 10,569 (31.3) 6,013 (63.9) 4,556 (18.7) < 0.001 Biologics 4,062 (12.0) 2,319 (24.6) 1,743 (7.2) < 0.001 Use of steroid < 0.001 No use 21,992 (21.7) 5,981 (17.7) 1,527 (16.2) 4,454 (18.3) < 30 days 51,195 (50.5) 10,507 (31.1) 2,603 (27.7) 7,904 (32.4) 30 to 89 days 19,116 (18.9) 7,736 (22.9) 2,464 (26.2) 5,272 (21.6) 90 to 179 days 5,657 (5.6) 4,577 (13.6) 1,516 (16.1) 3,061 (12.6) 180 to 364 days 2,191 (2.2) 2,853 (8.4) 757 (8.0) 2,096 (8.6) ≥ 1 year 1,114 (1.1) 2,124 (6.3) 541 (5.8) 1,583 (6.5) Follow up duration, yr 5.0 (3.0–9.0) 5.0 (2.0–8.0) 5.0 (2.0–8.0) 5.0 (2.0–8.0) - Values are presented as median (interquartile range) or number of patients (%). P vales are the results of the χ2 test and the Mann-Whitney U test. IBD = inflammatory bowel disease, CD = Crohn’s disease, UC = ulcerative colitis, CCI = Charlson Comorbidity Index, 5-ASA = 5-aminosalicylic acid. aCD vs. UC. Cumulative events of vertebral fractures Among IBD patients, 559 vertebral fractures occurred during the study period (444 in UC patients and 115 in CD patients). In the reference population, 1413 vertebral fractures occurred. During the follow-up period, the incidence rate ratio (IRR) of vertebral fractures in the IBD patients was 1.27 per 1,000 person-years (95% confidence interval [CI], 1.26–1.27). The risk of vertebral fracture was higher in CD and UC patients than in the matched reference group (hazard ratio [HR], 1.593; 95% CI, 1.317–1.927; P < 0.001 and HR, 1.269; 95% CI, 1.141–1.412; P < 0.001, respectively) (Fig. 1, Table 2). The prevalence of vertebral fractures according to age groups is listed in Table 3. Fig. 1 Cumulative risk of vertebral fractures in inflammatory bowel disease patients and the matched controls. CD = Crohn’s disease, UC = ulcerative colitis. Table 2 HR for vertebral fractures Variables HR 95% CI P value General population 1.000 CD 1.593 1.317–1.927 < 0.001 UC 1.269 1.141–1.412 < 0.001 Age Years 1.094 1.090–1.098 < 0.001 Sex Male 1.000 Female 2.073 1.892–2.272 < 0.001 HR = hazard ratio, CI = confidence interval, CD = Crohn’s disease, UC = ulcerative colitis. Table 3 Prevalence of vertebral fractures according to age groups Age group General population (N = 101,265) CD (n = 9,408) UC (n = 24,370) ≤ 29 31,503 (31.1) 5,025 (53.4) 5,485 (22.5) 30–39 20,377 (20.1) 1,782 (18.9) 5,017 (20.6) 40–49 19,730 (19.5) 1,144 (12.2) 5,431 (22.3) 50–59 17,227 (17.0) 808 (8.6) 4,938 (20.3) 60–69 9,016 (8.9) 437 (4.6) 2,586 (10.6) ≥ 79 3,412 (3.4) 212 (2.3) 913 (3.7) All values are expressed as numbers (%). CD = Crohn’s disease, UC = ulcerative colitis. Risk factors for vertebral fractures In univariate analyses, UC patients tended to have a higher HR for vertebral fracture than the general population (HR, 1.394; 95% CI, 1.253–1.551; P < 0.001). However, in a multivariate analysis, CD patients tended to have a higher HR of vertebral fracture than the general population (HR, 1.311; 95% CI, 1.081–1.591; P < 0.001). Older age, female sex, high CCI score, and duration of steroid use were significantly associated with vertebral fracture in univariate and multivariate analyses (Tables 4 and 5). Table 4 Risk factors for the occurrence of vertebral fractures in CD Variables Univariate analysis Multivariate analysis HR 95% CI P value HR 95% CI P value General population 1.000 1.000 CD 0.933 0.771–1.128 0.471 1.311 1.081–1.591 0.006 Age Years 1.096 1.092–1.101 < 0.001 1.090 1.086–1.095 < 0.001 Sex Male 1.000 1.000 Female 2.400 2.165–2.66 < 0.001 1.818 1.639–2.016 < 0.001 Duration of steroid use No use 1.000 1.000 < 30 days 1.552 1.275–1.888 < 0.001 1.455 1.196–1.771 < 0.001 30 to 89 days 2.102 1.712–2.581 < 0.001 1.766 1.436–2.172 < 0.001 90 to 179 days 2.905 2.308–3.657 < 0.001 2.180 1.727–2.752 < 0.001 180 to 364 days 3.850 2.962–5.005 < 0.001 2.587 1.984–3.373 < 0.001 ≥ 1 year 7.012 5.398–9.108 < 0.001 3.715 2.844–3.373 < 0.001 CCI score group 0 1.000 1.000 1 1.715 1.492–1.972 < 0.001 1.041 0.904–1.199 0.575 2 3.066 2.637–3.565 < 0.001 1.252 1.072–1.462 0.005 3+ 5.695 5.013–6.469 < 0.001 1.421 1.238–1.630 < 0.001 CD = Crohn’s disease, HR = hazard ratio, CI = confidence interval, CCI = Charlson Comorbidity Index. Table 5 Risk factors for the occurrence of vertebral fractures in UC Variables Univariate analysis Multivariate analysis HR 95% CI P value HR 95% CI P value General population 1.000 1.000 UC 1.394 1.253–1.551 < 0.001 1.055 0.945–1.178 0.343 Age Years 1.097 1.093–1.101 < 0.001 1.090 1.086–1.094 < 0.001 Sex Male 1.000 1.000 Female 2.472 2.25–2.717 < 0.001 2.020 1.838–2.221 < 0.001 Duration of steroid use No use 1.000 1.000 < 30 days 1.659 1.375–2.003 < 0.001 1.557 1.290–1880 < 0.001 30 to 89 days 2.301 1.892–2.798 < 0.001 1.879 1.543–2.288 < 0.001 90 to 179 days 3.169 2.556–3.931 < 0.001 2.216 1.782–2.755 < 0.001 180 to 364 days 3.881 3.062–4.919 < 0.001 2.527 1.987–3.213 < 0.001 ≥ 1 year 6.905 5.486–8.691 < 0.001 3.885 3.071–4.916 < 0.001 CCI score group 0 1.000 1.000 1 1.803 1.587–2.047 < 0.001 1.071 0.941–1.218 0.299 2 3.257 2.843–3.732 < 0.001 1.322 1.150–1.520 < 0.001 3+ 5.727 5.095–6.438 < 0.001 1.463 1.291–1.658 < 0.001 UC = ulcerative colitis, HR = hazard ratio, CI = confidence interval, CCI = Charlson Comorbidity Index. Risk factors for the severity of vertebral fractures We estimated risk factors for the severity of vertebral fractures by Cox regression analysis. In a multivariate analysis, CD and UC patients tended to have a higher HR for the severity of vertebral fracture than the general population (HR, 1.824; 95% CI, 1.173–2.836; P = 0.008 and HR, 1.495; 95% CI, 1.179–1.895; P < 0.001, respectively). Older age, but not female sex and the duration of steroid use, was associated with the severity of vertebral fractures (Table 6). Table 6 Risk factors for the severity of vertebral fractures Variables Univariate analysis Multivariate analysis HR 95% CI P value HR 95% CI P value General population 1.000 1.000 CD 1.374 0.888–2.126 0.154 1.824 1.173–2.836 0.008 UC 1.551 1.230–1.957 < 0.001 1.495 1.179–1.895 < 0.001 Age Years 1.066 1.054–1.078 < 0.001 1.064 1.051–1.076 < 0.001 Sex Male 1.000 1.000 Female 1.922 1.505–2.455 < 0.001 1.268 0.985–1.631 0.065 Duration of steroid use No use 1.000 1.000 < 30 days 0.944 0.601–1.485 0.804 0.925 0.587–1.465 0.735 30 to 89 days 0.920 0.578–1.466 0.726 0.956 0.598–1.527 0.851 90 to 179 days 0.929 0.558–1.549 0.779 0.911 0.544–1.526 0.723 180 to 364 days 1.027 0.604–1.746 0.923 0.910 0.532–1.556 0.730 ≥ 1 year 1.255 0.751–2.097 0.387 1.071 0.636–1.802 0.797 CCI score group 0 1.000 1.000 1 1.447 1.072–1.954 0.016 1.198 0.884–1.622 0.244 2 1.406 1.022–1.933 0.036 1.042 0.755–1.438 0.805 3+ 1.779 1.351–2.342 < 0.001 1.243 0.935–1.654 0.135 HR = hazard ratio, CI = confidence interval, CD = Crohn’s disease, UC = ulcerative colitis, CCI = Charlson Comorbidity Index. Cumulative events of vertebral fracture by duration of steroid use The risks of vertebral fractures in CD and UC patients by duration of steroid use are shown in Fig. 2 and Table 7. In CD and UC patients, older age, female sex, and long-term steroid use had higher HRs. Fig. 2 Cumulative risk of vertebral fractures in CD and UC patients. CD = Crohn’s disease, UC = ulcerative colitis. Table 7 HR for vertebral fractures in CD and UC Variables CD UC HR 95% CI P value HR 95% CI P value Age Years 1.088 1.075–1.101 < 0.001 1.093 1.084–1.101 < 0.001 Sex Male 1.000 1.000 Female 1.448 0.989–2.120 0.057 2.705 2.217–3.300 < 0.001 Duration of steroid use No use 1.000 1.000 < 30 days 1.152 0.499–2.658 0.741 2.205 1.305–3.725 < 0.001 30 to 89 days 1.538 0.659–3.590 0.319 2.696 1.590–4.573 < 0.001 90 to 179 days 2.134 0.890–5.116 0.089 2.931 1.698–5.058 < 0.001 180 to 364 days 3.356 1.383–8.146 0.007 3.507 2.015–6.106 < 0.001 ≥ 1 year 3.524 1.486–8.357 0.004 5.544 3.240–9.487 < 0.001 HR = hazard ratio, CD = Crohn’s disease, UC = ulcerative colitis, CI = confidence interval. DISCUSSION The risk for vertebral compression fractures was significantly higher in IBD patients than in the reference population. The risk of vertebral fracture in CD patients was higher than in UC patients (Table 2). We evaluated the risk of vertebral fracture in Asian patients with IBD; the results are consistent with previous large cohort studies of Western populations in which the risks of hip and any fractures were higher in IBD patients.111214 A recent study of the risk of vertebral and hip fractures in Asian populations yielded similar results.10 The risk of vertebral and hip fractures was higher in IBD patients (IRR, 1.24) compared with a matched reference group.10 However, previous studies have evaluated the risk of vertebral fractures and pelvic fractures without distinguishing between UC and CD patients.101112 In this study, to increase accuracy, we evaluated the risk of only vertebral fractures. Risk factors for vertebral fractures in IBD patients were old age, female sex, and steroid use, similar to previous reports.1012 The risk of vertebral fracture was higher in CD patients than in UC patients in a multivariate analysis. This is similar to previous reports that hip fractures and any fractures occur more frequently in CD patients.101112 In a large-scale study of 2,102 patients with IBD conducted in the United Kingdom, patients with CD had a higher risk of spine fractures than those with UC.11 In a population-based study of 83,435 patients with IBD conducted in Sweden, the relative risks of hip fracture were 1.7 for CD patients and 1.3 for UC patients, and those of any fracture were 1.2 for CD patients and 1.2 for UC patients.12 A recent population-based study of 18,228 patients with IBD from South Korea reported that among patients < 60 years of age at diagnosis of IBD, the IRR of fractures in CD and UC patients was 1.15 and 0.96, respectively. Among patients > 60 years of age at diagnosis of IBD, the IRR of fractures in CD and UC patients was 2.19, and 1.32 respectively.10 Although the cause of the difference in fracture risk between CD and UC patients is unclear, it might be a result of differences in time of disease onset, location, and/or severity; inflammatory activity; and inflammatory mediator levels.815 Compared with the matched reference population, severe vertebral fractures that required vertebral augmentation procedures such as vertebroplasty or kyphoplasty were more frequent in IBD patients. Other risk factors for severe vertebral fractures were older age and female sex. Interestingly, among IBD patients, the risk of severe vertebral fractures was higher in UC patients (HR, 1.551) than in CD patients (HR, 1.374) in a univariate analysis. This result might be because, compared with CD patients, UC patients were more numerous, had an older mean age, and a larger proportion of females. However, in multivariate analysis, the risk of severe vertebral fractures was higher in CD patients (HR, 1.824) than in UC patients (HR, 1.495). This is likely to be a result of greater disease activity and malabsorption of nutrients in CD compared to UC, increasing the risk of osteoporosis and severe vertebral fractures.7813 These results are consistent with a previous population-based study in which patients with CD had a higher risk of osteoporosis.16 Therefore, to prevent severe vertebral fractures in CD patients, more attention should be paid to CD treatment. Use of corticosteroids causes secondary osteoporosis and increases the risk of fractures.1718 In a recent meta-analysis of 470,541 IBD patients, those with fractures were more likely to be on corticosteroids than those without fractures (odds ratio, 1.47).19 Several large-scale studies have reported that the cumulative dose of corticosteroids increases the risk of fractures in IBD patients.111214 In this study, long-term steroid use was associated with the occurrence, but not the severity, of vertebral fractures. These findings suggest that long-term steroid use should be avoided to reduce vertebral fractures in IBD patients. In addition, biologics instead of steroids should be considered, especially for IBD patients who are older, female, or have comorbidities. Interestingly, this study showed the medication-using patterns of the UC and CD groups appear to be different. There are some possible explanations for this finding. Firstly, conventional drugs such as 5-ASA, corticosteroids, and immune modulators should be preceded first according to the guideline of Health Insurance Review and Assessment service (HIRA), which has some differences in the guideline for CD and UC. For CD, immunomodulators are used along with 5-ASA and steroids. For UC, 5-ASA is used as a base and steroids are recommended as an adjunctive therapy, and also the role of immunomodulators is less important than in CD. Secondly, HIRA allows the use of biologics for patients with moderate-to-severe IBD who are not responsive to, intolerant of, or contraindicated to conventional drugs. All Korean physicians follow this policy, which is why the majority of patients have used 5-ASA. CD has a higher proportion of moderate-severe cases than UC, so it is estimated that there will be more use of biological agents. There were limitations in our study. First, since this study was a big data study using The NHIS database, we could not differentiate refractures that occurred after diagnosis of IBD. We targeted the first vertebral fracture after diagnosis of IBD. Old vertebral fractures occurring prior to diagnosis of IBD were excluded. Second, there was no information on bone mineral density in this study. Considering that bone mineral density has a strong relationship with the risk of vertebral fractures, the lack of data on bone mineral density may be an inevitable limitation of big data research. Third, we were unable to collect data on the activity or location/extent of IBD from this limited database. Fourth, data on anemia, hypertension, diabetes, osteoporosis, and other metabolic diseases that may affect the patient’s disease and vertebral fracture were limited and could not be analyzed. Fifth, we could not analyze chronic diseases that may be present in the reference population. Additionally, because we were unable to stratify the matched reference population according to the status of corticosteroid exposure together with the age- and sex-matching method, the effect of corticosteroid exposure on the risk of fractures may have been lessened. Sixth, the effect of the cumulative steroid dose could not be proven since we set the length of steroid usage, not the cumulative dose, as a factor affecting vertebral fractures. Seventh, we did not analyze the effects of drugs such as 5-ASA, immunomodulators, and biological agents on vertebral fractures. Since it was difficult to see those drugs as the direct cause of osteoporosis-inducing factors, we excluded them from the analysis and focused on the main factor, steroids. Lastly, we used only the vertebroplasty or kyphoplasty codes to determine the severity of vertebral fractures and did not use surgical codes. The reason for not using surgical codes is that in the case of vertebral fracture surgery, the main surgical codes are for instrument fixation, decompression, and fusion. However, the same surgical codes are used for the treatment of other spinal diseases such as spinal stenosis and spondylolisthesis. It is very difficult to distinguish whether the patient has undergone surgery only for a vertebral fracture or for other spinal diseases. Therefore, we only used the vertebroplasty or kyphoplasty codes specific to vertebral fractures and their treatment. In conclusion, the risk of vertebral fractures and of severe vertebral fractures requiring augmentation procedures was higher in IBD patients, particularly those with CD, than in the general population. Old age, female sex and long-term steroid use were associated with vertebral fractures in IBD patients. Therefore, we suggest frequent monitoring of bone density, regular administration of vitamin D supplements, and reducing the duration of corticosteroid use to prevent vertebral fractures in IBD patients. Funding: This research was supported by a grant from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (grant number: HI19C1298). Disclosure: The authors have no potential conflicts of interest to disclose. Author Contributions: Conceptualization: Jung SH, Lee JS. Data curation: Choi A, Kim S. Formal analysis: Choi A, Kim S, Jung SH. Investigation: Jung SH, Lee JS Funding acquisition. Methodology: Jung SH, Lee JS. Project administration: Jung SH, Kim S. Software: Choi A. Validation: Jung SH, Lee JS. Writing - original draft: Choi A, Jung SH, Lee JS. Writing - review & editing: Jung SH, Lee JS. ==== Refs 1 Kendler DL Bauer DC Davison KS Dian L Hanley DA Harris ST Vertebral fractures: clinical importance and management Am J Med 2016 129 2 221.e1 221.10 2 Goldstein CL Chutkan NB Choma TJ Orr RD Management of the elderly with vertebral compression fractures Neurosurgery 2015 77 Suppl 4 S33 S45 26378356 3 Kim HY Ha YC Kim TY Cho H Lee YK Baek JY Healthcare costs of osteoporotic fracture in Korea: information from the National Health Insurance claims database, 2008-2011 J Bone Metab 2017 24 2 125 133 28642857 4 Dewar C Diagnosis and treatment of vertebral compression fractures Radiol Technol 2015 86 3 301 320 25739109 5 Emkey GR Epstein S Secondary osteoporosis: pathophysiology & diagnosis Best Pract Res Clin Endocrinol Metab 2014 28 6 911 935 25432361 6 Guan Q A comprehensive review and update on the pathogenesis of inflammatory bowel disease J Immunol Res 2019 2019 7247238 31886308 7 Filippi J Al-Jaouni R Wiroth JB Hébuterne X Schneider SM Nutritional deficiencies in patients with Crohn’s disease in remission Inflamm Bowel Dis 2006 12 3 185 191 16534419 8 Bischoff SC Herrmann A Göke M Manns MP von zur Mühlen A Brabant G Altered bone metabolism in inflammatory bowel disease Am J Gastroenterol 1997 92 7 1157 1163 9219790 9 Faye AS Colombel JF Aging and IBD: a new challenge for clinicians and researchers Inflamm Bowel Dis 2022 28 1 126 132 33904578 10 Ahn HJ Kim YJ Lee HS Park JH Hwang SW Yang DH High risk of fractures within 7 years of diagnosis in Asian patients with inflammatory bowel diseases Clin Gastroenterol Hepatol 2022 20 5 e1022 e1039 34216823 11 van Staa TP Cooper C Brusse LS Leufkens H Javaid MK Arden NK Inflammatory bowel disease and the risk of fracture Gastroenterology 2003 125 6 1591 1597 14724810 12 Ludvigsson JF Mahl M Sachs MC Björk J Michaelsson K Ekbom A Fracture risk in patients with inflammatory bowel disease: a nationwide population-based cohort study from 1964 to 2014 Am J Gastroenterol 2019 114 2 291 304 30730858 13 Klaus J Armbrecht G Steinkamp M Brückel J Rieber A Adler G High prevalence of osteoporotic vertebral fractures in patients with Crohn’s disease Gut 2002 51 5 654 658 12377802 14 Card T West J Hubbard R Logan RF Hip fractures in patients with inflammatory bowel disease and their relationship to corticosteroid use: a population based cohort study Gut 2004 53 2 251 255 14724159 15 Ghishan FK Kiela PR Advances in the understanding of mineral and bone metabolism in inflammatory bowel diseases Am J Physiol Gastrointest Liver Physiol 2011 300 2 G191 G201 21088237 16 Tsai MS Lin CL Tu YK Lee PH Kao CH Risks and predictors of osteoporosis in patients with inflammatory bowel diseases in an Asian population: a nationwide population-based cohort study Int J Clin Pract 2015 69 2 235 241 25472555 17 van Staa TP Leufkens HG Cooper C The epidemiology of corticosteroid-induced osteoporosis: a meta-analysis Osteoporos Int 2002 13 10 777 787 12378366 18 Buckley L Humphrey MB Glucocorticoid-Induced Osteoporosis N Engl J Med 2018 379 26 2547 2556 30586507 19 Komaki Y Komaki F Micic D Ido A Sakuraba A Risk of fractures in inflammatory bowel diseases: a systematic review and meta-analysis J Clin Gastroenterol 2019 53 6 441 448 29672437

PMC010xxxxxx/PMC10353916.txt

==== Front J Korean Med Sci J Korean Med Sci JKMS Journal of Korean Medical Science 1011-8934 1598-6357 The Korean Academy of Medical Sciences 10.3346/jkms.2023.38.e213 Original Article Medicine General & Policy Trends in the Prevalence of Blindness and Correlation With Health Status in Korean Adults: A 10-Year Nationwide Population-Based Study https://orcid.org/0000-0002-0951-9162 Na Kyeong Ik 1 https://orcid.org/0000-0002-3506-7139 Lee Won June 2 https://orcid.org/0000-0002-6037-8449 Kim Young Kook 3 1 Department of Ophthalmology, Kangdong Sacred Heart Hospital, Hallym University College of Medicine, Seoul, Korea. 2 Department of Ophthalmology, Hanyang University Seoul Hospital, Hanyang University College of Medicine, Seoul, Korea. 3 Department of Ophthalmology, Seoul National University Hospital, Seoul National University College of Medicine, Seoul, Korea. Address for Correspondence: Young Kook Kim, MD, PhD. Department of Ophthalmology, Seoul National University Hospital, Seoul National University College of Medicine, 101 Daehak-ro, Jongno-gu, Seoul 03080, Korea. md092@naver.com 17 7 2023 08 6 2023 38 28 e21316 11 2022 22 3 2023 © 2023 The Korean Academy of Medical Sciences. 2023 The Korean Academy of Medical Sciences https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Background Contemporary data on vision impairment form an important basis for public health policies. However, most data on the clinical epidemiology of blindness are limited by small sample sizes and focused not on systemic conditions but ophthalmic diseases only. In this study, we examined the ten-year trends of blindness prevalence and its correlation with systemic health status in Korean adults. Methods This study investigated 10,000,000 participants randomly extracted from the entire Korean population (aged ≥ 20 years) who underwent a National Health Insurance Service health checkup between 2009 and 2018. Participants with blindness, defined as visual acuity in the better-seeing eye of ≤ 20/200, were identified. The prevalence of blindness was assessed, and the systemic health status was compared between participants with blindness and without blindness. Results The mean prevalence of blindness was 0.473% (47,115 blindness cases) and tended to decrease over ten years (0.586% in 2009 and 0.348% in 2018; P < 0.001). The following factors were significantly associated with blindness: female sex, underweight (body mass index < 18.5), high serum creatinine (> 1.5 mg/dL), and bilateral hearing loss. In addition, except for those aged 30–39 and 40–49 years, high fasting glucose (≥ 126 mg/dL) and low hemoglobin (male: < 12 g/dL, female: < 10 g/dL) were significantly correlated with prevalent blindness. Conclusion Our ten-year Korean nationwide population-based study suggested a gradual decrease in the prevalence of blindness and its association with specific systemic health status. These conditions might be the cause or consequence of blindness and can be used as a reference for the prevention and/or rehabilitation of blindness to establish public health policies. Graphical Abstract Blindness National Health Insurance Service Health Checkup Prevalence Health Status Population-Based Study ==== Body pmcINTRODUCTION Blindness is an important public health issue that can severely influence personal and social well-being. Visual impairment causes an enormous global financial burden and significant annual global productivity losses.12345 According to a report from the World Health Organization (WHO) analyzing surveys from 39 countries since 2000, the estimated number of visually impaired individuals worldwide is 285 million (39 million blind and 246 million with low vision).6 Visual impairment is caused by several ophthalmic diseases. Recent population-based studies have revealed that cataracts, age-related macular degeneration (AMD), glaucoma, myopic degeneration, and diabetic retinopathy are major causes of visual impairment.789101112131415 It is well known that several systemic health problems such as hypertension, diabetes mellitus, and chronic kidney disease affect the development and progression of sight-threatening eye diseases.161718 Conversely, functional end-stage eye diseases can affect general health through lifestyle changes. In particular, blindness can negatively affect mental health and quality of life and experience difficulties in daily activities.19 For the prevention of visual impairment and rehabilitation of visually impaired individuals, accurate assessment of epidemiological features, including the clinical and demographic characteristics of patients with visual impairment, is required. Previous studies have generally focused on ophthalmic diseases only and have included only hospital-based sample data.78911131415 We used a nationwide cohort dataset of health checkups from 2009 to 2018 in South Korea to investigate the population-based prevalence of blindness and identify the systemic risk factors for blindness in the adult population. METHODS National Health Insurance Service (NHIS) database and health checkup program The NHIS, a single insurer that manages the National Health Insurance Program, has collected medical information from approximately 50 million Koreans (up to 97.0% of the Korean population’s health insurance claims).2021 The NHIS database contains patients’ demographic data, such as region, age, sex, medical utilization/transaction information, claims and deduction data, and insurers’ payment coverage. The NHIS also manages a biennial health checkup program for all insured Koreans ≥ 40 years of age, and heads of households and employee subscribers who are ≥ 20 years of age are recommended to undergo a NHIS health checkup every year.22 The NHIS health checkup programs include anthropometric measurements, hearing and visual acuity checks, and laboratory tests. Hospitals perform health checkups after being certified by the NHIS, which also regularly qualifies trained examiners. Study population We used the NHIS health checkup database from 2009 to 2018 (i.e., a ten-year period). The database was constructed by randomly extracting 1,000,000 people who had undergone health checkups each year, excluding results of those aged under 20 years. Finally, 10,000,000 subjects (5,407,349 males and 4,592,651 females) were included in this study. For a detailed analysis by age group, the data were categorized into seven subgroups (at ten-year-old intervals) as follows: 20–29 years, 30–39 years, 40–49 years, 50–59 years, 60–69 years, 70–79 years, and 80 years and older. Deidentified and anonymized data were used in the analyses. Definition of blindness The presenting visual acuity, which was measured using a standard visual acuity chart at a distance of 5 m, was recorded. Presenting visual acuity was defined as visual acuity without correction for those who did not use corrective lenses, or visual acuities with correction for those who wore corrective lenses in daily life. The definition of blindness was determined by modifying the WHO and United States (US) criteria. The WHO criteria defined blindness as best corrected visual acuity (BCVA) < 20/400 in the better eye, and the US criteria defined blindness as BCVA < 20/200 in the better eye. In the NHIS health checkup database, visual acuity is expressed as 0.1–2.5, and visual acuity less than 0.1 is expressed as 0.1. We defined blindness as visual acuity in the better-seeing eye of ≤ 0.1. Assessment of systemic health status A total of 14 systemic health status factors were tested: 1) body mass index (BMI), systemic blood pressure, and diastolic blood pressure; 2) blood tests including fasting glucose, total cholesterol, hemoglobin, creatinine, aspartate aminotransferase (AST), alanine aminotransferase (ALT), and gamma-glutamyl transpeptidase (GGT); 3) urinalysis for proteinuria; 4) hearing; and 5) questionnaire for smoking status and alcohol consumption. Based on the criteria suggested by the NHIS,23 the health checkup results were categorized as follows. All participants were categorized into four groups based on their BMI, which was calculated by dividing the weight in kilograms by the square of the height in meters: underweight (BMI < 18.5), normal weight (18.5 ≤ BMI < 23), overweight (23 ≤ BMI < 25), and obesity (25 ≤ BMI). Blood pressure was measured using a sphygmomanometer after resting for at least 5 minutes. If the systolic blood pressure was > 120 mmHg or diastolic blood pressure was > 80 mmHg, re-measurement was carried out after an interval of 2 minutes. Systolic blood pressure ≥ 140 mmHg and diastolic blood pressure ≥ 90 mmHg were defined as abnormal blood pressure. After confirming the participants’ fasting condition, a blood test was performed. The urine test strip-based proteinuria test was performed and graded from 0 (negative) to 4 in 5 steps. Abnormal results for each variable were determined as follows: fasting glucose (≥ 126 mg/dL), total cholesterol (≥ 240 mg/dL), hemoglobin (male: < 12 g/dL, female: < 10 g/dL), creatinine (> 1.5 mg/dL), AST (≥ 51 IU/L), ALT (≥ 46 IU/L), and GGT (male: ≥ 78 IU/L, female: ≥ 46 IU/L), and proteinuria (≥ 1+). Hearing was measured by pure-tone audiometry, but if the participant was 66 years or older, it could be measured using the whisper voice test. Cases were defined as having “hearing loss” where the whisper voice test result was less than three out of six numbers or pure-tone audiometry was ≥ 40 dB. Smoking history was categorized as follows: “Current smoker” is someone who has smoked over five packs (100 cigarettes) in their lifetime and who still smokes currently. “Ex-smoker” is someone who was a previous current smoker but is no longer smoking. “Non-smoker” is someone who has never smoked over five packs (100 cigarettes) in their lifetime. Alcohol consumption was divided into two groups: non-drinkers and drinkers. Establishment and scope of study subjects and controls After excluding subjects with missing visual acuity data, the prevalence of blindness was analyzed. To analyze the factors associated with blindness, subjects with missing variables were excluded again and then divided into the blindness and control groups. Statistical analysis The prevalence of blindness was analyzed for each year and age group. The long-term trend of blindness was evaluated using Poisson regression analysis. The prevalence of blindness according to age was analyzed using polynomial regression analysis. In this analysis, age was defined as the median age of each age group, and the 85 years and older group was set at 87 years. The general characteristics of the subjects are expressed as the mean ± standard deviation for continuous variables and percentage for categorical variables. Differences between the blindness and control groups were evaluated using the t-test for continuous variables and the χ2 test for categorical variables. Odds ratios (ORs) were calculated by multiple logistic regression analysis for variables in the blindness group, using the control group as a reference. The quantitative data were descriptively analyzed using the open-source statistical package R version 4.0.3 (R Project for Statistical Computing, Vienna, Austria). Ethics statement This study was approved by the Institutional Review Board (IRB) of Kangdong Sacred Heart Hospital and the requirement for informed consent was waived (IRB No. 2021-08-018). RESULTS Trends in prevalence of blindness over a decade A total of 10,000,000 subjects were analyzed from the national health checkup database between 2009 and 2018. Among them, 49,018 subjects (0.49%) with missing visual acuity data were excluded. Out of 9,950,982 subjects, 47,115 had visual acuity in the better-seeing eyes of ≤ 0.1. The mean prevalence of blindness was 0.473% in the adult population (age ≥ 20 years; Table 1). Table 2 shows the number of people with blindness by age group and year. The prevalence of blindness for each year, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, and 2018 were 0.586%, 0.647%, 0.611%, 0.474%, 0.461%, 0.434%, 0.420%, 0.393%, 0.363%, and 0.348%, respectively. The prevalence of blindness significantly decreased over ten years in the Poisson regression analysis after adjusting for age (OR, 0.913; 95% confidence interval [CI], 0.910–0.916). Fig. 1 show the number of people with blindness over ten years in the Korean population. Table 1 Prevalence of blindness in the Korean population Year Age, yr 20–24 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 65–69 70–74 75–79 80–84 ≥ 85 Total 2009 0.646 0.450 0.304 0.230 0.249 0.300 0.389 0.441 0.679 1.142 1.770 3.275 6.059 11.969 0.586 2010 0.593 0.579 0.503 0.474 0.350 0.452 0.455 0.430 0.668 1.000 1.618 2.590 4.545 13.651 0.647 2011 0.563 0.533 0.470 0.426 0.316 0.385 0.418 0.507 0.573 0.910 1.398 2.541 4.444 15.847 0.611 2012 0.520 0.354 0.231 0.192 0.214 0.251 0.336 0.356 0.578 0.840 1.287 2.030 3.898 11.195 0.474 2013 0.518 0.288 0.252 0.176 0.233 0.248 0.331 0.343 0.546 0.728 1.163 1.876 3.723 9.341 0.461 2014 0.453 0.297 0.228 0.152 0.192 0.211 0.291 0.315 0.481 0.646 1.112 1.843 3.297 11.571 0.434 2015 0.426 0.288 0.220 0.152 0.188 0.242 0.282 0.276 0.470 0.589 0.904 1.781 3.391 10.121 0.420 2016 0.416 0.268 0.196 0.127 0.169 0.220 0.277 0.268 0.382 0.551 0.814 1.479 2.953 10.354 0.393 2017 0.367 0.243 0.192 0.137 0.171 0.190 0.229 0.238 0.351 0.524 0.835 1.358 2.616 8.738 0.363 2018 0.400 0.251 0.194 0.138 0.139 0.188 0.211 0.213 0.349 0.468 0.725 1.288 2.390 7.574 0.348 Total 0.494 0.368 0.285 0.225 0.225 0.269 0.323 0.330 0.495 0.735 1.153 1.917 3.471 10.512 0.473 Table 2 Number of participants with blindness among National Health Insurance Service health checkup database of a 1,000,000 Korean population for each year Year Age, yr 20–24 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 65–69 70–74 75–79 80–84 ≥ 85 Total 2009 180 426 291 219 352 360 515 353 543 624 829 539 426 169 5,826 2010 134 484 493 458 502 540 627 378 526 531 736 451 345 218 6,423 2011 135 429 460 383 475 436 590 456 469 427 690 459 383 277 6,069 2012 119 247 223 169 316 281 493 339 507 391 668 387 367 208 4,715 2013 127 204 246 145 346 281 453 344 458 375 601 399 393 217 4,589 2014 109 198 212 131 272 242 404 324 430 330 567 397 396 305 4,317 2015 100 185 198 128 260 285 381 297 436 321 441 413 438 300 4,183 2016 99 177 161 106 227 262 364 295 387 293 407 344 433 357 3,912 2017 81 157 150 117 224 226 298 266 374 278 424 345 392 281 3,613 2018 85 164 150 120 170 216 277 237 374 272 382 359 396 266 3,468 Total 1,169 2,671 2,584 1,976 3,144 3,129 4,402 3,289 4,504 3,842 5,745 4,093 3,969 2,598 47,115 Fig. 1 Number of participants with blindness over ten years among National Health Insurance Service health checkup database of a 1,000,000 Korean population for each year. The blindness prevalence for each of the age groups, 20–24, 25–29, 30–34, 35–39, 40–44, 45–49, 50–54, 55–59, 60–64, 65–69, 70–74, 75–79, 80–84, and 85 years and older were 0.494%, 0.368%, 0.285%, 0.225%, 0.225%, 0.269%, 0.323%, 0.330%, 0.495%, 0.735%, 1.153%, 1.917%, 3.471%, and 10.512%, respectively. The prevalence of blindness significantly increased exponentially with age (Prevalence [%] = 238.65 − 0.114 × Year − 0.405 × Age + 0.005 Age2, P < 0.001, R2 = 0.712; Fig. 2). Fig. 2 Boxplot of blindness prevalence by age group in the Korean population. Demographics and systemic parameters of blindness Excluding subjects with missing data in the variables, the characteristics of the blindness group (38,758 subjects) and control group (8,109,132 subjects) are listed in Table 3. All demographic and systemic parameters, including age, sex, and BMI, showed significant differences between the two groups (all P < 0.001; Table 3). Table 3 Demographics and systemic parameters of the blindness and control groups Variables Control (n = 8,109,132) Blindness (n = 38,758) P value Age, yr 48.2 ± 14.0 57.8 ± 17.9 < 0.001a Male, % 54.8 38.3 < 0.001b Body mass index, kg/m2 23.7 ± 3.3 23.1 ± 3.5 < 0.001a Systolic blood pressure, mmHg 122.1 ± 14.7 124.8 ± 16.4 < 0.001a Diastolic blood pressure, mmHg 76.1 ± 9.9 76.3 ± 10.3 < 0.001a Fasting glucose, mg/dL 98.5 ± 23.7 102.8 ± 31.0 < 0.001a Total cholesterol, mg/dL 194.9 ± 37.1 191.9 ± 39.4 < 0.001a Hemoglobin, g/dL 14.1 ± 1.6 13.4 ± 1.7 < 0.001a Proteinuria 1.1 ± 0.4 1.1 ± 0.6 < 0.001a Creatinine, mg/dL 0.9 ± 0.9 1.5 ± 2.3 < 0.001a AST, mg/dL 25.5 ± 17.3 25.0 ± 16.8 < 0.001a ALT, U/L 25.3 ± 22.2 22.4 ± 19.7 < 0.001a GGT, U/L 36.8 ± 49.6 32.2 ± 48.3 < 0.001a Hearing, % < 0.001b Normal hearing 96.1 87.7 Unilateral hearing loss 2.5 5.5 Bilateral hearing loss 1.4 6.8 Smoking status, % < 0.001b Non-smoker 59.6 74.4 Ex-smoker 16.2 10.5 Current smoker 24.3 15.1 Alcohol consumption, % < 0.001b Non-drinker 45.5 58.6 Drinker 50.5 41.4 AST = aspartate aminotransferase, ALT = alanine aminotransferase, GGT = gamma-glutamyl transpeptidase. aUsing the t-test; bUsing the χ2 test. Factors associated with blindness (grouped by ten years) Table 4 shows the results of multiple logistic regression analysis for systemic parameters in the blindness group, using the control group as a reference for each of the seven age groups. In all age groups, males were negatively associated with prevalent blindness (OR, 0.47 [95% CI, 0.43–0.51]–OR, 0.87 [95% CI, 0.81–0.94]). Compared to normal weight, underweight was positively associated with prevalent blindness in all age groups (OR, 1.16, [95% CI, 1.05–1.27]–OR, 1.75 [95% CI, 1.61–1.90]). Overweight in all age groups except for the 40 to 49 years old group, and obesity in all age groups showed a negative association with prevalent blindness (OR, 0.66 [95% CI, 0.61–0.72]–OR, 0.93 [95% CI, 0.87–0.99]; OR, 0.56 [95% CI, 0.51–0.61]–OR, 0.88 [95% CI, 0.82–0.94], respectively). Systolic blood pressure ≥ 140 mmHg was positively associated with prevalent blindness in the 40 to 69 years old group (OR, 1.17 [95% CI, 1.09–1.26]–OR, 1.24 [95% CI, 1.13–1.36]), whereas it was negatively associated in the 80 years and older group (OR, 0.78 [95% CI, 0.72–0.85]). Diastolic blood pressure ≥ 90 mmHg was positively associated with prevalent blindness in the 70 years and older group (OR, 1.17 [95% CI, 1.09–1.26]–OR, 1.23 [95% CI, 1.11–1.35]). Table 4 Multiple logistic regression analysis for systemic parameters in the blindness group, using the control group as a reference Parameters Age, yr 20–29 30–39 40–49 50–59 60–69 70–79 ≥ 80 Sex (male) 0.52 (0.48–0.58) 0.70 (0.64–0.76) 0.87 (0.81–0.94) 0.73 (0.63–0.78) 0.57 (0.53–0.61) 0.52 (0.49–0.55) 0.47 (0.43–0.51) < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 BMI Normal weight (≥ 18.5, < 23 kg/m2) (ref.) Underweight (< 1.85 kg/m2) 1.16 (1.05–1.27) 1.22 (1.09–1.37) 1.33 (1.17–1.50) 1.48 (1.29–1.69) 1.63 (1.44–1.84) 1.65 (1.51–1.80) 1.75 (1.61–1.90) 0.002 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 Overweight (≥ 23, < 25 kg/m2) 0.89 (0.80–0.99) 0.90 (0.83–0.99) 0.94 (0.88–1.01) 0.93 (0.87–0.99) 0.80 (0.75–0.85) 0.72 (0.68–0.76) 0.66 (0.61–0.72) 0.038 0.022 0.083 0.017 < 0.001 < 0.001 < 0.001 Obesity (≥ 25 kg/m2) 0.75 (0.66–0.84) 0.80 (0.74–0.88) 0.88 (0.82–0.94) 0.88 (0.83–0.94) 0.82 (0.77–0.86) 0.69 (0.65–0.72) 0.56 (0.51–0.61) < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 Systolic blood pressure (≥ 140 mmHg) 0.94 (0.66–1.33) 0.92 (0.75–1.15) 1.18 (1.03–1.34) 1.24 (1.13–1.36) 1.17 (1.09–1.26) 1.02 (0.96–1.09) 0.78 (0.72–0.85) 0.733 0.471 0.017 < 0.001 < 0.001 0.441 < 0.001 Diastolic blood pressure (≥ 90 mmHg) 1.03 (0.75–1.43) 1.01 (0.83–1.23) 0.92 (0.81–1.05) 0.99 (0.90–1.09) 1.03 (0.95–1.12) 1.17 (1.09–1.26) 1.23 (1.11–1.35) 0.853 0.916 0.213 0.849 0.495 < 0.001 < 0.001 Fasting glucose (≥ 126 mg/dL) 1.91 (1.38–2.63) 1.14 (0.92–1.41) 1.33 (1.19–1.48) 1.38 (1.27–1.49) 1.45 (1.35–1.54) 1.37 (1.29–1.46) 1.27 (1.17–1.39) < 0.001 0.245 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 Total cholesterol (> 240 mg/dL) 0.90 (0.73–1.11) 0.85 (0.75–0.96) 0.98 (0.89–1.07) 1.00 (0.93–1.07) 0.93 (0.87–1.00) 1.01 (0.94–1.08) 0.86 (0.77–0.95) 0.331 0.012 0.600 0.931 0.043 0.812 0.003 Hemoglobin (M: < 12, F: < 10 g/dL) 1.44 (1.03–2.01) 0.70 (0.50–0.97) 1.13 (0.98–1.31) 1.65 (1.42–1.91) 1.92 (1.68–2.20) 1.62 (1.47–1.78) 1.66 (1.51–1.83) 0.034 0.030 0.093 < 0.001 < 0.001 < 0.001 < 0.001 Proteinuria (≥ 1+) 1.14 (0.90–1.45) 0.93 (0.74–1.16) 1.13 (0.97–1.31) 1.16 (1.02–1.31) 1.70 (1.53–1.88) 1.31 (1.19–1.45) 1.09 (0.95–1.25) 0.267 0.508 0.109 0.021 < 0.001 < 0.001 0.209 Creatinine (> 1.5 mg/dL) 9.21 (8.22–10.32) 32.57 (30.01–35.36) 20.84 (19.19–22.64) 11.14 (10.22–12.14) 2.45 (2.14–2.80) 1.70 (1.51–1.93) 1.23 (1.06–1.43) < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 0.007 AST (≥ 51 IU/L) 1.06 (0.77–1.47) 0.77 (0.60–0.98) 1.03 (0.86–1.23) 1.25 (1.09–1.45) 0.91 (0.78–1.05) 0.99 (0.85–1.16) 1.07 (0.84–1.36) 0.729 0.034 0.774 0.002 0.199 0.897 0.579 ALT (≥ 46 IU/L) 0.97 (0.79–1.19) 1.13 (1.00–1.28) 0.96 (0.86–1.08) 0.95 (0.85–1.06) 1.04 (0.93–1.17) 1.00 (0.87–1.15) 0.96 (0.74–1.24) 0.766 0.052 0.501 0.351 0.512 0.978 0.737 GGT (M: ≥ 78, F: ≥ 46 IU/L) 1.14 (0.92–1.41) 0.96 (0.84–1.09) 1.15 (1.05–1.26) 1.08 (1.00–1.17) 1.08 (1.00–1.17) 1.02 (0.94–1.12) 0.98 (0.86–1.13) 0.238 0.518 0.003 0.065 0.064 0.620 0.812 Hearing Normal hearing (ref.) Unilateral hearing loss 1.21 (0.67–2.19) 2.30 (1.74–3.04) 1.75 (1.47–2.09) 1.76 (1.57–1.97) 1.46 (1.33–1.60) 1.24 (1.15–1.34) 1.08 (0.98–1.20) 0.535 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 0.135 Bilateral hearing loss 3.92 (2.38–6.45) 3.90 (2.69–5.67) 3.65 (2.94–4.52) 2.81 (2.44–3.23) 2.69 (2.44–2.96) 1.94 (1.80–2.08) 1.95 (1.81–2.11) < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 Smoking status Non-smoker (ref.) Ex-smoker 0.88 (0.75–1.04) 0.82 (0.73–0.91) 0.68 (0.61–0.75) 0.66 (0.61–0.73) 0.68 (0.62–0.75) 0.91 (0.83–0.99) 0.75 (0.66–0.85) 0.126 < 0.001 < 0.001 < 0.001 < 0.001 0.022 < 0.001 Current smoker 0.96 (0.87–1.06) 0.92 (0.84–1.00) 0.82 (0.76–0.90) 0.81 (0.74–0.88) 0.84 (0.77–0.92) 0.93 (0.84–1.02) 0.73 (0.62–0.86) 0.446 0.058 < 0.001 < 0.001 < 0.001 0.120 < 0.001 Drinker 1.03 (0.95–1.10) 1.09 (1.02–1.17) 1.03 (0.97–1.09) 0.99 (0.94–1.04) 0.88 (0.83–0.93) 0.94 (0.89–1.00) 0.83 (0.73–0.93) 0.481 0.009 0.374 0.637 < 0.001 0.069 0.002 Values are presented as odd ratio (95% confidence interval) and P value. Shaded boxes (grey) indicate a P value of < 0.05. BMI = body mass index, AST = aspartate aminotransferase, ALT = alanine aminotransferase, GGT = gamma-glutamyl transpeptidase. In all age groups except for the 30 to 39 years old group, fasting glucose ≥ 126 mg/dL was positively associated with prevalent blindness (OR, 1.27 [95% CI, 1.17–1.39]–OR, 1.91 [95% CI, 1.38–2.63]). Total cholesterol > 240 mg/dL was negatively associated with prevalent blindness in the 30 to 39 years, 60 to 69 years, and 80 years and older groups (OR, 0.85 [95% CI, 0.75–0.96]–OR, 0.93 [95% CI, 0.87–1.00]). The OR of hemoglobin (male: < 12 g/dL, female: < 10 g/dL) to prevalent blindness was 1.44 (95% CI, 1.03–2.01)–1.92 (95% CI, 1.68–2.20) in the 20 to 29 years group and 50 years and older group, respectively, whereas it was 0.70 (95% CI, 0.50–0.97) in the 30 to 39 years old group. Proteinuria ≥ 1+ was positively associated with prevalent blindness in the 50 to 79 years old group (OR, 1.16 [95% CI, 1.02–1.31]–OR, 1.70 [95% CI, 1.53–1.88]). Creatinine > 1.5 mg/dL was positively associated with prevalent blindness in all age groups (OR, 1.23 [95% CI, 1.06–1.43]– OR, 32.57 [95% CI, 30.01–35.36]). AST ≥ 51 IU/L showed a negative association with prevalent blindness in the 30 to 39 years old group (OR, 0.77 [95% CI, 0.60–0.98]), whereas it showed a positive association in the 50–59 years old group (OR, 1.25 [95% CI, 1.09–1.45]). ALT ≥ 46 IU/L were not significantly associated with prevalent blindness in all age groups. GGT (male: ≥ 78 IU/L, female: ≥ 46 IU/L) showed a positive correlation in the 40–49 age group (OR, 1.15 [95% CI, 1.05–1.26]). Unilateral hearing loss was positively associated with prevalent blindness in the 30 to 79 years old group (OR, 1.24 [95% CI, 1.15–1.34]–OR, 2.30 [95% CI, 1.74–3.04]), and bilateral hearing loss was positively associated with all age groups (OR, 1.94 [95% CI, 1.80–2.08]–OR, 3.92 [95% CI, 2.38–6.45]). Ex-smokers showed negative associations in the 30 years and older group (OR, 0.66 [95% CI, 0.61–0.73]–OR, 0.91 [95% CI, 0.83–0.99]), and current smokers showed negative associations in the 40–69 years and 80 years and older groups (OR, 0.73 [95% CI, 0.62–0.86]–OR, 0.84 [95% CI, 0.77–0.92]). Drinkers were positively associated with prevalent blindness in the 30 to 39 years old group (OR, 1.09 [95% CI, 1.02–1.17]), whereas drinking was negatively associated in the 60 to 69 years and 80 years and older groups (OR, 0.83 [95% CI, 0.83–0.93] and OR, 0.88 [95% CI, 0.83–0.93], respectively). Factors associated with blindness (grouped by 20 years) The subjects were categorized into three subgroups (20–39, 40–59 years, and 60–79 years old) and multiple logistic regression analysis was performed (Table 5). The following factors were significantly associated with blindness in all age groups: male sex (OR, 0.55 [95% CI, 0.52–0.57]–OR, 0.79 [95% CI, 0.75–0.83]), underweight (OR, 1.22 [95% CI, 1.14–1.32]–OR, 1.79 [95% CI, 1.67–1.92]), overweight (OR, 0.74 [95% CI, 0.71–0.77]–OR, 0.95 [95% CI, 0.90–0.99]), obesity (OR, 0.73 [95% CI, 0.70–0.76]–OR, 0.89 [95% CI, 0.85–0.93]), fasting glucose (≥ 126 mg/dL; OR, 1.27 [95% CI, 1.06–1.52]–OR, 1.44 [95% CI, 1.37–1.50]), creatinine (> 1.5 mg/dL; OR, 2.05 [95% CI, 1.87–2.24]–OR, 19.70 [95% CI, 18.45–21.02]), unilateral hearing loss (OR, 1.45 [95% CI, 1.37–1.54]–OR, 1.98 [95% CI, 1.54–2.54]), bilateral hearing loss (OR, 2.53 [95% CI, 2.38–2.68]–OR, 3.88 [95% CI, 2.88–5.22]) and ex-smoker (OR, 0.67 [95% CI, 0.62–0.71]–OR, 0.84 [95% CI, 0.77–0.92]). Fig. 3 shows the alluvial diagrams for the overall health status of patients with blindness according to three age groups: sex, BMI, fasting glucose, hemoglobin, creatinine, hearing, and smoking status. Table 5 Multiple logistic regression analysis for systemic parameters in the blindness group by broad age groups, using the control group as a reference Parameters Age, yr 20–39 40–59 60–79 Sex (male) 0.60 (0.57–0.64) 0.79 (0.75–0.83) 0.55 (0.52–0.57) < 0.001 < 0.001 < 0.001 BMI Normal weight (≥ 18.5, < 23 kg/m2) (ref.) Underweight (< 1.85 kg/m2) 1.22 (1.14–1.32) 1.36 (1.24–1.49) 1.79 (1.67–1.92) < 0.001 < 0.001 < 0.001 Overweight (≥ 23, < 25 kg/m2) 0.89 (0.83–0.95) 0.95 (0.90–0.99) 0.74 (0.71–0.77) < 0.001 0.018 < 0.001 Obesity (≥ 25 kg/m2) 0.77 (0.72–0.83) 0.89 (0.85–0.93) 0.73 (0.70–0.76) < 0.001 < 0.001 < 0.001 Systolic blood pressure (≥ 140 mmHg) 0.93 (0.78–1.12) 1.25 (1.16–1.35) 1.18 (1.13–1.24) 0.439 < 0.001 < 0.001 Diastolic blood pressure (≥ 90 mmHg) 1.00 (0.84–1.18) 0.96 (0.88–1.03) 1.08 (1.02–1.14) 0.973 0.266 0.008 Fasting glucose (≥ 126 mg/dL) 1.27 (1.06–1.52) 1.39 (1.30–1.48) 1.44 (1.37–1.50) 0.008 < 0.001 < 0.001 Total cholesterol (≥ 240 mg/dL) 0.85 (0.76–0.95) 1.02 (0.97–1.08) 0.94 (0.90–0.99) 0.003 0.407 0.012 Hemoglobin (M: < 12, F: < 10 g/dL) 0.90 (0.71–1.14) 1.26 (1.13–1.39) 1.92 (1.77–2.08) 0.374 < 0.001 < 0.001 Proteinuria (≥ 1+) 1.07 (0.91–1.26) 1.14 (1.04–1.25) 1.51 (1.40–1.62) 0.439 0.007 < 0.001 Creatinine (> 1.5 mg/dL) 19.70 (18.45–21.02) 15.27 (14.39–16.21) 2.05 (1.87–2.24) < 0.001 < 0.001 < 0.001 AST (≥ 51 IU/L) 0.86 (0.71–1.04) 1.18 (1.05–1.32) 0.98 (0.88–1.09) 0.116 0.004 0.706 ALT (≥ 46 IU/L) 1.08 (0.98–1.20) 0.95 (0.87–1.03) 0.95 (0.87–1.04) 0.135 0.178 0.288 GGT (M: ≥ 78, F: ≥ 46 IU/L) 0.98 (0.88–1.09) 1.12 (1.05–1.19) 1.02 (0.96–1.09) 0.684 < 0.001 0.437 Hearing Normal hearing (ref.) Unilateral hearing loss 1.98 (1.54–2.54) 1.82 (1.65–2.00) 1.45 (1.37–1.54) < 0.001 < 0.001 < 0.001 Bilateral hearing loss 3.88 (2.88–5.22) 3.09 (2.75–3.48) 2.53 (2.38–2.68) < 0.001 < 0.001 < 0.001 Smoking status Non-smoker (ref.) Ex-smoker 0.84 (0.77–0.92) 0.67 (0.62–0.71) 0.77 (0.72–0.82) < 0.001 < 0.001 < 0.001 Current smoker 0.94 (0.88–1.01) 0.81 (0.76–0.86) 0.82 (0.77–0.88) 0.082 < 0.001 < 0.001 Drinker 1.08 (1.03–1.13) 0.99 (0.96–1.03) 0.88 (0.84–0.92) 0.003 0.705 < 0.001 Values are presented as odd ratio (95% confidence interval) and P value Shaded boxes (grey) indicate a P value of < 0.05. BMI = body mass index, AST = aspartate aminotransferase, ALT = alanine aminotransferase, GGT = gamma-glutamyl transpeptidase. Fig. 3 Alluvial diagram for the health status of blindness in the Korean population. DISCUSSION In the NHIS health checkup, the visual acuity is measured and recorded as presenting visual acuity. BCVA has been used in many previous population-based studies,7891122 but it has been pointed out that it does not reflect the effect of uncorrected or under-corrected refractive error, which is the main cause of treatable visual impairment.242526 Therefore, it has recently been recommended that presenting visual acuity should be used in population-based studies.2728 In this nationwide population-based study, the prevalence of blindness, defined as visual acuity in the better-seeing eye of ≤ 0.1, was 0.473% in Korean adults (aged 20 years and older). In previous population-based studies, the prevalence of blindness was reported to be 0.14–1.5%.78911 In the Tajimi Study, the prevalence of blindness (according to the WHO or US criteria) was 0.14% in subjects aged 40 years and older.7 The Taizhou Eye Study reported that the prevalence of blindness was 1.0% using the WHO criteria and 1.5% using the US criteria in subjects aged 45 years and older.11 In the Poisson regression analysis adjusted for age, the prevalence of blindness was found to have gradually decreased over the past decade. Community initiatives and medical advancements (i.e., increased awareness of sight-threatening eye diseases or neurological disorders, improved diagnostic strategies for early detection and diagnosis, and discovery and development of more effective treatments) would have had a significant impact on reducing the prevalence. Analyzing the prevalence of blindness by age group, it was found to increase exponentially from the age of 40 years. Previous studies have also reported that the prevalence of blindness is highly correlated with aging.789 However, in this study, we found that, before the age of 40 years, a higher prevalence of blindness was seen at a younger age. It is presumed that low presenting visual acuity due to uncorrected myopia improved with accurate correction of myopia; this was linked to the need for good eyesight for social activities at that age. To the best of our knowledge, this is the first time that the association between systemic health status and prevalent blindness has been thoroughly investigated. Seven clinico-demographic factors (sex, BMI, smoking status, fasting glucose, hemoglobin, creatinine, and hearing) associated with prevalent blindness were identified in this study. In this study, the prevalence of blindness was higher in females than in males (0.65% vs. 0.33%). In the Taizhou Eye Study, the prevalence of blindness was greater in females than in males (1.2% vs. 0.7% for WHO criteria).11 In the Beijing Eye Study, sex was not a statistically significant factor related to blindness.8 In contrast, the Barbados Eye Study reported that the prevalence of blindness was higher in males than in females (2.5% vs. 1.0% for the WHO criteria and 4.2% vs. 2.1% for the US criteria).13 A previous meta-analysis reported that among the causes of blindness, diabetic retinopathy and cataracts were more common in females than in males, and glaucoma and corneal opacity were more common in males than in females.12 It can be inferred that differences in the prevalence of eye diseases in each country may affect the sex ratio of patients with blindness. Compared to normal weight, overweight and obesity decreased the risk of blindness, while underweight increased the risk of blindness in this study. A previous study on a population with type 2 diabetes reported that visual impairment was associated with lower BMI.29 Furthermore, several studies have shown that a lower BMI is associated with a greater risk of glaucoma. Kim et al.30 reported that participants aged > 40 years with BMI > 25 had a low prevalence of primary open-angle glaucoma compared to those with BMI < 25. However, contradictory results have also been reported.31 Newman-Casey et al.32 reported that obese patients had a higher risk of glaucoma. The blindness group had lower smoking and alcohol consumption rates than the control group, and smoking was significantly associated with a lower OR for prevalent blindness in this study. These findings may indicate that visual impairment leads to smoking cessation and/or alcohol abstinence, resulting from lower socioeconomic status.3334 Fasting glucose levels were significantly higher in the blindness group than in the control group in almost all age groups. Fasting glucose is an important factor in the diagnosis of diabetes, and its relationship is well known.35 A previous population-based study also reported that higher fasting glucose levels were associated with vision-threatening diabetic retinopathy.18 It is also possible that if physical activity is restricted in patients with blindness and the amount of exercise they perform decreases, their risk of diabetes may increase due to the decrease in exercise. Hemoglobin is an important protein for the transport of oxygen in the body. In this study, blood hemoglobin concentration was significantly lower in the blindness group. A previous population-based study reported that glaucoma was associated with a low blood hemoglobin concentration.36 Diabetic patients with retinopathy also have lower hemoglobin levels and a higher frequency of anemia.37 Blood hemoglobin concentration is also decreased in patients with decreased renal function.38 Anemia was found to have harmful associations with AMD in a previous epidemiological study in a Korean population.39 These results suggest that low hemoglobin levels may influence the development and progression of ocular diseases, and systemic disease may be involved in this process. Serum creatinine is a product of muscle metabolism that is released at a constant rate and is an important indicator of renal function. Cheng et al.17 reported that the early AMD was associated with chronic kidney disease and AMD is also well known as an important cause of blindness.7891112 A population-based study reported that low estimated glomerular filtration rate levels were independently associated with primary open-angle glaucoma.40 Hearing loss is common in older adults and impairs activities of daily living.4142 Both unilateral and bilateral hearing loss were significantly associated with blindness in this study. There are various syndromes in which hearing loss and visual impairment appear simultaneously, such as pseudoexfoliation syndrome, oculo-auricular syndrome, Cogan’s syndrome, Vogt-Koyanagi-Harada syndrome, Behçet syndrome, congenital rubella syndrome, retinitis pigmentosa-related genetic syndromes, and Down syndrome.434445 According to an epidemiological study in Korea, the presence of proliferative diabetic retinopathy was significantly associated with hearing loss in the middle-aged group. The potential role of microvascular diseases in the development of hearing loss, especially in middle-aged patients, can be considered.46 The prevalence of concurrent visual and hearing impairment is reported to be 16.6% in adults aged 80 years and older.47 The SHELTER study reported that those with dual sensory impairment had higher rates of clinical problems than those without sensory impairment.48 In addition, this dual impairment has been reported to increase the risk of mortality.4950 This study has several limitations. First, it was not possible to classify the subjects according to the causative eye disease. Depending on the eye disease, the relationship with systemic health status is expected to be different. Second, the NHIS health checkup database was not created using a stratified multistage cluster sampling method. Therefore, the data in this study cannot be considered a perfect representation of the entire Korean population. Third, although this study from a single ethnic group (i.e., Korean) has the advantage of being a population-based design, it is necessary to reconfirm the study findings in various ethnicities. Fourth, since each patient is sampled every year, overlapping samples may occur, which may affect the statistical processing. Fifth, in this study, a visual acuity in the better-seeing eye of ≤ 0.1 was defined as blindness; this may be a broad definition, compared to previous studies which used WHO criteria. The visual acuity data of the NHIS was not subdivided below 0.1, and this is considered to be a limitation of the visual acuity test included in the general health checkup. Nevertheless, considering the US criteria, the definition of blindness as used in our study is thought to be sufficiently reasonable. In conclusion, over the past decade, the overall prevalence of blindness among Korean adults was 0.473%. While the prevalence of blindness decreased from 2009 to 2018, it was found to increase with age. Several systemic conditions associated with prevalent blindness were confirmed in the present study. Further studies are warranted to determine whether the systemic conditions identified are causative factors or consequences of blindness. ACKNOWLEDGMENTS We thank Prof. Sehyug Kwon (Department of Statistics, Hannam University, Daejeon, Korea) for assisting with the statistical analyses in this study. Disclosure: The authors have no potential conflicts of interest to disclose. Author Contributions: Conceptualization: Kim YK. Formal analysis: Na KI, Lee WJ. Investigation: Na KI, Lee WJ. Methodology: Kim YK. Writing - original draft: Na KI. Writing - review & editing: Na KI, Lee WJ, Kim YK. ==== Refs 1 Chuvarayan Y Finger RP Köberlein-Neu J Economic burden of blindness and visual impairment in Germany from a societal perspective: a cost-of-illness study Eur J Health Econ 2020 21 1 115 127 31493181 2 Köberlein J Beifus K Schaffert C Finger RP The economic burden of visual impairment and blindness: a systematic review BMJ Open 2013 3 11 e003471 3 Jacobs JM Hammerman-Rozenberg R Maaravi Y Cohen A Stessman J The impact of visual impairment on health, function and mortality Aging Clin Exp Res 2005 17 4 281 286 16285193 4 Klein BE Moss SE Klein R Lee KE Cruickshanks KJ Associations of visual function with physical outcomes and limitations 5 years later in an older population: the Beaver Dam eye study Ophthalmology 2003 110 4 644 650 12689880 5 Swanson MW McGwin G Visual impairment and functional status from the 1995 National Health Interview Survey on Disability Ophthalmic Epidemiol 2004 11 3 227 239 15370554 6 Pascolini D Mariotti SP Global estimates of visual impairment: 2010 Br J Ophthalmol 2012 96 5 614 618 22133988 7 Iwase A Araie M Tomidokoro A Yamamoto T Shimizu H Kitazawa Y Prevalence and causes of low vision and blindness in a Japanese adult population: the Tajimi Study Ophthalmology 2006 113 8 1354 1362 16877074 8 Xu L Wang Y Li Y Wang Y Cui T Li J Causes of blindness and visual impairment in urban and rural areas in Beijing: the Beijing Eye Study Ophthalmology 2006 113 7 1134.e1 1134.11 9 Gunnlaugsdottir E Arnarsson A Jonasson F Prevalence and causes of visual impairment and blindness in Icelanders aged 50 years and older: the Reykjavik Eye Study Acta Ophthalmol 2008 86 7 778 785 18513265 10 Limburg H Barria von-Bischhoffshausen F Gomez P Silva JC Foster A Review of recent surveys on blindness and visual impairment in Latin America Br J Ophthalmol 2008 92 3 315 319 18211928 11 Tang Y Wang X Wang J Huang W Gao Y Luo Y Prevalence and causes of visual impairment in a Chinese adult population: the Taizhou Eye Study Ophthalmology 2015 122 7 1480 1488 25986897 12 Flaxman SR Bourne RR Resnikoff S Ackland P Braithwaite T Cicinelli MV Global causes of blindness and distance vision impairment 1990-2020: a systematic review and meta-analysis Lancet Glob Health 2017 5 12 e1221 e1234 29032195 13 Hyman L Wu SY Connell AM Schachat A Nemesure B Hennis A Prevalence and causes of visual impairment in the Barbados Eye Study Ophthalmology 2001 108 10 1751 1756 11581045 14 Zheng Y Lavanya R Wu R Wong WL Wang JJ Mitchell P Prevalence and causes of visual impairment and blindness in an urban Indian population: the Singapore Indian Eye Study Ophthalmology 2011 118 9 1798 1804 21621261 15 Liang YB Friedman DS Wong TY Zhan SY Sun LP Wang JJ Prevalence and causes of low vision and blindness in a rural Chinese adult population: the Handan Eye Study Ophthalmology 2008 115 11 1965 1972 18684506 16 Wong TY Mitchell P The eye in hypertension Lancet 2007 369 9559 425 435 17276782 17 Cheng Q Saaddine JB Klein R Rothenberg R Chou CF Il’yasova D Early age-related macular degeneration with cardiovascular and renal comorbidities: an analysis of the National Health and Nutrition Examination Survey, 2005-2008 Ophthalmic Epidemiol 2017 24 6 413 419 28891729 18 Sasongko MB Widyaputri F Agni AN Wardhana FS Kotha S Gupta P Prevalence of diabetic retinopathy and blindness in Indonesian adults with type 2 diabetes Am J Ophthalmol 2017 181 79 87 28669781 19 Varma R Wu J Chong K Azen SP Hays RD Los Angeles Latino Eye Study Group Impact of severity and bilaterality of visual impairment on health-related quality of life Ophthalmology 2006 113 10 1846 1853 16889831 20 Song SO Jung CH Song YD Park CY Kwon HS Cha BS Background and data configuration process of a nationwide population-based study using the Korean National Health Insurance system Diabetes Metab J 2014 38 5 395 403 25349827 21 Kim YH Han K Son JW Lee SS Oh SW Kwon HS data analytic process of a nationwide population-based study on obesity using the National Health Information Database presented by the National Health Insurance Service 2006-2015 J Obes Metab Syndr 2017 26 1 23 27 31089490 22 Ministry of Health and Welfare (KR) The Health Screening Implementation Standards (Notification No. 2020-313) Sejong, Korea Ministry of Health and Welfare 2020 23 Ministry of Health and Welfare (KR) Table 4. Criteria for determining diagnoses based on the results of general health screening The Health Screening Implementation Standards (Notification No. 2020-313) Sejong, Korea Ministry of Health and Welfare 2020 24 Vitale S Cotch MF Sperduto RD Prevalence of visual impairment in the United States JAMA 2006 295 18 2158 2163 16684986 25 Robinson B Feng Y Woods CA Fonn D Gold D Gordon K Prevalence of visual impairment and uncorrected refractive error - report from a Canadian urban population-based study Ophthalmic Epidemiol 2013 20 3 123 130 23713914 26 Dandona R Dandona L Refractive error blindness Bull World Health Organ 2001 79 3 237 243 11285669 27 World Health Organization (WHO) Programme for the Prevention of Blindness and Deafness Consultation on Development of Standards for Characterization of Vision Loss and Visual Functioning: Geneva, 4–5 September 2003 Geneva, Switzerland World Health Organization 2003 28 World Health Organization World Report on Vision Geneva, Switzerland World Health Organization 2019 29 Cui Y Zhang L Zhang M Yang X Zhang L Kuang J Prevalence and causes of low vision and blindness in a Chinese population with type 2 diabetes: the Dongguan Eye Study Sci Rep 2017 7 1 11195 28894238 30 Kim KE Kim MJ Park KH Jeoung JW Kim SH Kim CY Prevalence, awareness, and risk factors of primary open-angle glaucoma: Korea National Health and Nutrition Examination Survey 2008-2011 Ophthalmology 2016 123 3 532 541 26746594 31 Cheung N Wong TY Obesity and eye diseases Surv Ophthalmol 2007 52 2 180 195 17355856 32 Newman-Casey PA Talwar N Nan B Musch DC Stein JD The relationship between components of metabolic syndrome and open-angle glaucoma Ophthalmology 2011 118 7 1318 1326 21481477 33 Kuper H Polack S Eusebio C Mathenge W Wadud Z Foster A A case-control study to assess the relationship between poverty and visual impairment from cataract in Kenya, the Philippines, and Bangladesh PLoS Med 2008 5 12 e244 19090614 34 Mojon-Azzi SM Sousa-Poza A Mojon DS Impact of low vision on employment Ophthalmologica 2010 224 6 381 388 20606492 35 American Diabetes Association (2) Classification and diagnosis of diabetes Diabetes Care 2015 38 Suppl S8 16 36 Nangia V Jonas JB Matin A Bhojwani K Sinha A Kulkarni M Prevalence and associated factors of glaucoma in rural central India. The Central India Eye and Medical Study PLoS One 2013 8 9 e76434 24098790 37 Bahar A Kashi Z Ahmadzadeh Amiri A Nabipour M Association between diabetic retinopathy and hemoglobin level Caspian J Intern Med 2013 4 4 759 762 24294469 38 Clyne N Jogestrand T Lins LE Pehrsson SK Progressive decline in renal function induces a gradual decrease in total hemoglobin and exercise capacity Nephron J 1994 67 3 322 326 39 Park SJ Lee JH Woo SJ Ahn J Shin JP Song SJ Age-related macular degeneration: prevalence and risk factors from Korean National Health and Nutrition Examination Survey, 2008 through 2011 Ophthalmology 2014 121 9 1756 1765 24813632 40 Shim SH Sung KC Kim JM Lee MY Won YS Kim JH Association between renal function and open-angle glaucoma: the Korea National Health and Nutrition Examination Survey 2010-2011 Ophthalmology 2016 123 9 1981 1988 27432204 41 Gopinath B Rochtchina E Wang JJ Schneider J Leeder SR Mitchell P Prevalence of age-related hearing loss in older adults: Blue Mountains Study Arch Intern Med 2009 169 4 415 416 19237727 42 Gopinath B Schneider J McMahon CM Teber E Leeder SR Mitchell P Severity of age-related hearing loss is associated with impaired activities of daily living Age Ageing 2012 41 2 195 200 22130560 43 Abou-Elhamd KA ElToukhy HM Al-Wadaani FA Syndromes of hearing loss associated with visual loss Eur Arch Otorhinolaryngol 2014 271 4 635 646 23632871 44 Papadopoulos TA Naxakis SS Charalabopoulou M Vathylakis I Goumas PD Gartaganis SP Exfoliation syndrome related to sensorineural hearing loss Clin Exp Ophthalmol 2010 38 5 456 461 20649615 45 Cahill M Early A Stack S Blayney AW Eustace P Pseudoexfoliation and sensorineural hearing loss Eye (Lond) 2002 16 3 261 266 12032714 46 Shin YU Park SH Chung JH Lee SH Cho H Diabetic retinopathy and hearing loss: results from the Fifth Korean National Health and Nutrition Survey J Clin Med 2021 10 11 2398 34071684 47 Caban AJ Lee DJ Gómez-Marín O Lam BL Zheng DD Prevalence of concurrent hearing and visual impairment in US adults: the National Health Interview Survey, 1997-2002 Am J Public Health 2005 95 11 1940 1942 16195516 48 Yamada Y Vlachova M Richter T Finne-Soveri H Gindin J van der Roest H Prevalence and correlates of hearing and visual impairments in European nursing homes: results from the SHELTER study J Am Med Dir Assoc 2014 15 10 738 743 24984787 49 Lam BL Lee DJ Gómez-Marín O Zheng DD Caban AJ Concurrent visual and hearing impairment and risk of mortality: the National Health Interview Survey Arch Ophthalmol 2006 124 1 95 101 16401790 50 Miyawaki A Kobayashi Y Kawachi I Self-reported hearing/visual loss and mortality in middle-aged and older adults: findings from the Komo-Ise Cohort, Japan J Epidemiol 2020 30 2 67 73 30662042

PMC010xxxxxx/PMC10353917.txt

==== Front J Korean Med Sci J Korean Med Sci JKMS Journal of Korean Medical Science 1011-8934 1598-6357 The Korean Academy of Medical Sciences 10.3346/jkms.2023.38.e212 Original Article Laboratory Medicine Associations of Perioperative Red Blood Cell Transfusion With Outcomes of Kidney Transplantation in Korea Over a 16-Year Period https://orcid.org/0000-0002-4370-4265 Kim Yoonjung 1* https://orcid.org/0000-0003-4888-774X Kim Banseok 2* https://orcid.org/0000-0002-8020-135X Kang Minjin 3 https://orcid.org/0000-0001-8696-4625 Nam HyunJun 1 https://orcid.org/0000-0002-9781-0928 Ko Dae-Hyun 4 https://orcid.org/0000-0001-5668-4120 Park Yongjung 1 1 Department of Laboratory Medicine, Gangnam Severance Hospital, Yonsei University College of Medicine, Seoul, Korea. 2 Department of Laboratory Medicine, National Health Insurance Service Ilsan Hospital, Goyang, Korea. 3 Research Institute, National Health Insurance Service Ilsan Hospital, Goyang, Korea. 4 Department of Laboratory Medicine, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea. Address for Correspondence: Yongjung Park, MD, PhD. Department of Laboratory Medicine, Gangnam Severance Hospital, Yonsei University College of Medicine, 211 Eonju-ro, Gangnam-gu, Seoul, Korea. ypark119@yuhs.ac Address for Correspondence: Dae-Hyun Ko, MD, PhD. Department of Laboratory Medicine, Asan Medical Center, University of Ulsan College of Medicine, 88 Olympic-ro 43-gil, Songpa-gu, Seoul, Korea. daehyuni1118@amc.seoul.kr *Yoonjung Kim and Banseok Kim contributed equally to this work. 17 7 2023 08 6 2023 38 28 e21210 11 2022 22 3 2023 © 2023 The Korean Academy of Medical Sciences. 2023 The Korean Academy of Medical Sciences https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Background This study investigated the associations between transfusion of different types of red blood cell (RBC) preparations and kidney allograft outcomes after kidney transplantation (KT) over a 16-year period in Korea using a nationwide population-based cohort. Methods We investigated the reported use of RBCs during hospitalization for KT surgery, rejection, and graft failure status using nationwide data from the National Health Information Database (2002–2017). The associations between the type of perioperative RBC product and transplant outcomes were evaluated among four predefined groups: no RBC transfusion, filtered RBCs, washed RBCs, and packed RBCs (pRBCs). Results A total of 17,754 KT patients was included, among which 8,530 (48.0%) received some type of RBC transfusion. Of the patients who received RBC transfusion, 74.9%, 19.7%, and 5.4% received filtered RBCs, pRBCs, or washed RBCs, respectively. Regardless of the type of RBC products, the proportions of acute rejection and graft failure was significantly greater in patients receiving transfusion (P < 0.001). Cox proportional hazards regression analyses showed that the filtered RBC and pRBC groups were significantly associated with both rejection and graft failure. The washed RBC group also had hazard ratios greater than 1.0 for rejection and graft failure, but the association was not significant. Rejection-free survival of the pRBC group was significantly lower than that of the other groups (P < 0.001, log-rank test), and graft survival for the no RBC transfusion group was significantly greater than in the other groups (P < 0.001, log-rank test). Conclusion Perioperative RBC transfusion was associated with poor graft outcomes. Notably, transfusion of pRBCs significantly increased transplant rejection. Therefore, careful consideration of indications for RBC transfusion and selection of the appropriate type of RBCs is necessary, especially for patients at high risk of rejection or graft failure. Graphical Abstract Kidney Transplantation Blood Transfusion Red Blood Cell Graft Survival National Health Insurance Ilsan Hospital grant 2019-20-002 ==== Body pmcINTRODUCTION The prevalence of end-stage renal disease (ESRD) in the United States was 2,242 cases per million people in 2018, the highest prevalence in the world. Kidney transplantation (KT) has been the treatment of choice for a minority of patients with ESRD since the 1960s, and at the end of 2018, there were 229,887 patients with a functioning KT in the United States.1 The prevalence of anemia in people with chronic kidney disease and that in patients with ESRD was 15.4% and 53.4%, respectively, in the United States.2 Therefore, red blood cell (RBC) transfusion is considered unavoidable during the peri-transplant period due to the high prevalence of anemia and blood loss during surgery.34 Sensitization to human leukocyte antigens (HLAs) expressed on the surface of leukocytes and platelets in blood units can be caused by factors including previous transplant, pregnancy, or blood transfusion.56 Several studies have shown that RBC transfusion is an important cause of alloimmunization and is associated with increased rejection and graft loss.789101112 Based on these results, it has been suggested that active efforts to minimize blood transfusion are required to prevent HLA sensitization and to improve kidney allograft outcomes. Appropriate management of perioperative RBC transfusion is a critical factor in KT patient outcomes as more than half of all kidney recipients receive transfusions. The consequences of perioperative transfusion on graft outcomes have been continuously evaluated.131415 however, there remain unexplored factors that are potentially relevant to patient prognosis. With respect of the type of RBC unit preparation, the risk of allosensitization in patients with chronic kidney diseases is considered to be lower with leuko-reduced RBCs.9 Methods for leukocyte depletion from RBC products include centrifugal precipitation, filtration, and red cell washing. In the past, centrifugal precipitation and red cell washing were generally used. In these days, the filtering method is used to reduce leukocytes conveniently and effectively in blood components and achieves residual leucocyte counts of < 5 × 106.1617 However, in practice, packed RBCs (pRBCs), filtered RBCs, and washed RBCs have been heterogeneously transfused into KT recipients.14 Moreover, the impact of transfusion of different types of RBC preparations on transplant outcomes is unclear, and to our knowledge, the clinical significance of the type of RBC unit preparation and graft outcome has not been evaluated in existing studies. Therefore, to better answer this practical question about which type of RBC products are suitable for patients, we investigated the associations between the transfusion of different types of RBC preparations and kidney allograft outcome using a nationwide database linked to the Korean National Health Insurance Service (KNHIS).18 Thus, the current study aimed to investigate the outcomes of KT in South Korea over a 16-year period using a KNHIS nationwide database and to evaluate the association with poor prognosis after transplantation according to the type of RBC product transfused. METHODS Data source and study subjects The KNHIS was implemented in 1988 and controls all medical costs among individuals, health care providers, and the government in South Korea. Medical data, including personal information, diagnosis, medical treatment, and demographics of patients, are centralized in the National Health Information Database (NHID) of the KNHIS.18 The NHID provides de-identified data for research purposes, and we collected detailed patient characteristics of all transplant recipients within this registry. The KNHIS-NHID includes the diagnosis of patients according to the Korean Classification of Diseases codes, which is the Korean version of the International Classification of Diseases (ICD). All insurance claims are classified based on Electronic Data Interchange (EDI) codes. We extracted data on patients who underwent KT between 2002 and 2017 from the KNHIS-NHID using the specific EDI code (R3280) for KT. Among 18,331 KT patients, we excluded patients who received transfusions of two or more types of RBCs. Finally, we analyzed 17,754 KT cases. Variable definitions We investigated the sex, age, type of donor (living or deceased), income level, type of hospital, year of surgery, length of hospital stay, regimen of induction treatment, types of initial immunosuppressant and steroid regimens, and occurrence of acute kidney allograft rejection or graft failure. Since the characteristics of the KNHIS claim data make it difficult to specify the exact time of kidney allograft rejection, acute rejection was defined as any case in which a diagnosis of kidney allograft rejection, as identified by ICD-10 codes T86 and/or T86.1, was recorded during the KT-related hospitalization period. Graft failure was defined as a KT recipient undergoing repeated dialysis for three months or longer during the post-KT follow-up period.19 The Organ Transplantation Act of South Korea requires a recipient to pay for the cost of donor nephrectomy in the case of living-donor KT, while the government covers the primary cost for organ donation in the case of a deceased donor. Therefore, the donor type was able to be classified as a living donor when the EDI code for donor nephrectomy ‘R3272’ was charged to a recipient. In addition, among living-donor KT recipients, ABO-incompatible KT was defined as a transplant procedure where ABO antibody tests (EDI code B2080) were performed two or more times during the KT-related hospitalization, and plasma exchange (EDI code X2505) was performed concurrently. The proportions of deceased- or living-donor KT recipients in this study were consistent with the statistics from the Korean Network for Organ Sharing (KONOS). The codes or criteria for defining variables in building a database have been described in previous studies.1920 Because death certificates are automatically reported to the KNHIS, mortality was detected when healthcare coverage by the KNHIS was terminated. The number and types of RBC products used in the subjects during hospitalization were analyzed. However, the NHID did not provide pre- and post-transplant information separately. Therefore, we investigated the associations between the type of perioperative RBC product received and short- and long-term transplant outcomes. We divided the subjects into four groups based on whether they had undergone RBC transfusion and the type of RBC product transfused. The four groups consisted of patients without RBC transfusion and patients transfused respectively with filtered RBCs, washed RBCs, or pRBCs. One unit of pRBCs contains approximately 200 or 250 mL of blood products separated from 320 or 400 mL, respectively, of donor whole blood. The types of RBCs were identified based on the KNHIS EDI codes for medical procedures regarding the types of RBC products; X2021, X2022, X2031, X2032, X2111, and X2112. Statistical analyses The Mann-Whitney U and Kruskal-Wallis tests were used to compare continuous variables among subject groups, and the χ2 test was used to compare categorical variables. Graft survival and rejection-free survival were calculated using Kaplan-Meier survival analyses. Data were censored at the time of death or at the last available follow-up. Cox’s proportional hazard regression was conducted to construct multivariate models for identifying factors associated with occurrence of acute kidney allograft rejection or chronic kidney allograft failure, and hazard ratios (HRs) for risk factors with 95% confidence intervals were calculated. All statistical analyses were performed using SAS 7.15 (SAS Institute Inc., Cary, NC, USA) and RStudio v1.1.463 (RStudio Inc., Boston, MA, USA), and P values less than 0.05 were considered to be statistically significant. Ethics statement This was a retrospective cohort study, and the protocol was implemented after approval from the Institutional Review Board (IRB) of National Health Insurance Service Ilsan Hospital (Approval No. NHIMC 2021-09-020). Informed consent was waived by the IRB. The administration number of the National Health Insurance Sharing Service was NHIS-2022-1-101 (REQ202104307-004). RESULTS Patients and baseline characteristics We reviewed a total of 17,754 KT recipients included in the KNHIS-NHID between 2002 and 2017. The median post-KT follow-up period was 66 months (mean 73.1 months; range, 0 to 194 months; 1st to 3rd quartiles, 28 to 109 months). The proportion of males was greater in the group without RBC transfusion. The most common age group among KT recipients was 40–59 years old, representing 56.9% to 64.5% of patients depending on the type of RBC product received (Table 1). Table 1 Kidney transplant patient characteristics according to the type of perioperative RBC transfusion Characteristics No RBC transfusion (n = 9,224) Filtered RBCs (n = 6,392) Washed RBCs (n = 459) Packed RBCs (n = 1,679) P value No. of males (%) 5,939 (64.4) 3,470 (54.3) 271 (59.0) 927 (55.2) < 0.001 Mean age ± SD 46.48 ± 12.18 47.08 ± 12.97 48.53 ± 10.55 46.20 ± 11.63 < 0.001 Age group, yr (%) < 0.001 < 20 206 (2.2) 202 (3.2) 2 (0.4) 25 (1.5) 20–39 2,376 (25.8) 1,504 (23.5) 91 (19.8) 438 (26.1) 40–59 5,365 (58.2) 3,639 (56.9) 296 (64.5) 1,024 (61.0) ≥ 60 1,277 (13.8) 1,047 (16.4) 70 (15.3) 192 (11.4) Income level, No. (%) 0.312 Bottom 50% 8,667 (94.0) 6,045 (94.6) 437 (95.2) 1,578 (94.0) Top 50% 557 (6.0) 347 (5.4) 22 (4.8) 101 (6.0) Hospital type, No. (%) < 0.001 General 1,298 (14.1) 601 (9.4) 151 (32.9) 529 (31.5) Tertiary 7,926 (85.9) 5,791 (90.6) 308 (67.1) 1,150 (68.5) Year of surgery, No. (%) < 0.001 2002–2009 2,922 (31.7) 1,894 (29.6) 60 (13.1) 861 (51.3) 2010–2017 6,302 (68.3) 4,498 (70.4) 399 (86.9) 818 (48.7) Donor type, No. (%) < 0.001 Deceased 3,391 (36.8) 2,371 (37.1) 107 (23.3) 790 (47.1) Living - ABO compatible 5,363 (58.1) 3,581 (56.0) 172 (37.5) 858 (51.1) Living - ABO incompatible 470 (5.1) 440 (6.9) 180 (39.2) 31 (1.8) Perioperative RBC transfusion, unitsa 0 (0 to 0) 2 (2 to 4) 2 (2 to 4) 2 (2 to 3) < 0.001 Acute rejection, No. (%) 654 (7.1) 614 (9.6) 44 (9.6) 192 (11.4) < 0.001 Length of stay, daysa 23 (18–30) 26 (20–34) 31 (23–42) 26 (19–38) < 0.001 Medical costs, USDa 13,445 (10,705–16,305) 15,172 (12,075–20,029) 18,108 (15,393–21,409) 12,007 (9,587–15,512) < 0.001 RBC = red blood cell, SD = standard deviation, USD = United States dollar (1 USD = 1200 Korean Won). aShown as ‘median (1st to 3rd quartiles).’ Significant differences among the four groups according to the types of RBC product administered to KT recipients were found for sex, age group, hospital type, year of surgery, donor type, history of acute rejection, length of hospital stay, and medical costs for hospitalization (P < 0.001). Compared to general hospitals, tertiary hospitals more frequently did not transfuse any RBC products and, when needed, tended to use filtered RBCs rather than other types. Washed RBCs were more frequently transfused in cases of ABO-incompatible living-donor KT. Acute rejection was more frequently diagnosed in KT recipients transfused with pRBCs, whereas it occurred less frequently in patients without RBC transfusion. In addition, patients who did not receive RBC transfusion had shorter length of hospital stay (Table 1). RBC transfusion depended on the type of preparation All patients were divided into four groups depending on the type of RBC preparation received. A total of 9,224 patients did not receive RBC transfusion during the perioperative period, and 8,530 (48.0%) of 17,754 KT recipients were transfused with some type of RBC product during the perioperative period. Among these 8,530 patients, 74.9% (n = 6,392) received filtered RBCs, 19.7% (n = 1,679) received pRBCs, and 5.4% (n = 459) received washed RBCs. Transfused patients received a median of 2 RBC units during the perioperative period, regardless of type of RBC preparation (Table 1). A total of 30,889 RBC units was transfused into patients perioperatively in 77 hospitals between 2002 and 2017. Filtered RBCs were most frequently used, followed by pRBCs. Washed RBCs were used in 27 of the 77 institutions from 2002 to 2017. Regardless of the type of RBC preparation transfused, the average amount of RBCs used per patient and that of RBCs used per hospital were 3.6 and 406.4 units, respectively. The median transfusion incidence among hospitals was 50.6% (1st to 3rd: 37.7% to 72.0%; Table 2). Table 2 Perioperative RBC use according to the type of transfused units Parameters Groups Total No transfusion Filtered RBCs Washed RBCs Packed RBCs Any RBCs No. of KT patients 9,224 6,392 459 1,679 8,530 17,754 No. of hospitals 72 69 27 64 76 77 Transfused RBCs (units) 0 24,083 1,757 5,049 30,889 30,889 Mean RBCs per patient (units) 0.0 3.8 3.8 3.0 3.6 1.7 Mean RBCs per hospital (units) 0.0 349.0 65.1 78.9 406.4 401.2 Mean RBCs per patient by hospital (units)a 0.0 (0.0–0.0) 3.4 (2.9–4.8) 2.7 (2.0–3.2) 2.6 (2.0–3.0) 3.2 (2.7–4.2) 1.8 (1.2–2.6) Transfusion rate by hospital (%)a 0.0 (0.0–0.0) 35.3 (10.5–54.7) 0.0 (0.0–1.5) 9.8 (1.6–28.2) 50.6 (37.7–72.0) 50.6 (37.7–72.0) aShown as ‘median (1st to 3rd quartiles).’ RBC = red blood cell, KT = kidney transplantation. Acute rejection and rejection-free survival after KT The incidence rate of acute rejection in KT patients during hospitalization over the study period was 8.5% in this study. The proportion of patients with acute rejection was significantly greater in women; in the 20–39 and 40–59 age groups; in patents with an earlier year of transplantation, receiving an allograft from a deceased donor, or treated with anti-thymocyte globulin (ATG); and in those receiving cyclosporine rather than tacrolimus or dexa/betamethasone or fludro/hydrocortisone as initial immunosuppressants (Supplementary Table 1). Types of RBC preparations were transfused inconsistently to KT recipients depending on the institution. We investigated the association between transfusion of different types of RBC preparations and acute rejection. Regardless of RBC product (filtered RBCs, washed RBCs, or pRBCs), the proportion of patients with acute rejection was significantly greater among patients receiving transfusion (P < 0.001) (Supplementary Table 1). Multivariate analyses indicated that the HRs for rejection after KT in patients with incomes exceeding the median, KT at general hospitals or in more recent year of transplantation, and KT received from living ABO-compatible donors were independently significant at HRs of 1.242, 1.197, 0.422, and 0.848, respectively. The HRs in the patients receiving ATG, basiliximab, rituximab, and dexa/betamethasone compared to deflazacort were also statistically significant at 2.990, 1.418, 2.456, and 1.613, respectively (Table 3). The HRs for rejection in the filtered RBC and pRBC groups compared to the KT recipients without RBC transfusion were 1.192 and 1.359, respectively. Although it was not statistically significant, the washed RBC group also had an increased HR of 1.208 compared with the no RBC transfusion group (Table 3). Table 3 Multivariate Cox proportional hazard model of risk factors of rejection during post-kidney-transplantation follow-up Variables Hazard ratio 95% Confidence interval P value Female 1.077 0.970–1.196 0.165 Age group, yr < 20 1.000 20–39 1.155 0.790–1.690 0.456 40–59 1.160 0.798–1.686 0.436 ≥ 60 1.057 0.711–1.571 0.786 Income level Bottom 50% 1.000 Top 50% 1.242 1.014–1.521 0.037 Hospital type Tertiary 1.000 General 1.197 1.036–1.383 0.015 Year of surgery 2002–2009 1.000 2010–2017 0.422 0.371–0.480 < 0.001 Donor type Deceased 1.000 Living – ABO-compatible 0.848 0.756–0.952 0.005 Living – ABO-incompatible 0.784 0.612–1.005 0.055 Perioperative RBC transfusion None 1.000 Filtered RBCs 1.192 1.062–1.337 0.003 Washed RBCs 1.208 0.881–1.658 0.240 Packed RBCs 1.359 1.151–1.605 < 0.001 Perioperative use of ATG 2.990 2.538–3.522 < 0.001 Perioperative use of basiliximab 1.418 1.237–1.626 < 0.001 Perioperative use of rituximab 2.456 2.054–2.938 < 0.001 Initial calcineurin inhibitor Cyclosporine 1.000 Tacrolimus 0.875 0.759–1.009 0.066 Initial steroid agent Deflazacort 1.000 Dexa/betamethasone 1.613 1.273–2.043 < 0.001 Fludro/hydrocortisone 1.182 0.960–1.456 0.115 (Methyl)prednisolone 0.873 0.734–1.039 0.127 Bold letters denote statistical significance at the P < 0.05 level. RBC = red blood cell, ATG = anti-thymocyte globulin. In Kaplan-Meier analyses of the four groups (no transfusion, filtered RBCs, washed RBCs, and pRBCs), the log-rank test indicated poorer rejection-free survival of the KT patients being transfused with pRBCs than other groups (no transfusion, filtered RBCs, and washed RBCs) during the post-KT follow-up period (P < 0.001) (Fig. 1). Fig. 1 Rejection-free survival according to the type of RBCs transfused perioperatively at kidney transplantation. Patients transfused with packed RBCs showed poorer rejection-free survival than the other groups (log-rank test, P < 0.001). RBC = red blood cell. Graft failure Graft failure occurred in 7.2% of kidney transplant recipients during the follow-up period. The proportion of patients with graft failure was significantly greater in males; in the 20–39 age group; and in patients with an early year of transplantation, receiving an allograft from a deceased donor, or receiving cyclosporine treatment. Patients who developed cytomegalovirus (CMV) infection within 1 year after KT and those with a history of acute rejection comprised significantly larger proportions of the graft failure group (Supplementary Table 2). In the Cox regression model, an inverse association with graft failure was found for females, with an HR of 0.816, and for patients who received a KT from a living ABO-compatible donor (HR, 0.820). The most important risk factor for graft failure was a recent year of transplantation, with an HR of 9.393. The HRs for acute rejection history; CMV infection within 1 year after KT; and perioperative treatments such as ATG, basiliximab, and rituximab ranged from 1.357 to 1.957 (Table 4). Table 4 Multivariate Cox proportional hazard model for risk factors of graft failure among kidney transplant recipients Variables Hazard ratio 95% Confidence interval P value Female 0.816 0.726–0.916 < 0.001 Age group, yr < 20 1.000 20–39 1.086 0.784–1.503 0.621 40–59 0.894 0.648–1.234 0.496 ≥ 60 1.257 0.879–1.800 0.211 Income level Bottom 50% 1.000 Top 50% 0.826 0.650–1.049 0.117 Hospital type Tertiary 1.000 General 1.109 0.943–1.305 0.212 Year of surgery 2002–2009 1.000 2010–2017 9.393 7.406–11.913 < 0.001 Donor type Deceased 1.000 Living – ABO-compatible 0.820 0.724–0.928 0.002 Living – ABO-incompatible 0.768 0.525–1.122 0.172 Perioperative RBC transfusion None 1.000 Filtered RBCs 1.240 1.091–1.410 0.001 Washed RBCs 1.167 0.781–1.744 0.452 Packed RBCs 1.363 1.158–1.604 < 0.001 Perioperative use of ATG 1.957 1.560–2.456 < 0.001 Perioperative use of basiliximab 1.824 1.578–2.108 < 0.001 Perioperative use of rituximab 1.357 1.014–1.816 0.040 Initial calcineurin inhibitor Cyclosporine 1.000 Tacrolimus 1.324 1.155–1.518 < 0.001 Initial steroid agent Deflazacort 1.000 Dexa/betamethasone 1.024 0.758–1.383 0.879 Fludro/hydrocortisone 0.984 0.758–1.276 0.902 (Methyl)prednisolone 0.861 0.710–1.044 0.128 Acute rejection 1.428 1.224–1.666 < 0.001 CMV infection within 1 yr after KT 1.468 1.243–1.734 < 0.001 Bold letters denote statistical significance at the P < 0.05 level. RBC = red blood cell, ATG = anti-thymocyte globulin, CMV = cytomegalovirus, KT = kidney transplantation. The proportion of patients with graft failure was significantly greater in those receiving transfusion (P < 0.001) (Supplementary Table 2). The HRs for graft failure in filtered RBC and pRBC groups compared with those without RBC transfusion were 1.240 and 1.363, respectively. The washed RBC group showed an HR of 1.167 for graft failure, but this finding was not statistically significant (P = 0.452; Table 4). Patients who did not receive perioperative RBC transfusion had significantly greater overall graft survival than the patients transfused with filtered RBCs, washed RBCs, or pRBCs (P < 0.001 for each comparison) according to Kaplan-Meier analysis. Graft survival was not significantly different among the groups transfused with different RBC products (Fig. 2). Fig. 2 Kidney allograft survival according to the type of RBCs transfused perioperatively at kidney transplantation. The patients who did not receive any RBC product perioperatively showed longer overall graft survival than the groups being transfused with filtered RBCs, washed RBCs, or packed RBCs (P < 0.001 for each comparison). Graft survival was not significantly different among the groups transfused with different types of RBC preparations. RBC = red blood cell. DISCUSSION The results of this study using nationwide data from the KNHIS-NHID (2002-2017) indicate differences in transplant outcomes among Korean KT recipients according to the type of RBC product transfused during the perioperative period. Given the high prevalence of intraoperative anemia and bleeding during the peri-transplant period, RBC transfusion is often unavoidable.34 The reported percentage of KT patients receiving post-transplant transfusion has ranged from 28.0% to 37.2% in Canada and France,1521 and 49.7% of KT recipients received RBC transfusion during the perioperative period in Korea.19 When we excluded patients who received transfusions of two or more types of RBCs, 48.0% of 17,754 KT recipients received RBC transfusion during the perioperative period between 2002 and 2017. Generally, in order to reduce the risk of allosensitization, leuko-reduced RBCs are preferable for KT recipients9; however, in practice, several additional types of RBC preparations including pRBCs, filtered RBCs, and washed RBCs have been transfused into Korean patients.19 Of the 8,530 patients who received RBC transfusion, 74.9% received filtered RBCs, 19.7% received pRBCs, and 5.4% received washed RBCs. Among the 77 institutions included in this study, 64 and/or 27 respectively used pRBCs and/or washed RBCs (Table 2). In this study, the percentage of patients with both acute rejection and graft failure was significantly greater among those receiving RBC transfusion (P < 0.001). Also, regardless of perioperatively transfused RBC product, similar patterns of association were found in patients with rejection or graft failure. The filtered RBC and pRBC groups were significantly associated with both rejection during follow-up after KT and graft failure in the long term. The washed RBC group showed HRs of 1.208 and 1.167 for rejection and graft failure, respectively, although these were not statistically significant (Tables 3 and 4); this finding may have been due to insufficient statistical power, since the washed RBC group accounted for only a small fraction (2.6%) of the total cases in this study. Cox’s multivariate models indicated that transfusion of pRBCs was associated with the worst transplant outcomes among the four groups in this study. The impact of transfusion of different types of RBC preparations on long-term transplant outcomes is uncertain. However, a few previous studies have evaluated the potential adverse outcomes associated with transfusion regardless of RBC product type.1521222324 We assessed the association between the type of transfused RBC products and acute rejection using a nationwide population-based database. In the survival analyses, rejection-free survival was significantly lower in the pRBC group than in the no RBC transfusion or filtered or washed RBC groups (Fig. 1). The use of pRBCs is declining, as they have been increasingly replaced by filtered RBCs, but pRBCs were still used in 6.7% of KT recipients between the 2014 and 2017 in Korea.14 Therefore, when perioperative RBC transfusion is necessary in KT recipients, transfusion of leuko-reduced RBCs should be recommended to lower the risk of kidney allograft rejection. We also found graft survival to be significantly better in patients without RBC transfusion than in the cases transfused with any type of RBCs. Few prior studies have reported the effects of perioperative blood transfusions on transplantation outcomes.152124 In a study using data from the national database of the French transfusion service, Gaiffe et al.15 reported that both pre- and early post-transplant transfusions were associated with increased transplant failure. Our data also showed that KT recipients without RBC transfusion in the perioperative period had better graft survival. Although perioperative RBC transfusion was significantly associated with poor outcomes, it cannot be concluded that transfusion of RBCs is the direct cause of graft failure, since patients who were in poorer clinical condition at the time of KT are more likely to receive RBC transfusion. On the other hand, there was no clear associations between a specific type of RBCs and graft survival. Exposure to non-self HLA through RBC transfusions may lead to the development of donor-specific HLA antibodies (DSAs) against the kidney allograft donor. Avoiding transfusions or using HLA-matched blood could reduce graft failure.2223 Therefore, further evaluation of additional data, such as utilization of HLA-matched blood and DSA results, is required to more thoroughly assess the associations between transfusion with specific types of RBCs and long-term KT outcomes, including graft failure. Washed RBCs have been considered for reducing exogenous anti-A or anti-B antibodies and HLA sensitization in ABO-incompatible transplantation.2526 Washed RBCs were transfused into KT recipients at 27 of the 77 institutions in our nationwide cohort. However, there was insufficient evidence to support the effectiveness of the use of washed RBCs in KT recipients. In addition, Aston et al. reported that washed RBCs did not further reduce patient HLA sensitization over the use of filtered RBCs,27 and our findings showed that transfusion with washed RBCs did not lead to better graft outcomes than that with filtered RBCs, although this finding was not statistically significant (Fig. 2). Therefore, in consideration of the clinical efficacy, risk of infection due to contamination, and labor and time required to manufacture washed RBCs, our findings do not support the use of washed RBCs for KT recipients. This study analyzed KNHIS-NHID data from a 16-year period to determine the associations between the type of RBC preparation transfused and KT outcomes among 17,754 patients who underwent KT for the first time. Due to the limitations of the data related to health insurance claims, we were not able to distinguish between pre- and post-operative RBC transfusions, and we were also unable to obtain detailed clinical information about the KT patients, such as the exact timing of acute rejection occurrence and results of laboratory tests which are not included in the national database. Many factors would influence KT outcomes in the pre- and post-transplantation period, such as renal function, hemoglobin level, allosensitization, pre-existing diseases, and case volume of centers.132829303132 Unfortunately, these data could not be analyzed in this study because they have not been collected and organized in the KNHIS-NHID. The impact of post-transplantation anemia and the clinical significance of de novo DSA on KT outcomes were also studied.12293334 However, the relationship between these factors and types of transfused RBCs has not been sufficiently studied. Further researches with comprehensive clinical data from each medical institution would be necessary. In addition, we used an operational definition to determine whether the KT recipient and donor were ABO-compatible and whether the donor type was living or deceased. Despite these shortcomings, our data have identified an association between RBC transfusion and short- and long-term graft outcomes among Korean patients who underwent KT during a recent 16-year period. Perioperative RBC transfusion was associated with an increased risk of kidney allograft rejection and long-term chronic graft failure. Notably, the transfusion of pRBC preparations increased the likelihood of rejection. Therefore, careful consideration of indications for RBC transfusion and selection of the appropriate type of RBC is necessary, especially for patients at high risk of rejection or graft failure. In addition, our data may support future revision of guidelines for clinicians regarding RBC transfusion of KT recipients and the development of computerized order entry system alerts when ordering pRBCs for KT recipients. SUPPLEMENTARY MATERIALS Supplementary Table 1 Characteristics of patients with or without acute rejection during hospitalization period for kidney transplantation Supplementary Table 2 Characteristics of patients with or without graft failure during follow-up after kidney transplantation Funding: This work was performed as a follow-up study of initial research supported by a National Health Insurance Ilsan Hospital grant (2019-20-002). This study used NHIS-NHID data (administration no. NHIS-2022-1-101 & REQ202104307-004) compiled by the Korean National Health Insurance Service. Disclosure: The authors have no potential conflicts of interest to disclose. Author Contributions: Conceptualization: Ko DH, Park Y. Data curation: Kang M, Park Y. Formal analysis: Kang M, Nam H, Park Y. Funding acquisition: Park Y. Investigation: Kim Y, Kim B, Ko DH, Park Y. Methodology: Ko DH, Park Y. Project administration: Kim B, Park Y. Resources: Kim B, Park Y. Software: Kang M. Supervision: Ko DH, Park Y. Validation: Kim Y, Nam H, Ko DH. Visualization: Kang M, Park Y. Writing - original draft: Kim Y, Park Y. Writing - review & editing: Ko DH, Park Y. ==== Refs 1 Johansen KL Chertow GM Foley RN Gilbertson DT Herzog CA Ishani A US Renal Data System 2020 Annual Data Report: epidemiology of kidney disease in the United States Am J Kidney Dis 2021 77 4 Suppl 1 A7 A8 33752804 2 Stauffer ME Fan T Prevalence of anemia in chronic kidney disease in the United States PLoS One 2014 9 1 e84943 24392162 3 Koo BN Kwon MA Kim SH Kim JY Moon YJ Park SY Korean clinical practice guideline for perioperative red blood cell transfusion from Korean Society of Anesthesiologists Korean J Anesthesiol 2019 72 2 91 118 30513567 4 Vanrenterghem Y Ponticelli C Morales JM Abramowicz D Baboolal K Eklund B Prevalence and management of anemia in renal transplant recipients: a European survey Am J Transplant 2003 3 7 835 845 12814475 5 Akalin E A new treatment option for highly sensitized patients awaiting kidney transplantation Am J Kidney Dis 2018 71 4 458 460 29352605 6 Snyder EL Prevention of HLA alloimmunization: role of leukocyte depletion and UV-B irradiation Yale J Biol Med 1990 63 5 419 427 2293501 7 Scornik JC Meier-Kriesche HU Blood transfusions in organ transplant patients: mechanisms of sensitization and implications for prevention Am J Transplant 2011 11 9 1785 1791 21883910 8 Balasubramaniam GS Morris M Gupta A Mesa IR Thuraisingham R Ashman N Allosensitization rate of male patients awaiting first kidney grafts after leuko-depleted blood transfusion Transplantation 2012 93 4 418 422 22228416 9 Fidler S Swaminathan R Lim W Ferrari P Witt C Christiansen FT Peri-operative third party red blood cell transfusion in renal transplantation and the risk of antibody-mediated rejection and graft loss Transpl Immunol 2013 29 1-4 22 27 24090807 10 Hong D Wu S Pu L Wang F Wang J Wang Z Abdominal aortic calcification is not superior over other vascular calcification in predicting mortality in hemodialysis patients: a retrospective observational study BMC Nephrol 2013 14 1 120 23738982 11 Leffell MS Kim D Vega RM Zachary AA Petersen J Hart JM Red blood cell transfusions and the risk of allosensitization in patients awaiting primary kidney transplantation Transplantation 2014 97 5 525 533 24300013 12 Ferrandiz I Congy-Jolivet N Del Bello A Debiol B Trébern-Launay K Esposito L Impact of early blood transfusion after kidney transplantation on the incidence of donor-specific anti-HLA antibodies Am J Transplant 2016 16 9 2661 2669 26998676 13 Lee K Lee S Jang EJ Kim GH Yoo S Lee M The association between peri-transplant RBC transfusion and graft failure after kidney transplantation: a nationwide cohort study J Clin Med 2021 10 16 3750 34442041 14 Kim B Kang M Lee JK Lee HS Park Y Perioperative blood usage and therapeutic plasma exchange in kidney transplantation during a 16-year period in South Korea Blood Transfus 2021 19 2 102 112 32530400 15 Gaiffe E Vernerey D Bardiaux L Leroux F Meurisse A Bamoulid J Early post-transplant red blood cell transfusion is associated with an increased risk of transplant failure: a nationwide French study Front Immunol 2022 13 854850 35711440 16 Bianchi M Vaglio S Pupella S Marano G Facco G Liumbruno GM Leucoreduction of blood components: an effective way to increase blood safety? Blood Transfus 2016 14 2 214 227 26710353 17 Sharma RR Marwaha N Leukoreduced blood components: advantages and strategies for its implementation in developing countries Asian J Transfus Sci 2010 4 1 3 8 20376259 18 Cheol Seong S Kim YY Khang YH Heon Park J Kang HJ Lee H Data resource profile: The National Health Information Database of the National Health Insurance Service in South Korea Int J Epidemiol 2017 46 3 799 800 27794523 19 Kim B Kang M Kim Y Lee HS Kim B Lee JJ De novo cancer incidence after kidney transplantation in South Korea from 2002 to 2017 J Clin Med 2021 10 16 3530 34441826 20 Lee HS Kang M Kim B Park Y Outcomes of kidney transplantation over a 16-year period in Korea: an analysis of the National Health Information Database PLoS One 2021 16 2 e0247449 33606787 21 Massicotte-Azarniouch D Sood MM Fergusson DA Chassé M Tinmouth A Knoll GA Blood transfusion and adverse graft-related events in kidney transplant patients Kidney Int Rep 2021 6 4 1041 1049 33912754 22 Hassan S Regan F Brown C Harmer A Anderson N Beckwith H Shared alloimmune responses against blood and transplant donors result in adverse clinical outcomes following blood transfusion post-renal transplantation Am J Transplant 2019 19 6 1720 1729 30582278 23 Jalalonmuhali M Carroll RP Tsiopelas E Clayton P Coates PT Development of de novo HLA donor specific antibodies (HLA-DSA), HLA antibodies (HLA-Ab) and allograft rejection post blood transfusion in kidney transplant recipients Hum Immunol 2020 81 7 323 329 32327243 24 O’Brien FJ Lineen J Kennedy CM Phelan PJ Kelly PO Denton MD Effect of perioperative blood transfusions on long term graft outcomes in renal transplant patients Clin Nephrol 2012 77 6 432 437 22595384 25 Cardigan R New HV Tinegate H Thomas S Washed red cells: theory and practice Vox Sang 2020 115 8 606 616 32633823 26 West LJ Pollock-Barziv SM Dipchand AI Lee KJ Cardella CJ Benson LN ABO-incompatible heart transplantation in infants N Engl J Med 2001 344 11 793 800 11248154 27 Aston A Cardigan R Bashir S Proffitt S New H Brown C Washing red cells after leucodepletion does not decrease human leukocyte antigen sensitization risk in patients with chronic kidney disease Pediatr Nephrol 2014 29 10 2005 2011 24777534 28 Richie RE Niblack GD Johnson HK Green WF MacDonell RC Turner BI Factors influencing the outcome of kidney transplants Ann Surg 1983 197 6 672 677 6344816 29 Jung HY Kim SH Seo MY Cho SY Yang Y Choi JY Characteristics and clinical significance of de novo donor-specific anti-HLA antibodies after kidney transplantation J Korean Med Sci 2018 33 34 e217 30127706 30 Kim HW Kang SW Lee HY Choi DH Shim WH Kim SI Correlates of the severity of coronary atherosclerosis in long-term kidney transplant patients J Korean Med Sci 2010 25 5 706 711 20436705 31 Kwon CH Lee SK Ha J Trend and outcome of Korean patients receiving overseas solid organ transplantation between 1999 and 2005 J Korean Med Sci 2011 26 1 17 21 21218024 32 Oh HW Jang EJ Kim GH Yoo S Lee H Lim TY Effect of institutional kidney transplantation case-volume on post-transplant graft failure: a retrospective cohort study J Korean Med Sci 2019 34 40 e260 31625292 33 Gafter-Gvili A Gafter U Posttransplantation anemia in kidney transplant recipients Acta Haematol 2019 142 1 37 43 30970356 34 Chhabra D Grafals M Skaro AI Parker M Gallon L Impact of anemia after renal transplantation on patient and graft survival and on rate of acute rejection Clin J Am Soc Nephrol 2008 3 4 1168 1174 18463170

PMC004xxxxxx/PMC4508896.txt

==== Front J Cell Biol J Cell Biol jcb jcb The Journal of Cell Biology 0021-9525 1540-8140 The Rockefeller University Press 26169354 201411025 10.1083/jcb.201411025 Research Articles Article CIN85 modulates TGFβ signaling by promoting the presentation of TGFβ receptors on the cell surface The adaptor protein CIN85 modulates TGFβ signaling Yakymovych Ihor 1 Yakymovych Mariya 1 Zang Guangxiang 2 Mu Yabing 2 Bergh Anders 2 Landström Maréne 12* Heldin Carl-Henrik 1* 1 Science for Life Laboratory, Ludwig Institute for Cancer Research Ltd., Uppsala University, SE-75124 Uppsala, Sweden 2 Department of Medical Biosciences and Pathology, Umeå University, SE-90185 Umeå, Sweden Correspondence to Carl-Henrik Heldin: c-h.heldin@licr.uu.se * M. Landström and C.-H. Heldin contributed equally to this paper. 20 7 2015 210 2 319332 07 11 2014 03 6 2015 © 2015 Yakymovych et al. 2015 https://creativecommons.org/licenses/by-nc-sa/3.0/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 3.0 Unported license, as described at http://creativecommons.org/licenses/by-nc-sa/3.0/). The adaptor CIN85 enhances TGFβ-induced signaling and cellular responses to TGFβ by promoting the expression of TGFβ receptors on the surface in a Rab11-dependent manner. Members of the transforming growth factor β (TGFβ) family initiate cellular responses by binding to TGFβ receptor type II (TβRII) and type I (TβRI) serine/threonine kinases, whereby Smad2 and Smad3 are phosphorylated and activated, promoting their association with Smad4. We report here that TβRI interacts with the SH3 domains of the adaptor protein CIN85 in response to TGFβ stimulation in a TRAF6-dependent manner. Small interfering RNA–mediated knockdown of CIN85 resulted in accumulation of TβRI in intracellular compartments and diminished TGFβ-stimulated Smad2 phosphorylation. Overexpression of CIN85 instead increased the amount of TβRI at the cell surface. This effect was inhibited by a dominant-negative mutant of Rab11, suggesting that CIN85 promoted recycling of TGFβ receptors. CIN85 enhanced TGFβ-stimulated Smad2 phosphorylation, transcriptional responses, and cell migration. CIN85 expression correlated with the degree of malignancy of prostate cancers. Collectively, our results reveal that CIN85 promotes recycling of TGFβ receptors and thereby positively regulates TGFβ signaling. ==== Body pmcIntroduction Members of the TGFβ family of multifunctional cytokines govern key cellular functions via binding to transmembrane serine/threonine kinases named TGFβ receptor type I (TβRI) and type II (TβRII; Heldin and Moustakas, 2012; Xu et al., 2012). Ligand binding initiates signaling by activation of the Smad family of transcription factors, which are central mediators of TGFβ signaling to the nucleus. In addition, TGFβ receptors activate non-Smad signaling pathways, such as extracellular signal-regulated kinase p38 and JNK MAPKs, AKT (Mu et al., 2012), and the small GTPases Rho, Rac, and Cdc42 (Kardassis et al., 2009). The initiation and regulation of TGFβ signaling is achieved by posttranslational modifications of signaling components, which determine the subcellular localization, activity, and duration of the signal. Several receptor-interacting proteins, such as Smad7, ELF, and SARA, play critical roles in the proper control of Smad access to the receptors (Mishra and Marshall, 2006). The ubiquitin ligase tumor necrosis factor receptor-associated factor 6 (TRAF6) mediates activation of p38 and JNK by TGFβ (Sorrentino et al., 2008; Yamashita et al., 2008). Other receptor-associated proteins, such as cPML and Dab2, have roles in vesicular trafficking of the receptors (Lin et al., 2004; Penheiter et al., 2010). CIN85 (Cbl-interacting protein of 85 kD, also called SH3 domain kinase binding protein 1 [SH3KBP1]) is a ubiquitously expressed adaptor protein that has been shown to associate with several signaling proteins, thus linking it to many cellular compartments and processes, including signal transduction, vesicle-mediated transport, cytoskeleton remodeling, programmed cell death, and viral infection (Dikic, 2002; Kowanetz et al., 2004; Havrylov et al., 2010). The N terminus of CIN85 is composed of three SH3 domains that mediate interactions with various proteins, typically containing proline-rich sequences (Dikic, 2002). It was also demonstrated that all three SH3 domains bind ubiquitin (Bezsonova et al., 2008). The proline-rich region of CIN85, localized between SH3 domains and the C terminus, is a recognition site for other SH3 domain–containing proteins, such as the p85 subunit of phosphatidylinositol-3′-kinase (Gout et al., 2000), kinases of Src family (Dikic, 2002), p130Cas, and cortactin (Lynch et al., 2003). The C-terminal coiled-coil region of CIN85 mediates its dimerization (Watanabe et al., 2000) and binds to phosphatidic acid on cell membranes (Zhang et al., 2009). CIN85 has been implicated in the control of internalization of receptor tyrosine kinases (Szymkiewicz et al., 2004), including the receptors for EGF (Soubeyran et al., 2002), hepatocyte growth factor (Petrelli et al., 2002), platelet-derived growth factor, and stem cell factor (Szymkiewicz et al., 2002), as well as the dopamine receptor (Shimokawa et al., 2010). Besides, CIN85 participates in post-endocytic EGF receptor (EGFR) trafficking and degradation (Schroeder et al., 2010, 2012; Rønning et al., 2011). In addition to affecting endocytosis and trafficking of transmembrane proteins, CIN85 has been reported to control the level of the nonreceptor tyrosine kinase Syk (Peruzzi et al., 2007) and to link B cell receptor signaling to the canonical NF-κB pathway (Kometani et al., 2011). In this study, we have investigated the role of CIN85 in the regulation of TGFβ signaling. We found that CIN85 enhances TGFβ-induced signaling and cellular responses to TGFβ by promoting the expression of TGFβ receptors on the surface in a Rab11-dependent manner. We have also shown that CIN85 interacts with TβRI in a TRAF6-dependent manner. Results CIN85 augments TGFβ-induced intracellular signaling events, activation of transcription, and cell motility As CIN85 has been shown to interact with many components of signaling pathways affected by TGFβ, we investigated its effect on TGFβ signaling. We found that TGFβ treatment caused ∼1.5 times stronger phosphorylation of Smad2 in PC-3U cells overexpressing CIN85 than in control cells (Fig. 1 A). Moreover, down-regulation of CIN85 by siRNA transfection reduced TGFβ-dependent Smad2 phosphorylation (Fig. 1 B). TGFβ-induced phosphorylation of p38 MAPK was also enhanced by CIN85 overexpression in human embryonic kidney (HEK) 293T and PC-3U cells (Fig. 1 C). However, because the background phosphorylation of p38 was enhanced about twofold in CIN85 overexpressing cells, the fold increase after TGFβ stimulation was less affected. It is possible that overexpression of CIN85 makes cells more sensitive to endogenously synthesized TGFβ, or CIN85 may activate p38 through other mechanisms that are not directly connected to TGFβ signaling (Aissouni et al., 2005; Kim et al., 2013). Figure 1. CIN85 enhances TGFβ signaling. (A) PC-3U cells were transfected with empty vector (pCDNA3) or Flag-CIN85 plasmid and stimulated with 5 ng/ml TGFβ, as indicated. Cell lysates were prepared and the levels of phosphorylated Smad2, total Smad2, and Flag-CIN85 were analyzed by immunoblotting. (B) PC-3U cells were transfected with control siRNA or different amounts of CIN85-siRNA and incubated with 5 ng/ml TGFβ for 15 min, as indicated. Cell lysates were prepared and the level of phosphorylated Smad2 and the efficiency of CIN85 knockdown were analyzed by immunoblotting. The membrane was stripped and probed for tubulin to confirm equal loading. (C) PC-3U cells (top) or HEK293T cells (bottom) were transfected with control plasmid or Flag-CIN85 and stimulated with 5 ng/ml TGFβ, for the indicated time periods. Cell lysates were prepared and the level of phosphorylated p38 MAPK was analyzed by immunoblotting. The efficiency of transfection was evaluated by blotting with Flag antibodies and loading was assessed by probing for tubulin. (D) PC-3U cells were transfected with control or Flag-CIN85 plasmids. At 48 h after transfection, TGFβ (5 ng/ml) was added to the culture media for an additional 72 h. The expression of E-cadherin, N-cadherin, fibronectin, and Snail was analyzed by immunoblotting. The efficiency of transfection was evaluated by incubation of the membrane with Flag antibodies and loading was determined by probing for tubulin. In addition, other known effects of TGFβ on the protein expression in cells were also strengthened by CIN85. Thus, in PC-3U cells the TGFβ-induced expression of fibronectin and Snail, and the down-regulation of E-cadherin, were enhanced in cells overexpressing CIN85 compared with untransfected cells (Fig. 1 D). N-Cadherin expression in PC-3U cells was not affected by TGFβ treatment or by CIN85 overexpression (Fig. 1 D). To further investigate the role of CIN85 in TGFβ signaling, we performed Smad-specific promoter reporter assays. In agreement with the aforementioned results, transfection of HEK293T and PC-3U cells with increasing amounts of CIN85 enhanced TGFβ-dependent activation of CAGA12-Luc and ARE-Luc reporters in a dose-dependent manner (Fig. 2 A). The level of endogenous PAI-1 mRNA was also higher in PC-3U cells overexpressing CIN85 after treatment with TGFβ (Fig. 2 B). The effect of CIN85 on TGFβ-induced activation of transcription was dependent on the receptor kinase activity; treatment of the cells with a specific inhibitor of the TβRI kinase (SB505124) completely blocked activation of the CAGA12-Luc reporter in control cells as well as in the cells overexpressing CIN85 (Fig. 2 C). Figure 2. CIN85 enhances TGFβ transcriptional activity. (A) HEK293T cells, transfected with CAGA12-Luc (left) or xFAST/ARE-Luc (right) and different amounts of Flag-CIN85 plasmid, were not treated (open bars) or treated with 5 ng/ml TGFβ (shaded bars) for 20 h, and luciferase activity was measured. The results are presented as the mean (in percentage of control) of six independent experiments. Error bars represent SD. (B) qRT-PCR analysis for expression for PAI-1 was performed on mRNA extracted from PC-3U cells transiently transfected with empty pCDNA3 vector or Flag-CIN85 and treated with 5 ng/ml TGFβ for the indicated time periods. The results were normalized on the basis of GAPDH mRNA expression. The data are plotted as the mean fold induction of TGFβ-stimulated mRNA levels relative to unstimulated levels (0 h set to 1) with SD determined from triplicate measurements. Results of one representative experiment out of three performed are shown. (C) HEK293T cells transfected with CAGA12-Luc reporter and control plasmid (open bars) or Flag-CIN85 plasmid (shaded bars). The cells were pretreated for 1 h with DMSO or TβRI kinase inhibitor SB505124 (10 µM) and then incubated with 5 ng/ml TGFβ for 20 h as indicated. Luciferase activity was measured and transfection efficiency was normalized to β-galactosidase activity. The results are presented as the mean (in percentage of control) of three independent experiments. Error bars represent SD. *, P < 0.05 cells transfected with CIN85 compared with cells transfected with empty vector. Additionally, CIN85 enhanced TGFβ-induced migration of PC-3U cells. In a tissue culture wound-healing assay PC-3U cells overexpressing CIN85 closed the wound faster in response to TGFβ stimulation compared with control cells (Fig. 3). These observations further support the conclusion that CIN85 enhances TGFβ signaling. Figure 3. CIN85 enhances TGFβ-induced cell migration. (A) PC-3U cells were transfected with either an empty pcDNA3 vector or a Flag-CIN85 expression construct. To increase the percentage of transfected cells, they were cultured in medium with 0.4 mg/ml G418 for 10 d. Expression of Flag-CIN85 in the cells was evaluated by blotting with Flag and CIN85 antibodies (right). For the in vitro wound healing assay, cells were grown to confluence in 6-well plates and TGFβ (5 ng/ml) was added to culture medium supplemented with 0.1% FBS immediately after wounding. Phase-contrast images were captured after 24 and 30 h. Representative images for control (PC-3U/pCDNA3) and Flag-CIN85 expressing (PC-3U/CIN85) cells at different time points are shown (left). Bar, 200 µm. (B) Quantification of the motility of PC-3U/pCDNA3 and PC-3U/CIN85 cells in the cell culture wound healing assay was performed on 12 measurements in each experimental condition and expressed as a percentage of the open wound area. Data represent means ± SD. Brackets indicate the comparisons that showed significant differences in migration of control cells and cells overexpressing CIN85 under TGFβ treatment. Three independent experiments were done. *, P < 0.05. CIN85 interacts physically with TGFβ receptors To explore the mechanisms for CIN85-mediated enhancement of TGFβ signaling, we investigated whether CIN85 binds to TGFβ receptors. Lysate of HEK293T cells overexpressing Flag-CIN85 and HA-TβRI was subjected to immunoprecipitation using an HA antibody, followed by immunoblotting using a Flag antibody. A band with the expected size of Flag-CIN85 was seen in the lane corresponding to a lysate of cells stimulated with TGFβ, but not in the lane corresponding to unstimulated cells (Fig. 4 A), suggesting that TβRI and CIN85 make a complex after ligand treatment. Endogenous CIN85 and TβRI were also reciprocally coimmunoprecipitated from lysates of PC-3U cells by antibodies against TβRI and CIN85, respectively, in a TGFβ-dependent manner (Fig. 4 B). CIN85 was also coimmunoprecipitated with TβRII; however, in this case interaction did not depend on the treatment of cells with TGFβ (Fig. 4 C). These results suggest that both TGFβ receptors can form a complex with CIN85 and that interaction between TβRI and CIN85 is enhanced by ligand treatment. Figure 4. CIN85 interacts with TGFβ receptors. (A) HEK293T cells were transfected with Flag-CIN85 in the absence or presence of HA-TβRI and incubated with 5 ng/ml TGFβ for 15 min, as indicated. The cell lysates were subjected to immunoprecipitation with HA antibodies followed by blotting with Flag antibodies. The levels of HA-TβRI and Flag-CIN85 expression were determined by immunoblotting of total cell lysates. (B) PC-3U cells were treated with 5 ng/ml TGFβ for 15 min and endogenous TβRI (left) or CIN85 (right) were immunoprecipitated with 1 µg of goat anti-TβRI or rabbit anti-CIN85 antibodies, respectively. Immunoprecipitates were analyzed by immunoblotting for the presence of CIN85 or TβRI, as indicated. Total cell lysate was also subjected to immunoblotting for CIN85 and TβRI, and for phosphorylated Smad2 to determine activation of TGFβ signaling (bottom). (C) HEK293T cells were transfected with Flag-CIN85 in the absence or presence of His-TβRII and incubated with 5 ng/ml TGFβ for 15 min, as indicated. TβRII was immunoprecipitated from the cell lysates with His antibodies. Coimmunoprecipitated CIN85 and immunoprecipitated TβRII were detected by immunoblotting with Flag and His antibodies, respectively. The levels of His-TβRII and Flag-CIN85 expression were determined by immunoblotting of total cell lysates. (D) Schematic illustration of the CIN85 molecule. The full-length molecule (FL) and the parts of CIN85 included into the deletion mutants containing three SH3 domains (3SH3) or proline-rich and coiled-coil domains (PcC) are indicated. (E) TβRI interacts with the N-terminal part of CIN85. HA-TβRI was transfected into HEK293T cells together with full-length Flag-CIN85 or with deletion mutants including three SH3 domains (3SH3) or proline-rich and coiled coil domains (PcC). The cells were incubated with 5 ng/ml TGFβ for 15 min and TβRI was immunoprecipitated from the cell lysates with HA antibodies. The coimmunoprecipitated CIN85 or CIN85 deletion mutants were detected by immunoblotting with Flag antibodies. The expression of HA-TβRI and Flag-CIN85 molecules was determined by immunoblotting of total cell lysates. (F) TβRII interacts with the N-terminal part of CIN85. HA-TβRII was transfected into HEK293T cells together with full-length Flag-CIN85 or with its deletion mutants. The cells were treated as in E and TβRII was immunoprecipitated from the cell lysates with HA antibodies. The coimmunoprecipitated CIN85 or CIN85 deletion mutants were detected by immunoblotting with Flag antibodies. (G) The transcriptional activity of TGFβ is enhanced by full-length CIN85 but not by its N- or C-terminal fragments. HEK293T cells, transfected with CAGA12-Luc and full-length CIN85 or its N- or C-terminal parts, were treated (shaded bars) or not (open bars) with TGFβ (5 ng/ml) for 20 h. Luciferase activity was measured and transfection efficiency was normalized to β-galactosidase activity. The results are presented as the mean (in percentage of control) of three independent experiments. Error bars are SD. *, P < 0.05. (H) COS7 cells were grown on cover glasses and transfected with HA-TβRI and Flag-CIN85. 24 h after transfection, the cells were transferred to serum-low culture medium and starved for 16 h. Then cells were treated with TGFβ (5 ng/ml; 15 min), fixed, permeabilized, and incubated with rabbit HA antibodies and mouse Flag antibodies. HA-tagged receptor (green) and Flag-CIN85 (red) were detected by binding Alexa Fluor 488–conjugated anti-rabbit IgG and Alexa Fluor 555–conjugated anti–mouse IgG secondary antibodies, respectively. Sites of colocalization are indicated by arrowheads. Bars, 10 µm. We explored if the kinase activities of the TGFβ receptors are important for the interaction between TβRI and CIN85. We pretreated the HEK293T cells, transfected or not by CIN85, with an inhibitor of the TβRI kinase (SB505124) or with an inhibitor of both TβRI and TβRII kinases (LY2109761), and immunoprecipitated TβRI from the lysates of these cells. TGFβ enhanced the interaction between TβRI and CIN85, but the kinase inhibitors had no effect on the amount of CIN85 coprecipitated with TβRI (Fig. S1 A). In addition, inhibition of internalization by low temperature or by the dynamin inhibitor dynasore did not prevent TGFβ-induced TβRI interaction with CIN85 (Fig. S1, B and C). These results suggest that neither the kinase activities of the receptors nor endosomal trafficking are needed for ligand-enhanced TβRI binding to CIN85. CIN85 consists of three SH3 domains in the N-terminal half of the molecule and a proline-rich region and coiled-coil domain in the C terminus (Fig. 4 D). Coimmunoprecipitation experiments revealed that a deletion mutant of CIN85 that contained three SH3 domains coprecipitated with TβRI, whereas a mutant that encompassed the proline-rich and coiled-coil domains did not (Fig. 4 E). Similarly, only full-length CIN85 and the SH3 domains containing deletion mutant, but not the deletion mutant that contained only the C-terminal part, were coprecipitated with HA-TβRII (Fig. 4 F). To further delineate which part of CIN85 is critical for the interaction with TβRI, we performed GST pull-down assays. Consistent with the results of coimmunoprecipitation experiments (Fig. 4 E), only the N-terminal part of CIN85 was found to interact with TβRI (Fig. S2 A). GST fusion proteins that consisted of only the SH3-A or only the SH3-B domain of CIN85 both interacted with TβRI, but less efficiently than the GST fusion construct that included all three SH3 domains of CIN85 (Fig. S2 A). The N-terminal part of CIN85 was less efficient in enhancing TGFβ-induced activation of the CAGA12 luciferase reporter, when compared with full-length CIN85 (Fig. 4 G), suggesting that the proline-rich and/or coiled-coil domains are also important for the enhancement of TGFβ signaling by CIN85. To explore whether CIN85 and TβRI coexist in the same intracellular compartments, we performed immunofluorescence staining of COS7 cells expressing both proteins. Colocalization of CIN85 with TβRI was detected in the perinuclear region, intracellular vesicles, and membrane ruffles (Fig. 4 H). Quantitation of the immunostained images suggested that 14 ± 3% of the TβRI were colocalized with CIN85 at the cell periphery; this number did not change in cells treated with TGFβ, suggesting that the distribution of either of the proteins was not affected by TGFβ treatment. In contrast, the increased coimmunoprecipitation of TβRI and CIN85 in lysates from cells treated with TGFβ (Fig. 4, A and B) suggests that ligand stimulation somehow modifies TβRI and/or CIN85 in a manner that enhances the affinity of the proteins for each other. TRAF6 ubiquitinates CIN85 and enhances its binding to TβRI Next, we studied possible means by which TGFβ stimulation promotes the interaction between TβRI and CIN85. Because all three SH3 domains of CIN85 bind to ubiquitin (Bezsonova et al., 2008) and TGFβ stimulation, via the E3 ubiquitin ligase TRAF6, causes polyubiquitination of TβRI (Mu et al., 2011; Sundar et al., 2015), we investigated whether TRAF6 plays a role in the interaction between CIN85 and TβRI. We found that TGFβ stimulation of HEK293T cells overexpressing CIN85 and TRAF6 induced ubiquitination of CIN85 (Fig. 5 A). Moreover, inhibition of TRAF6 expression by siRNA decreased ubiquitination of CIN85 in cells treated with TGFβ (Fig. 5 B). Figure 5. TRAF6 ubiquitinates CIN85 and enhances the interaction between TβRI and CIN85. (A) HEK293T cells were cotransfected with Flag-CIN85, Flag-TRAF6, and HA-ubiquitin and treated with 5 ng/ml TGFβ for the indicated time periods. The cell lysates were subjected to immunoprecipitation with CIN85 antibodies followed by immunoblotting with polyubiquitin antibodies. The expression of the proteins under investigation was determined by immunoblotting total cell lysates. (B) HEK293T cells were transfected with control siRNA or TRAF6-siRNA, as well as with HA-ubiquitin, and treated with 5 ng/ml TGFβ for 10 min. The cell lysates were subjected to immunoprecipitation using CIN85 antibodies followed by immunoblotting with polyubiquitin antibodies. The efficiency of TRAF6 knockdown and equality of protein loading were analyzed by immunoblotting of total cell lysates. (C) HEK293T cells were cotransfected with HA-TβRI and Flag-CIN85, in the absence or presence of Flag-TRAF6 or E3-ligase-deficient Flag-TRAF6(C70A) mutant, and treated with 5 ng/ml TGFβ for 15 min, as indicated. Cell lysates were subjected to immunoprecipitation with HA antibodies followed by immunoblotting with Flag or HA antibodies. The expression of the proteins under investigation was determined by immunoblotting of total cell lysates. (D) HEK293T cells were cotransfected with Flag-CIN85 and HA-TβRI or HA-TβRI(E161A) mutant and treated with 5 ng/ml TGFβ for 15 min, as indicated. Cell lysates were subjected to immunoprecipitation with HA antibodies followed by immunoblotting with Flag or HA antibodies. The expression of proteins was evaluated by immunoblotting of total cell lysates. (E) Wild-type or TRAF6-deficient mouse embryonic fibroblasts (MEFs) were treated with 5 ng/ml TGFβ for 15 min. Cell lysates were subjected to immunoprecipitation with anti-TβRI antibodies, followed by immunoblotting for CIN85. Total cell lysates were also subjected to immunoblotting for endogenous TRAF6, CIN85, and TβRI. To explore further the effect of TRAF6 on TβRI–CIN85 complex formation, HA-TβRI was immunoprecipitated from the lysates of HEK293T cells transfected with Flag-CIN85 and HA-TβRI, in the presence or absence of Flag-TRAF6. The TGFβ-induced coprecipitation of TβRI and CIN85 was enhanced upon coexpression of TRAF6 in the cells (Fig. 5 C). In contrast, TGFβ treatment did not affect coimmunoprecipitation of CIN85 with mutant TβRI(E161A), which is unable bind to TRAF6 (Sorrentino et al., 2008; Fig. 5 D). Moreover, treatment of TRAF6-deficient mouse embryonic fibroblasts with TGFβ did not enhance coimmunoprecipitation of endogenous CIN85, whereas a coimmunoprecipitation was seen in wild-type mouse embryonic fibroblasts (Fig. 5 E). The interaction of TβRII with CIN85 was not affected by overexpression of TRAF6, with or without stimulation with TGFβ (Fig. S2 C). To elucidate the importance of the enzymatic activity of TRAF6 for the CIN85–TβRI interaction, we used a ligase-inactive mutant of TRAF6 in the coimmunoprecipitation experiments. In contrast to wild-type TRAF6, mutant TRAF6 was unable to promote the interaction between the receptor and CIN85 (Fig. 5 C). Moreover, GST-TβRI ubiquitinated by TRAF6 in vitro precipitated CIN85 more efficiently than nonubiquitinated GST-TβRI (Fig. S2 B). Thus, our observations support the notion that the enzymatic activity of TRAF6 is important for TGFβ-induced interaction between TβRI and CIN85. CIN85 regulates the intracellular localization of TβRs Similar to many other adaptor proteins, CIN85 has been reported to take part in multiple cellular processes including internalization of several activated receptor tyrosine kinases and their vesicle-mediated transport (Szymkiewicz et al., 2004; Havrylov et al., 2010). We therefore studied if the effect of CIN85 on TGFβ signaling could be explained by modulation of TβRI location. We separated the postnuclear supernatants of PC-3U cells transfected with HA-TβRI alone or together with Flag-CIN85 by centrifugation in discontinuous iodixanol gradient (Fig. 6 A). Immunoblot analysis of the fractions showed that most of HA-TβRI (∼80%) was localized at the 30–20% or the 20–10% iodixanol interfaces enriched for EEA1- and Rab7-positive vesicles, respectively (Fig. 6, A and B). Overexpression of CIN85 resulted in the appearance of detectable amounts of HA-TβRI in light fractions at the 10–0% iodixanol interface enriched for Rab11-positive vesicles (Fig. 6, A and B). Analysis of these fractions for the presence of biotinylated cell surface proteins revealed that they contained plasma membrane as well (Fig. 6 C), which is in accordance with a previous study (Zhu et al., 2012). We also obtained a similar effect of CIN85 overexpression on the distribution of HA-TβRI in an iodixanol gradient using postnuclear supernatants from HEK293T cells (Fig. S3, A and B). Treatment of these cells with TGFβ for 15 min did not change the distribution of the receptor (Fig. 3, A and B). In addition, immunofluorescence staining showed that in PC-3U cells deprived of CIN85 by siRNA transfection, endogenous TβRI was concentrated in intracellular compartments, whereas in cells with normal level of CIN85, the distribution of the receptor was more even throughout the cell (Fig. 6 D). These results suggested that CIN85 may be involved in the regulation of the subcellular localization of TβRI. Figure 6. CIN85 alters the compartmentalization of TβRI. (A) PC-3U cells were transfected with HA-TβRI alone or in combination with Flag-CIN85. 48 h after transfection, cells were homogenized and subcellular membrane vesicles were separated by iodixanol density gradient ultracentrifugation, as described in Materials and methods. Equal volumes of each fraction were separated by SDS-PAGE and analyzed by immunoblotting with antibodies against HA to detect HA-TβRI. Marker proteins for different membrane vesicles, including EEA1 for early endosomes, Rab7 for late endosomes, and Rab11 for recycling endosomes, were used to locate the migration of these membrane vesicles along the density gradients. CIN85 was analyzed in the fractions by blotting with Flag antibodies. (B) Quantification of the immunoblots of HA-TβRI presented in A showing that coexpression of CIN85 resulted in an increased amount of the receptor in the upper fractions of iodixanol gradient. The intensity of HA-positive signal in each fraction was quantified using the Quantity One software. The sum of HA-positive signals from all fractions was set as 100%. The location of each marker protein is indicated. PM, plasma membrane. The data shown are from a representative experiment out of four repeats. (C) Cell surface proteins migrate at the top of the density gradient. Proteins on the cell surface of PC-3U cells transfected with HA-TβRI were labeled with biotin and then the subcellular membrane vesicles were separated by iodixanol density gradient ultracentrifugation. Biotinylated proteins from each fraction were precipitated with neutravidin-agarose beads and analyzed by immunoblotting with HA and EGFR antibodies. The total amount of HA-TβRI was determined by subjecting the whole fraction to immunoblotting using HA antibodies. (D) Endogenous TβRI is sequestered in intracellular compartments in cells with down-regulated CIN85 expression. PC-3U cells were grown on cover glasses and transfected with control siRNA or CIN85-siRNA. 72 h after transfection the cells were fixed, permeabilized, and incubated with goat anti-TβRI and rabbit anti-CIN85 antibodies. TβRI (green channel) and CIN85 (red channel) were visualized by binding Alexa Fluor 488–conjugated anti–goat and Alexa Fluor 555–conjugated anti–rabbit antibodies, respectively. Cells with a lower amount of CIN85 are indicated by arrows. Arrowheads indicate cells containing a normal level of CIN85. Bars, 10 µm. By biotinylating cell surface proteins, we found that more TβRI and TβRII were pulled down with neutravidin agarose beads from lysates of HEK293T cells (Fig. 7 A) or PC-3U cells (Fig. 7 B), which overexpressed CIN85, compared with control cells. These data were consistent with the results we obtained using iodixanol density gradient membrane flotation assays and suggested that CIN85 promotes the presentation of TGFβ receptors on the cell surface. Notably, the biotinylation assay showed that the amount of HA-TβRI on the surface of both control and CIN85-transfected cells was not affected by treatment with TGFβ, at least for the first two hours (Fig. 7 C). Figure 7. CIN85 increases the amount of TβRI and TβRII on the cell surface. (A) Overexpression of CIN85 in HEK293T cells increases the amount of TβRI and TβRII at the cell surface. HEK293T cells were transfected with HA-TβRI and His-TβRII in the absence or presence of Flag-tagged CIN85, as indicated. 48 h after transfection, cell surface proteins were biotinylated and isolated by incubation with neutravidin-agarose beads, as described in Materials and methods, and the amount of HA-TβRI and His-TβRII was analyzed by immunoblotting with HA and His antibodies, respectively. The levels of HA-TβRI, His-TβRII, and Flag-CIN85 expression were determined by immunoblotting total cell lysates. (B) Overexpression of CIN85 in PC-3U cells increases the amount of endogenous TβRI on the cell surface. PC-3U cells that either overexpressed CIN85 or did not were prepared, as described in Fig. 3 A. The cell surface proteins were biotinylated and precipitated by neutravidin-agarose beads. The precipitates were analyzed for the presence of endogenous TβRI by immunoblotting with rabbit anti-TβRI antibodies. Expression of CIN85 and TβRI was evaluated by immunoblotting of total cell lysates with the indicated antibodies. CIN85-overexpressing cells not labeled with biotin were used as a control for pull-down specificity. (C) The effect of CIN85 on cell surface expression of HA-TβRI does not depend on ligand treatment. HEK293T cells were transfected with HA-TβRI in the absence or presence of Flag-tagged CIN85 and incubated with 5 ng/ml TGFβ for 0, 60, and 120 min. Cell surface proteins were biotinylated and isolated by neutravidin-agarose pull-down, and the amount of the HA-TβRI was analyzed by immunoblotting with HA antibodies. The levels of HA-TβRI and Flag-CIN85 expression were determined by immunoblotting of total cell lysates. (D) Full-length CIN85 is necessary to promote TβRI expression at the cell surface. HEK293T cells were transfected with HA-TβRI and full-length CIN85 or its truncated mutants, as indicated. 48 h after transfection biotinylated cell surface proteins were isolated by neutravidin-agarose pull-down, and the amount of HA-TβRI brought down was analyzed by immunoblotting. HA-TβRI–overexpressing cells that had not been treated with biotin were used as a control for pull-down specificity. The expression of transfected proteins was evaluated by immunoblotting of total cell lysates. Increased amount of available TGFβ receptors on the surface of CIN85-overexpressing cells could explain the stronger response to ligand stimulation. We expressed the C- or N-terminal fragments of CIN85 in HEK293T cells and determined their effects on the amount of HA-TβRI on the cell surface, compared with the effect of full-length CIN85. None of the CIN85 fragments could increase cell surface expression of TβRI (Fig. 7 D). These results are consistent with the observation that only full-length CIN85 strengthens cellular responses to TGFβ (Fig. 4 E). They also support the notion that the enhancing effect of CIN85 on TGFβ signaling is caused by the higher amount of the receptors on the cell surface. The effect of CIN85 on TGFβ signaling is blocked by dominant-negative Rab11 We did not observe any effect of CIN85 overexpression on the total amount of TGFβ receptors (see, e.g., Fig. 7 A, bottom), whereas the results of fractionation in density gradient (Fig. 6) supported the possibility that CIN85 alters their intracellular localization. It has been proposed that clathrin-mediated endocytosis of TGFβ receptors promotes TGFβ signaling, whereas internalization through lipid rafts/caveolae negatively modulates TGFβ signaling by promoting TGFβ receptor degradation (Di Guglielmo et al., 2003; Chen, 2009). As CIN85 has been shown to interact with components of clathrin-coated pits, we examined whether its enhancing effect on TGFβ signaling can be explained by a shift to clathrin-mediated internalization of TβRI. However, we could not detect any significant effect of CIN85 overexpression on the distribution of TβRI between caveolin-1–positive fractions and nonraft membrane fractions (Fig. S3 C). TGFβ receptors internalized via clathrin-coated pits are recycled back to the plasma membrane through a Rab11-dependent mechanism (Mitchell et al., 2004). The possibility that TβRI recycling was affected by CIN85 was investigated by a biotinylation assay (Fig. S4). TβRI at the cell surface was first biotinylated by sulfo-NHS-SS-biotin. Cells were then incubated for 30 min at 37°C, and thereafter biotin was stripped from extracellular proteins by incubation in a reducing agent. 87% of the biotinylated receptor remained internalized in control cells (Fig. S4, lane 3), whereas cells that overexpressed CIN85 contained only 66% of the biotinylated receptor intracellularly (Fig. S4, lane 8), suggesting that CIN85 promoted recycling of the receptor back to the cell surface. We therefore investigated whether enhancement of TGFβ signaling by CIN85 was affected by interfering with Rab11 activity. Overexpression of a dominant-negative myc-Rab11(S25N) mutant in HEK293T cells decreased the amount of biotinylated HA-TβRI on the cell surface under steady-state conditions (Fig. 8 A), consistent with previously reported data (Mitchell et al., 2004). In cells transfected with myc-Rab11(S25N), overexpression of CIN85 did not increase the level of TβRI on the cell surface (Fig. 8 A). Moreover, the enhancement of TGFβ-induced CAGA12-Luc reporter activation by CIN85 overexpression was abolished by expression of the dominant-negative Rab11 mutant (Fig. 8 B). Rab11 was coimmunoprecipitated with wild-type CIN85, but not with the truncated mutants of CIN85 (Fig. 8 C). A dominant-negative mutant of Rab4, which does not regulate TGFβ receptor recycling (Mitchell et al., 2004), did not prevent CIN85-regulated increasing of TβRI on the cell surface (Fig. 8 D). Together, these data suggest that CIN85 is involved in the Rab11-dependent trafficking of TβRI to the plasma membrane. Figure 8. Dominant-negative Rab11 inhibits the effect of CIN85 on TGFβ signaling. (A) HEK293T cells were transfected with HA-TβRI, Flag-CIN85, and the dominant-negative myc-Rab11(S25N) mutant, as indicated. 48 h after transfection, cell surface proteins were biotinylated and isolated by neutravidin-agarose pull-down. The amount of biotinylated HA-TβRI was analyzed by immunoblotting with HA antibodies. Cells transfected with HA-TβRI, but not treated with biotin, were used as a control for specificity of neutravidin-agarose pull-down. The levels of HA-TβRI, Flag-CIN85, and myc-Rab11(S25N) expression were determined by immunoblotting of total cell lysates. (B) HEK293T cells that had been transfected with CAGA12-Luc, Flag-CIN85, and myc-Rab11(S25N), as indicated, were either treated with TGFβ (5 ng/ml) for 20 h (shaded bars) or not (open bars), and the luciferase activity in the cell lysates was measured. Transfection efficiency was normalized to β-galactosidase activity. The data are presented as the mean (in percentage of control) of six independent experiments. Error bars are SD. *, P < 0.05. (C) HEK293T cells were transfected with full-length Flag-CIN85 or with its deletion mutants that encompassed the three SH3 domains (3SH3) or the proline-rich and coiled-coil domains (PcC). The cell lysates were subjected to immunoprecipitation with Flag antibodies, followed by blotting with Rab11 antibodies. The expression levels were determined by immunoblotting of total cell lysates with Rab11 and Flag antibodies. (D) Dominant-negative Rab4 does not inhibit the effect of CIN85 on the cell surface expression of TβRI. HEK293T cells were transfected with HA-TβRI, Flag-CIN85, and the dominant-negative Rab4 mutant, as indicated. 48 h after transfection, cell surface proteins were biotinylated and isolated by neutravidin-agarose pull-down. The amount of biotinylated HA-TβRI on the cell surface was analyzed by immunoblotting with HA antibodies. The levels of HA-TβRI, Flag-CIN85, and Rab4 expression were determined by immunoblotting of total cell lysates. Expression of CIN85 is increased in prostate tumors It is known that prostate tumors express high levels of TGFβ (Thompson et al., 1992) and that TGFβ signaling is activated in many tumors, especially in the late stages of tumorigenesis (Massagué, 2008; Heldin and Moustakas, 2012; Mu et al., 2012). We investigated whether expression of CIN85 correlates with the development of prostate cancer. Histological analysis of prostate tumor tissues from patients showed that expression of CIN85 positively correlated with the severity of the disease (Fig. 9, A and B). We also observed much higher levels of phospho-Smad2 in aggressive prostate cancer tissues when compared with normal prostate, and a high level of phosphorylated Smad2 corresponded with high levels of CIN85 in the same tissue (Fig. S5). These results suggest a correlation between the level of CIN85 expression and activation of TGFβ signaling in prostate cancers. Figure 9. CIN85 expression correlates with malignancy of prostate cancers. (A) Normal prostate and tumor tissues at different malignancy grades were stained with anti-CIN85 antibodies. Bar, 20 µm. GS, Gleason score. (B) Quantification of CIN85 expression in different prostate tissues shows a positive correlation between tumor grade and amount of CIN85. Error bars represent SEM. *, P < 0.05, compared with normal tissue. Interestingly, we found that treatment of PC-3U cells with TGFβ resulted in higher levels of CIN85 mRNA (Fig. 10 A) and protein (Fig. 10 B). These observations suggest the existence of a positive feedback mechanism by which activation of TGFβ receptors results in elevation of CIN85 expression, which in turn enhances TGFβ signaling. Figure 10. TGFβ induces CIN85 mRNA and protein expression. (A) PC-3U cells were stimulated with TGFβ for 0, 6, and 24 h. The DNA was extracted and the expression of CIN85 mRNA (SH3KBP1; GenBank accession no. NM_031892.2) was analyzed by qRT-PCR. The values were normalized to GAPDH mRNA and plotted as the mean fold induction of TGFβ-stimulated mRNA levels relative to unstimulated levels (0 h was set to 1). Error bars correspond to SD values determined from triplicate measurements. *, P < 0.05 cells treated with TGFβ compared with untreated cells. (B) PC-3U cells were incubated in the presence of 5 ng/ml TGFβ for the indicated time periods. Cell lysates were analyzed for the expression of CIN85 by immunoblotting with anti-CIN85 antibodies. The membrane was stripped and probed with β-tubulin antibodies for loading control. (C) Schematic illustration of the proposed role of CIN85 in TGFβ signaling. Upon ligand-induced activation of TGFβ receptors, TRAF6 becomes autoubiquitinated and ubiquitinates TβRI and CIN85. CIN85 binds to TβRI and this complex is stabilized by the ubiquitin chains on TβRI, CIN85, and/or TRAF6. Formation of such a complex enhances the recycling of receptors through a Rab11-dependent route and increases their amount on the plasma membrane. This results in higher responsiveness of the cells to TGFβ. The activation of the TGFβ signaling pathway also increases expression of CIN85, providing a positive feedback loop. Discussion We have shown in this paper that CIN85 forms a complex with TGFβ receptors and that the E3 ubiquitin ligase TRAF6 promotes its interaction with TβRI in a TGFβ-dependent manner (Fig. 4, A–C; and Fig. 5, C and D). Overexpression of CIN85 resulted in a greater amount of TGFβ receptors on the cell surface (Fig. 7, A and B), but did not cause any significant changes in the total amount of TGFβ receptors. Dominant-negative Rab11 abolished the effect of CIN85 (Fig. 8, A and B), which, together with the results of a TβRI cell surface biotinylation assay (Fig. S4), supports the notion that CIN85 regulates TβRI recycling. The inability of dominant-negative Rab4 mutant to prevent the effect of CIN85 suggests that CIN85 does not direct TβRI to the fast recycling route that is regulated by Rab4, but rather enhances its recycling through the Rab11-controlled route. We cannot exclude the possibility that the increased number of TGFβ receptors on the cell surface is also a result of enhanced trafficking of newly synthesized receptors to the plasma membrane, which has been proposed to be modulated by CIN85 for other proteins (Havrylov et al., 2010). The increased number of TGFβ receptors at the cell surface may explain why CIN85 expression makes cells more sensitive to TGFβ treatment (Figs. 1, 2, and 3), as TGFβ responsiveness has been shown to correlate with the cell surface receptor levels (Xu et al., 2012). Knockdown of CIN85 by siRNA resulted in trapping of TβRI in intracellular compartments, but not on the cell surface (Fig. 6 D). This observation further supports the notion that CIN85 is primarily involved in the regulation of TβRI recycling, but not in its internalization. In a similar manner, the recycling of TβRII was blocked by down-regulation of Dab2, another adaptor protein (Penheiter et al., 2010). It has been shown that Dab2 makes complexes with CIN85 on the plasma membrane and that they dissociate from each other upon EGF stimulation (Kowanetz et al., 2003). We explored the possibility that Dab2 and CIN85 work together to control TβRI recycling. However, overexpression of mutant Dab2R699A, which cannot bind CIN85 (Kowanetz et al., 2003), had no effect on the enhancement of TGFβ signaling by CIN85 (unpublished data). Thus, it seems that these proteins regulate different events in the trafficking of TGFβ receptors, which agrees with the finding that the transport of TβRI and TβRII back to the cell surface is dependent on two distinct recycling pathways (Gleason et al., 2014). It has been shown that CIN85 affects the recycling of EGFR through interaction with the Arf GTPase-activating protein ASAP1 (Kowanetz et al., 2004). It remains to be determined if ASAP1 is also involved in CIN85-modulated trafficking of TβRI. The E3 ubiquitin ligase TRAF6 strongly enhanced the TGFβ-induced interaction between CIN85 and TβRI, as demonstrated by coimmunoprecipitation experiments (Fig. 5, C and D). CIN85 has previously been shown to interact with TRAF1 and TRAF2, but not with TRAF6 (Narita et al., 2005). We found that the interaction between TβRI and CIN85 was enhanced only by wild-type TRAF6, but not by a TRAF6 ubiquitin ligase–deficient mutant (Fig. 5 C). TGFβ-induced oligomerization of TGFβ receptors leads to autoubiquitination of TRAF6, which in turn activates the MAPK kinase kinase TAK1 and the MKK3/6–p38 pathway (Sorrentino et al., 2008). We have shown that TβRI is also ubiquitinated by TRAF6 (Mu et al., 2011; Sundar et al., 2015); given that all three SH3 domains of CIN85 bind to ubiquitin (Bezsonova et al., 2008), it is possible that polyubiquitin chains mediate the binding of CIN85 to the receptor complex. The role of TβRI ubiquitination in the regulation of its intracellular trafficking remains to be explored further, as it can also promote the cleavage of the receptor’s intracellular domain and its translocation to the nucleus (Mu et al., 2011). The interaction between TβRI and CIN85 was found to be promoted by TGFβ stimulation (Figs. 4 and 5). However, the finding that overexpression of CIN85 by itself made cells more sensitive to TGFβ stimulation (Fig. 1) suggests that there may be a low affinity interaction between TβRI and CIN85 before TGFβ stimulation. The TGFβ-induced stabilization of the interaction would then serve as an amplification mechanism. We found that TGFβ induces expression of CIN85 in cultured cells (Fig. 10, A and B). This suggests the existence of a positive feedback loop, where TGFβ induces the expression of CIN85, which in turn enhances the responsiveness of cells to TGFβ. Interestingly, expression of CIN85 correlates with the malignancy of prostate cancers (Fig. 9, A and B) and with increased TGFβ activity, as visualized by the staining of phosphorylated Smad2 in aggressive prostate cancer tissues (Fig. S5). This further supports the notion that CIN85 is part of an amplification mechanism that contributes to progression of prostate cancers by increasing TGFβ signaling. In summary, we have demonstrated that CIN85 enhances TGFβ-induced signaling and cellular responses to TGFβ (Fig. 10 C). CIN85 interacts with TGFβ receptors in a TRAF6-dependent manner and increases the exposure of TβRI on the cell surface by promoting receptor recycling. The positive effect of CIN85 on TGFβ signaling together with the observed expression of CIN85 in malignant prostate cancer cells are of potential importance for the understanding of the mechanisms that control cancer progression. Materials and methods Cell culture and transfection HEK293T and COS7 cells (ATCC) were cultured in DMEM supplemented with 10% FBS, glutamine, and antibiotics. The PC-3U human prostatic carcinoma cell line was obtained from S. Nilsson (University Hospital, Uppsala, Sweden) and represents a clone from the original PC-3 cell line (ATCC; Franzén et al., 1993). The cells were cultured in RPMI medium supplemented with 10% FBS, 1% glutamine, and 1% penicillin-streptomycin. All cell cultures were incubated at 37°C in the presence of 5% CO2. For TGFβ stimulation experiments, PC-3U cells were starved for 16 h in 0.1% FBS in RPMI. HEK293T and COS7 cells were starved for 16 h in 1% FBS in DMEM. Cells were then stimulated with 5 ng/ml TGFβ. Plasmids and siRNAs Expression vectors encoding Flag-CIN85, Flag-CIN85(3SH3), and Flag-CIN85(PcC) in pcDNA3 vector, and GST-CIN85(SH3ABC), GST-CIN85(SH3A), GST-CIN85(SH3B), and GST-CIN85(PcC) in pGEX-4T-1 vector, were provided by I. Dikic (Goethe University, Frankfurt am Main, Germany; Soubeyran et al., 2002). In brief, the FLAG-CIN85 construct encodes amino acids 1–665 of CIN85 (UniProt accession no. Q96B97), the Flag-CIN85(3SH3) construct encodes amino acids 1–332, and the Flag-CIN85(PcC) construct encodes amino acids 327–665 of CIN85. GST-CIN85(SH3A) and GST-CIN85(SH3B) encode individual SH3 domains of CIN85 (amino acids 1–83 and 79–217, respectively), the GST-CIN85(SH3ABC) construct encodes all three SH3 domains of CIN85 (amino acids 1–332), and GST-CIN85(PcC) contains only the proline-rich region and the coiled-coil motif of CIN85 (amino acids 327–665). The plasmid pCDNA3-TβRI-HA for expression of full-length TβRI (UniProt accession no. P36897) with an HA tag fused to the C terminus was provided by P. ten Dijke (University of Leiden, Leiden, Netherlands) and plasmid pCMV5-TβRII-His for expression of full-length TβRII (UniProt accession no. P37173) with 6XHis tag at the C terminus was obtained from J. Massagué (Memorial Sloan-Kettering Cancer Center, New York, NY). Constructs of N-terminal Flag-tagged mouse wild-type TRAF6 (UniProt accession no. P70196) and its E3 ligase–deficient mutant (Flag-TRAF6 C70A) cloned into pCDNA3-1 vector were gifs from Zhijian J. Chen (University of Texas Southwestern Medical Center, Dallas, TX). Plasmid pCDNA3.1-3XHA ubiquitin for expression of N-terminal HA-tagged ubiquitin was a gift from V.M. Dixit (Genentech, San Francisco, CA). The plasmid PCMV-intron-myc-Rab11S25N for expression of a dominant-negative mutant of Rab11A (UniProt accession no. P62491 was purchased from Addgene. The plasmid pEGFP-Rab4AS22N for expression of GTPase-defective Rab4A (UniProt accession no. P20338) was a gift from C. Hellberg (University of Birmingham, Birmingham, UK; Karlsson et al., 2006). The CAGA12 luciferase reporter vector was generated using the pGL3 basic plasmid (Promega) by inserting 12 tandem copies of the CAGA Smad binding element upstream of the adenovirus major late promoter driving luciferase gene expression (Dennler et al., 1998). The ARE luciferase reporter vector was generated using pGL5ti plasmid by inserting two tandem copies of the activin-responsive element (Chen et al., 1996). Plasmid pCMV-LacZ for expression of β-galactosidase was purchased from Takara Bio Inc. To knock down CIN85 expression, we used Silencer Select siRNA (5′-GACUGUUACCAUAUCCCAAtt-3′; Life Technologies), and nontargeting Silencer Select Negative Control No.1 siRNA (Life Technologies) was used as negative control. SMARTpool siGENOME Human TRAF6 (7189) siRNA (M-004712-00-0020) for the knockdown of TRAF6 and siGENOME Non-Targeting siRNA #4 (D-001210-04-05) used as a control were obtained from GE Healthcare. For transient transfections, cells were seeded at 60% density in the corresponding culture medium. Cells were transfected with plasmid DNAs using FuGENE HD (Roche) and with siRNAs using siLentFect reagent (Bio-Rad Laboratories) according to the manufacturers’ protocols. Transfection media were replaced with fresh media after 24 h. Antibodies and reagents Antibodies against the following proteins were used for immunoblotting (IB) or immunofluorescence (IF): rabbit anti-CIN85 antibody (diluted 1:500 for IB and 1:200 for IF) was obtained from Proteintech; mouse anti-His antibody (diluted 1:2,000) was obtained from Takara Bio Inc.; rabbit anti-HA (Y-11; diluted 1:500 for IB and 1:50 for IF), mouse anti-ubiquitin (P4D1; diluted 1:400), and rabbit anti-TβRI (V22; diluted 1:500 for IB and 1:50 for IF) antibodies were obtained from Santa Cruz Biotechnology, Inc.; rabbit anti–phospho-p38 MAPK (D3F9, 4511; diluted 1:1,000) and mouse anti-p38 MAPK (5F11, 9217; diluted 1:1,000) antibodies were obtained from Cell Signaling Technology. Mouse anti-Rab11 antibody (diluted 1:1,000 for IB and 1:500 for IF) was obtained from BD and rabbit anti-Smad2 antibody (diluted 1:1,000) was obtained from Epitomics. Rabbit antibody against TRAF6 (diluted 1:500) was obtained from Life Technologies. Mouse antibody against β-tubulin (diluted 1:1,000), mouse anti-HA (HA7; 1:1,000), and mouse anti-Flag (M2; diluted 1:1,000) antibodies were obtained from Sigma-Aldrich. Rabbit anti–phospho-Smad2 antiserum (diluted 1:500) was generated in house by immunizing rabbits with a peptide KKK-SSpMSp coupled to keyhole limpet hemocyanin, as described previously (Persson et al., 1998). Alexa Fluor–labeled anti-goat antibodies for IF were purchased from Invitrogen and used at a dilution of 1:1,000. RPMI and DMEM were purchased from Sigma-Aldrich. FBS was obtained from Biowest. Recombinant human TGFβ1 was obtained from PeproTech. The inhibitors of the kinases of TGFβ receptors SB505124 and LY2109761 were purchased from Sigma-Aldrich and Cayman Chem, respectively. Protein G–Sepharose 4B conjugate was purchased from Invitrogen, Clarity western ECL Substrate was purchased from Bio-Rad Laboratories, complete protease inhibitor cocktail was purchased from Roche, and prestained protein molecular mass markers for SDS-PAGE were purchased from Thermo Fisher Scientific (PageRuler). DAPI fluorescent dye to visualize cell nuclei by microscopy was purchased from Merck. Immunoblotting, immunoprecipitation, GST binding, and ubiquitination assays Cell lysates were prepared in buffer containing 50 mM Tris-HCl, pH 7.4, 0.5% Triton X-100, 10 mM NaCl, 2 mM NaF, 2 mM sodium pyrophosphate, 2 mM β-glycerophosphate, 2 mM sodium orthovanadate, and protease inhibitors. The lysates were cleared by centrifugation at 13,000 rpm for 15 min at 4°C and protein concentrations were measured by Bradford assay (Bio-Rad Laboratories). For protein expression analysis, total cell lysates with adjusted protein concentration were separated by SDS-PAGE and transferred to nitrocellulose membranes, which were incubated with the indicated antibodies. Bound antibodies were visualized by enhanced chemiluminescence. For coimmunoprecipitation analysis, cell lysates were first cleared by incubation with protein G beads for 20 min, and then incubated with the indicated antibody for 2 h at 4°C before incubation with protein G beads for 1 h at 4°C. The beads were washed four times with ice-cold lysis buffer and twice with PBS, and immunocomplexes were eluted from the beads by adding 2× Laemmli SDS sample buffer and boiling for 5 min. Immunoblot analysis of the precipitated proteins was performed using indicated primary antibodies and corresponding anti–rabbit or anti–mouse IgG light chain–specific antibodies (1:10,000 dilution; Jackson ImmunoResearch Laboratories, Inc.). Immunoblots were analyzed by Quantity One software (Bio-Rad Laboratories). For GST binding assays, GST fusion proteins were adsorbed on glutathione superflow agarose beads (Thermo Fisher Scientific), incubated with the lysates for 2 h at 4°C, and washed in the lysis buffer, and bound proteins were eluted and analyzed by Western blotting, as described in the previous paragraph. In vivo ubiquitination assay was performed, as previously described (Sorrentino et al., 2008). In brief, HEK293T cells were transfected as described in the figure legends, washed once in PBS, scraped in 1 ml PBS, and centrifuged at 400 g for 5 min. Non-covalent protein interactions were dissociated by boiling for 5–10 min in 1% SDS. Samples were diluted in PBS (1:10) containing 0.5% NP-40 and cleared by centrifugation at 12,000 g for 10 min. The supernatants were subjected to immunoprecipitation with a CIN85 antibody, followed by immunoblotting with a ubiquitin antibody. Cell migration assay PC-3U cells were transfected with either empty pCDNA3 vector or Flag-CIN85 expression construct. At 48 h after transfection, transfected cells were selected by growing them in culture medium with 0.5 mg/ml G418 for 10 d. Cells were grown to confluence in 6-well plates (Sarstedt AG & Co.) and incubated for 16 h in medium with 0.1% FBS, and wounds were made using a pipette tip. 5 ng/ml TGFβ was added to the culture media immediately after wounding. Phase-contrast images were captured at indicated times and the imaging data were analyzed by the TScratch software (Gebäck et al., 2009). Motility of the cells was quantitated by 12 measurements in each experimental category and expressed as a percentage of the original open wound area. TGFβ transcriptional activity analysis Quantitative RT-PCR (qRT-PCR) analysis for expression of PAI-1 was performed on mRNA extracted from PC-3U cells transiently transfected with empty pCDNA3 (vector) or Flag-CIN85 and treated with TGFβ (5 ng/ml) for indicated time periods. Total cellular RNA was purified with the RNeasy mini kit (QIAGEN) and cDNA was synthesized from 0.5 µg RNA using the iQ SYBR Green supermix (Bio-Rad Laboratories). The results were normalized on the basis of GAPDH mRNA expression. The data were plotted as the mean fold induction of TGFβ-stimulated mRNA levels relative to unstimulated levels (0 h set to 1) with SDs determined from triplicate measurements. qRT-PCR conditions were as described previously (Mu et al., 2011) with primers used as follows: human CIN85 mRNA (SH3KBP1; GenBank accession no. NM_031892.2), 5′-TGCAGATGGAAGTGAACGAC-3′ (forward) and 5′-TGGGGCAGAAAATTTGAGTC-3′ (reverse); human PAI-1, 5′-CTCTCTCTGCCCTCACCAAC-3′ (forward) and 5′-GTGGAGAGGCTCTTGGTCTG-3′ (reverse); human GAPDH, 5′-GGAGTCAACGGATTTGGTCGTA-3′ (forward) and 5′-GGCAACAATATCCACTTTACCA-3′ (reverse). The promoter-reporter assays were performed as previously described (Yakymovych et al., 2001). In brief, cells were transiently transfected with luciferase reporter vectors and pCMV-LacZ together with protein expression plasmids, as indicated in the figure legends. 24 h after transfection, the cells were starved for 20 h in culture medium with 1% FBS, and then stimulated with TGFβ for the next 24 h. The luciferase activity was analyzed with the Luciferase Assay System (Promega) using an EnSpire Multimode Plate Reader (PerkinElmer) and normalized on the basis of β-galactosidase activity. The data were plotted as mean values from triplicate determinations with SDs. Biotinylation of cell surface TβRI and recycling assay HEK293T cells or PC-3U were transfected with HA-TβRI in the absence or presence of Flag-tagged CIN85. 48 h after transfection, cells were starved in serum-low medium for 16 h and then treated with or without TGFβ (5 ng/ml) at 37°C. After indicated time periods, cells were washed with ice-cold PBS and incubated with 0.25 mg/ml sulfo-NHS-SS-biotin (Thermo Fisher Scientific) dissolved in PBS, supplemented with 0.9 mM CaCl2 and 0.5 mM MgCl2 (PBS2+) for 30 min at 4°C. All unreacted biotin was removed by washing with 50 mM glycine and 0.5% BSA dissolved in PBS2+, and cells were lysed in the RIPA buffer (50 mM Tris-HCl, pH 8, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 25 mM NaF, 25 mM β-glycerophosphate, and protease inhibitors). Biotinylated proteins were precipitated with neutravidin-agarose beads (Thermo Fisher Scientific) and analyzed by immunoblotting using HA antibody. The levels of HA-TβRI and Flag-CIN85 expression were determined by immunoblotting of total cell lysates. To study the effect of CIN85 overexpression on the TβRI recycling, we used biotinylation recycling assay according to Arancibia-Carcamo et al. (2006). Transfected COS7 cells were labeled with sulfo-NHS-SS-biotin at 4°C, as in the previous paragraph. Upon completion of biotin labeling, the culture media were changed and cells were incubated at 37°C for 60 min to resume membrane trafficking of the receptor. After the internalization step, all remaining surface sulfo-NHS-SS-biotin was stripped by extensive washing in 50 mM glutathione dissolved in 75 mM NaCl and 10 mM EDTA, pH 7.5, at 4°C. The cells were then incubated at 37°C for an additional 30 min in DMEM with 50 mM glutathione, pH 7.5. Control dishes were kept all the time at 4°C to prevent internalization and to monitor the efficacy of biotin stripping. The biotin of any sulfo-NHS-SS-biotinylated receptors that were recycled back to the cell surface was cleaved off by the glutathione present in the medium. At the end of incubation, biotinylated proteins were purified using streptavidin affinity chromatography and analyzed by immunoblotting. Immunoblots were analyzed by Quantity One software. In the lysates of the cells incubated with glutathione, all biotinylated HA-TβRI was internalized but not recycled during incubation, thus, inversely correlating with the rate of receptor recycling. The loss of biotinylated internalized receptors after the second biotin cleavage provides a measure of receptor recycling. Subcellular fractionation Subcellular fractionation was performed as described previously (Zhu et al., 2012). In short, the cells were washed with ice-cold PBS and homogenized in buffer (20 mM Hepes, pH 7.4, 1 mM EDTA, and 250 mM sucrose with protease and phosphatase inhibitors) 20 times with a 22-gauge needle and 5 times with a 26-gauge needle. The nuclei were pelleted at 3,000 g for 10 min at 4°C. After measurement of protein concentration, a 10–20–30% discontinuous gradient was generated by mixing postnuclear supernatant 1:1 with 60% iodixanol (OptiPrep; Sigma-Aldrich) and layered under 1.2 ml of 20% iodixanol with 1.2 ml of 10% iodixanol and 0.4 ml of homogenization buffer at the top of the centrifuge tubes. The samples were centrifuged at 260,000 g for 3 h at 4°C. Fractions (200–400 µl) were collected from the top of the gradient and equal volumes of each fraction were separated by electrophoresis in polyacrylamide gels and analyzed by immunoblotting for the presence of the endosomal markers (EEA1, Rab7, and Rab11) and HA-TβRI. Lipid rafts were isolated using sucrose density centrifugation, as described by Ostrom and Insel (2006). In brief, HEK293T cells transfected with HA-TβRI alone or with Flag-CIN85 were grown to near confluence in 100-mm dishes. After two washes with ice-cold PBS, two confluent dishes were scraped into 1 ml of lysis buffer (1% Triton X-100, 25 mM 2-(N-morpholino) ethanesulfonic acid, 150 mM NaCl, and protease inhibitors, pH 6.0). Cells were homogenized by forcing lysates 10 times through a 25-gauge needle. The homogenates were adjusted to 45% sucrose, placed at the bottom of an ultracentrifuge tube, and overlaid with 35% and 5% sucrose. After centrifugation at 41,000 rpm for 24 h in a rotor (SW50.1; Beckman Coulter), 0.5-ml fractions were collected from the top of the tubes and a portion of each fraction was analyzed by SDS-PAGE. Immunofluorescence assays For all immunofluorescence studies, cells were seeded on sterile coverslips in 6-well plates, transfected, and treated, as indicated in figure legends. The coverslips were washed with PBS, fixed for 10 min in 4% paraformaldehyde, and permeabilized with 0.2% Triton X-100 in PBS for 5 min. Cells were blocked by 5% normal goat serum and incubated with primary antibody in the antibody dilution buffer (1% BSA and 0.3% Triton X-100 in PBS). The proteins of interest were visualized by subsequent binding with Alexa Fluor–conjugated antibodies in the antibody dilution buffer. Control samples were labeled with the individual fluorophores and exposed identically as the dual-labeled samples at each wavelength to verify that there was no crossover among emission channels. After washing in PBS, coverslips were mounted with Fluoromount-G (SouthernBiotech) mounting medium. Standard fluorescence microscopy digital images were acquired at room temperature using an oil immersion 60× objective (numerical aperture 1.40) on a fluorescence microscope (Axio Imager.M2; Carl Zeiss) equipped with a Retiga EXi camera (QImaging) and controlled by ZEN software (Carl Zeiss). Acquired images were processed and analyzed using ImageJ software (National Institutes of Health). Histology of human prostate cancer tissues Tissue sections of normal and malignant tissues were provided by Umeå University Hospital and ethical permit was granted by the Umeå ethical review board in full agreement with the Swedish Ethical Review Act (540/03, Dnr 03-482). For immunohistochemistry staining the sections were rehydrated in xylene two times each for 10 min, 100% ethanol for 10 min, and 95%, 80%, and 70% ethanol and deionized H2O each for 5 min. Then the sections were treated with Antigen Retrieval Reagent (R&D systems) at 95°C for 5 min, rinsed with PBS, kept in 3% H2O2/100% methanol for 10 min, washed twice with PBS, blocked in 5% normal goat serum for 1 h at room temperature, and incubated with primary antibodies CIN85 (1:100; Sigma-Aldrich) overnight at 4°C. After washing with PBS three times each for 5 min, the sections were incubated with secondary antibody (DAKO) for 45 min at room temperature followed by three washes in PBS. Then, the sections were developed with AEC (Vector Laboratories), counterstained with hematoxylin, and mounted in aqueous mounting medium (Vector Laboratories). For immunohistofluorescence staining, the sections were deparaffinized and retrieved in the same way as described in the previous paragraph. After blocking in 1% horse serum in PBS for 1 h at room temperature, the sections were incubated with primary antibodies (CIN85 [Sigma-Aldrich] diluted 1:50 and phospho-Smad2 diluted 1:50) overnight at 4°C. After washing in PBS three times each for 5 min, the sections were incubated with NorthernLights secondary antibody (R&D Systems) for 45 min at room temperature. Finally, the sections were washed with PBS three times each for 5 min, and thereafter mounted with mounting medium with DAPI (Vector Laboratories). Digital images of stainings were acquired by scanning with Pannoramic 250 Flash II (3DHistech) or a fluorescence microscope (Axioplan 2; Carl Zeiss, and analyzed by ImageJ software. Grayscale values (pixel intensities) were measured and calculated to compare the intensity of different stainings. Means ± SEM were obtained from different groups, and Students’ t tests were used to evaluate the differences between the groups compared. Online supplemental material Fig. S1 provides evidence that interaction between TβRI and CIN85 is not altered by TGFβ receptors kinase inhibitors, dynamin inhibitor, or low temperature. Fig. S2 shows that HA-TβRI interacts with the SH3 domains of CIN85 in a GST pull-down assay and that in vitro ubiquitination of GST-TβRI increases the amount of CIN85 precipitated in a GST pull-down assay; overexpression of TRAF6 has no effect on the interaction between TβRII and CIN85. Fig. S3 shows that overexpression of CIN85 alters TβRI distribution in an iodixanol density gradient membrane flotation assay, but does not change the localization of TβRI to lipid rafts. Fig. S4 provides evidence that overexpression of CIN85 increases the recycling rate of HA-TβRI in COS7 cells. Fig. S5 demonstrates that CIN85 expression correlates with higher level of phosphorylated Smad2 in prostate cancer tissue. Online supplemental material is available at http://www.jcb.org/cgi/content/full/jcb.201411025/DC1. Supplementary Material Supplemental Material Acknowledgments We thank Pernilla Andersson for skillful technical assistance with human prostate tissue samples, Dr. Wanzhong Wang for expert histopathological evaluation of prostate cancer tissues, and Dr. Nikolay Aksenov for selection of CIN85 primers used in qRT-PCR. This work was supported by the Ludwig Institute for Cancer Research and grants to M. Landström from the Knut and Alice Wallenberg Foundation (2012.0090), Swedish Medical Research Council (K2013-66X-15284-04-4), and the Swedish Cancer Society (13 0688). The authors declare no competing financial interests. Abbreviations used in this paper: EGFR EGF receptor HEK human embryonic kidney qRT-PCR quantitative RT-PCR TβRI TGFβ receptor type I TRAF6 tumor necrosis factor receptor-associated factor 6 ==== Refs Aissouni, Y., G. Zapart, J.L. Iovanna, I. Dikic, and P. Soubeyran. 2005. CIN85 regulates the ability of MEKK4 to activate the p38 MAP kinase pathway. Biochem. Biophys. Res. Commun. 338 :808–814. 10.1016/j.bbrc.2005.10.032 16256071 Arancibia-Carcamo, I.L., B.P. Fairfax, S.J. Moss, and J.T. Kittler. 2006. Studying the localization, surface stability and endocytosis of neurotransmitter receptors by antibody labeling and biotinylation approaches. In The Dynamic Synapse: Molecular Methods in Ionotropic Receptor Biology. J.T. Kittler and S.J. Moss, editors. CRC Press, Boca Raton, FL. 91–118. Bezsonova, I., M.C. Bruce, S. Wiesner, H. Lin, D. Rotin, and J.D. Forman-Kay. 2008. Interactions between the three CIN85 SH3 domains and ubiquitin: implications for CIN85 ubiquitination. Biochemistry. 47 :8937–8949. 10.1021/bi800439t 18680311 Chen, Y.G. 2009. Endocytic regulation of TGF-β signaling. Cell Res. 19 :58–70. 10.1038/cr.2008.315 19050695 Chen, X., M.J. Rubock, and M. Whitman. 1996. A transcriptional partner for MAD proteins in TGF-β signalling. Nature. 383 :691–696. 10.1038/383691a0 8878477 Dennler, S., S. Itoh, D. Vivien, P. ten Dijke, S. Huet, and J.-M. Gauthier. 1998. Direct binding of Smad3 and Smad4 to critical TGFβ-inducible elements in the promoter of human plasminogen activator inhibitor-type 1 gene. EMBO J. 17 :3091–3100. 10.1093/emboj/17.11.3091 9606191 Di Guglielmo, G.M., C. Le Roy, A.F. Goodfellow, and J.L. Wrana. 2003. Distinct endocytic pathways regulate TGF-β receptor signalling and turnover. Nat. Cell Biol. 5 :410–421. 10.1038/ncb975 12717440 Dikic, I. 2002. CIN85/CMS family of adaptor molecules. FEBS Lett. 529 :110–115. 10.1016/S0014-5793(02)03188-5 12354621 Franzén, P., H. Ichijo, and K. Miyazono. 1993. Different signals mediate transforming growth factor-β1-induced growth inhibition and extracellular matrix production in prostatic carcinoma cells. Exp. Cell Res. 207 :1–7. 10.1006/excr.1993.1156 7686495 Gebäck, T., M.M. Schulz, P. Koumoutsakos, and M. Detmar. 2009. TScratch: a novel and simple software tool for automated analysis of monolayer wound healing assays. Biotechniques. 46 :265–274.19450233 Gleason, R.J., A.M. Akintobi, B.D. Grant, and R.W. Padgett. 2014. BMP signaling requires retromer-dependent recycling of the type I receptor. Proc. Natl. Acad. Sci. USA. 111 :2578–2583. 10.1073/pnas.1319947111 24550286 Gout, I., G. Middleton, J. Adu, N.N. Ninkina, L.B. Drobot, V. Filonenko, G. Matsuka, A.M. Davies, M. Waterfield, and V.L. Buchman. 2000. Negative regulation of PI 3-kinase by Ruk, a novel adaptor protein. EMBO J. 19 :4015–4025. 10.1093/emboj/19.15.4015 10921882 Havrylov, S., M.J. Redowicz, and V.L. Buchman. 2010. Emerging roles of Ruk/CIN85 in vesicle-mediated transport, adhesion, migration and malignancy. Traffic. 11 :721–731. 10.1111/j.1600-0854.2010.01061.x 20331533 Heldin, C.-H., and A. Moustakas. 2012. Role of Smads in TGFβ signaling. Cell Tissue Res. 347 :21–36. 10.1007/s00441-011-1190-x 21643690 Kardassis, D., C. Murphy, T. Fotsis, A. Moustakas, and C. Stournaras. 2009. Control of transforming growth factor β signal transduction by small GTPases. FEBS J. 276 :2947–2965. 10.1111/j.1742-4658.2009.07031.x 19490100 Karlsson, S., K. Kowanetz, A. Sandin, C. Persson, A. Ostman, C.H. Heldin, and C. Hellberg. 2006. Loss of T-cell protein tyrosine phosphatase induces recycling of the platelet-derived growth factor (PDGF) β-receptor but not the PDGF α-receptor. Mol. Biol. Cell. 17 :4846–4855. 10.1091/mbc.E06-04-0306 16971512 Kim, J., D. Kang, B.K. Sun, J.-H. Kim, and J.J. Song. 2013. TRAIL/MEKK4/p38/HSP27/Akt survival network is biphasically modulated by the Src/CIN85/c-Cbl complex. Cell. Signal. 25 :372–379. 10.1016/j.cellsig.2012.10.010 23085457 Kometani, K., T. Yamada, Y. Sasaki, T. Yokosuka, T. Saito, K. Rajewsky, M. Ishiai, M. Hikida, and T. Kurosaki. 2011. CIN85 drives B cell responses by linking BCR signals to the canonical NF-κB pathway. J. Exp. Med. 208 :1447–1457. 10.1084/jem.20102665 21708930 Kowanetz, K., J. Terzic, and I. Dikic. 2003. Dab2 links CIN85 with clathrin-mediated receptor internalization. FEBS Lett. 554 :81–87. 10.1016/S0014-5793(03)01111-6 14596919 Kowanetz, K., K. Husnjak, D. Höller, M. Kowanetz, P. Soubeyran, D. Hirsch, M.H.H. Schmidt, K. Pavelic, P. De Camilli, P.A. Randazzo, and I. Dikic. 2004. CIN85 associates with multiple effectors controlling intracellular trafficking of epidermal growth factor receptors. Mol. Biol. Cell. 15 :3155–3166. 10.1091/mbc.E03-09-0683 15090612 Lin, H.-K., S. Bergmann, and P.P. Pandolfi. 2004. Cytoplasmic PML function in TGF-β signalling. Nature. 431 :205–211. 10.1038/nature02783 15356634 Lynch, D.K., S.C. Winata, R.J. Lyons, W.E. Hughes, G.M. Lehrbach, V. Wasinger, G. Corthals, S. Cordwell, and R.J. Daly. 2003. A Cortactin-CD2-associated protein (CD2AP) complex provides a novel link between epidermal growth factor receptor endocytosis and the actin cytoskeleton. J. Biol. Chem. 278 :21805–21813. 10.1074/jbc.M211407200 12672817 Massagué, J. 2008. TGFβ in cancer. Cell. 134 :215–230. 10.1016/j.cell.2008.07.001 18662538 Mishra, L., and B. Marshall. 2006. Adaptor proteins and ubiquinators in TGF-β signaling. Cytokine Growth Factor Rev. 17 :75–87. 10.1016/j.cytogfr.2005.09.001 16359909 Mitchell, H., A. Choudhury, R.E. Pagano, and E.B. Leof. 2004. Ligand-dependent and -independent transforming growth factor-β receptor recycling regulated by clathrin-mediated endocytosis and Rab11. Mol. Biol. Cell. 15 :4166–4178. 10.1091/mbc.E04-03-0245 15229286 Mu, Y., R. Sundar, N. Thakur, M. Ekman, S.K. Gudey, M. Yakymovych, A. Hermansson, H. Dimitriou, M.T. Bengoechea-Alonso, J. Ericsson, 2011. TRAF6 ubiquitinates TGFβ type I receptor to promote its cleavage and nuclear translocation in cancer. Nat. Commun. 2 :330. 10.1038/ncomms1332 21629263 Mu, Y., S.K. Gudey, and M. Landström. 2012. Non-Smad signaling pathways. Cell Tissue Res. 347 :11–20. 10.1007/s00441-011-1201-y 21701805 Narita, T., T. Nishimura, K. Yoshizaki, and T. Taniyama. 2005. CIN85 associates with TNF receptor 1 via Src and modulates TNF-α-induced apoptosis. Exp. Cell Res. 304 :256–264. 10.1016/j.yexcr.2004.11.005 15707590 Ostrom, R.S., and P.A. Insel. 2006. Methods for the study of signaling molecules in membrane lipid rafts and caveolae. Methods Mol. Biol. 332 :181–191.16878693 Penheiter, S.G., R.D. Singh, C.E. Repellin, M.C. Wilkes, M. Edens, P.H. Howe, R.E. Pagano, and E.B. Leof. 2010. Type II transforming growth factor-β receptor recycling is dependent upon the clathrin adaptor protein Dab2. Mol. Biol. Cell. 21 :4009–4019. 10.1091/mbc.E09-12-1019 20881059 Persson, U., H. Izumi, S. Souchelnytskyi, S. Itoh, S. Grimsby, U. Engström, C.H. Heldin, K. Funa, and P. ten Dijke. 1998. The L45 loop in type I receptors for TGF-β family members is a critical determinant in specifying Smad isoform activation. FEBS Lett. 434 :83–87. 10.1016/S0014-5793(98)00954-5 9738456 Peruzzi, G., R. Molfetta, F. Gasparrini, L. Vian, S. Morrone, M. Piccoli, L. Frati, A. Santoni, and R. Paolini. 2007. The adaptor molecule CIN85 regulates Syk tyrosine kinase level by activating the ubiquitin-proteasome degradation pathway. J. Immunol. 179 :2089–2096. 10.4049/jimmunol.179.4.2089 17675467 Petrelli, A., G.F. Gilestro, S. Lanzardo, P.M. Comoglio, N. Migone, and S. Giordano. 2002. The endophilin–CIN85–Cbl complex mediates ligand-dependent downregulation of c-Met. Nature. 416 :187–190. 10.1038/416187a 11894096 Rønning, S.B., N.M. Pedersen, I.H. Madshus, and E. Stang. 2011. CIN85 regulates ubiquitination and degradative endosomal sorting of the EGF receptor. Exp. Cell Res. 317 :1804–1816. 10.1016/j.yexcr.2011.05.016 21635887 Schroeder, B., S.G. Weller, J. Chen, D. Billadeau, and M.A. McNiven. 2010. A Dyn2-CIN85 complex mediates degradative traffic of the EGFR by regulation of late endosomal budding. EMBO J. 29 :3039–3053. 10.1038/emboj.2010.190 20711168 Schroeder, B., S. Srivatsan, A. Shaw, D. Billadeau, and M.A. McNiven. 2012. CIN85 phosphorylation is essential for EGFR ubiquitination and sorting into multivesicular bodies. Mol. Biol. Cell. 23 :3602–3611. 10.1091/mbc.E11-08-0666 22833562 Shimokawa, N., K. Haglund, S.M. Hölter, C. Grabbe, V. Kirkin, N. Koibuchi, C. Schultz, J. Rozman, D. Hoeller, C.-H. Qiu, 2010. CIN85 regulates dopamine receptor endocytosis and governs behaviour in mice. EMBO J. 29 :2421–2432. 10.1038/emboj.2010.120 20551902 Sorrentino, A., N. Thakur, S. Grimsby, A. Marcusson, V. von Bulow, N. Schuster, S. Zhang, C.-H. Heldin, and M. Landström. 2008. The type I TGF-β receptor engages TRAF6 to activate TAK1 in a receptor kinase-independent manner. Nat. Cell Biol. 10 :1199–1207. 10.1038/ncb1780 18758450 Soubeyran, P., K. Kowanetz, I. Szymkiewicz, W.Y. Langdon, and I. Dikic. 2002. Cbl–CIN85–endophilin complex mediates ligand-induced downregulation of EGF receptors. Nature. 416 :183–187. 10.1038/416183a 11894095 Sundar, R., S.K. Gudey, C.H. Heldin, and M. Landström. 2015. TRAF6 promotes TGFβ-induced invasion and cell-cycle regulation via Lys63-linked polyubiquitination of Lys178 in TGFβ type I receptor. Cell Cycle. 14 :554–565. 10.4161/15384101.2014.990302 25622187 Szymkiewicz, I., K. Kowanetz, P. Soubeyran, A. Dinarina, S. Lipkowitz, and I. Dikic. 2002. CIN85 participates in Cbl-b-mediated down-regulation of receptor tyrosine kinases. J. Biol. Chem. 277 :39666–39672. 10.1074/jbc.M205535200 12177062 Szymkiewicz, I., O. Shupliakov, and I. Dikic. 2004. Cargo- and compartment-selective endocytic scaffold proteins. Biochem. J. 383 :1–11. 10.1042/BJ20040913 15219178 Thompson, T.C., L.D. Truong, T.L. Timme, D. Kadmon, B.K. McCune, K.C. Flanders, P.T. Scardino, and S.H. Park. 1992. Transforming growth factor β1 as a biomarker for prostate cancer. J. Cell. Biochem. Suppl. 16H :54–61. 10.1002/jcb.240501212 1289674 Watanabe, S., H. Take, K. Takeda, Z.X. Yu, N. Iwata, and S. Kajigaya. 2000. Characterization of the CIN85 adaptor protein and identification of components involved in CIN85 complexes. Biochem. Biophys. Res. Commun. 278 :167–174. 10.1006/bbrc.2000.3760 11071869 Xu, P., J. Liu, and R. Derynck. 2012. Post-translational regulation of TGF-β receptor and Smad signaling. FEBS Lett. 586 :1871–1884. 10.1016/j.febslet.2012.05.010 22617150 Yakymovych, I., P. Ten Dijke, C.-H. Heldin, and S. Souchelnytskyi. 2001. Regulation of Smad signaling by protein kinase C. FASEB J. 15 :553–555.11259364 Yamashita, M., K. Fatyol, C. Jin, X. Wang, Z. Liu, and Y.E. Zhang. 2008. TRAF6 mediates Smad-independent activation of JNK and p38 by TGF-β. Mol. Cell. 31 :918–924. 10.1016/j.molcel.2008.09.002 18922473 Zhang, J., X. Zheng, X. Yang, and K. Liao. 2009. CIN85 associates with endosomal membrane and binds phosphatidic acid. Cell Res. 19 :733–746. 10.1038/cr.2009.51 19417776 Zhu, L., L. Wang, X. Luo, Y. Zhang, Q. Ding, X. Jiang, X. Wang, Y. Pan, and Y. Chen. 2012. Tollip, an intracellular trafficking protein, is a novel modulator of the transforming growth factor-β signaling pathway. J. Biol. Chem. 287 :39653–39663. 10.1074/jbc.M112.388009 23027871

PMC004xxxxxx/PMC4689169.txt

==== Front J Exp Med J Exp Med jem jem The Journal of Experimental Medicine 0022-1007 1540-9538 The Rockefeller University Press 26621454 20151093 10.1084/jem.20151093 Research Articles Brief Definitive Report Melanocyte antigen triggers autoimmunity in human psoriasis Melanocyte-directed autoimmunity in psoriasis Arakawa Akiko 1* Siewert Katherina 2* Stöhr Julia 1* Besgen Petra 1 Kim Song-Min 1 Rühl Geraldine 2 Nickel Jens 1 Vollmer Sigrid 1 Thomas Peter 1 Krebs Stefan 3 Pinkert Stefan 5 Spannagl Michael 4 Held Kathrin 2 Kammerbauer Claudia 1 Besch Robert 1 Dornmair Klaus 2 Prinz Jörg C. 1 1 Department of Dermatology, Ludwig-Maximilian-University, D-80337 Munich, Germany 2 Institute of Clinical Neuroimmunology, Ludwig-Maximilian-University, D-82152 Planegg-Martinsried, Germany 3 Gene Center Munich, Ludwig-Maximilian-University, D-81377 Munich, Germany 4 Laboratory of Immunogenetics and Molecular Diagnostics, Ludwig-Maximilian-University, D-81377 Munich, Germany 5 German Cancer Research Center, D-69120 Heidelberg, Germany Correspondence to Jörg C. Prinz: joerg.prinz@med.uni-muenchen.de * A. Arakawa, K. Siewert, and J. Stöhr contributed equally to this paper and are listed in alphabetic order. 14 12 2015 212 13 22032212 02 7 2015 04 11 2015 © 2015 Arakawa et al. 2015 https://creativecommons.org/licenses/by-nc-sa/3.0/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 3.0 Unported license, as described at http://creativecommons.org/licenses/by-nc-sa/3.0/). Arakawa et al. discovered that the autoimmune response in psoriasis is directed against melanocytes. They show that the main psoriasis risk allele HLA-C*06:02 mediates melanocyte-specific autoimmunity and identify ADAMTSL5 as a melanocyte autoantigen, which stimulates IL-17 and IFN-γ production in CD8+ T cells. Psoriasis vulgaris is a common T cell–mediated inflammatory skin disease with a suspected autoimmune pathogenesis. The human leukocyte antigen (HLA) class I allele, HLA-C*06:02, is the main psoriasis risk gene. Epidermal CD8+ T cells are essential for psoriasis development. Functional implications of HLA-C*06:02 and mechanisms of lesional T cell activation in psoriasis, however, remained elusive. Here we identify melanocytes as skin-specific target cells of an HLA-C*06:02–restricted psoriatic T cell response. We found that a Vα3S1/Vβ13S1 T cell receptor (TCR), which we had reconstituted from an epidermal CD8+ T cell clone of an HLA-C*06:02–positive psoriasis patient specifically recognizes HLA-C*06:02–positive melanocytes. Through peptide library screening, we identified ADAMTS-like protein 5 (ADAMTSL5) as an HLA-C*06:02–presented melanocytic autoantigen of the Vα3S1/Vβ13S1 TCR. Consistent with the Vα3S1/Vβ13S1-TCR reactivity, we observed numerous CD8+ T cells in psoriasis lesions attacking melanocytes, the only epidermal cells expressing ADAMTSL5. Furthermore, ADAMTSL5 stimulation induced the psoriasis signature cytokine, IL-17A, in CD8+ T cells from psoriasis patients only, supporting a role as psoriatic autoantigen. This unbiased analysis of a TCR obtained directly from tissue-infiltrating CD8+ T cells reveals that in psoriasis HLA-C*06:02 directs an autoimmune response against melanocytes through autoantigen presentation. We propose that HLA-C*06:02 may predispose to psoriasis via this newly identified autoimmune pathway. ==== Body pmcPsoriasis vulgaris (OMIM no. MIM177900) is among the most frequent T cell–mediated disorders, affecting 120–180 million people worldwide by a chronically relapsing hyperproliferative skin inflammation (Griffiths and Barker, 2007; Lowes et al., 2007). Within a complex genetic predisposition, HLA-C*06:02 on psoriasis susceptibility locus PSORS1 (6p21.33) is the main psoriasis risk allele (Nair et al., 2006). HLA-C*06:02 is present in more than 60% of patients, increases the risk for psoriasis 9- to 23-fold, and decides an earlier onset and more severe disease course (Gudjonsson et al., 2003). A direct contribution of HLA-C*06:02 to psoriasis manifestation, however, could not be determined as the result of a strong linkage disequilibrium within the PSORS1 locus (Lowes et al., 2007) and a lack of experimental systems for analyzing its function in psoriasis. HLA class I molecules present peptide antigens to CD8+ T cells. Novel psoriasis lesions develop upon epidermal influx (Conrad et al., 2007) and clonal expansion of CD8+ T cells, indicating persistent CD8+ T cell recruitment and activation by locally presented autoantigens (Chang et al., 1994; Kim et al., 2012). Potential psoriatic autoantigens have been proposed by us and others mainly based on the hypothesis that the lesional CD8+ T cells react against keratinocytes (Valdimarsson et al., 2009; Besgen et al., 2010; Lande et al., 2014). Nevertheless, the target cells and antigens that drive pathogenic CD8+ T cell responses in psoriasis lesions are still unproven. Accordingly, an autoimmune pathogenesis of psoriasis remained hypothetical to date. We formerly established an unbiased technique to characterize αβ TCRs of single T cells (Kim et al., 2012). By this method, we identified dominant CD8+ T cell clones in psoriasis lesions and determined the molecular structure of their paired TCR α- and β-chain rearrangements. Clonal T cell expansions in autoimmune lesions result from a T cell response to locally presented autoantigens (Kent et al., 2005). Epidermal psoriatic CD8+ T cells preferentially rearrange TCR Vβ13S1 (Chang et al., 1994). Here, we reconstitute a Vα3S1/Vβ13S1 TCR from an epidermal CD8+ T cell clone isolated from a psoriasis lesion of an HLA-C*06:02–positive patient in a T hybridoma cell line. Along with human CD8αβ and NFAT-sGFP transfection, this TCR hybridoma reports on TCR signaling by robust sGFP expression (Seitz et al., 2006; Siewert et al., 2012). Assuming that the Vα3S1/Vβ13S1-TCR hybridoma carries the antigen specificity of pathogenic psoriatic CD8+ T cells, we used it to explore the mechanisms of lesional psoriatic T cell activation. RESULTS AND DISCUSSION Melanocytes are HLA-C*06:02–restricted autoimmune target cells of the Vα3S1/Vβ13S1 TCR We first analyzed the reactivity of the Vα3S1/Vβ13S1 TCR in co-culture experiments with various skin cell types in association with HLA-C*06:02. We observed that primary melanocytes from both HLA-C*06:02–positive psoriasis patients and healthy donors, but not HLA-C*06:02–negative psoriasis patients or healthy individuals, activated the Vα3S1/Vβ13S1-TCR hybridoma (Fig. 1, A and B). Hybridoma activation was enhanced by preincubation of melanocytes with IFN-γ to increase the otherwise low HLA-C surface expression (McCutcheon et al., 1995) and inhibited by an HLA class I–blocking antibody (Fig. 1, B and C). To specify the role of HLA-C*06:02 in Vα3S1/Vβ13S1-TCR ligation, we co-cultured the TCR hybridoma with two inherently HLA-C*06:02–positive melanoma cell lines, WM278 (Fig. 1 D) and 1205Lu (not depicted) as melanocyte surrogates. Both of them activated the TCR hybridoma when preincubated with IFN-γ to induce HLA-C (Fig. 1 C). Two HLA-C*06:02–negative melanoma cell lines, WM9 (Fig. 1 E) and WM1232 (not depicted), activated the Vα3S1/Vβ13S1-TCR hybridoma only upon transfection with HLA-C*06:02, but not HLA-A*02:01. This effect was independent from IFN-γ and suppressed by HLA class I blockade. Figure 1. HLA-C*06:02-positive melanocytes are skin-specific target cells of the Vα3S1/Vβ13S1 TCR. (A and B) TCR hybridoma (round cells) activation by HLA-C*06:02–positive (left) but not HLA-C*06:02–negative (right) primary human melanocytes (elongated) in the presence of IFN-γ, as determined by combined UV-light microscopy (A) or flow cytometry and effect of melanocyte preincubation with IFN-γ or an HLA class I–blocking antibody (B). Data represent three HLA-C*06:02–positive (healthy n = 2, psoriasis n = 1) and nine HLA-C*06:02–negative subjects (healthy n = 7, psoriasis n = 2) as mean ± SD. Numbers give stimulating/total samples. *, P < 0.05; ***, P < 0.005; two tailed Fisher’s exact. Bars, 20 µm. (C) Western blot analysis of HLA-C expression in primary melanocytes or WM278 cells cultured without/with IFN-γ or HLA-C*06:02–negative WM9 cells without/with HLA-C*06:02 transfection. (D and E) Hybridoma activation by an HLA-C*06:02–positive (WM278) or HLA-C*06:02–negative (WM9) melanoma cell line transfected with HLA-C*06:02 or HLA-A*02:01 and effect of a blocking HLA antibody. (F and G) Hybridoma activation by HLA-C*06:02–negative primary keratinocytes (F) or the HLA-C*06:02–negative epithelial carcinoma cell line A431 (G) transfected with HLA-C*06:02 or HLA-A*02:01 or cotransfected with HLA-C*06:02 and mimotope #6. Data represent three (C, E, and F), five (D), and two (G) independent experiments as mean ± SD. No hybridoma activation was seen with HLA-C*06:02–transfected primary human keratinocytes (Fig. 1 F), epithelial A431 cells (Fig. 1 G), the keratinocyte cell line, HaCaT, or HLA-C*06:02–positive primary skin fibroblasts (not depicted). This was surprising because former disease models had proposed keratinocytes as immune targets in psoriasis (Valdimarsson et al., 2009). Keratinocytes and epithelial cell lines, however, could present antigen and stimulate the TCR hybridoma upon cotransfection with HLA-C*06:02 and an appropriate Vα3S1/Vβ13S1-TCR ligand, mimotope #6 (Fig. 1, F and G) identified by peptide library screening (see Fig. 3, A and B). Together, these experiments identified melanocytes as HLA-C*06:02–restricted autoimmune target cells of the Vα3S1/Vβ13S1 TCR. CD8+ T cells in lesional psoriatic epidermis target melanocytes To determine whether the reactivity of the Vα3S1/Vβ13S1 TCR reflected a common pathway of psoriatic T cell activation, we assessed the epidermal distribution of CD8+ T cells in chronic psoriatic plaques (n = 23), healthy skin (n = 5), and nickel contact eczema (n = 5), another T cell–mediated inflammatory skin disease induced by a defined antigen. Tissue sections were immunostained for CD8 and the melanocyte markers MART-1 or c-Kit (Fig. 2, A and B). Consistent with a previous study (Bovenschen et al., 2005), the numbers of CD8+ T cells were significantly higher in the epidermis of psoriasis lesions and nickel eczema than in normal skin (Fig. 2 C). Significantly more CD8+ T cells exhibited direct contacts with melanocytes in psoriasis (median 37.2%, SD 15.4) than in nickel eczema (median 6.9%, SD 4.92; Fig. 2 C). In psoriasis biopsies where HLA-C*06:02 typing was available, contact frequencies in HLA-C*06:02–positive (n = 7) and negative (n = 3) samples were in a similar range (Fig. 2 D). The psoriatic CD8+ T cells polarized lytic granules containing Granzyme B toward the melanocyte contact sites (Fig. 2, E and F), indicating TCR-mediated activation. Lytic granules constitute an important effector mechanism of CD8+ T cells, but their directed release does not necessarily induce apoptosis of target cells (Knickelbein et al., 2008). Indeed, an antibody against cleaved Caspase 3 (cCASP3), a crucial mediator of apoptosis (Porter and Jänicke, 1999) detected no signs of cell death in melanocytes in psoriatic epidermis (Fig. 2 G). Instead, melanocyte numbers in psoriasis lesions were reportedly increased (Wang et al., 2013). Thus, melanocytes are likely targets of a noncytotoxic CD8+ T cell–mediated autoimmune response in psoriasis. According to the selective epidermal localization of melanocytes, a melanocyte-directed autoimmune response may explain why psoriatic inflammation is primarily confined to the skin. Figure 2. Melanocytes are targets of a noncytotoxic CD8+ T cell response in psoriasis lesions. (A and B) Representative immunohistologic images of CD8+ T cells (green) and melanocytes (red) stained for c-Kit (A) or MART-1 (B) in healthy skin (n = 5), nickel contact eczema (n = 5), and psoriasis (n = 23). Asterisks designate CD8+ T cells contacting melanocytes. Overlay of red and green shows as yellow and of red, green, and blue as white. Dashed lines indicate basal membrane. (C) Incidence of CD8+ T cells (left) and percentage of CD8+ T cells in contact with c-Kit+ melanocytes (right) in normal skin, nickel eczema, and psoriasis. Each dot represents one subject, and bars mark the median; *, P < 0.05; ***, P < 0.005; ****, P < 0.001; two tailed Mann-Whitney U test. (D) Percentage of CD8+ T cells in contact with c-Kit+ melanocytes in three HLA-C*06:02–negative and seven HLA-C*06:02–positive psoriasis patients. Bars indicate median. (E and F) CD8+ T cells (blue) polarizing lytic granules (Granzyme B: green; overlay with red: yellow; overlay with red and blue: white; see arrowheads) toward contact sites with c-Kit+ (E) or MART-1+ (F) melanocytes (red) in psoriasis. Data are representative of eight psoriasis lesions. (G) Staining for MART-1 (red) and cCASP3 (green) in normal skin, nickel eczema, and psoriasis (n = 3). Only keratinocytes in nickel eczema focally stained positive for cCASP3 (arrowheads). Dashed lines indicate basal membrane. Bars: (A and G) 40 µm; (B) 20 µm; (E and F) 5 µm. HLA-C*06:02 mediates melanocyte-directed autoimmunity through autoantigen presentation To identify a potential melanocyte autoantigen of the Vα3S1/Vβ13S1 TCR, we screened plasmid-encoded combinatorial nonamer peptide libraries (PECPLs; total library size: 4.75 × 108) for peptides that are recognized by this TCR in the context of HLA-C*06:02. Screening with a completely randomized PECPL or PECPLs with predefined library positions according to known HLA-C*06:02 anchor positions (Fig. 3 A; Falk et al., 1993; Dionne et al., 2004; Rasmussen et al., 2014) and newly identified TCR ligands recovered eight TCR peptides stimulating the Vα3S1/Vβ13S1-TCR hybridoma (Fig. 3 B). The mimotopes were remarkably similar by distinct residue preferences and shared Arg at positions P2 and P8, Leu at P9, and Arg or Leu at P7. Six mimotopes carried Arg at P5, whereas P1, P3, P4, and P6 exhibited greater diversity. This polyspecificity is consistent with the inherent property of TCRs to react with distinct, though similar peptide ligands (Birnbaum et al., 2014). Figure 3. ADAMTSL5 is an HLA-C*06:02–presented melanocytic autoantigen of the Vα3S1/Vβ13S1 TCR. (A) Design of PECPLs #1–3. Predefined amino acid residues are in pink; X indicates randomized. (B and C) Mimotopes derived from PECPLs #1–3 (B) and natural human TCR peptide ligands stimulating the Vα3S1/Vβ13S1 TCR (C). HLA-C*06:02 anchor positions are labeled in yellow, and other conserved amino acids in green, blue, or pink. (D) TCR hybridoma stimulation by plasmid-encoded human peptide epitopes cotransfected with HLA-C*06:02 into COS-7 cells as normalized to CD3 stimulation. (E and F) Changes in TCR hybridoma activation by full-length proteins (E) or by wild-type, epitope-mutated (Ala58-65) or epitope-deleted (Del58-65) full-length ADAMTSL5 transfected into WM278 cells preincubated with IFN-γ as compared with vector control (F). Data represent three (D and E) and five (F) independent experiments as mean ± SD; *, P < 0.05; ***, P < 0.005; two tailed Mann-Whitney U test. (G) TCR hybridoma activation by WM278 cells transfected with two different control siRNAs or siRNAs targeting ADAMTSL5 and incubated with IFN-γ as compared with mock control. Data represent three biological triplicates as mean ± SD. (H) Validation of ADAMTSL5 knockdown by qPCR in triplicates and normalized to PBGD. Data are given as mean ± SD. Bioinformatic searches against the human proteome and the transcriptome of 1205Lu cells did not identify proteins containing sequences identical to the mimotopes. We therefore used the convergent amino acid pattern to select 180 natural human peptides for testing as candidate antigens of the Vα3S1/Vβ13S1 TCR (not depicted). Six of them stimulated the Vα3S1/Vβ13S1-TCR hybridoma when cotransfected with HLA-C*06:02 into COS-7 cells (Fig. 3, C and D). All of them corresponded to the conserved amino acid motif of the mimotopes and contained Arg at P5 and P8, Arg or Leu at P7, and the HLA-C*06:02–anchor residues Arg at P2 and Leu at P9 (Fig. 3 C). Thus, the conserved amino acid pattern of the library mimotopes allowed for the identification of several HLA-C*06:02–restricted natural self-epitopes of the Vα3S1/Vβ13S1 TCR. The antigenicity of full-length proteins differs from their peptide epitopes To distinguish potential psoriatic autoantigens, we investigated whether the parent proteins of the candidate epitopes would activate the Vα3S1/Vβ13S1-TCR hybridoma, i.e., whether the epitopes would be generated from the full-length parent proteins by proteasomal degradation. Unlike the antigenic peptides, none of the corresponding full-length proteins activated the TCR hybridoma when coexpressed with HLA-C*06:02 in COS-7 cells or the human cell line, HEK293FT, although Western blot confirmed ectopic protein expression (not depicted). When overexpressed in genuinely stimulatory HLA-C*06:02–positive WM278 (Fig. 3 E) or 1205Lu cells (not depicted), however, one of the six parent proteins, ADAMTSL5, enhanced Vα3S1/Vβ13S1-TCR hybridoma activation substantially beyond the basal activation induced by vector-transfected cells. Conversely, ADAMTSL5 protein lost its ability for TCR hybridoma stimulation upon deletion (Del58–65) or alanine substitution (Ala58–65) of the peptide epitope when expressed in both HLA-C*06:02–positive WM278 (Fig. 3 F) and 1205Lu cells or HLA-C*06:02–negative WM9 cells cotransfected with HLA-C*06:02 (not depicted). Knockdown of ADAMTSL5 in WM278 cells diminished TCR hybridoma activation (Fig. 3, G and H). Thus, only ADAMTSL5 retained the antigenicity of the peptide epitope as a full-length protein in HLA-C*06:02–positive or HLA-C*06:02–transfected melanocytic target cells, corroborating its role as psoriatic autoantigen of the Vα3S1/Vβ13S1 TCR in melanocytes. These results further indicate that the number of potential autoantigens is probably much smaller than the broad TCR reactivity against different self-peptide ligands would predict (Siewert et al., 2012; Birnbaum et al., 2014), and it supports that processing of particular autoantigenic epitopes may be cell type dependent (Kniepert and Groettrup, 2014). ADAMTSL5 is a melanocyte autoantigen in psoriasis To examine whether ADAMTSL5 might represent a public autoantigen of CD8+ T cells in psoriasis, we tested the in vitro response of CD8+ T cells to a synthetic ADAMTSL5 peptide. As determined by multiparametric flow cytometry analysis, ADAMTSL5 stimulation of PBMCs markedly increased the frequencies of CD8+ T cells expressing IL-17A and IFN-γ in psoriasis patients (n = 47), whereas healthy individuals (n = 11) did not respond (each P < 0.005; Fig. 4, A and B). Analysis of culture supernatants confirmed that significantly higher proportions of psoriasis patients produced IL-17A (15/42, P < 0.05) and IFN-γ (20/42, P < 0.005) in response to ADAMTSL5 peptide stimulation compared with healthy individuals (Fig. 4 C). A control peptide lacking homology to human proteins had no effects on IL-17A and IFN-γ production, excluding unspecific peptide effects. Overall, 61.9% of psoriasis patients responded to ADAMTSL5 stimulation in terms of IL-17A and/or IFN-γ production. Given that IL-17A and IFN-γ are key mediators of psoriasis pathogenesis (Zaba et al., 2007), these responses confirmed a role of ADAMTL5 as autoantigen in a substantial proportion of psoriasis patients. Figure 4. ADAMTSL5 is a psoriasis autoantigen. (A) Representative flow cytometry dot plots for IL-17A– and IFN-γ–producing CD8+ T cells as the proportion of total CD8+ T cells in a healthy and psoriasis subject. (B and D) Change in the percentage of IL-17A– and IFN-γ–positive CD8+ T cells after ADAMTSL5 peptide stimulation of PBMCs from healthy and psoriasis subjects over medium control (B) and differentiated according to HLA-C*06:02 status (D). Data were assessed by multiparametric flow cytometry analysis after CD8 and intracellular cytokine staining and compared by two-tailed Mann-Whitney U test. Each dot represents one subject; *, P < 0.05; ***, P < 0.005. (C and E) Change in IL-17A and IFN-γ secretion of PBMCs stimulated with ADAMTSL5 or control peptide over medium control as evaluated by ELISA (C) and differentiated according to HLA-C*06:02 status (E). Dashed lines indicate the positive threshold. *, P < 0.05; ***, P < 0.005; two tailed Fisher’s exact test. Approximately one third of psoriasis patients are HLA-C*06:02 negative (Table S1; Gudjonsson et al., 2003). PBMCs from both HLA-C*06:02–positive and –negative patients reacted similarly to ADAMTSL5 peptide stimulation (Fig. 4, D and E). A single autoantigenic epitope can be presented by different HLA molecules (Martin et al., 1991). The peptide-binding motifs of HLA-C*06:02 and several other HLA molecules overlap (Rasmussen et al., 2014). Similar CD8+ T cell reactivities against ADAMTSL5 and melanocytes in HLA-C*06:02–positive and –negative psoriasis patients support that HLA class I molecules other than HLA-C*06:02 may present ADAMTSL5 and promote melanocyte-specific autoimmunity as well. Thus, the specificity of the Vα3S1/Vβ13S1 TCR actually designates a common autoimmune pathway in psoriasis beyond HLA-C*06:02. We finally examined ADAMTSL5 expression in skin and melanocytes. Melanocytes and WM278 cells highly expressed ADAMTSL5 mRNA compared with whole skin (Fig. 5 A). By immunohistology, we observed that in both healthy and lesional psoriatic epidermis, only melanocytes expressed ADAMTSL5 (Fig. 5 B). In psoriasis lesions, CD8+ T cells were seen in direct contact with ADAMTSL5-positive melanocytes (Fig. 5 C). Figure 5. ADAMTSL5 is selectively expressed by melanocytes in epidermis. (A) ADAMTSL5 mRNA expression in healthy skin (n = 2), primary melanocytes (n = 3), and WM278 cells was assessed by qPCR in triplicates and normalized to PBGD. Data are given as mean ± SD. (B) Immunostaining of ADAMTSL5 (green; overlay with red: yellow) and MART-1 (red) in healthy skin (n = 3) and psoriasis (n = 3). Dashed lines indicate basal membrane, and arrowheads indicate ADAMTSL5 expression by melanocytes. (C) Representative image of a CD8+ T cell (blue) contacting a melanocyte (red) expressing ADAMTSL5 (green; overlay with red: yellow) in psoriasis (n = 2). Bars: (B, left) 20 µm; (B [right] and C) 10 µm. Overall, the autoreactivity of a TCR obtained directly from lesion-infiltrating CD8+ T cells reveals that HLA-C*06:02 may direct an autoimmune response against melanocytes in psoriasis. Identification of ADAMTSL5 as an HLA-C*06:02–restricted melanocytic autoantigen for the Vα3S1/Vβ13S1 TCR experimentally verifies that HLA-C*06:02 mediates melanocyte-specific autoimmunity through tissue-specific autoantigen presentation. Unveiling melanocytes, but not keratinocytes as previously hypothesized (Valdimarsson et al., 2009), as autoimmune target cells of the lesional psoriatic T cell response identifies a formerly unknown pathway of skin-specific psoriatic immune activation by which HLA-C*06:02 may predispose to psoriasis. The autoimmune response against melanocytes does not necessarily exclude other cellular targets. Previous hypothesis-driven studies reported T cell responses against various self-peptides from the cathelicidin LL37 (Lande et al., 2014) or other keratinocyte proteins in psoriasis (Valdimarsson et al., 2009; Besgen et al., 2010). Because single TCRs can recognize hundreds of different peptides (Siewert et al., 2012; Birnbaum et al., 2014), HLA class I–restricted peptide autoantigens should be validated in the context of the parent proteins within the target cells. Here, we identify melanocytes as targets of the Vα3S1/Vβ13S1 TCR in an unbiased fashion and finally differentiate ADAMTSL5 as HLA-C*06:02–restricted melanocyte autoantigen from various other natural TCR peptide ligands. Thus, our data provide conclusive evidence that psoriasis is an autoimmune disease. Beyond psoriasis, our approach allowed the unequivocal determination that HLA-based autoimmunity in humans arises through autoantigen presentation. Furthermore, this study provides experimental strategies for exploring T cell–mediated immune responses in autoimmunity, tumor immunity, or infections. MATERIALS AND METHODS Patients Lesional biopsies were obtained from patients with psoriasis or nickel eczema patch test reactions, and normal skin specimens were obtained from discarded healthy skin of donors undergoing plastic surgery. Psoriasis patients with chronic plaque psoriasis were included (Table S1). Healthy donors were defined as persons without a history of psoriasis or other inflammatory or autoimmune diseases (five females and six males, median age 41.7 yr). Psoriasis patients under immunosuppressive treatments such as methotrexate, ciclosporin, or fumaric acid esters were excluded. There are no statistically significant differences in age and sex between groups. Patients and healthy individuals participated voluntarily and gave written informed consent. The study was performed in accordance with the Helsinki Declaration and approved by the Ethics Committee of the Ludwig-Maximilian-University, Munich. Generation of Vα3S1/Vβ13S1-TCR CD8+ reporter T hybridoma Identification of the matching Vα3S1 and Vβ13S1-TCR chains of a CD8+ T cell clone (Vα3S1-NN-Jα 45.1: CA TDAL YSGG, Vβ13S1-N(D)N-Jβ 1.1: CASSY SEGED EAFF; Arden nomenclature [Arden et al., 1995]) has been described previously (Kim et al., 2012). TCR hybridoma generation was performed as described previously (Seitz et al., 2006; Siewert et al., 2012). In brief, Vα and Vβ regions were cloned into expression plasmids pRSV-hygro (α chain) and pRSV-neo (β chain) using restriction sites SalI–PvuII or SalI–BIpI. The resulting plasmids were linearized (XmnI and NdeI, respectively) and electroporated into 58 α−β− T hybridoma cells. Functional clones were supertransfected with pLPCX-CD8α-IRES-β and pcDNA-NFAT-sGFP. TCR activation–induced NFAT-sGFP expression was determined by CD3 cross-linking with anti–mouse CD3 antibody (clone 17A2; eBioscience), flow cytometry, and fluorescence microscopy (AxioVert200M [Carl Zeiss], 520/35 BrightLine filter, Semrock, and 605/70 filter). Hybridoma batches were frequently recloned to minimize sporadic sGFP expression and decrease of activation rates that occur with prolonged culture. Clones with the highest frequencies of responding cells (usually between 30% and 60% NFAT-sGFP–positive cells after CD3 stimulation) were expanded in T hybridoma medium (see “Primary cells and cell lines” section). The different proportions of activatable hybridoma cells in different batches contribute to a certain interexperimental quantitative variability in activation rates without affecting specificity of TCR hybridoma activation (Siewert et al., 2012). Construction of PECPLs The nonameric PECPLs #1–3 were prepared as described previously (Siewert et al., 2012). Library designs are shown in Fig. 3 A. Sequences for all oligonucleotide primers are listed in Table S2. Identification of Vα3S1/Vβ13S1-TCR mimotopes PECPL screening was performed as described previously (Siewert et al., 2012). COS-7 cells were cotransfected with PECPLs and HLA-C*06:02 and co-cultured with Vα3S1/Vβ13S1-TCR hybridoma cells. After 16 h, COS-7 cells in close contact to fluorescent hybridoma cells were isolated and library peptides were amplified by nested PCR. PCR products were cloned into pcDNA.3.1D/V5-His-TOPO and transformed into Escherichia coli. The mimotope-enriched library plasmids were cotransfected with pRSV-HLA-C*06:02 into COS-7 cells, and the mimotope-containing plasmid was isolated by subcloning and sequencing. Transcriptome analysis of 1205Lu cells Total RNA was prepared from 6 × 106 1205Lu cells using 1.5 ml TRIzol. RNA was used for library generation after assessment for purity by UV-VIS spectrometry (NanoDrop) and for integrity by Bioanalyzer 2100 (Agilent Technologies). RNA-seq libraries were prepared from 100 ng of total quality-controlled RNA with a strand-specific protocol (RNA-seq complete kit, NuGEN). In brief, RNA was reverse transcribed with a reduced set of hexamer primers, avoiding excessive representation of rRNA in the cDNA. Second strand cDNA synthesis was performed in the presence of dUTP. After ultrasonic cDNA fragmentation, end repair Illumina-compatible adapters were ligated. Adapters contained uracil in one strand, allowing complete digestion of the second strand–derived DNA. After strand selection, the libraries were amplified, assessed for correct insert size on the Agilent Technologies Bioanalyzer, and diluted to 10 nM. Barcoded libraries were mixed in equimolar amounts and sequenced on a total of three lanes on a Genome Analyser IIx (Illumina) in single-read mode with a read length of 84 bp. Raw reads were demultiplexed by sorting for and simultaneously trimming a leading 4-bp barcode sequence at the 5′-end of each read. Obtained 80-bp reads were mapped to the hg19 release of the human genome using spliced-read mapper Tophat and the UCSC gene annotation. Mapped reads overlapping with known genes were counted with HTseq count, available from the developers of DESeq. Identification of natural candidate peptide antigens based on mimotope sequences Two different search strategies were used to identify human candidate peptide epitopes based on the conserved amino acid motifs of the mimotopes. We used two tools from the European Molecular Biology Open Software Suite (EMBOSS); PROPHECY was used to create a frequency matrix from sequences of positively tested antigenic peptides. With this matrix and the PROFIT tool, the taxa-specific Homo sapiens [9606] UniProt database was scanned. The list of results was further refined with information on anchor amino acids of HLA-C*06:02 ligands (Falk et al., 1993; Dionne et al., 2004; SYFPEITHI database). The matrix was constantly adjusted according to newly isolated mimotopes or tested candidate antigen peptides. 1205Lu transcriptome data were used for a selective search in melanocytic proteins. The MOTIF Search web server was searched for human genes in the “KEGG Genes” dataset containing peptide motifs that were defined according to mimotope sequences and general HLA-C*06:02 anchor positions. The motifs used for the search were (encoded in PROSITE format): R-X(4,5)-[RL]-R-[LIVY], R-X(5)-R-[LFMIVY], R-X(4)-[RLFS]-R-[LFMIVY], [AFGIRSHM]-R-[ACHNPQSTWF]-[ASRVWYC]-[NRQT]-[STVYA]-[LRFS]-R-[LFM], R-[ACHNPQSTWF]-[ASRVWYC]-[NRQT]-[STVYA]-[LRFS]-R-[LFM], and R-X-X-[NRQT]-[STVYA]-[LRFS]-R-[LFMIVY]. Cloning of HLA-C*06:02, candidate peptides, full-length proteins, and mutated ADAMTSL5 HLA-C*06:02 was amplified by PCR from cDNA of an HLA-C*06:02–positive psoriasis patient and cloned into pRSV5-neo via EcoRI and XhoI restriction sites. pRSV–HLA-A*02:01 has been described previously (Siewert et al., 2012). For expression of short peptides, forward and reverse oligonucleotides were annealed and ligated into pcDNA3.1D/V5-His-TOPO using the Directional TOPO Expression kit (Invitrogen). Forward primers carried a 5′-CACCATG overhang and a stop codon at the 3′-end of the target sequence. Open reading frames of THEM6 (UniProt accession no. Q8WUY1), RASSF10 (A6NK89), and C2CD4B (A6NLJ0) were PCR-amplified from 1205Lu cDNA using Taq and Pwo polymerases (both from Roche). Because of limitations in protein size for overexpression, we restricted cloning of ASH1L (Q9NR48) to a partial protein fragment corresponding to amino acids 2074–2564, comprising the ASH1L catalytic domain and the BROMO domain. Clone IRCM10B06 (BioScience) was used as template for amplification of the Hepacam (Q14CZ8) ORF. An ADAMTSL5 (NCBI Protein accession no. NP_998769) template was provided by S.S. Apte (Lerner Research Institute, Cleveland Clinic, Cleveland, OH; Bader et al., 2012). C-terminal stop codons were omitted in all constructs yielding a vector-encoded C-terminal V5-His6-tag. PCR products were cloned into pcDNA3.1D/V5-His-TOPO and confirmed by sequencing. Internal reverse primers were designed to introduce deletion or alanine substitutions of ADAMTSL5 at amino acid positions 58–65. Mutations were introduced by PCR in combination with the ADAMTSL5 forward primer using Taq polymerase. After blunting, resulting DNA fragments were used as forward primers in a second PCR in combination with the ADAMTSL5 reverse primer. After a second blunting reaction, resulting DNA was cloned into pcDNA3.1D/V5-His-TOPO. Western blotting Cell lysates were separated by 10% SDS-PAGE and transferred onto nitrocellulose membranes (Schleicher & Schuell) by semi-dry electroblotting. Expression of V5-His6–tagged proteins was detected using an anti–V5-AP antibody. HLA-C expression was analyzed using an HLA-C antibody in combination with AP-conjugated anti–mouse IgG secondary antibody. Western blot for β-actin served as loading control. The Alkaline Phosphatase Conjugate Substrate kit (Bio-Rad Laboratories) was used for colorimetric detection. Primary cells and cell lines Primary human melanocytes were prepared as described previously (Hsu and Herlyn, 1996). Skin samples were incubated with 2.5 U/ml dispase (grade II, Boehringer Mannheim) overnight at 4°C. Epidermis was removed from dermis and incubated for 5 min at 37°C in HBSS without Ca2+ and Mg2+ containing 0.25% trypsin and 0.1% EDTA. Epidermal single-cell suspensions were expanded in melanocyte growth medium M2 (PromoCell). Primary cultured melanocytes were stained with p75NTR antibody and anti–mouse secondary antibody conjugated with Alexa Fluor 597 overnight. Human melanoma cell lines, WM9, 1205Lu (NCBI BioSample accession no. SAMN03471797), WM1232, and WM278 (SAMN03471796) were originally obtained from the Wistar Institute. They were cultured in 2% TU medium containing MCDB153, 20% Leibovitz’s L15, 5 µg/ml insulin, 2% FCS, and 1.68 mM CaCl2. Human neonatal epidermal keratinocytes (Invitrogen) were cultured in EpiLife medium supplemented with EpiLife Defined Growth Supplement (Thermo Fisher Scientific). COS-7, HEK293FT, HaCaT (NCBI EST accession no. LIBEST_003731) and A431 (LIBEST_000407) cells were cultured in DMEM supplemented with 100 U/ml penicillin and 100 µg/ml streptomycin, 1 mM sodium pyruvate, 1× MEM nonessential amino acids, 10% FCS (Biochrom), and 10 µg/ml ciprofloxacin. HEK293FT medium additionally contained 500 µg/ml G-418. Vα3S1/Vβ13S1-TCR hybridoma cells were cultured in RPMI 1640 containing 10% FCS, 100 U/ml penicillin, 100 µg/ml streptomycin, 1 mM sodium pyruvate, 1× MEM nonessential amino acids, and 10 µg/ml ciprofloxacin, supplemented with selection antibiotics G418 (1.5 mg/ml), hygromycin (300 µg/ml), puromycin (1 µg/ml), and blasticidin (3 µg/ml). For co-culture experiments, TCR hybridoma cells were pelleted and resuspended in hybridoma growth medium without selection antibiotics before addition to antigen-presenting cells. Vα3S1/Vβ13S1-TCR hybridoma activation assays sGFP induction in Vα3S1/Vβ13S1-TCR hybridoma cells was examined after 24-h co-culture with antigen-presenting cells by UV-fluorescence microscopy and/or flow cytometry. As controls, Vα3S1/Vβ13S1-TCR hybridoma cells were cultured in anti–mouse CD3 antibody–coated (clone 17A2, 2 µg/ml in PBS; eBioscience) culture plates. The percentage of NFAT-sGFP+ cells was assessed by flow cytometry and normalized to the CD3 activation value in the same experiment. pRSV-GFP or pcDNA-GFP served as transfection controls. To increase HLA-C surface expression, 100 ng/ml IFN-γ was added 24 h before co-cultivation with TCR hybridoma cells. Anti–human HLA class I antibody (clone W6/32, LEAF grade, 1.5 µg/ml; BioLegend) was used to block HLA class I–restricted TCR hybridoma activation. 1205Lu and WM278 were seeded in 24-well plates in TU 2% medium at densities of 2.5 × 104 or 5 × 104 cells/well, respectively. HLA-C*06:02–negative cell line WM9 or WM1232 (105 cells/well) was transfected with 250 ng pRSV–HLA-C*06:02 or pRSV–HLA-A*02:01 using FuGENE HD reagent (Promega) according to the manufacturer’s instructions. Plasmid-encoded peptides were cotransfected with pRSV–HLA-C*06:02 (250 ng each) into COS-7 or HEK293FT using FuGENE. After 24 h, medium was replaced with fresh medium containing Vα3S1/Vβ13S1-TCR hybridoma cells. For stimulation of Vα3S1/Vβ13S1-TCR hybridoma cells by ectopic expression of full-length proteins, antigen-presenting cell lines were transfected with 250 ng expression plasmid, as described above for 1205Lu and WM278 cells. 100 ng/ml IFN-γ was added to HLA-C*06:02–positive melanocytic cell lines. HLA-C*06:02–negative antigen-presenting cell lines were cotransfected with pRSV–HLA-C*06:02. ADAMTSL5 knockdown ADAMTSL5-1 siRNA was designed and purchased from QIAGEN. ADAMTSL5-2 siRNA and control siRNAs 1 and 2 were obtained from MWG Eurofins. siRNA target sequences are given in Table S2. 24 h after siRNA transfection of WM278 cells, 100 ng/ml IFN-γ was added. 24 h later, medium was replaced and hybridoma cells were added and co-cultured for 24 h. For validation of ADAMTSL5 knockdown, ADAMTSL5 mRNA levels were determined by real-time quantitative PCR (qPCR) in triplicates using Light Cycler 2.0 (Roche). Porphobilinogen deaminase (PBGD) mRNA was used as an internal standard. Immunofluorescence staining of paraffin sections Sections (5 µm) from formalin-fixed and paraffin-embedded tissue samples were cut on adhesive glass slides (SuperFrost plus; Menzel-Glaeser), dewaxed, and rehydrated. Heat-induced antigen retrieval used Tris-EDTA buffer (10 mM Tris Base, 1 mM EDTA, and 0.05% Tween 20, pH 9.0) at 120°C for 15 min. Sections were incubated with primary antibody or isotype controls at 4°C for 12–60 h, washed, and incubated with fluorescence-conjugated secondary antibody for 90 min. Nuclei were counterstained with DAPI. ADAMTSL5-blocking peptide was added at 5 ng/ml. For multiple stainings, primary antibodies were used in conditions established in single-color stainings. For triple staining, anti–Granzyme B antibody was stained using the tyramide signal amplification method (TSA; PerkinElmer). All antibodies are listed in Table S3. Epidermal CD8+ T cells were defined as reported previously (Bovenschen et al., 2005). Microscopic analysis was performed on an Axio Observer microscope (Carl Zeiss) with a Chroma ET filter-set and a CoolSNAP-HQ2 digital camera system (Photometrics). Images were merged using Basic MetaMorph Imaging Software (Visitron). Semiquantification of CD8+ T cell to melanocyte contacts Epidermal areas of skin sections were completely photodocumented at a 100-fold magnification with GFP, Texas Red, and DAPI filters, respectively (Visitron), and images were merged. Epidermal segments were evaluated in each photographic image using Fiji/ImageJ software (National Institutes of Health). Stained cells and direct cell contacts were counted in each overlay independently by two authors. Resulting median values in each individual were used for statistical analyses. PBMC peptide stimulation, intracellular cytokine staining, and ELISA PBMCs were separated by Ficoll density gradient centrifugation. Freshly isolated PBMCs were seeded to 96-well flat-bottom plates (106 cells/well) and cultured in RPMI 1640 medium supplemented with 5 IU/ml IL-2 and 10% human AB serum. Highly purified ADAMTSL5 peptide (VRSRRCLRL, purity >95%; Thermo Fisher Scientific) and an unrelated peptide (ELQGLKDD) lacking homology to human proteins were used at 10 ng/ml. Intracellular cytokine staining was performed as described previously (Fujii et al., 2011). After 48 h of peptide stimulation, monensin (Golgi stop [BD], 0.35 µl for 200 µl culture medium) was added for 5 h. Cells were surface stained for CD8, fixed, and permeabilized in fixation buffer (eBioscience) and intracellularly stained using antibodies against IL-17A and IFN-γ. Data were analyzed by FlowJo software 887 (Tree Star). Lymphocyte gates were defined using SSC and FSC channels. Positive/negative cut-off levels for IFN-γ and IL-17A were defined by isotype control stainings of CD8+ T cells in unstimulated samples from healthy donors in each experiment. To determine cytokine secretion, PBMCs were stimulated with peptides for 48 h as described above for intracellular cytokine staining. Culture supernatants were harvested, and IFN-γ and IL-17A were quantified in triplicates using ELISA kits (Mabtech). During the experiments, the investigator was blinded for the sample group allocation. Thresholds for positive cytokine induction by ADAMTSL5 peptide are set at mean + 3SD of healthy control samples. HLA typing HLA haplotypes were determined by sequence-based typing at the Laboratory for Immunogenetics and Molecular Diagnostics, University of Munich. In select experiments, HLA-C*06.02 typing was performed by PCR restriction fragment length polymorphism analysis as described previously (Tazi Ahnini et al., 1999). Statistical analysis Kruskal-Wallis H-test was used for multiple comparisons, and Bonferroni correction was applied. There is more than one group without normal distributions in each comparison (Shapiro-Wilk W test). Accordingly, when the p-value of Kruskal-Wallis H-test was significant, two-group comparison was performed using the Mann-Whitney U test. Different proportions in the two groups were compared using Fisher’s exact test. Two-tailed P < 0.05 was considered significant. All statistical analyses were performed using GraphPad Software version 4 and R software. Sample size was determined based on preliminary data (mean and variation) and previous publications, as well as observed effect sizes. No samples were excluded from analysis. Online supplemental material Table S1, included as a separate Excel file, provides patients’ data and HLA-C*06:02 status. Table S2, included as a separate Excel file, provides primer and siRNA sequences. Antibodies are listed in Table S3, included as a separate Excel file. Online supplemental material is available at http://www.jem.org/cgi/content/full/jem.20151093/DC1. Supplementary Material Supplemental Materials ACKNOWLEDGMENTS We thank Mr. Adrian Galinski, Mr. Sebastian Harrasser, Ms. Ursula Puchta, and Ms. Marija Rozman for their support and Prof. Ulla Knaus and Mr. Takanobu Tagawa for helpful discussion. We thank Prof. Suneel S. Apte for the ADAMTSL5 plasmid. This project was supported by the Deutsche Forschungsgemeinschaft, grants Pr 241/4-1 and CRC-128-A5. The Ludwig-Maximilian-University filed a patent application for the peptides with J.C. Prinz and K. Dornmair as inventors. The authors declare no additional competing financial interests. Abbreviations used: PBGD porphobilinogen deaminase PECPL plasmid-encoded combinatorial nonamer peptide library qPCR quantitative PCR ==== Refs Arden, B., S.P. Clark, D. Kabelitz, and T.W. Mak. 1995. Human T-cell receptor variable gene segment families. Immunogenetics. 42 :455–500.8550092 Bader, H.L., L.W. Wang, J.C. Ho, T. Tran, P. Holden, J. Fitzgerald, R.P. Atit, D.P. Reinhardt, and S.S. Apte. 2012. A disintegrin-like and metalloprotease domain containing thrombospondin type 1 motif-like 5 (ADAMTSL5) is a novel fibrillin-1-, fibrillin-2-, and heparin-binding member of the ADAMTS superfamily containing a netrin-like module. Matrix Biol. 31 :398–411. 10.1016/j.matbio.2012.09.003 23010571 Besgen, P., P. Trommler, S. Vollmer, and J.C. Prinz. 2010. Ezrin, maspin, peroxiredoxin 2, and heat shock protein 27: potential targets of a streptococcal-induced autoimmune response in psoriasis. J. Immunol. 184 :5392–5402. 10.4049/jimmunol.0903520 20363977 Birnbaum, M.E., J.L. Mendoza, D.K. Sethi, S. Dong, J. Glanville, J. Dobbins, E. Ozkan, M.M. Davis, K.W. Wucherpfennig, and K.C. Garcia. 2014. Deconstructing the peptide-MHC specificity of T cell recognition. Cell. 157 :1073–1087. 10.1016/j.cell.2014.03.047 24855945 Bovenschen, H.J., M.M. Seyger, and P.C. Van de Kerkhof. 2005. Plaque psoriasis vs. atopic dermatitis and lichen planus: a comparison for lesional T-cell subsets, epidermal proliferation and differentiation. Br. J. Dermatol. 153 :72–78. 10.1111/j.1365-2133.2005.06538.x 16029329 Chang, J.C., L.R. Smith, K.J. Froning, B.J. Schwabe, J.A. Laxer, L.L. Caralli, H.H. Kurland, M.A. Karasek, D.I. Wilkinson, D.J. Carlo, 1994. CD8+ T cells in psoriatic lesions preferentially use T-cell receptor V beta 3 and/or V beta 13.1 genes. Proc. Natl. Acad. Sci. USA. 91 :9282–9286. 10.1073/pnas.91.20.9282 7937756 Conrad, C., O. Boyman, G. Tonel, A. Tun-Kyi, U. Laggner, A. de Fougerolles, V. Kotelianski, H. Gardner, and F.O. Nestle. 2007. α1β1 integrin is crucial for accumulation of epidermal T cells and the development of psoriasis. Nat. Med. 13 :836–842. 10.1038/nm1605 17603494 Dionne, S.O., D.F. Lake, W.J. Grimes, and M.H. Smith. 2004. Identification of HLA-Cw6.02 and HLA-Cw7.01 allele-specific binding motifs by screening synthetic peptide libraries. Immunogenetics. 56 :391–398. 10.1007/s00251-004-0710-1 15309347 Falk, K., O. Rötzschke, B. Grahovac, D. Schendel, S. Stevanović, V. Gnau, G. Jung, J.L. Strominger, and H.G. Rammensee. 1993. Allele-specific peptide ligand motifs of HLA-C molecules. Proc. Natl. Acad. Sci. USA. 90 :12005–12009. 10.1073/pnas.90.24.12005 8265661 Fujii, H., A. Arakawa, A. Kitoh, M. Miyara, M. Kato, S. Kore-eda, S. Sakaguchi, Y. Miyachi, M. Tanioka, and M. Ono. 2011. Perturbations of both nonregulatory and regulatory FOXP3+ T cells in patients with malignant melanoma. Br. J. Dermatol. 164 :1052–1060. 10.1111/j.1365-2133.2010.10199.x 21198537 Griffiths, C.E., and J.N. Barker. 2007. Pathogenesis and clinical features of psoriasis. Lancet. 370 :263–271. 10.1016/S0140-6736(07)61128-3 17658397 Gudjonsson, J.E., A. Karason, A. Antonsdottir, E.H. Runarsdottir, V.B. Hauksson, R. Upmanyu, J. Gulcher, K. Stefansson, and H. Valdimarsson. 2003. Psoriasis patients who are homozygous for the HLA-Cw*0602 allele have a 2.5-fold increased risk of developing psoriasis compared with Cw6 heterozygotes. Br. J. Dermatol. 148 :233–235. 10.1046/j.1365-2133.2003.05115.x 12588373 Hsu, M.Y., and M. Herlyn. 1996. Cultivation of normal human epidermal melanocytes. Methods Mol. Med. 2 :9–20.21359729 Kent, S.C., Y. Chen, L. Bregoli, S.M. Clemmings, N.S. Kenyon, C. Ricordi, B.J. Hering, and D.A. Hafler. 2005. Expanded T cells from pancreatic lymph nodes of type 1 diabetic subjects recognize an insulin epitope. Nature. 435 :224–228. 10.1038/nature03625 15889096 Kim, S.M., L. Bhonsle, P. Besgen, J. Nickel, A. Backes, K. Held, S. Vollmer, K. Dornmair, and J.C. Prinz. 2012. Analysis of the paired TCR α- and β-chains of single human T cells. PLoS One. 7 :e37338. 10.1371/journal.pone.0037338 22649519 Knickelbein, J.E., K.M. Khanna, M.B. Yee, C.J. Baty, P.R. Kinchington, and R.L. Hendricks. 2008. Noncytotoxic lytic granule-mediated CD8+ T cell inhibition of HSV-1 reactivation from neuronal latency. Science. 322 :268–271. 10.1126/science.1164164 18845757 Kniepert, A., and M. Groettrup. 2014. The unique functions of tissue-specific proteasomes. Trends Biochem. Sci. 39 :17–24. 10.1016/j.tibs.2013.10.004 24286712 Lande, R., E. Botti, C. Jandus, D. Dojcinovic, G. Fanelli, C. Conrad, G. Chamilos, L. Feldmeyer, B. Marinari, S. Chon, 2014. The antimicrobial peptide LL37 is a T-cell autoantigen in psoriasis. Nat. Commun. 5 :5621. 10.1038/ncomms6621 25470744 Lowes, M.A., A.M. Bowcock, and J.G. Krueger. 2007. Pathogenesis and therapy of psoriasis. Nature. 445 :866–873. 10.1038/nature05663 17314973 Martin, R., M.D. Howell, D. Jaraquemada, M. Flerlage, J. Richert, S. Brostoff, E.O. Long, D.E. McFarlin, and H.F. McFarland. 1991. A myelin basic protein peptide is recognized by cytotoxic T cells in the context of four HLA-DR types associated with multiple sclerosis. J. Exp. Med. 173 :19–24. 10.1084/jem.173.1.19 1702137 McCutcheon, J.A., J. Gumperz, K.D. Smith, C.T. Lutz, and P. Parham. 1995. Low HLA-C expression at cell surfaces correlates with increased turnover of heavy chain mRNA. J. Exp. Med. 181 :2085–2095. 10.1084/jem.181.6.2085 7760000 Nair, R.P., P.E. Stuart, I. Nistor, R. Hiremagalore, N.V. Chia, S. Jenisch, M. Weichenthal, G.R. Abecasis, H.W. Lim, E. Christophers, 2006. Sequence and haplotype analysis supports HLA-C as the psoriasis susceptibility 1 gene. Am. J. Hum. Genet. 78 :827–851. 10.1086/503821 16642438 Porter, A.G., and R.U. Jänicke. 1999. Emerging roles of caspase-3 in apoptosis. Cell Death Differ. 6 :99–104. 10.1038/sj.cdd.4400476 10200555 Rasmussen, M., M. Harndahl, A. Stryhn, R. Boucherma, L.L. Nielsen, F.A. Lemonnier, M. Nielsen, and S. Buus. 2014. Uncovering the peptide-binding specificities of HLA-C: a general strategy to determine the specificity of any MHC class I molecule. J. Immunol. 193 :4790–4802. 10.4049/jimmunol.1401689 25311805 Seitz, S., C.K. Schneider, J. Malotka, X. Nong, A.G. Engel, H. Wekerle, R. Hohlfeld, and K. Dornmair. 2006. Reconstitution of paired T cell receptor α- and β-chains from microdissected single cells of human inflammatory tissues. Proc. Natl. Acad. Sci. USA. 103 :12057–12062. 10.1073/pnas.0604247103 16882720 Siewert, K., J. Malotka, N. Kawakami, H. Wekerle, R. Hohlfeld, and K. Dornmair. 2012. Unbiased identification of target antigens of CD8+ T cells with combinatorial libraries coding for short peptides. Nat. Med. 18 :824–828. 10.1038/nm.2720 22484809 Tazi Ahnini, R., N.J. Camp, M.J. Cork, J.B. Mee, S.G. Keohane, G.W. Duff, and F.S. di Giovine. 1999. Novel genetic association between the corneodesmosin (MHC S) gene and susceptibility to psoriasis. Hum. Mol. Genet. 8 :1135–1140. 10.1093/hmg/8.6.1135 10332047 Valdimarsson, H., R.H. Thorleifsdottir, S.L. Sigurdardottir, J.E. Gudjonsson, and A. Johnston. 2009. Psoriasis—as an autoimmune disease caused by molecular mimicry. Trends Immunol. 30 :494–501. 10.1016/j.it.2009.07.008 19781993 Wang, C.Q., Y.T. Akalu, M. Suarez-Farinas, J. Gonzalez, H. Mitsui, M.A. Lowes, S.J. Orlow, P. Manga, and J.G. Krueger. 2013. IL-17 and TNF synergistically modulate cytokine expression while suppressing melanogenesis: potential relevance to psoriasis. J. Invest. Dermatol. 133 :2741–2752. 10.1038/jid.2013.237 23732752 Zaba, L.C., I. Cardinale, P. Gilleaudeau, M. Sullivan-Whalen, M. Suárez-Fariñas, J. Fuentes-Duculan, I. Novitskaya, A. Khatcherian, M.J. Bluth, M.A. Lowes, and J.G. Krueger. 2007. Amelioration of epidermal hyperplasia by TNF inhibition is associated with reduced Th17 responses. J. Exp. Med. 204 :3183–3194. (published erratum appears in J. Exp. Med. 2008. 205:1941) 10.1084/jem.20071094 18039949

PMC004xxxxxx/PMC4845687.txt

==== Front J Gen Physiol J Gen Physiol jgp jgp The Journal of General Physiology 0022-1295 1540-7748 The Rockefeller University Press 27114612 201511499 10.1085/jgp.201511499 Research Articles Research Article Structure–function relationships of peptides forming the calcin family of ryanodine receptor ligands Structure–function relationship of calcins Xiao Liang 13* Gurrola Georgina B. 23* Zhang Jing 3 Valdivia Carmen R. 3 SanMartin Mario 3 Zamudio Fernando Z. 23 Zhang Liming 1 Possani Lourival D. 2 Valdivia Héctor H. 3 1 Department of Marine Biotechnology, Faculty of Naval Medicine, Second Military Medical University, Shanghai 200433, China 2 Departamento de Medicina Molecular y Bioprocesos, Instituto de Biotecnología, Universidad Nacional Autónoma de México, Cuernavaca, Morelos 62271, México 3 Center for Arrhythmia Research, Cardiovascular Division, Department of Internal Medicine, University of Michigan, Ann Arbor, MI 48109 Correspondence to Héctor H. Valdivia: hvaldiv@umich.edu * L. Xiao and G.B. Gurrola contributed equally to this paper. 5 2016 147 5 375394 14 8 2015 21 3 2016 © 2016 Xiao et al. 2016 https://creativecommons.org/licenses/by-nc-sa/3.0/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 3.0 Unported license, as described at http://creativecommons.org/licenses/by-nc-sa/3.0/). Calcins are a new and expanding family of ryanodine receptor agonists found in scorpion venom. Xiao and colleagues characterize the structure–function relationships of all known calcins and show both similarities and differences in their form and function. Calcins are a novel family of scorpion peptides that bind with high affinity to ryanodine receptors (RyRs) and increase their activity by inducing subconductance states. Here, we provide a comprehensive analysis of the structure–function relationships of the eight calcins known to date, based on their primary sequence, three-dimensional modeling, and functional effects on skeletal RyRs (RyR1). Primary sequence alignment and evolutionary analysis show high similarity among all calcins (≥78.8% identity). Other common characteristics include an inhibitor cysteine knot (ICK) motif stabilized by three pairs of disulfide bridges and a dipole moment (DM) formed by positively charged residues clustering on one side of the molecule and neutral and negatively charged residues segregating on the opposite side. [3H]Ryanodine binding assays, used as an index of the open probability of RyRs, reveal that all eight calcins activate RyR1 dose-dependently with Kd values spanning approximately three orders of magnitude and in the following rank order: opicalcin1 > opicalcin2 > vejocalcin > hemicalcin > imperacalcin > hadrucalcin > maurocalcin >> urocalcin. All calcins significantly augment the bell-shaped [Ca2+]-[3H]ryanodine binding curve with variable effects on the affinity constants for Ca2+ activation and inactivation. In single channel recordings, calcins induce the appearance of a subconductance state in RyR1 that has a unique fractional value (∼20% to ∼60% of the full conductance state) but bears no relationship to binding affinity, DM, or capacity to stimulate Ca2+ release. Except for urocalcin, all calcins at 100 nM concentration stimulate Ca2+ release and deplete Ca2+ load from skeletal sarcoplasmic reticulum. The natural variation within the calcin family of peptides offers a diversified set of high-affinity ligands with the capacity to modulate RyRs with high dynamic range and potency. National Institutes of Health 10.13039/100000002 R01-HL120108 R01-HL055438 Dirección General Asuntos del Personal Académico, Universidad Nacional Autónoma de México 10.13039/501100006087 IN200113 ==== Body pmcINTRODUCTION Excitation–contraction coupling is the process that facilitates Ca2+ release from the SR of cardiac or skeletal muscle fibers after electrical excitation of the surface/transverse (t-) tubule membrane (Tanabe et al., 1988; Bers, 2002). During this process, the release of Ca2+ from the SR is caused by the activation of the Ca2+ release channel/ryanodine receptors (RyRs), which form homotetrameric assemblies and constitute the largest ion channels known to date, with molecular masses of ∼2.2 MD and each monomer consisting of ∼5,000 amino acid residues (Lai et al., 1988; Yuchi et al., 2012). In mammals, RyRs are expressed in three different forms, which help them gain the functional flexibility to respond to different triggering signals: RyR1 (Takeshima et al., 1989) is predominantly expressed in fast- and slow-twitch skeletal muscle and in cerebellar Purkinje cells; RyR2 (Nakai et al., 1990) is found in cardiac muscle but is also robustly expressed in brain and in visceral and arterial smooth muscle; RyR3 (Hakamata et al., 1992) is the least understood of the RyR isoforms and appears to play its most important role during development, although in mature cells is found in diaphragm, epithelial cells, brain, and smooth muscle (Fill and Copello, 2002). Structurally, the RyR channel is the center of a massive macromolecular complex whose function is modulated by secondary messengers (Ca2+, Mg2+, ATP) and accessory proteins (e.g., calmodulin, FK506-binding protein, sorcin, triadin, junctin, calsequestrin; Bers, 2004; Ather et al., 2013). Exogenous chemicals and peptides binding to RyRs and modulating their activities have become invaluable tools for the structural and functional profiling of RyRs as well as the development of new drugs for RyR-associated diseases. The plant alkaloid ryanodine, for which this receptor was named, binds RyRs preferentially in the open state (Sutko et al., 1997), which allows experimenters to use [3H]ryanodine as a high-affinity (nanomolar) probe of the functional state of the channel. Ryanodine was instrumental in the isolation of the RyR1 and RyR2 channels and continues to aid in the characterization of key pharmacologic properties of RyRs, although some undesirable features like concentration-dependent biphasic effects, slow association and dissociation rate (which make its effect practically irreversible), loss of affinity upon derivatization, etc. hamper its use in cellular and biochemical settings. Caffeine is another classical ligand of RyRs with low affinity (Kd ∼300 µM) that is widely used to assess the presence and size of RyR-gated Ca2+ stores by increasing the sensitivity of RyR channels to both cytosolic and luminal Ca2+ (Rousseau and Meissner, 1989; Kong et al., 2008) and eliciting Ca2+ release. Other ligands like doxorubicin (Abramson et al., 1988), tetracaine (Curran et al., 2007), ruthenium red (Lukyanenko et al., 2000), K201 (Wehrens et al., 2004), and carvedilol (Zhou et al., 2011) bind to and modulate the activity of RyRs, helping to “tame” their activity in pathological settings (Valdivia, 2014). Except for ryanodine at low concentrations, all of the aforementioned ligands of RyRs exhibit off-target effects. In search of novel ligands of RyRs with high affinity and specificity, we found in the venom of selected scorpions peptides that selectively mobilize Ca2+ from RyR-gated stores. Imperatoxin A (from the scorpion Pandinus imperator) is a small (3.7-kD), highly basic (pI 8.9), globular, and thermostable peptide that activates RyRs with high affinity (Kd ≈ 5–10 nM) and specificity (no other target proteins known to date; Valdivia et al., 1992; El-Hayek et al., 1995; Zamudio et al., 1997; Nabhani et al., 2002). Imperatoxin A spawned the discovery of calcins, a small but growing group of scorpion peptide agonists of RyRs, and to conform to its structural and functional characteristics, we hereby rename it imperacalcin, which also avoids confusion with other scorpion toxin blockers of ion channels. The defining functional feature of calcins is their capacity to stabilize RyR openings in a long-lasting, subconductance state (Tripathy et al., 1998). This effect is nearly analogous to that of ryanodine, but unlike ryanodine, calcins bind rapidly to RyRs (fast association rate), freely dissociate from their binding site (reversible effect), display a dose- and sequence-variable effect, and are amenable for derivatization without undergoing major loss in receptor affinity (El-Hayek et al., 1995; Samsó et al., 1999; Shtifman et al., 2000; Dulhunty et al., 2004; Gurrola et al., 2010). As a native family of intracellular RyR ligands, calcins are also appealing by their capacity to penetrate cell membranes with high efficiency, resulting in extremely rapid effects (lag time 2–3 s) on intracellular Ca2+ signaling, as indicated by an acute increase in the amplitude of the [Ca2+]i transient, indicative of enhanced Ca2+ release from the SR (Schwartz et al., 2009; Gurrola et al., 2010). Moreover, calcins are capable of carrying large, membrane-impermeable cargo across the plasma membrane, a finding with exciting implications for intracellular drug delivery (Altafaj et al., 2005; Boisseau et al., 2006), particularly in the treatment of RyR channelopathies (Benkusky et al., 2004). Despite the importance of calcins as selective, high-affinity membrane-permeable ligands of RyRs, key structural features of this family of scorpion peptides remain unclear, and there are no studies to date that have systematically investigated the role of their structural domains in relation to RyR affinity and single RyR channel effects. Here we present all native calcins known to date, including four previously reported (imperacalcin, maurocalcin, hemicalcin, and hadrucalcin), three previously uncharacterized (opicalcin1, opicalcin2, and urocalcin), and one novel (vejocalcin) peptides, and provide a comprehensive analysis of their structure–function relationship based on primary sequence examination, physicochemical characterization, three-dimensional structure modeling, and their functional effects on RyR1 channels using [3H]ryanodine binding, single channel recordings, and Ca2+ release from SR vesicles. Natural variations in the calcin family of peptides offer a diversified set of RyR ligands with the capacity to modulate RyR channels with variable potency and high dynamic range. MATERIALS AND METHODS Animals Animal care and handling conformed to the Guide for Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH publication no. 85-23, revised 1996), and experimental protocols were approved by the local institutional ethical committee. The New Zealand White rabbits used in this study were maintained in environmentally controlled rooms with 12-h light/dark cycles and adequate food and water, where they was assessed daily for wellbeing by trained veterinary personnel. Back and leg skeletal muscle was obtained after intraperitoneal injection of 100 mg/kg pentobarbital. Adequacy of anesthesia was assessed by pedal reflex and evaluation of jaw and muscle tone before proceeding with surgery (Schwartz et al., 2009). Venom source and purification procedures for vejocalcin Venom from Vaejovis mexicanus (Scorpiones: Vaejovidae) for vejocalcin was obtained in the laboratory by electrical stimulation (15–25 V, 3 ms) in the articulation of the telson (last post-abdominal segment). Scorpions were milked using a Grass stimulator. The venom was collected in Eppendorf LoBind tubes and, when needed, dissolved in double-distilled water and then centrifuged at 14,000 g for 15 min at 4°C. The soluble supernatant was either lyophilized or stored at −20°C and later separated by HPLC essentially as described previously (Schwartz et al., 2006). In brief, whole venom was injected into a C18 reverse-phase semi-preparative column (Vydac) and separated by using a linear gradient from solvent A (0.12% trifluoroacetic acid [TFA] in water) to 60% solvent B (0.10% TFA in acetonitrile) run for 60 min at a flow rate of 1 ml·min−1. The fractions of interest from several independent runs were pooled and further purified on a C18 reverse-phase analytic column (Vydac) to obtain vejocalcin in homogeneous form, as shown by mass spectrometry analysis and amino acid sequencing. Amino acid sequence determination and mass spectrometry analysis The amino acid sequence of vejocalcin was obtained by automatic Edman degradation in an LF 3000 Protein Sequencer (Beckman Coulter) using the chemicals and procedures previously described for other peptides of scorpion venom (Schwartz et al., 2006). The molecular mass of vejocalcin was obtained by mass spectrometer analysis, using an LCQ Duo (Finnigan), also as described previously (Schwartz et al., 2006). Chemical synthesis, folding, and purification of calcins Calcins were synthesized by an automated peptide synthesizer (Applied Biosystems model 432A) as described previously (Zamudio et al., 1997; Seo et al., 2011). In brief, a linear analogue of calcin (imperacalcin, for example) was synthesized by the solid-phase methodology with Fmoc–amino acids. After cleavage with 90% TFA for 4 h at room temperature, the crude linear peptide was extracted with 5% acetic acid and dried by vacuum centrifugation. The cyclization reaction to form disulfide bonds in the molecule was performed in the folding buffer (20 mM Na2HPO4 + 0.1 M NaCl + 5 mM GSH + 0.5 mM GSSG). After pH adjustment to 7.9 with 1 M NaOH, the reaction mixture was allowed to oxidize at room temperature for 4 h. After oxidation of the disulfide bonds, the peptide solution was acidified with formic acid to pH 3.0 and pumped onto a C18 column. The sample was eluted with a linear gradient from 0 to 60% B in 60 min (buffer A = H2O containing 0.12% TFA, and buffer B = acetonitrile containing 0.1% TFA). Finally, the resulting peptide was analyzed by mass spectrometry and amino acid sequence. Peptide purity was >98% for all calcins. Rabbit skeletal heavy SR preparation Heavy SR vesicles were purified from rabbit white back and leg skeletal muscle, as described previously (Meissner, 1984; Dyck et al., 1987). Protein concentration was measured by the Bradford method. [3H]Ryanodine binding assay [3H]Ryanodine binding to rabbit skeletal SR was performed as described previously (El-Hayek et al., 1995; Gurrola et al., 1999; Schwartz et al., 2009). In brief, calcins were diluted directly into the incubation medium (0.2 M KCl, 10 µM CaCl2, and 10 mM Na-HEPES, pH 7.2) to achieve a final concentration of 1 pM to 20 µM, as indicated in graphs. To determine the effect of calcins on Ca2+ sensitivity of [3H]ryanodine binding activity, each calcin was added at fixed concentration (100 nM) to the standard incubation medium, which contained 0.2 M KCl, 1 mM Na2EGTA, 10 mM Na-PIPES, pH 7.2, and CaCl2 to set free [Ca2+] in the range of 10 nM to 10 mM. The Ca2+/EGTA ratio was calculated with the computer program MAXCHELATOR. [3H]Ryanodine (95 Ci·mmol−1; PerkinElmer) was diluted directly in the incubation medium to a final concentration of 5 nM. Protein concentration of heavy SR was 0.3 mg·ml−1 and was determined by the Bradford method. Incubation lasted 120 min at 36°C. 100-µl samples were always run in duplicate, filtered on GF/C glass filters (Whatman) and washed twice with 5 ml of distilled water using an M24-R cell harvester (Brandel). Nonspecific binding was determined in the presence of 20 µM unlabeled ryanodine, reached no more than 5–10% of the total binding, and was subtracted from each sample. Single channel recordings Single channel recordings of rabbit skeletal RyR1 incorporated into planar lipid bilayers were performed as described previously (Schwartz et al., 2009). Planar lipid bilayers, composed of phosphatidylethanolamine (Avanti Polar Lipids, Inc.) and phosphatidylserine (Avanti Polar Lipids, Inc.; 1:1), were painted with a glass rod across an aperture of 150-µm diameter in a Delrin cup. The cis chamber represented the cytosolic side, which was held at virtual ground and contained the reference electrode. The trans chamber corresponded to the luminal side and contained the voltage command electrode. Electrodes were connected to the head stage of a 200A Axopatch amplifier. Both the cis (0.7 ml) and trans (0.7 ml) chambers were filled with 300 mM caesium methanesulphonate and 20 mM MOPS, pH 7.2. Cs+ was selected as the charge carrier to increase the channel conductance and to avoid any contribution from potassium channels present in the SR membrane. Chloride channels were inhibited by using the impermeant anion methanesulfonate. Contaminant Ca2+, determined by a Ca2+ electrode, was ≈5 µM and served to activate RyR1 channels. Heavy SR vesicles (20–50 µg) were added to the cis chamber. Channel recordings were collected at different holding potentials (40 mV) before and after the addition of calcins. Single channel recordings were filtered with an 8-pole low-pass Bessel filter set at 2 kHz and digitized at a rate of 4 kHz by using a Digidata 1440A AD/DA interface. Data acquisition and analysis were performed with Axon Instruments hardware and software (pCLAMP 10) and graphed by using Microcal Inc. Origin 9.0. SR Ca2+ release measurements Ca2+ release from heavy SR was measured using the Ca2+-sensitive dye Arsenazo III (Sigma-Aldrich), modified as previously described (Gurrola et al., 1999; Chen et al., 2003). In brief, the absorbance was monitored at 650 nm by a compact UV-visible spectrophotometer (BioPhotometer Plus; Eppendorf). Heavy SR vesicles (20 µg) were actively loaded with Ca2+ at room temperature in a 1-ml basal buffer containing 100 mM KCl, 7.5 mM sodium pyrophosphate, and 20 mM MOPS, pH 7.0, supplemented with 25 µM Arsenazo III, 1 mM ATP/MgCl2 (Sigma-Aldrich), 5 mM creatine phosphate disodium salt tetrahydrate (ICN Biomedicals Inc.), and 12 µg/ml creatine phosphokinase (porcine heart; EMD Millipore). Ca2+ loading was started by consecutive additions of 50 nmol twice and 20 nmol thrice of CaCl2 before the addition of calcins. The total [Ca2+] loaded into the SR was quantified by the addition of the Ca2+ ionophore A23187 (5 µM), and the absorbance was converted to nmol [Ca2+] by a standard curve generated by sequential additions of CaCl2 (10–20 µM). Bioinformatics Five calcin sequences (mature peptide and/or precursors), namely opicalcin1 (P60252; Zhu et al., 2003), opicalcin2 (P60253; Zhu et al., 2003), maurocalcin (P60254; Fajloun et al., 2000), hadrucalcin (B8QG00; Schwartz et al., 2009), and urocalcin (AGA82762; Luna-Ramírez et al., 2013), were obtained by similarity with imperacalcin (formerly imperatoxin A; P59868; Zamudio et al., 1997) with BLAST search. The exact mature sequences were further confirmed from their original papers. The sequence of hemicalcin was obtained from the corresponding paper (Shahbazzadeh et al., 2007). Vejocalcin was purified and sequenced from its natural venom in the current study. Multiple sequence alignment tool Clustal Omega was used to align the calcin sequences and analyze their identity. Evolutionary history, using the “minimum evolution” method (Kumar, 1996), was conducted in MEGA 5.2 (Tamura et al., 2011). The evolutionary distance (ED) was computed using the Poisson correction method starting from the assumption that the rate of amino acid substitution at each site follows the Poisson distribution. Physical and chemical characterization, net charge versus pH plots, and hydrophobicity plots were analyzed by ProtParam, WebLab, ViewerPro, and the online peptide calculator (http://www.chinapeptides.com/tool.php?isCalu=1) in their corresponding functions. The secondary structures were analyzed by Ramachandran plots according to the structure of imperacalcin (MMDB ID: 23273), which was first resolved by NMR (Lee et al., 2004). Similarly, using imperacalcin as template, three-dimensional simulation was performed by Swiss-PdbViewer 4.1.0 with the mutagenesis function following the minimum energy principle. Discovery Studio 3.5 was used to display the modeling molecules. Solvent-accessible polar surface area (PSA) for each calcin was calculated by using the molecular visualization program PyMOL (DeLano Scientific), and the dipole moment (DM) was analyzed by the online Protein Dipole Moments Server. Statistics Statistical analyses were performed by using either the computer software program Origin (version 9.0) or SigmaStat computer software (version 5.0; Rockware Inc.). Data are presented as the mean ± SEM. Data were compared using a one-way ANOVA. Differences of P < 0.05 between sample means were considered significant. Online supplemental material Fig. S1 shows the backbone structure of the eight calcins. Fig. S2 shows the dose–response effect of imperacalcin on single RyR channel kinetics. Table S1 lists the amino acid composition of all eight known calcins. Online supplemental material is available at http://www.jgp.org/cgi/content/full/jgp.201511499/DC1. RESULTS Isolation and primary structure determination of vejocalcin Initial fractionation of soluble venom from V. mexicanus by HPLC separated >60 different components (Fig. 1 A). The fraction eluting at 20.96 min (vejocalcin) was further purified in an analytical C18 reverse-phase column, yielding the component of interest (Fig. 1 B, peak labeled with asterisk). Mass spectrometry analysis of this peak revealed a single peptide with 3,774.92 atomic mass units (a.m.u.). Amino acid sequence of the reduced and alkylated peptides yielded a single sequence up to residue 25. The digestion of reduced vejocalcin with AspN-endopeptidase permitted separation by HPLC of several subpeptides that under sequence analysis by Edman degradation and mass spectrometry fragmentation allowed the elucidation of a unique primary structure. An AspN-endopeptidase peptide yielded the overlapping sequence from position D15 to R33. The theoretical mean mass of the amino acid sequence shown in Fig. 1 C is 3,774.5 (M+H)+, which corresponds well with the experimental value of the native peptide. We have termed this peptide according to the genus of the scorpion and their great similarity to other peptide members of the calcin family (see Fig. 2). Figure 1. Purification and characterization of vejocalcin. (A) 1.5 mg of soluble venom from V. mexicanus was separated by HPLC using a semi-preparative C18 reverse-phase column, eluted with a linear gradient from solution A to 60% solution B, run for 60 min. The fraction labeled with the asterisk was rechromatographed in an analytical C18 reverse-phase column and run from solution A to 40% solution B in 60 min. (B) The component with retention time of 28.09 min (marked with an asterisk) is vejocalcin. (C) The complete amino acid sequence of vejocalcin was obtained by a combination of direct Edman degradation and mass spectrometry, as described in Materials and methods. Amino acid sequence alignment and evolutionary analysis of the calcin family Fig. 2 A shows the amino acid sequence of all peptide members of the calcin family known to date. All calcins are made up of 33 amino acids, except hadrucalcin, which prolongs its N terminus by adding Ser and Glu for a total of 35 residues. Six cysteines are highly conserved and may be used for precise alignment of these peptides; i.e., the number of residues interspaced between cysteines is identical for all calcins. Determination of disulfide bonds in maurocalcin (Mosbah et al., 2000) and imperacalcin (Lee et al., 2004) indicates that the pairings Cys3-Cys17, Cys10-Cys21, and Cys16-Cys32 (Fig. 2 B) contribute significantly to the inhibitor cysteine knot (ICK) motif (see Fig. S1) that characterizes this family of peptides. In terms of homology, calcins may be grossly split into halves for a distinctive pattern of variation: the N-terminal segment 1G-N14 contains only four residues that are identical in all calcins (two of them are cysteines), whereas the C-terminal segment 15D-R33, comprising the highly positively charged motifs 19KKCKRR24 and 30KRCR33, is relatively conserved among calcins (there are only nine variations in a total of 152 amino acids; 13 columns of amino acids are identical). Thus, by primary sequence analysis only, one might be tempted to estimate the hierarchical role of calcins domains, with the C-terminal half playing a prominent role based on high similarity. However, this hierarchy may turn equivocal because the overall variation of calcins is relatively minor, with identity descending only to 78.8% in the case of hadrucalcin. Also, as will be shown in three-dimensional representations of calcins, variable segments may intertwine with highly conserved segments, breaking up what may be considered “canonical” domains. Figure 2. Calcin sequence alignment and evolutionary analysis. (A) The eight calcins known to date are aligned with opicalcin1 as reference by Clustal Omega. The identity values to opicalcin1 are shown on the right side. Positively charged residues lysine (K) and arginine (R), negatively charged residues aspartic acid (D) and glutamic acid (E), and the disulfide bond-forming cysteine (C) are highlighted by the colors blue, red, and gray, respectively. Columns with identical residue are marked by asterisks, whereas columns with only one difference are marked by colons on the bottom. (B) Three pairs of highly conserved disulfide bonds (Cys3-Cys17, Cys10-Cys21, and Cys16-Cys32) and the residues forming part of β strands appear connected to form an ICK motif. (C) The evolutionary tree is built with the principle minimum evolution by MEGA 5.2. The genetic distance is measured by Poisson correction method with the formula dAB = −ln(1 − fAB), where fAB is the fraction of different amino acids after the Poisson distribution between two sequences (dissimilarity). Fig. 2 C shows the evolutionary relationship of the calcin family obtained with the minimum evolution method. The evolutionary value was determined with the Poisson correction method, where 0 and 1 represent identity and divergence, respectively, between two molecules. Opicalcin2 is naturally the closest to opicalcin1 (only one residue different) with ED value = 0.015. In the middle, imperacalcin, hemicalcin, and maurocalcin display a similar ED to opicalcin1 (0.065), followed by vejocalcin with ED = 0.123. Urocalcin and hadrucalcin are the farthest and hence the most divergent among calcins; yet the ED is still small, with maximal ED = 0.142, indicating that the ancestor of this family of peptides is relatively close to these eight known calcins. Physicochemical features of calcins The molecular mass of calcins ranges from 3,758.5 D (imperacalcin) to 4,190.8 D (hadrucalcin), whereas the molecular volume falls between 2,692.7 Å3 (vejocalcin) and 3,001.9 Å3 (hadrucalcin; Table 1). As for the composition of amino acids (Fig. 2 and Table S1), positively charged residues Lys and Arg are abundant and range from 9 (27%) for vejocalcin to 13 (39%) for urocalcin. Conversely, negatively charged residues Asp and Glu are sparse, ranging from three (9%) for vejocalcin to five (14%) for hadrucalcin. This high positive/negative amino acid ratio gives calcins high isoelectric point (pI = 9.3–10.1) and net positive charge at pH 7.0 that ranges from 5.8 (vejocalcin) to 8.7 (urocalcin). By the same token, the richness of charged residues determines the excellent solubility in water of calcins (hydrophilicity = 0.7–1.2) despite the fact that they are membrane permeable and should intuitively be hydrophobic in nature. However, as is the case for most membrane-permeable peptides, hydrophilicity and membrane permeability may be oddly compatible. The most notable deviation from the group is vejocalcin, which holds the lowest pI (9.3), poorest hydrophilicity (0.7), and least positive charges at pH 7.0 (5.8). On the other extreme, urocalcin has the highest number of positively charged residues (13) and the highest pI (10.1), is the most hydrophilic (1.2), and holds the most charges at pH 7.0 (8.7; Table 1 and Fig. 3, A and B). As we will discuss, both extremes turn out to be detrimental for receptor affinity. Table 1. Physical and chemical characteristics of calcins Calcins Formula Amino acids Molecular mass Molecular volume Negatively charged residues Positively charged residues Theoretical isoelectric point Hydrophilicity Ratio of hydrophilic residues Net charge at pH 7.0 PSA D Å3 Å2 OpCa1 C153H261N59O47S6 33 3,871.5 2,762.6 4 (12%) 11 (33%) 9.7 1.1 58% 6.8 2,761 OpCa2 C152H259N59O46S6 33 3,841.5 2,742.0 4 (12%) 11 (33%) 9.7 1.1 58% 6.8 2,733 IpCa C148H254N58O45S6 33 3,758.4 2,783.7 4 (12%) 11 (33%) 9.7 1.0 52% 6.8 2,721 MCa C156H270N56O46S6 33 3,858.6 2,781.9 4 (12%) 11 (33%) 9.6 1.0 55% 6.8 2,821 HmCa C153H263N55O45S6 33 3,785.5 2,721.9 4 (12%) 11 (33%) 9.6 1.0 52% 6.8 2,765 VjCa C149H254N56O47S6 33 3,774.4 2,692.7 3 (9%) 9 (27%) 9.3 0.7 55% 5.8 2,790 UrCa C158H282N56O47S6 33 3,910.7 2,840.9 4 (12%) 13 (39%) 10.1 1.2 64% 8.7 2,965 HdCa C164H282N64O53S6 35 4,190.8 3,001.9 5 (14%) 12 (34%) 9.7 1.2 66% 6.8 3,227 Summary of the physical and chemical characteristics of all eight known calcins, including formula, quantity of amino acids, molecular mass, molecular volume, negatively and positively charged amino acids, theoretical isoelectric point, hydrophilicity, ratio of hydrophilic residues, and net charge at pH 7.0. HdCa, hadrucalcin; HmCa, hemicalcin; IpCa, imperacalcin; MCa, maurocalcin; OpCa1, opicalcin1; OpCa2, opicalcin2; UrCa, urocalcin; VjCa, vejocalcin. Figure 3. Net charge versus pH plot and hydrophobicity plot of calcins. (A) Net charge versus pH plot of calcins. The theoretical isoelectric points of calcins are highly basic with values ranging from 9.3 (vejocalcin) to 10.1 (urocalcin), whereas the net charges at pH 7.0 are positive for all calcins and vary from 5.8 (vejocalcin) to 8.7 (urocalcin). (B) Hydrophobicity plot of calcins. The ratio of hydrophilic residues (green) of calcins is between 52% (imperacalcin and hemicalcin) and 66% (hadrucalcin), indicating high solubility in water for all calcins. Secondary and spatial structures of the calcin family of peptides The three-dimensional structures of maurocalcin (Mosbah et al., 2000) and imperacalcin (Lee et al., 2004) have been determined by 1H-NMR. We used their structural templates and modeled the six remainder calcins as described in Materials and methods (only imperacalcin-modeled structures are presented; similar results were obtained using maurocalcin as a template). All calcins are predicted to fold into an ICK motif, whose defining features include the presence of three disulfide bridges along with sections of polypeptide between them forming β strands. In the case of calcins, the first (Cys3-Cys17) and the second (Cys10-Cys21) disulfide bonds form a loop, through which the third (Cys16-Cys32) disulfide bond passes across, forming a knot (hence the term cysteine knot; Fig. 4 and Fig. S1). Accordingly, calcins contain three β strands and one α helix interspaced between β strands 1 and 2 (βαββ). β strands 2 and 3 form an antiparallel sheet (Table 2 shows that that 20KCK22, the second β strand, is 20KCS22 in vejocalcin, thus reducing its propensity to form β strand 2). The ICK motif confers to these peptides a globular, highly stable structure that is likely responsible for their resistance to heat denaturation and proteolysis. Despite their coiled, compact globular structure, the solvent-accessible PSA of calcins is relatively high, owing to their abundance of negatively and positively charged residues, ranging from 2,721 Å2 (imperacalcin) to 2,965 Å2 (urocalcin). Hadrucalcin breaks into the PSA >3,000 Å2 because of its additional N-terminal polar amino acids Ser and Glu, which stick out prominently from the main protein core (Fig. 4 H). Figure 4. Three-dimensional modeling of calcins. (A) Imperacalcin, resolved by 1H-NMR (Lee et al., 2004), was taken as the template molecule. (B–H) Other calcins, including opicalcin1 (B), opicalcin2 (C), maurocalcin (D), hemicalcin (E), vejocalcin (F), urocalcin (G), and hadrucalcin (H), were simulated by Swiss-PdbViewer 4.1.0 and viewed by Discovery Studio 3.5. Solvent-accessible PSAs were calculated using PyMOL, whereas DMs were analyzed by the online Protein Dipole Moments Server. For each calcin, the solid ribbon with line atom model may be found in Fig. S1, and the charged CPK model with frontal side (middle) and dorsal side (right) are displayed here. Positively charged residues (Lys and Arg), negatively charged residues (Asp and Glu), and neutral residues are colored by blue, red, and gray, respectively. Table 2. Imperacalcin-based secondary structure prediction Calcins Secondary structure (amino acids) β sheet1 (8–10) β sheet2 (20–22) β sheet3 (30–33) 3/10-helix (13–15) OpCa1 KRC KCK RCR NND OpCa2 KRC KCK RCR NND MCa KLC KCK RCR NKD IpCa KRC KCK RCR DND HmCa KLC KCK RCR DKD VjCa KLC – RCR NND UrCa KLC SCK RCR NKD HdCa QRC KCS RCR NKD Three β sheets (8–10, 20–22, and 30–33) and one 3/10-helix (13–15) are predicted for all calcins except vejocalcin, which lacks the β2 sheet. HdCa, hadrucalcin; HmCa, hemicalcin; IpCa, imperacalcin; MCa, maurocalcin; OpCa1, opicalcin1; OpCa2, opicalcin2; UrCa, urocalcin; VjCa, vejocalcin. Another distinguishing feature of calcins is their segregated arrangement of charged residues, where most of the positively charged amino acids are clustered in one side of the molecule, whereas neutral and negatively charged residues crowd the opposite side (Fig. 4), generating a discrete DM. The best examples of this segregated electrostatic distribution are imperacalcin (Fig. 4 A), opicalcin1 (Fig. 4 B), and opicalcin2 (Fig. 4 C), with DM (in Debye lengths [D]) = 152, 173, and 175 D, respectively. As observed, the “frontal” side of these calcins presents the majority (but not all) of the basic residues that form part of clusters 8KRCK11, 19KKCKRR24, and 30KRCR33, whereas the “dorsal” side presents only neutral and acidic residues, creating a largely negative sector that is electrically countered only in the flanks by basic residues. On the other extreme, vejocalcin (Fig. 4 F) and hadrucalcin (Fig. 4 H) exhibit the smallest charge segregation, with DM = 40 and 65 D, respectively. In these calcins, disruption of basic clusters by neutral and acidic residues in the frontal side of the molecules and, conversely, encroachment of basic residues in the negative sector of the dorsal side contribute substantially to equalize the spatial distribution of electrical charges. Interestingly, imperacalcin and hadrucalcin both penetrate isolated ventricular myocytes and induce intracellular Ca2+ release with comparable kinetics (Schwartz et al., 2009; Gurrola et al., 2010), indicating that a large DM is not required for calcins to enter cells and exert their pharmacological effect. We tested direct interaction of calcins with RyR1 using [3H]ryanodine binding assays to determine whether DM is instead related to calcin affinity for RyRs (see next section). Calcins stimulate [3H]ryanodine binding in a dose-dependent manner and at all Ca2+ levels We compared side by side the property of all eight calcins to stimulate [3H]ryanodine binding to RyR1, the skeletal isoform of RyRs (Fig. 5) and contrasted their effect with that of caffeine. Imperacalcin, the founder member of the calcin family, was discovered by its capacity to activate [3H]ryanodine binding to RyR1 (Valdivia et al., 1992). Three other calcins have since been characterized, namely, maurocalcin, hemicalcin, and hadrucalcin (Zamudio et al., 1997; Fajloun et al., 2000; Shahbazzadeh et al., 2007; Schwartz et al., 2009), which also stimulate [3H]ryanodine binding to RyR1. Opicalcin1 and opicalcin2 (Zhu et al., 2003) as well as urocalcin (Luna-Ramírez et al., 2013) and vejocalcin (this study) have never been tested for biological activity. As expected from their structural homology, all calcins stimulate [3H]ryanodine binding to RyR1, but also as expected from their unique amino acid sequence, they exhibit different affinity (apparent Kd; Fig. 5 A) and potency (maximal stimulation as percentage of control; Fig. 5 B). The apparent Kd surprisingly spanned ≈3 orders of magnitude, with the following ranking order (nM): opicalcin1 0.3 ± 0.04 > opicalcin2 3.2 ± 0.4 > vejocalcin 3.7 ± 0.4 > hemicalcin 6.9 ± 0.7 > imperacalcin 8.7 ± 0.9 > hadrucalcin 14.8 ± 1.9 > maurocalcin 26.4 ± 3.9 >> urocalcin 376 ± 45 (mean ± SEM; n = 3–5). Especially remarkable are opicalcin1 and opicalcin2, which differ by only one amino acid (the highly conserved Thr26 is replaced by Ala in opicalcin2), yet they display an ≈10-fold difference in Kd. An identical replacement in synthetic imperacalcin also decreases its Kd by ≈10-fold (Gurrola et al., 1999), indicating that Thr26 forms part, or is close to, the active site of calcins. Another notable deviation is urocalcin, with the lowest Kd of all calcins but with multiple variations in apparently key domains of calcins (see Discussion). Figure 5. [3H]Ryanodine binding stimulation by calcins. (A) Dose-dependent activation of RyR1 by calcins, with boundaries set at 1 pM to 20 µM (n = 3–5). Heavy SR from rabbit skeletal muscle was incubated with 5 nM [3H]ryanodine in the absence (control) and the presence of the indicated concentrations of calcins. Binding conditions were specified in Materials and methods. The Kd was determined with the formula B = (Bmax)/[1 + (Kd/[calcin])nH), where B is specific binding of [3H]ryanodine, Bmax is the maximum binding stimulated by calcin, and nH is the Hill coefficient. (B) Effect of calcins (100 nM each; n = 3–5) on the Ca2+-dependent [3H]ryanodine binding curve. Ca2+-dependent activation and inactivation were fitted with the formula B = ((Bmax·[Ca2+])/(Ka + [Ca2+]))·(Ki/(Ki + [Ca2+])), where B is the specific binding of [3H]ryanodine, Bmax is the maximum binding stimulated by calcin, Ka is the activation constant, and Ki is the inactivation constant. The specific [3H]ryanodine binding was standardized with the value of the control at pCa5 as 100%. (C–E) Radar charts illustrating the effect of calcins on Ka, Ki, and maximal amplitude (Amax; percentage of stimulation with respect to control). The black line corresponds to control (no calcin) and is included in all charts for reference. Caffeine (C, dashed line) moves Ka but has little effect on Amax and no effect on Ki. This effect contrasts with that of calcins (C–E, color lines as indicated), which affect Ka, Ki, and Amax. We also tested the effect of calcins over a wide range of [Ca2+] to determine how they affect the Ca2+-dependent activation and inactivation of RyR1 (Fig. 5, B–E). Specific binding in the absence of calcins (Fig. 5 B, control, black squares) had a threshold for detection at 100 nM [Ca2+] (pCa 7) and was maximal at 10–100 µM [Ca2+] (pCa 5–4). Higher [Ca2+] decreased binding. This dual effect of Ca2+ gave rise to a typical bell-shaped curve, which could be fitted with the equation B = ((Bmax · [Ca2+])/(Ka + [Ca2+]) · (Ki/(Ki + [Ca2+])), where B = specific binding, Bmax is the maximal amount of binding (normalized to 100% in the absence of calcins), and Ka (558 ± 116 nM) and Ki (389 ± 86 µM) represent the apparent affinity of Ca2+ to an activation and inactivation site, respectively. In the presence of a near-saturating concentration of calcin (3 µM for urocalcin and 100 nM for the remainder calcins), the binding curve was also bell-shaped, but it was dramatically augmented in absolute values (Fig. 5 B). In general, the augmentation of [3H]ryanodine binding produced by calcins increased with [Ca2+], the threshold for activation decreased to ≈pCa 8, and the optimal binding shifted to ∼pCa 5. Thus, for all calcins except vejocalcin, Ka was lower than control and ranged between ∼50 and 530 nM [Ca2+], indicating that calcins potentiate the activating effect of Ca2+ on RyR1 (Fig. 5, C–E). Conversely, more [Ca2+] was required to inactivate RyR1 in the presence of calcins; thus, Ki was invariably lower than control, ranging from 0.42 ± 0.05 (vejocalcin) to 5.52 ± 0.64 (urocalcin) mM [Ca2+]. In contrast, 3 mM caffeine increased almost exclusively the ascending limb of the curve, i.e., it “sensitized” RyRs to activating Ca2+ (Fig. 5 B); hence, Ka is exclusively affected with negligible effect on Ki (Fig. 5 C). Thus, in contrast to caffeine, calcins widen both limbs of the Ca2+-dependent [3H]ryanodine binding curve of RyR1, indicating that they act by different mechanisms. If [3H]ryanodine binding is proportional to channel activity (Pessah et al., 1987), the results indicate that calcins, too, bind to a conformational-sensitive state of the channel, requiring Ca2+ to open the channel and expose its binding site (see Discussion). Single channel recordings To further investigate the functional properties of calcins, we reconstituted RyR1 channels in planar lipid bilayers and determined the effect of calcins on single channel gating. Because a full description of the imperacalcin effect on single RyR1 channel gating has already been provided (Tripathy et al., 1998), followed by isolated accounts on maurocalcin (Mosbah et al., 2000), hemicalcin (Shahbazzadeh et al., 2007), and hadrucalcin (Schwartz et al., 2009) effects also on the RyR1 channel, our main goal here was to compare the amplitude of the subconductance state induced by each calcin under identical experimental conditions and to relate this effect to their physicochemical characteristics (Fig. 6). Because the probability of calcin binding to RyR1 is Po and voltage dependent (Tripathy et al., 1998), we conducted experiments at “Po-clamped” conditions (by fixing [Ca2+] at quasi-optimal levels) and positive holding potential (40 mV), which favor the natural (lumen to cytosol) flow of ions through RyRs. Under these conditions, control (no calcin) RyR1 channel recordings showed the characteristic fast-flickering (mean open time 1–3 ms), low Po (0.03–0.15), and high-conductance (≈600 pS) gating, with occasional sojourns to native subconducting states that were barely noticeable in the current amplitude histogram (Fig. 6 A). There were also bursts of high Po, reflecting the characteristic intermodal gating of all RyR isoforms (Rosales et al., 2004). Calcins (100 nM each) all interacted with RyR1 and induced a “signature” event, consisting of long-lasting, reversible, transient subconductance state. However, fractional conductance of the calcin-induced event differed for each peptide. Opicalcin1 (Fig. 6 B) and opicalcin2 (Fig. 6 C) induced subconductance states corresponding to 0.35 and 0.40 of the full-conductance level, respectively. Imperacalcin (Fig. 6 D), maurocalcin (Fig. 6 E), and hadrucalcin (Fig. 6 F) induced substates of fractional value that are in good agreement with those previously reported, i.e., 0.30 (Tripathy et al., 1998), 0.48 (Fajloun et al., 2000), and 0.35 (Schwartz et al., 2009) of full conductance state, respectively. However, the subconducting state induced by hemicalcin (Fig. 6 G) was only 0.20 in our hands, which is lower than the 0.38 value previously reported (Shahbazzadeh et al., 2007). The subconductance position induced by vejocalcin was the highest, with a value of 0.60 (Fig. 6 H). In contrast, urocalcin induced a subconductance state = 0.55, but in line with its low affinity in [3H]ryanodine binding assays (Fig. 5 A), it was difficult to detect it at concentrations similar to the other calcins (100 nM), so we increased its concentration to 1 µM (Fig. 6 I) in this assay. Figure 6. Fractional subconductance value induced by calcins on single RyR1 channels. RyR1 channel from rabbit skeletal muscle SR vesicles was reconstituted in lipid bilayers and recorded and analyzed as described in Materials and methods. (A) Single RyR1 channel in the absence of calcin is presented as control. (B–I) The eight calcins were added separately into the cis (cytosolic) chamber at the indicated concentration. c represents the zero current level when the channel is in the fully closed state, whereas s and o display the subconductance state and full opening of the channel, respectively. Current histograms from 10,000–30,000-ms segments of channel activity are shown at the right side of the traces for each calcin. The number of experiments varied for each calcin. Imperacalcin, maurocalcin, and hadrucalcin: n = 12, 8, and 5 different channels. All other calcins were tested at least two or three times. To better ascertain whether calcins bind exclusively to the open state of the RyR, as postulated before (Tripathy et al., 1998; Lukács et al., 2008), we performed additional single channel experiments using increasing and nonsaturating concentrations of imperacalcin and examined (a) the probability of substate conductance, (b) the mean open and mean closed times of the channel in “bursting mode” (calcin-free), and (c) the mean duration of the subconductance state (calcin dwelling time). Fig. S2 A shows representative single RyR1 channel recordings in the presence of 5, 30, and 100 nM imperacalcin, and Fig. S2 C shows the probability of substate occurrence (Psubstate) as a function of [imperacalcin]. The data can be fitted by a hyperbolic function, and a Lineweaver–Burk plot (not depicted) indicates that the data fit Michaelis–Menten-type kinetics with a maximal Psubstate close to 1.0 and a Kd = 12 nM (in very close agreement to the [3H]ryanodine binding assay of Fig. 5 A). Mean open and mean closed times of the channel in bursting mode or those in the “calcin mode” (obtained by exponential fits of the bursts in the fully conducting or subconducting channel, respectively) were not significantly modified by [imperacalcin], indicating that the kinetics of gating of the calcin-free channel remain impervious to [imperacalcin] as expected and, more importantly, that the dwelling time of imperacalcin is also unaffected by [imperacalcin]. Thus, a simplified scheme emerges, in which the RyR transits between close (C) and open (O) states and calcins bind to the open channel to produce a calcin-modified RyR (Osubstate), as follows: C↔O↔Osubstrate. Because Psubstate follows Michaelis–Menten-type kinetics and is strictly dependent on [imperacalcin], the on reaction O → Osubstate corresponds to kon, the bimolecular process governed by open RyR · [imperacalcin]. Conversely, the off reaction O ← Osubstate is independent of [imperacalcin] and corresponds to a process governed exclusively by the intrinsic properties of calcin binding to the channel. Because the off reaction is the same at low and saturating concentration of imperacalcin, this strongly suggests that, at least within the range of [imperacalcin] tested, the process is unimolecular, that is, only one calcin dissociates from the channel after inducing the substate. Thus, we favor the notion that only one calcin molecule is required to exert the “signature effect” of these group of peptides. We also analyzed single channel recordings and visually inspected each of 48 calcin-induced subconductance states in the channel of Fig. S2 A, at 100 nM imperacalcin. We scored 1 for each channel transition from discretely open to subconducting (open, on) and the same for a transition from subconducting to discretely open again (open, off; examples marked by asterisks in Fig. S2 B) and 0 for subconductance transitions from and to closed channel. We then plotted these events cumulatively as a function of event number. A perfect straight line (correlation coefficient = 1.0, dashed line) is expected if each and all of the subconductance transitions occur from and to an open channel, whereas a flat line with linear regression coefficient = 0 would be expected if none of the subconductance transitions occurs from and to open channel. Results (Fig. S2 D) clearly show that there is a strong dependence on a channel being open to initiate and terminate a calcin-induced substate. Most, but not all (open, on = 39/48 and open, off = 37/48), of the calcin-induced transitions occurred from and to open channel. An insufficient recording bandwidth may be a potential source of false negatives (missed open events before or after subconductance), and it is possible that the channel did open before or after subconductances, but the opening was so fast that we could not record it. Ca2+ release from heavy SR We also tested the capacity of calcins to induce Ca2+ release from rabbit skeletal heavy SR vesicles (Fig. 7). SR vesicles (1 µg/ml) were loaded with two consecutive additions of 50 µM Ca2+ followed by 25 µM Ca2+ thrice. Typically, each bolus addition of Ca2+ was actively taken up by the SR vesicles, and by the fifth bolus, extravesicular [Ca2+] increased to ≈30 µM, indicating that the vesicles in the system were loaded to full capacity (Fig. 7 A). We then added each calcin and tested their ability to elicit Ca2+ release. Fig. 7 A shows a typical trace in which cumulative addition of imperacalcin elicits gradual Ca2+ release from SR vesicles. Addition of 10 nM imperacalcin produced initially modest Ca2+ release, but a second bolus that elevated imperacalcin to 20 nM elicited sharp Ca2+ release, and further imperacalcin additions yielded little additional release (Fig. 7 A). Addition of the Ca2+ ionophore A23187 (5 µM) to the reaction mixture indicated that imperacalcin released a fraction, only, of the total amount of Ca2+ trapped into the SR vesicles (≈70%; Fig. 7 C). Fig. 7 B show that this capacity of imperacalcin to release Ca2+ in a sharp, stepwise manner once a critical concentration has been reached, followed by incomplete emptying of the SR Ca2+ load at higher concentration, is a general property of all calcins (except urocalcin, see Fig. 7 C). Not surprisingly, the threshold for sharp Ca2+ release was the lowest for opicalcin1 (2.1 ± 0.1 nM), followed by maurocalcin (6.6 ± 0.5 nM), imperacalcin (11.7 ± 0.6 nM), hadrucalcin (11.8 ± 0.4 nM), vejocalcin (31.0 ± 0.6 nM), opicalcin2 (64.2 ± 5.5 nM), hemicalcin (68 ± 1.7 nM), and urocalcin (unable to release Ca2+ at concentration up to 1 µM). Thus, with the notable exception of opicalcin2, calcins elicit Ca2+ release with roughly the same ranking order exhibited in [3H]ryanodine binding experiments. Fig. 7 C shows that the total Ca2+ released by 100 nM calcin varied between 45% (hemicalcin) and 67% (opicalcin1). Figure 7. Calcin-induced Ca2+ release from skeletal heavy SR vesicles. (A) Typical trace of Ca2+ release by calcins. A concentration of 20 nM imperacalcin elicited abrupt Ca2+ release from SR vesicles, and further additions had incremental effects only. The fraction of calcin-induced Ca2+ release was calculated by adding the Ca2+ ionophore A23187 (5 µM). (B) Except urocalcin, all calcins elicited abrupt Ca2+ release within 100 nM (n = 3–4). (C) At 100 nM, all calcins except urocalcin elicited Ca2+ release to ∼60% of total Ca2+ load from heavy SR. Mean ± SEM is shown. **, P < 0.01; calcins versus A23187, t test. DISCUSSION In this study, we present all native calcins known to date, including one novel (vejocalcin) and three previously uncharacterized (opicalcin1, opicalcin2, and urocalcin) peptides, and make a systematic, side by side comparison of their structural and functional attributes using an array of experimental ([3H]ryanodine binding, Ca2+ release, and single RyR1 channel) assays, as well as bioinformatics analysis and three-dimensional modeling. Calcins are a relatively novel family of small (∼3.7–4.2 kD), highly basic (pI = 9.3–10.1), compact peptides from scorpion venoms that target RyRs with high affinity and exquisite selectivity (no other targets known to date). Their remarkable stability is due to the fact that they fold along an evolutionarily conserved structural motif consisting of a triple-stranded, antiparallel β sheet stabilized by a ring formed of three disulfide bonds, known as an ICK motif. The ICK motif has been found in snail and spider toxins blockers of Ca2+ channels (Zhu et al., 2003), but not in other classes of scorpion toxins known to date (Na+ channel modifiers, K+ channel blockers, antimicrobial peptides, defensins), underscoring the unique arrangement of calcins among scorpion peptides. Primary sequence alignment and evolutionary analysis show a very close relationship among calcins, yet their natural variability yields peptides with binding affinity spanning ≈3 orders of magnitude, highlighting the importance of discrete domains in their interaction with RyR1 channels. We also show that the defining functional characteristic of all calcins is their capacity to stabilize RyR1 openings in a long-lasting subconducting state. This effect is nearly analogous to that of ryanodine, but unlike ryanodine, calcins rapidly bind to RyR1 channels and freely dissociate from their binding site (reversible effect), displaying a dose- and sequence-variable effect. Finally, although not assessed in this study, others (Estève et al., 2005; Boisseau et al., 2006) and we (Schwartz et al., 2009; Gurrola et al., 2010) have shown that, despite their highly ionized nature, calcins may penetrate cellular membranes, reach their intended target, and elicit Ca2+ release with several degrees of potency. Thus, this is a comprehensive study of calcins as a group of RyR-specific, cell-penetrating peptides (CPPs) that mobilize Ca2+ from intracellular stores with high dynamic range. Ryanodine and calcins may bind independently and simultaneously to RyRs It is well established that ryanodine binds to a conformationally sensitive (open) state of the RyR, so that the simplified Scheme 1, where C is closed, O is open, and O·Ry is the ryanodine-bound and subconducting state of the RyR, correctly describes the main kinetic steps of the reaction. A pertinent question here, however, is whether [3H]ryanodine may properly monitor the binding of calcins given that they appear to bind to the same binding site in the RyR because of their highly similar effect. Ryanodine and calcins do bind to independent and nonoverlapping binding sites in the RyR, as indicated by the fact that addition of imperacalcin to ryanodine-modified RyR produces a calcin-induced substate of the ryanodine-modified substate (Tripathy et al., 1998), and hence, [3H]ryanodine may suitably report the binding of calcin. (Scheme 1) By using in our assays a concentration of [3H]ryanodine (7 nM) close to the Kd of the RyR–Ry complex at optimal [Ca2+], we ensure that there is “room” for activators and inhibitors to exert their effect on the binding of [3H]ryanodine. In principle, then, because [3H]ryanodine binds at half of Bmax, there should be approximately a twofold (100%) maximal increment of binding by a given agonist. Why, then, do calcins increase maximal binding to 300–500% above control? The most likely explanation is that they increase the “openness” of the channel above and beyond what Ca2+ alone is capable of doing. The calcin-induced substate of long duration promotes more effectively the binding of [3H]ryanodine than the frequent but brief openings (flickering) induced by Ca2+ alone. Thus, by requiring an open channel to bind and simultaneously promoting opening of the channel, calcins increase the dynamic range of the [3H]ryanodine binding assay, and the latter efficiently reports the binding of the calcins. Scheme 2, therefore, describes the binding of ryanodine (Ry) and calcin (Cn) to open RyR (O) in simplified terms, where O·Cn is the calcin-bound open RyR and O·Cn·Ry is the calcin- and ryanodine-bound open RyR. Because of the slow association rate of ryanodine to RyRs, transitions from O to doubly occupied O·Cn·Ry probably go through O·Cn first, but Scheme 2 depicts the complex at equilibrium. Simultaneous occupation and additive effects of ryanodine and calcins are thus possible, and our [3H]ryanodine binding assays here suitably reported this combined interaction. However, a caveat is that by using [Ry] at a concentration equal to its Kd, we impose the equality P(O) = P(O·Ry) (where P(O) and P(O·Ry) are occupancies or probabilities of the corresponding states). Thus, the value of bound Ry relative to a well-determined Bmax should then be a direct measure of P(O). This could be different from the probability of opening as observed in reconstituted RyR channels and constitutes a limitation of our binding approach. (Scheme 2) Primary sequence and natural variations Previous studies have addressed the effect of artificial mutations on calcins’ properties, i.e., alanine scanning on imperacalcin (Lee et al., 2004) and selected monosubstitutions on maurocalcin (Mabrouk et al., 2007). Here we examined nature’s own experimentation that includes subtle (and others seemingly drastic) mono-, di-, and multi-variations from an apparently common pattern of design. Comparison of calcins’ primary sequence permitted discernment of some of their most salient structural attributes and unique domains among scorpion peptides (Fig. 2). First, six highly conserved cysteines serve as the framework to flawlessly align calcins, with an identical number of amino acids interspaced between cysteines (no need to introduce gaps or delete residues to maximize homology). 1H-NMR of maurocalcin (Mosbah et al., 2000) and imperacalcin (Lee et al., 2004) confirmed that these cysteines are oxidized in the mature peptide and form disulfide bonds with the precise pairing Cys3-Cys17, Cys10-Cys21, and Cys16-Cys32. Maintenance of this disulfide-bonded core is essential not only to support the ICK motif, the signature folding of this family of scorpion peptides, but improperly paired or linear (disulfide-less) peptides are pharmacologically inert (Zamudio et al., 1997; Lee et al., 2004; Ram et al., 2008), indicating that the disulfide bonds buttress the tertiary structure that forms the calcins’ active site. Second, an abundance of basic amino acids is another remarkable feature of calcins (Figs. 2 A and 4). All peptides in this group are composed of at least 27% (vejocalcin) and as much as 39% (urocalcin) of positively charged amino acids. In the linear sequence, these basic residues (bolded) appear in three main clusters, namely, 8KRCR11 (cluster 1), 19KKCKRR24 (cluster 2), and 30KRCR33 (cluster 3). Notice how cysteines (C) “break” the continuity of basic residues in the clusters, perhaps not fortuitously, because the disulfide bonds that they form will “pull” together these residues in the tertiary structure (Fig. S1), contributing to expand even more the size of the clusters. Except for relatively rare exceptions, clusters 2 and 3 are remarkably conserved, stressing the prominent hierarchy of these motifs, whereas cluster 1 is interrupted by Leu or Gln in five out of eight calcins, also suggesting a lesser structural rank for this cluster. Indeed, in the first mutational study performed on calcins, we found that the mutation R23E, at the heart of cluster 2, completely abolished the capacity of imperacalcin to stimulate [3H]ryanodine binding and Ca2+ release and to induce subconductance states in RyR1 (Gurrola et al., 1999), an effect that was also observed with an even milder mutation (R23A) in imperacalcin (Lee et al., 2004) and maurocalcin (Chen et al., 2003; Lukács et al., 2008). In contrast, mutation K8E in imperacalcin had only modest effects (Gurrola et al., 1999), supporting the notion that cluster 1 plays a lesser role in the calcin–RyR1 interaction. Gurrola et al. (1999) also proposed that cluster 2 in extension to Thr26 (19KKCKRRGT26) was structurally similar to a segment of peptide A of the α1 subunit of the DHPR1 that was presumed to interact directly with RyR1 in the excitation–contraction coupling of skeletal muscle (Saiki et al., 1999) and showed that the mutation T26A decreased the affinity of imperacalcin by 10-fold. Although the participation of peptide A in excitation–contraction coupling is debated today, several studies have found a mechanism of activation of RyR1 that is common to peptide A and calcins (specifically, imperacalcin [Dulhunty et al., 2004] and maurocalcin [Chen et al., 2003]), again underscoring the importance of cluster 2 as integral part of calcins’ active site. The latter most likely encompasses Thr26 also, as in the present study we demonstrate that opicalcin1 and opicalcin2, which differ only by one amino acid (T26A), display a 10-fold difference in affinity. However, it is unlikely that the calcins’ active site is wholly contained in cluster 2 and Thr26, as the three-dimensional model of calcins clearly shows that cluster 3 intertwines with cluster 2 to form a “mega-cluster” of basic residues that crowds the lower end of the frontal face of calcins and is largely responsible for its highly positive charge (Fig. 4). In support of cluster 3 participation in the calcin–RyR1 interaction, Ala substitutions of K30, R31, and R33 drastically decreased the affinity of imperacalcin for RyR1 (from 28 to >1,000,000 nM [Dulhunty et al., 2004]). Again, in sharp contrast, cluster 1 residues populate the upper end of the molecule and are more prominent on the flank of the dorsal face (Fig. 4), apparently far from the residues of clusters 2 and 3 whose substitution dramatically decreases the affinity of calcins. Overall, then, clusters 2 and 3 appear of primordial importance for calcin interaction with RyRs by means of their coalescing into a single mega-cluster of positive charges in one face of the peptide. Because electrostatic interactions govern the calcin–RyR1 interaction (Tripathy et al., 1998), it is likely that this mega-cluster forms a cloud of positive charges that interacts with negatively charged residues in RyRs. Thus, in general terms, the calcin sequence can be roughly split in half to yield a highly variable N-terminal segment (1G-N14) and a remarkably conserved C-terminal segment (15D-R33). Previous studies on imperacalcin (Lee et al., 2004) and maurocalcin (Mabrouk et al., 2007) have shown that, as a general rule of thumb, mutations affecting positively charged residues dramatically decrease or even abolish [3H]ryanodine binding, whereas mutations affecting negatively charged residues slightly increase the affinity of these calcins. Mutations of neutral residues can increase, decrease, or have no significant effect (Lee et al., 2004; Mabrouk et al., 2007). However, examination of the array of calcins presented here reveals more nuances and complexity in this general rule. For example, let’s compare the primary sequences of opicalcin1 and urocalcin (Fig. 2), the calcins with the highest (∼0.3 nM) and the lowest (∼300 nM) affinity for RyR1 (Fig. 5), to attempt rationalization of their wildly diverse pharmacological potency. Urocalcin presents the natural substitutions R9L (cluster 1) and K20S (cluster 2), which according to the general rule and our discussion above should separately have modest and major effects, respectively, on binding affinity, and their combined presence might explain the drastic reduction of urocalcin’s affinity. However, similar substitutions are present in vejocalcin (R9L in cluster 1 and K22S in cluster 2), and yet its Kd is 3.7 ± 0.4 nM, which is as high as that of opicalcin2. In fact, a third substitution in cluster 3 (K30Q) should theoretically decrease the affinity of vejocalcin even more than that of urocalcin. Obviously, simply following the rule of thumb will not predict the degree of affinity change brought about by a set of mutations, even if they occur at domains that are presumably part of the active site. Instead, it seems that a more comprehensive approach must be followed to ascertain the role of substitutions that appear innocuous in the calcins’ landscape. What is peculiar about urocalcin (compared with opicalcin1) is not only the substitutions of critical residues in clusters 1 and 2 but also the presence of four additional basic residues in the N-terminal half of the molecule (Fig. 2). Intuitively, these extra positive charges should have resulted in a superior calcin following the rule of thumb, but analysis of their three-dimensional arrangement (Fig. 4 G) shows that they all protrude in the upper end of the frontal side of the molecule and intertwine with hydrophobic residues to dissipate a hydrophobic core that dominates this part of most calcins. Hence, an integral approach that considers variations in domains established to play primordial roles in receptor recognition (clusters 2 and 3) plus evaluation of the calcin’s overall amphipathicity (integrity of its hydrophobic and hydrophilic cores) are among the structural features that should be considered in the rational design of competent calcins. DM of calcins In addition to the signature ICK motif folding promoted by the three highly conserved disulfide bridges, a marked anisotropy of electrostatic charge distribution, where most of the positively charged residues are segregated in one end of the molecule, appears as a prevalent feature in the calcin family of peptides. This coalescence of basic residues presents a large functional surface area to interact with the RyRs (Mosbah et al., 2000; Lee et al., 2004) and generates a discrete DM. In principle, DM could be related to binding affinity because a larger DM could imply larger surface area to interact with RyRs, but other possibilities could undermine this relationship. In this study, we found that opicalcin1, opicalcin2, and imperacalcin have similar DM values, but the affinity of opicalcin1 is 10.7-fold and 29-fold higher than that of opicalcin2 and imperacalcin, respectively. Also, vejocalcin, which lacks the positively charged residues K22 and K30 (K22S and K30Q, instead) in the predicted functional face of calcins, has the smallest DM (40 D) but similar activity as opicalcin2, which presents the largest DM (175 D). Hadrucalcin, whose relatively small DM (65 D) is mainly the result of an increase of positively charged residues in the nonfunctional face, still has strong affinity for RyR1 (14.8 ± 1.9 nM), whereas urocalcin, which also scatters its positively charged residues to the upper end of the frontal face of the molecule and thus decreases its DM value (102 D), has the weakest affinity for RyR1 (376 ± 45 nM), over 1,000-fold lower than that of opicalcin1 and 25-fold lower than that of hadrucalcin. Thus, there appears to be no correlation between DM and affinity, and a plot of DM values versus Kd (as determined in [3H]ryanodine binding assays) yields a correlation coefficient r2 = 0.15 (Fig. 8 A), indicating that DMs or electrical amphiphilicity is not associated with the interaction with RyRs. In support of this conclusion, d-maurocalcin, a chiral analogue of maurocalcin composed of d–amino acids (instead of l–amino acids) that has identical charge distribution as the parent maurocalcin, completely loses the ability to stimulate [3H]ryanodine binding and Ca2+ release (Poillot et al., 2010). Thus, simply characterizing the calcins by charge anisotropy or electrical amphiphilicity is insufficient to formulate a model of calcin–RyR1 interaction, and the spatial conformation of calcins also plays as critical a role as the aggregation of positively charged residues in the calcin–RyR interacting process. Figure 8. DM, fractional conductance, and Ca2+ release appear unrelated to the Kd of calcins. (A) Plot of DM versus Kd of [3H]ryanodine binding. The DMs have a poor correlation with the binding affinity (r2 = 0.15). (B) Subconductance versus Kd of [3H]ryanodine binding. The position of subconductance induced by calcin is unrelated to the Kd of [3H]ryanodine binding (r2 = 0.07). (C) Ca2+ release versus [3H]ryanodine binding. No correlation is seen between calcin concentration that induces Ca2+ release and Kd of [3H]ryanodine binding (r2 = 0.04). Single channel effect and mechanism Several calcins have been tested on single RyRs, but a mechanistic understanding of their common biophysical effect has been mainly derived from experiments with imperacalcin (Tripathy et al., 1998) and maurocalcin (Lukács et al., 2008). When added to the cytosolic side of the channel, imperacalcin induces voltage- and concentration-dependent subconductance states in both skeletal and cardiac RyRs. Analysis of voltage and concentration dependence and kinetics of substate formation suggests that induction of subconductance states corresponds to reversible, voltage-dependent binding and unbinding of imperacalcin at a single site located at ∼23% of the voltage drop from the cytosolic side (Tripathy et al., 1998). The mechanism of substate formation by imperacalcin is not entirely clear, but several scenarios are applicable to calcins in general: steric mechanisms (i.e., calcin binding to residues within the ion conduction pathway to restrict ion flow), allosteric mechanisms (i.e., calcin binding to a site distant to the conduction pathway and altering the shape or charge of the RyR’s conduction pathway), and a combination thereof. However, because imperacalcin accesses its binding site by entering the channel through a cytosolically accessible opening (Tripathy et al., 1998), we favor a mechanism whereby calcin reaches its binding site located in the SR membrane–adjacent ion conduction pathway by entering the channel through a vestibule’s pore with a diameter ≥2.5 nm (the diameter of globular imperacalcin). Our data do not allow us to discard any of the mechanisms above, but interestingly, recent near-atomic resolution of the three-dimensional structure of RyR1 reveals an opening of ∼5 nm in the center of the channel, most precisely in the convergence of the N-terminal domains that form part of the most external layer of the channel facing the t-tubules (Efremov et al., 2015; Yan et al., 2015; Zalk et al., 2015), thus evidencing the existence of a wide vestibule in the cytosolic side of the RyR1. Accordingly, we favor the notion that globular calcins, all with surface diameter ≤3 nm, induce their characteristic subconductance state by entering the channel through this cytosolic opening and accessing their binding site deep in the core of the channel. This possibility needs to be tested experimentally. In an alternative scenario, Samsó et al. (1999), using cryo-electron microscopy and three-dimensional single-particle image analysis, found that imperacalcin binds to a cytoplasmic crevice in each RyR1 subunit (four binding sites/channel tetramer) located near the calmodulin-binding site and far (<11 nm) from the center of the transmembrane region of the channel (Samsó et al., 1999), supporting an allosteric mechanism of action of imperacalcin as opposed to the mechanism of direct positioning of the peptide within the ion conduction pathway. It is worth noting, however, that the cryo-electron microscopy experiments of Samsó et al. (1999) required imperacalcin covalently bound to biotin, which subsequently bound to streptavidin-coated gold particles to facilitate localization of this ligand in single channel images and that these additional tags increased the surface area of imperacalcin, probably beyond the maximum allowed to enter the vestibule of the channel and reach the binding site described above. In other words, the imperacalcin–biotin–streptavidin complex could have been too bulky to have accessed the binding site within the ion conduction pathway that we believe induces the signature subconductance state. Instead, we favor the notion that imperacalcin–biotin–streptavidin likely bound to a secondary and external binding site (the crevice described in the previous paragraph) that is also accessed by peptide A (Gurrola et al., 1999; Chen et al., 2003; Dulhunty et al., 2004; Altafaj et al., 2005). Only by binding to external sites of the RyR1 (one for each monomer) does the idea that imperacalcin and peptide A mimic a segment of the DHPR1 that interacts with the RyR1 during e-c coupling appear to make sense. In this scenario, occupancy of these external binding sites could activate the RyR1, but induction of the characteristic subconductance state described here appears unlikely because the latter follows the rules of a bimolecular reaction (one calcin–one channel) and requires proximity to the SR-adjacent voltage drop of the channel (Tripathy et al., 1998). Regardless of their precise mechanism of action, the reversible, concentration- and voltage-dependent, long-lasting subconductance state constitutes the signature effect of calcins on RyRs, and this substate has been measured up to now for imperacalcin (Tripathy et al., 1998; Gurrola et al., 1999), maurocalcin (Fajloun et al., 2000; Chen et al., 2003), hemicalcin (Shahbazzadeh et al., 2007), and hadrucalcin (Schwartz et al., 2009), with fractional values of the full conductance state corresponding to 0.35, 0.48, 0.38, and 0.35, respectively. Here we report for the first time the fractional conductance induced by the remaining four calcins opicalcin1 (0.35), opicalcin2 (0.40), vejocalcin (0.60), and urocalcin (0.55; Fig. 5). Furthermore, in our experiments, the subconductance induced by hemicalcin was only 0.2 of full conductance, lower than that reported previously (Shahbazzadeh et al., 2007) and most likely caused by the rectifying conductance induced by the calcin (i.e., the proportion of the substate is lower at higher positive voltages [Tripathy et al., 1998]). Considering the high structural similarity among calcins, it is likely that they bind to RyR1 at the same site but engage the receptive amino acids with varying degrees of affinity because of the charge distribution and structural conformation that is unique to each calcin. An interesting example is urocalcin, whose subconductance is induced only at the cumulative dose of 1 µM or higher, with low frequency and short duration (<500 ms), indicating that urocalcin loosely associates to, and readily dissociates from, the calcin site in RyR1 (Fig. 6). We also correlated the subconductance value with the binding activity and the ability to stimulate Ca2+ release; however, no relationship is seen (Fig. 8 B), suggesting that the capacity of calcins to stimulate [3H]ryanodine binding and Ca2+ release may be associated with the dwell time of the calcin in the RyR1 (mean duration of subconductance) rather than the fractional value of the substate, but more experiments are needed to confirm this. Ca2+ release from heavy SR Calcins induce Ca2+ release at concentrations that are crudely correlated with their [3H]ryanodine binding affinity, with opicalcin1 and urocalcin displaying the strongest and weakest capacity, respectively, to stimulate both [3H]ryanodine binding and Ca2+ release. But the correlation is not perfect for most calcins in between; for example, hemicalcin and maurocalcin show deviations in correlation, and opicalcin2 displays strong binding affinity but weaker ability to stimulate Ca2+ release (Fig. 8 C). This may be explained by the fact that Ca2+-induced Ca2+ release (CICR) is a nonlinear process, with inherent self-reinforcing features and sensitive to total SR Ca2+ load (Stern and Cheng, 2004). Hence, once a calcin “traps” a critical number of RyRs in its specific subconducting state (a step dependent on calcin affinity), it will induce Ca2+ release that in turn will induce more Ca2+ release (a step independent of calcin affinity). Thus, the calcin-induced fractional Ca2+ release (Ca2+ released/total SR Ca2+ content) will depend not only on the affinity of calcins for RyRs, but also in the self-amplifying nature of CICR and the amount of Ca2+ load in the SR. Although the relationship affinity-fractional Ca2+ release held as expected for the highest and weakest calcin, the assay was not sensitive enough to discriminate flawlessly for calcins of medium affinity. An interesting phenomenon that is not seen with other agonists of RyRs (i.e., caffeine or 4-chloro-m-cresol) is that calcin-induced Ca2+ release will consistently be a fraction of the total SR Ca2+ load; that is, a residual amount of Ca2+ will remain in the SR vesicles even at saturating calcin concentration (Fig. 7). The mechanism underlying this phenomenon is unclear but might be related to the property of calcins to induce a partial rather than a complete opening of RyRs, so that even when all RyRs are calcin-occupied, the total Ca2+ flow out of the SR is limited by the subconducting value and may be in balance with Ca2+ reuptake by the Ca2+ pump. Summary In summary, we have characterized by several functional assays the most salient structural features of all calcins known to date. Natural variations in amino acid sequence endow these peptides with a range of RyR affinity that spans ∼3 orders of magnitude and variable capacity to stimulate [3H]ryanodine binding and Ca2+ release. Despite their variations in amino acid sequence, all calcins fold along an ICK motif that maintains a globular and highly compact structure. The relatively small surface area of calcins suggests that their structure is already minimized, with all residues contributing at least partly to RyR1 binding and/or membrane permeation. Still, analysis of natural calcin variations recognizes prominent roles for discrete structural domains, with two clusters of basic residues in the carboxyl end (19KKCKRR24 and 30KRCR33) appearing essential for RyR1 pharmacology. At the single RyR1 channel level, calcins, unlike the effect of ryanodine, induce the appearance of a long-lasting but reversible subconductance state. This subconductance state is dose dependent and varies in amplitude for each calcin. Thus, the calcin family of peptides offers a diversified set of RyR ligands with the capacity to modulate RyRs in situ and with high dynamic range and potency. Supplementary Material Supplemental Materials ACKNOWLEDGMENTS We are grateful to reviewer 1 for suggesting a potential mechanism of interaction of calcin with the RyR. This study was supported by National Institutes of Health grants R01-HL120108 and R01-HL055438 (to H.H. Valdivia) and Dirección General Asuntos del Personal Académico, Universidad Nacional Autónoma de México (DGAP, UNAM) grant IN200113 (to L.D. Possani). The authors declare no competing financial interests. Eduardo Ríos served as editor. Abbreviations used in this paper: DM dipole moment ED evolutionary distance ICK inhibitor cysteine knot PSA polar surface area RyR ryanodine receptor TFA trifluoroacetic acid ==== Refs Abramson, J.J., E. Buck, G. Salama, J.E. Casida, and I.N. Pessah. 1988. Mechanism of anthraquinone-induced calcium release from skeletal muscle sarcoplasmic reticulum. J. Biol. Chem. 263 :18750–18758.3198599 Altafaj, X., W. Cheng, E. Estève, J. Urbani, D. Grunwald, J.M. Sabatier, R. Coronado, M. De Waard, and M. Ronjat. 2005. Maurocalcine and domain A of the II-III loop of the dihydropyridine receptor Cav 1.1 subunit share common binding sites on the skeletal ryanodine receptor. J. Biol. Chem. 280 :4013–4016. 10.1074/jbc.C400433200 15591063 Ather, S., J.L. Respress, N. Li, and X.H. Wehrens. 2013. Alterations in ryanodine receptors and related proteins in heart failure. Biochim. Biophys. Acta. 1832 :2425–2431. 10.1016/j.bbadis.2013.06.008 23770282 Benkusky, N.A., E.F. Farrell, and H.H. Valdivia. 2004. Ryanodine receptor channelopathies. Biochem. Biophys. Res. Commun. 322 :1280–1285. 10.1016/j.bbrc.2004.08.033 15336975 Bers, D.M. 2002. Cardiac excitation-contraction coupling. Nature. 415 :198–205. 10.1038/415198a 11805843 Bers, D.M. 2004. Macromolecular complexes regulating cardiac ryanodine receptor function. J. Mol. Cell. Cardiol. 37 :417–429. 10.1016/j.yjmcc.2004.05.026 15276012 Boisseau, S., K. Mabrouk, N. Ram, N. Garmy, V. Collin, A. Tadmouri, M. Mikati, J.M. Sabatier, M. Ronjat, J. Fantini, and M. De Waard. 2006. Cell penetration properties of maurocalcine, a natural venom peptide active on the intracellular ryanodine receptor. Biochim. Biophys. Acta. 1758 :308–319. 10.1016/j.bbamem.2006.02.007 16545341 Chen, L., E. Estève, J.M. Sabatier, M. Ronjat, M. De Waard, P.D. Allen, and I.N. Pessah. 2003. Maurocalcine and peptide A stabilize distinct subconductance states of ryanodine receptor type 1, revealing a proportional gating mechanism. J. Biol. Chem. 278 :16095–16106. 10.1074/jbc.M209501200 12586831 Curran, J., M.J. Hinton, E. Ríos, D.M. Bers, and T.R. Shannon. 2007. β-Adrenergic enhancement of sarcoplasmic reticulum calcium leak in cardiac myocytes is mediated by calcium/calmodulin-dependent protein kinase. Circ. Res. 100 :391–398. 10.1161/01.RES.0000258172.74570.e6 17234966 Dulhunty, A.F., S.M. Curtis, S. Watson, L. Cengia, and M.G. Casarotto. 2004. Multiple actions of imperatoxin A on ryanodine receptors: interactions with the II-III loop “A” fragment. J. Biol. Chem. 279 :11853–11862. 10.1074/jbc.M310466200 14699105 Dyck, J.D., T.E. David, B. Burke, G.D. Webb, M.A. Henderson, and R.S. Fowler. 1987. Management of coronary artery disease in Hutchinson-Gilford syndrome. J. Pediatr. 111 :407–410. 10.1016/S0022-3476(87)80466-3 2957478 Efremov, R.G., A. Leitner, R. Aebersold, and S. Raunser. 2015. Architecture and conformational switch mechanism of the ryanodine receptor. Nature. 517 :39–43. 10.1038/nature13916 25470059 El-Hayek, R., A.J. Lokuta, C. Arévalo, and H.H. Valdivia. 1995. Peptide probe of ryanodine receptor function. Imperatoxin A, a peptide from the venom of the scorpion Pandinus imperator, selectively activates skeletal-type ryanodine receptor isoforms. J. Biol. Chem. 270 :28696–28704. 10.1074/jbc.270.48.28696 7499390 Estève, E., K. Mabrouk, A. Dupuis, S. Smida-Rezgui, X. Altafaj, D. Grunwald, J.C. Platel, N. Andreotti, I. Marty, J.M. Sabatier, 2005. Transduction of the scorpion toxin maurocalcine into cells. Evidence that the toxin crosses the plasma membrane. J. Biol. Chem. 280 :12833–12839. 10.1074/jbc.M412521200 15653689 Fajloun, Z., R. Kharrat, L. Chen, C. Lecomte, E. Di Luccio, D. Bichet, M. El Ayeb, H. Rochat, P.D. Allen, I.N. Pessah, 2000. Chemical synthesis and characterization of maurocalcine, a scorpion toxin that activates Ca2+ release channel/ryanodine receptors. FEBS Lett. 469 :179–185. 10.1016/S0014-5793(00)01239-4 10713267 Fill, M., and J.A. Copello. 2002. Ryanodine receptor calcium release channels. Physiol. Rev. 82 :893–922. 10.1152/physrev.00013.2002 12270947 Gurrola, G.B., C. Arévalo, R. Sreekumar, A.J. Lokuta, J.W. Walker, and H.H. Valdivia. 1999. Activation of ryanodine receptors by imperatoxin A and a peptide segment of the II-III loop of the dihydropyridine receptor. J. Biol. Chem. 274 :7879–7886. 10.1074/jbc.274.12.7879 10075681 Gurrola, G.B., E.M. Capes, F.Z. Zamudio, L.D. Possani, and H.H. Valdivia. 2010. Imperatoxin A, a cell-penetrating peptide from scorpion venom, as a probe of Ca-release channels/ryanodine receptors. Pharmaceuticals (Basel). 3 :1093–1107. 10.3390/ph3041093 20668646 Hakamata, Y., J. Nakai, H. Takeshima, and K. Imoto. 1992. Primary structure and distribution of a novel ryanodine receptor/calcium release channel from rabbit brain. FEBS Lett. 312 :229–235. 10.1016/0014-5793(92)80941-9 1330694 Kong, H., P.P. Jones, A. Koop, L. Zhang, H.J. Duff, and S.R. Chen. 2008. Caffeine induces Ca2+ release by reducing the threshold for luminal Ca2+ activation of the ryanodine receptor. Biochem. J. 414 :441–452. 10.1042/BJ20080489 18518861 Kumar, S. 1996. A stepwise algorithm for finding minimum evolution trees. Mol. Biol. Evol. 13 :584–593. 10.1093/oxfordjournals.molbev.a025618 8882501 Lai, F.A., H.P. Erickson, E. Rousseau, Q.Y. Liu, and G. Meissner. 1988. Purification and reconstitution of the calcium release channel from skeletal muscle. Nature. 331 :315–319. 10.1038/331315a0 2448641 Lee, C.W., E.H. Lee, K. Takeuchi, H. Takahashi, I. Shimada, K. Sato, S.Y. Shin, D.H. Kim, and J.I. Kim. 2004. Molecular basis of the high-affinity activation of type 1 ryanodine receptors by imperatoxin A. Biochem. J. 377 :385–394. 10.1042/bj20031192 14535845 Lukács, B., M. Sztretye, J. Almássy, S. Sárközi, B. Dienes, K. Mabrouk, C. Simut, L. Szabó, P. Szentesi, M. De Waard, 2008. Charged surface area of maurocalcine determines its interaction with the skeletal ryanodine receptor. Biophys. J. 95 :3497–3509. 10.1529/biophysj.107.120840 18621823 Lukyanenko, V., I. Györke, S. Subramanian, A. Smirnov, T.F. Wiesner, and S. Györke. 2000. Inhibition of Ca2+ sparks by ruthenium red in permeabilized rat ventricular myocytes. Biophys. J. 79 :1273–1284. 10.1016/S0006-3495(00)76381-8 10968991 Luna-Ramírez, K., V. Quintero-Hernández, L. Vargas-Jaimes, C.V. Batista, K.D. Winkel, and L.D. Possani. 2013. Characterization of the venom from the Australian scorpion Urodacus yaschenkoi: Molecular mass analysis of components, cDNA sequences and peptides with antimicrobial activity. Toxicon. 63 :44–54. 10.1016/j.toxicon.2012.11.017 23182832 Mabrouk, K., N. Ram, S. Boisseau, F. Strappazzon, A. Rehaim, R. Sadoul, H. Darbon, M. Ronjat, and M. De Waard. 2007. Critical amino acid residues of maurocalcine involved in pharmacology, lipid interaction and cell penetration. Biochim. Biophys. Acta. 1768 :2528–2540. 10.1016/j.bbamem.2007.06.030 17888395 Meissner, G. 1984. Adenine nucleotide stimulation of Ca2+-induced Ca2+ release in sarcoplasmic reticulum. J. Biol. Chem. 259 :2365–2374.6698971 Mosbah, A., R. Kharrat, Z. Fajloun, J.G. Renisio, E. Blanc, J.M. Sabatier, M. El Ayeb, and H. Darbon. 2000. A new fold in the scorpion toxin family, associated with an activity on a ryanodine-sensitive calcium channel. Proteins. 40 :436–442. 10.1002/1097-0134(20000815)40:3<436::AID-PROT90>3.0.CO;2-9 10861934 Nabhani, T., X. Zhu, I. Simeoni, V. Sorrentino, H.H. Valdivia, and J. García. 2002. Imperatoxin a enhances Ca2+ release in developing skeletal muscle containing ryanodine receptor type 3. Biophys. J. 82 :1319–1328. 10.1016/S0006-3495(02)75487-8 11867448 Nakai, J., T. Imagawa, Y. Hakamat, M. Shigekawa, H. Takeshima, and S. Numa. 1990. Primary structure and functional expression from cDNA of the cardiac ryanodine receptor/calcium release channel. FEBS Lett. 271 :169–177. 10.1016/0014-5793(90)80399-4 2226801 Pessah, I.N., R.A. Stambuk, and J.E. Casida. 1987. Ca2+-activated ryanodine binding: mechanisms of sensitivity and intensity modulation by Mg2+, caffeine, and adenine nucleotides. Mol. Pharmacol. 31 :232–238.2436032 Poillot, C., K. Dridi, H. Bichraoui, J. Pêcher, S. Alphonse, B. Douzi, M. Ronjat, H. Darbon, and M. De Waard. 2010. d-Maurocalcine, a pharmacologically inert efficient cell-penetrating peptide analogue. J. Biol. Chem. 285 :34168–34180. 10.1074/jbc.M110.104919 20610396 Ram, N., N. Weiss, I. Texier-Nogues, S. Aroui, N. Andreotti, F. Pirollet, M. Ronjat, J.M. Sabatier, H. Darbon, V. Jacquemond, and M. De Waard. 2008. Design of a disulfide-less, pharmacologically inert, and chemically competent analog of maurocalcine for the efficient transport of impermeant compounds into cells. J. Biol. Chem. 283 :27048–27056. 10.1074/jbc.M804727200 18621738 Rosales, R.A., M. Fill, and A.L. Escobar. 2004. Calcium regulation of single ryanodine receptor channel gating analyzed using HMM/MCMC statistical methods. J. Gen. Physiol. 123 :533–553. 10.1085/jgp.200308868 15111644 Rousseau, E., and G. Meissner. 1989. Single cardiac sarcoplasmic reticulum Ca2+-release channel: activation by caffeine. Am. J. Physiol. 256 :H328–H333.2537030 Saiki, Y., R. El-Hayek, and N. Ikemoto. 1999. Involvement of the Glu724-Pro760 region of the dihydropyridine receptor II-III loop in skeletal muscle-type excitation-contraction coupling. J. Biol. Chem. 274 :7825–7832. 10.1074/jbc.274.12.7825 10075674 Samsó, M., R. Trujillo, G.B. Gurrola, H.H. Valdivia, and T. Wagenknecht. 1999. Three-dimensional location of the imperatoxin A binding site on the ryanodine receptor. J. Cell Biol. 146 :493–500. 10.1083/jcb.146.2.493 10427100 Schwartz, E.F., C.A. Schwartz, F. Gómez-Lagunas, F.Z. Zamudio, and L.D. Possani. 2006. HgeTx1, the first K+-channel specific toxin characterized from the venom of the scorpion Hadrurus gertschi Soleglad. Toxicon. 48 :1046–1053. 10.1016/j.toxicon.2006.08.009 17030052 Schwartz, E.F., E.M. Capes, E. Diego-García, F.Z. Zamudio, O. Fuentes, L.D. Possani, and H.H. Valdivia. 2009. Characterization of hadrucalcin, a peptide from Hadrurus gertschi scorpion venom with pharmacological activity on ryanodine receptors. Br. J. Pharmacol. 157 :392–403. 10.1111/j.1476-5381.2009.00147.x 19389159 Seo, I.R., D.E. Kang, D.W. Song, and D.H. Kim. 2011. Both basic and acidic amino acid residues of IpTxa are involved in triggering substate of RyR1. J. Biomed. Biotechnol. 2011 :386384. 10.1155/2011/386384 22007141 Shahbazzadeh, D., N. Srairi-Abid, W. Feng, N. Ram, L. Borchani, M. Ronjat, A. Akbari, I.N. Pessah, M. De Waard, and M. El Ayeb. 2007. Hemicalcin, a new toxin from the Iranian scorpion Hemiscorpius lepturus which is active on ryanodine-sensitive Ca2+ channels. Biochem. J. 404 :89–96. 10.1042/BJ20061404 17291197 Shtifman, A., C.W. Ward, J. Wang, H.H. Valdivia, and M.F. Schneider. 2000. Effects of imperatoxin A on local sarcoplasmic reticulum Ca2+ release in frog skeletal muscle. Biophys. J. 79 :814–827. 10.1016/S0006-3495(00)76338-7 10920014 Stern, M.D., and H. Cheng. 2004. Putting out the fire: what terminates calcium-induced calcium release in cardiac muscle? Cell Calcium. 35 :591–601. 10.1016/j.ceca.2004.01.013 15110149 Sutko, J.L., J.A. Airey, W. Welch, and L. Ruest. 1997. The pharmacology of ryanodine and related compounds. Pharmacol. Rev. 49 :53–98.9085309 Takeshima, H., S. Nishimura, T. Matsumoto, H. Ishida, K. Kangawa, N. Minamino, H. Matsuo, M. Ueda, M. Hanaoka, T. Hirose, 1989. Primary structure and expression from complementary DNA of skeletal muscle ryanodine receptor. Nature. 339 :439–445. 10.1038/339439a0 2725677 Tamura, K., D. Peterson, N. Peterson, G. Stecher, M. Nei, and S. Kumar. 2011. MEGA5: molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance, and maximum parsimony methods. Mol. Biol. Evol. 28 :2731–2739. 10.1093/molbev/msr121 21546353 Tanabe, T., K.G. Beam, J.A. Powell, and S. Numa. 1988. Restoration of excitation-contraction coupling and slow calcium current in dysgenic muscle by dihydropyridine receptor complementary DNA. Nature. 336 :134–139. 10.1038/336134a0 2903448 Tripathy, A., W. Resch, L. Xu, H.H. Valdivia, and G. Meissner. 1998. Imperatoxin A induces subconductance states in Ca2+ release channels (ryanodine receptors) of cardiac and skeletal muscle. J. Gen. Physiol. 111 :679–690. 10.1085/jgp.111.5.679 9565405 Valdivia, H.H. 2014. Structural and molecular bases of sarcoplasmic reticulum ion channel function. In Cardiac Electrophysiology: From Cell to Bedside. Sixth edition. D.P. Zipes, and J. Jalife, editors. Saunders Elsevier, Philadelphia. 55–69. 10.1016/B978-1-4557-2856-5.00006-6 Valdivia, H.H., M.S. Kirby, W.J. Lederer, and R. Coronado. 1992. Scorpion toxins targeted against the sarcoplasmic reticulum Ca2+-release channel of skeletal and cardiac muscle. Proc. Natl. Acad. Sci. USA. 89 :12185–12189. 10.1073/pnas.89.24.12185 1334561 Wehrens, X.H., S.E. Lehnart, S.R. Reiken, S.X. Deng, J.A. Vest, D. Cervantes, J. Coromilas, D.W. Landry, and A.R. Marks. 2004. Protection from cardiac arrhythmia through ryanodine receptor-stabilizing protein calstabin2. Science. 304 :292–296. 10.1126/science.1094301 15073377 Yan, Z., X.C. Bai, C. Yan, J. Wu, Z. Li, T. Xie, W. Peng, C.C. Yin, X. Li, S.H. Scheres, 2015. Structure of the rabbit ryanodine receptor RyR1 at near-atomic resolution. Nature. 517 :50–55. 10.1038/nature14063 25517095 Yuchi, Z., K. Lau, and F. Van Petegem. 2012. Disease mutations in the ryanodine receptor central region: crystal structures of a phosphorylation hot spot domain. Structure. 20 :1201–1211. 10.1016/j.str.2012.04.015 22705209 Zalk, R., O.B. Clarke, A. des Georges, R.A. Grassucci, S. Reiken, F. Mancia, W.A. Hendrickson, J. Frank, and A.R. Marks. 2015. Structure of a mammalian ryanodine receptor. Nature. 517 :44–49. 10.1038/nature13950 25470061 Zamudio, F.Z., G.B. Gurrola, C. Arévalo, R. Sreekumar, J.W. Walker, H.H. Valdivia, and L.D. Possani. 1997. Primary structure and synthesis of Imperatoxin A (IpTxa), a peptide activator of Ca2+ release channels/ryanodine receptors. FEBS Lett. 405 :385–389. 10.1016/S0014-5793(97)00227-5 9108323 Zhou, Q., J. Xiao, D. Jiang, R. Wang, K. Vembaiyan, A. Wang, C.D. Smith, C. Xie, W. Chen, J. Zhang, 2011. Carvedilol and its new analogs suppress arrhythmogenic store overload–­induced Ca2+ release. Nat. Med. 17 :1003–1009. 10.1038/nm.2406 21743453 Zhu, S., H. Darbon, K. Dyason, F. Verdonck, and J. Tytgat. 2003. Evolutionary origin of inhibitor cystine knot peptides. FASEB J. 17 :1765–1767.12958203

PMC004xxxxxx/PMC4925020.txt

==== Front J Exp Med J Exp Med jem jem The Journal of Experimental Medicine 0022-1007 1540-9538 The Rockefeller University Press 27242167 20151776 10.1084/jem.20151776 Research Articles Article Increased expression of AT-1/SLC33A1 causes an autistic-like phenotype in mice by affecting dendritic branching and spine formation AT-1/SLC33A1 and autism spectrum disorder Hullinger Rikki 12* Li Mi 1* Wang Jingxin 23 Peng Yajing 1 Dowell James A. 4 Bomba-Warczak Ewa 256 Mitchell Heather A. 7 Burger Corinna 8 Chapman Edwin R. 56 Denu John M. 4 Li Lingjun 3 Puglielli Luigi 159 1 Department of Medicine, University of Wisconsin-Madison, Madison, WI 53705 2 Neuroscience Training Program, University of Wisconsin-Madison, Madison, WI 53705 3 School of Pharmacy and Department of Chemistry, University of Wisconsin-Madison, Madison, WI 53705 4 Department of Biomolecular Chemistry and the Wisconsin Institute for Discovery, University of Wisconsin-Madison, Madison, WI 53705 5 Department of Neuroscience, University of Wisconsin-Madison, Madison, WI 53705 6 Howard Hughes Medical Institute, University of Wisconsin-Madison, Madison, WI 53705 7 Rodent Models Core, Waisman Center, University of Wisconsin-Madison, Madison, WI 53705 8 Department of Neurology, University of Wisconsin-Madison, Madison, WI 53705 9 Geriatric Research Education Clinical Center, Veterans Affairs Medical Center, Madison, WI 53705 Correspondence to Luigi Puglielli: lp1@medicine.wisc.edu * R. Hullinger and M. Li contributed equally to this paper. 27 6 2016 213 7 12671284 11 11 2015 15 4 2016 © 2016 Hullinger et al. 2016 https://creativecommons.org/licenses/by-nc-sa/3.0/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 3.0 Unported license, as described at http://creativecommons.org/licenses/by-nc-sa/3.0/). Increased expression of the ER membrane acetyl-CoA transporter AT-1 can cause an autism-like phenotype in mice. The import of acetyl-CoA into the lumen of the endoplasmic reticulum (ER) by AT-1/SLC33A1 regulates Nε-lysine acetylation of ER-resident and -transiting proteins. Specifically, lysine acetylation within the ER appears to influence the efficiency of the secretory pathway by affecting ER-mediated quality control. Mutations or duplications in AT-1/SLC33A1 have been linked to diseases such as familial spastic paraplegia, developmental delay with premature death, and autism spectrum disorder with intellectual disability. In this study, we generated an AT-1 Tg mouse model that selectively overexpresses human AT-1 in neurons. These animals demonstrate cognitive deficits, autistic-like social behavior, aberrations in synaptic plasticity, an increased number of dendritic spines and branches, and widespread proteomic changes. We also found that AT-1 activity regulates acetyl-CoA flux, causing epigenetic modulation of the histone epitope H3K27 and mitochondrial adaptation. In conclusion, our results indicate that increased expression of AT-1 can cause an autistic-like phenotype by affecting key neuronal metabolic pathways. National Institutes of Health 10.13039/100000002 NS094154 GM065386 DK071801 S10RR029531 National Institute of Child Health and Human Development 10.13039/100000071 P30 HD03352 ==== Body pmcAcetyl-CoA is an essential substrate for a wide range of biochemical reactions that occur within the cell (Pietrocola et al., 2015). Cytosolic acetyl-CoA is produced predominantly by the conversion of citrate and coenzyme A (CoA) by ATP-citrate lyase (ACLY) and the condensation of free acetate and CoA by acetyl-CoA synthetase (ACESS2, also referred to as AceCS; Pehar and Puglielli, 2013; Shi and Tu, 2015). Acetyl-CoA is then actively imported into the lumen of the ER by the ER membrane transporter AT-1 (also referred to as SLC33A1; Jonas et al., 2010), where it serves as donor of the acetyl group in the Nε-lysine acetylation of ER-resident and -transiting proteins (Choudhary et al., 2009; Pehar et al., 2012b). Recent reports suggest that lysine acetylation within the ER is required for ER-mediated quality control. Specifically, the ER-based acetyltransferases ATase1 and ATase2 associate with the oligosaccharyl transferase complex to acetylate properly folded glycoproteins (Ding et al., 2014). In addition, studies performed with two type I membrane proteins indicate that the acetylation status of nascent secretory proteins in the ER regulates the efficiency of transport along the secretory pathway (Costantini et al., 2007; Mak et al., 2014). Decreased influx of acetyl-CoA into the ER lumen leads to aberrant induction of autophagy in both cell-based (Jonas et al., 2010; Pehar et al., 2012a) and animal (Peng et al., 2014) models. At the mechanistic level, the induction of autophagy is linked to the acetylation status of autophagy-related protein 9A (Pehar et al., 2012a; Peng et al., 2014). Haploinsufficiency of AT-1 in the animal results in neurodegeneration, inflammation, and propensity to infections and cancer (Peng et al., 2014). Heterozygous mutations in AT-1 have been identified in patients affected by an autosomal dominant form of spastic paraplegia (Lin et al., 2008), whereas homozygous mutations have been identified in patients affected by severe developmental delay and childhood death (Huppke et al., 2012). Chromosomal duplications affecting the 3q25.31 locus harboring AT-1/SLC33A1 have been associated with autism spectrum disorder (ASD) and intellectual disability (SFARI Autism Database; http://sfari.org/; Sanders et al., 2011; Prasad et al., 2012; Krumm et al., 2013). Additionally, a gain of ∼1.1–1.5 Mb in 3q25.2-3q25.31, which contained AT-1/SLC33A1 and Guanine Monphosphate Synthase (GMPS, associated with myeloid leukemia) was found in three male children with autism, seizure, abnormal electroencephalogram, and facial dysmorphism (Swisshelm, K., et al. 2014. ASHG Annual Meeting; Abstract 3205T). In this study, we sought to investigate the consequences of increased AT-1 activity. Specifically, we generated an AT-1 transgenic (Tg) mouse model that selectively overexpresses human AT-1 in neurons. These animals demonstrate cognitive deficits, autistic-like social behavior, aberrations in synaptic plasticity, increased number of dendritic spines and branches, and widespread proteomic changes. The synaptic phenotype appears to be caused by increased trafficking of nascent proteins along the secretory pathway. In addition, we found that AT-1 Tg animals display increased expression of mitochondrial enzymes related to acetyl-CoA production, suggesting that increased movement of acetyl-CoA into the ER causes downstream compensatory mechanisms in mitochondrial activity. Furthermore, this apparent mitochondrial adaptation appears to be driven by changes in the acetylation/methylation status of Lys27 on the histone protein H3 suggesting that this site acts as a sensor to adapt supply of citrate from the mitochondria to rapidly compensate for changes in cytosolic acetyl-CoA, as induced by increased AT-1 activity. In conclusion, our results indicate that increased expression of AT-1 can cause an autistic-like phenotype by affecting key neuronal metabolic pathways. RESULTS AT-1 Tg animals display cognitive deficits and autistic-like behaviors To explore the role of AT-1 in the brain, we generated Tg mice with an inducible neuron-specific overexpression Tet-Off system driven by the CamK2 promoter (Fig. 1 A). For the purpose of this study, the animals (referred to as AT-1 Tg) were maintained in the absence of doxycycline (Dox); as such, they overexpressed human AT-1 during development, as well as after birth. Direct assessment of both mRNA and protein levels in neurons isolated from the adult brain confirmed successful up-regulation of AT-1 in Tg mice (Fig. 1, B–D). Figure 1. AT-1 Tg mice selectively overexpress AT-1 in forebrain neurons. (A) AT-1 Tg mice were generated with an inducible neuron-specific overexpression Tet-Off system driven by the CamK2 promoter. (B) Western blot showing isolation of neurons from the brain of adult AT-1 Tg mice. Total cells dissociated from the brain (Input), neuronal, and nonneuronal fractions are shown. The following markers were used: Iba1 for microglia; S100 for astrocytes; βIII Tubulin, L1CAM, and NFL 160 for neurons. (C and D) mRNA (C) and protein (D) levels of AT-1 in AT-1 Tg mice. Data in C represent three independent determinations of seven mice/group. **, P < 0.005; #, P < 0.0005. Student’s t test and one-way ANOVA, followed by Tukey-Kramer multiple comparisons test. Bars represent mean ± SD. To determine whether selective overexpression of AT-1 in neurons would impact learning and memory, mice were assessed with the fear conditioning (FC), novel object recognition (NOR), and Morris water maze (MWM) tasks. We found that AT-1 Tg animals displayed significant deficits in each of these hippocampus-dependent tasks (Fig. 2, A–F). Specifically, we observed significant impairments in contextual FC (Fig. 2 A) and NOR (Fig. 2 B). The animals did not display differences in freezing behavior during the initial exposure to the behavior arena (Fig. 2 A, Training) and did not display increased anxiety in the open field test (unpublished data). In the NOR test, WT mice spent >40% of their total investigation time exploring the novel object, whereas AT-1 Tg mice displayed no preference for the novel object relative to the trained objects. During the training phase of the MWM task, Tg mice had a significantly higher escape latency (Fig. 2 C) and swam greater distances before locating the hidden platform than WT controls (Fig. 2 D). Furthermore, they spent less time in the target quadrant (Fig. 2 E) and displayed fewer platform crossings during the probe trial than WT animals (Fig. 2 F). Figure 2. AT-1 Tg animals display cognitive deficits and autistic-like social behaviors. (A) Contextual FC. (WT, n = 6; AT-1 Tg, n = 6). (B) NOR (WT, n = 11; AT-1 Tg, n = 12). (C–F) MWM (WT, n = 10; AT-1 Tg, n = 10). (G) Marble burying task (WT, n = 13; AT-1 Tg, n = 13). Number of marbles that were at least 50% covered (total) and number of marbles that were completely buried (100%) are shown. (H) Social interaction test. Preference for investigating novel versus familiar mouse is shown. (WT, n = 13; AT-1 Tg, n = 13). *, P < 0.05; **, P < 0.005. Student’s t test and one-way ANOVA, followed by Tukey-Kramer multiple comparisons test. Bars represent mean ± SD. Given the possible association of AT-1/SLC33A1 with ASD and intellectual disability, we decided to test AT-1 Tg mice with the marble burying and social preference paradigms, two commonly used models of autistic behavior. We found that Tg mice displayed decreased repetitive behavior in the marble burying task (Fig. 2 G) and failed to show a preference for investigating the novel mouse in the social preference paradigm (Fig. 2 H). These tests indicate aberrations in repetitive behavior and social tendencies, suggesting that our Tg mouse line shares similarities with behavioral models of ASD (Gkogkas et al., 2013; Kouser et al., 2013; Dere et al., 2014; Lugo et al., 2014; Pucilowska et al., 2015; Speed et al., 2015). To assess whether the aforementioned behavioral changes were solely the result of developmental events, we also generated AT-1 Tg mice in the presence of Dox to repress expression of the transgene. The Dox was suspended at weaning and the animals were analyzed 6 mo later (Fig. 3 A). Under these conditions, we did not observe any differences with WT mice (Fig. 3, B–D), indicating that the aberrations in repetitive behavior and social tendencies described in Fig. 2 are solely—or mostly—developmental. Figure 3. Repression of AT-1 by Dox prevents cognitive deficits and autistic-like social behaviors. (A) Schematic view of the experimental setting. Mice were tested 6 mo after suspension of Dox and induction of AT-1. (B) Contextual FC of Dox-treated AT-1 Tg mice (WT, n = 12; AT-1 Tg, n = 12). (C) NOR of Dox-treated AT-1 Tg mice (WT, n = 12; AT-1 Tg, n = 12). (D) Marble burying task of Dox-treated AT-1 Tg mice (WT, n = 13; AT-1 Tg, n = 12). Number of marbles that were at least 50% covered (total) and number of marbles that were completely buried (100%) are shown. Student’s t test and one-way ANOVA, followed by Tukey-Kramer multiple comparisons test. Bars represent mean ± SD. Although there is no specific behavioral profile that defines ASD in the mouse, the combined deficiency in FC, NOR, MWM, marble burying, and social preference paradigms observed in AT-1 Tg mice has been described in other mouse models of ASD (Gkogkas et al., 2013; Kouser et al., 2013; Dere et al., 2014; Lugo et al., 2014; Pucilowska et al., 2015; Speed et al., 2015). Because chromosomal duplications of the 3q25.31 locus, which harbors AT-1/SLC33A1, have been associated with ASD and intellectual disability (see later in the Discussion section), we conclude that AT-1 Tg mice display an autistic-like phenotype. Tg mice demonstrate changes in neuronal morphology and aberrant synaptic plasticity We next sought to investigate whether increased expression of AT-1 in neurons would lead to morphological and/or synaptic changes that could explain the behavioral phenotype. We found that cultured neurons from AT-1 Tg mice displayed a drastic increase in the number of dendritic branches and spines compared with WT controls (Fig. 4, A–D). Similar results were obtained in vivo by analyzing hippocampal tissue from 6-mo-old mice (Fig. 4 E), thus confirming that overexpression of AT-1 causes drastic changes in neuronal morphology. Figure 4. AT-1 Tg animals display changes in neuronal morphology and imbalanced synaptic plasticity. (A–D) Morphological assessment of hippocampal neurons in culture. Phalloidin staining (top) and unbiased computer-driven reconstruction (bottom) of both WT and AT-1 Tg neurons (A and B) are shown. Sholl analysis (C) and spines quantification (D) are shown. Bars, 25 µm. Number of determinations for Sholl analysis: WT, n = 4; AT-1 Tg, n = 5. Number of independent segments analyzed for spine density: WT, n = 16; AT-1 Tg, n = 36. (E) Golgi staining of the hippocampus from 6-mo-old WT and AT-1 Tg mice. Bar, 50 µm. (F) 3xTBS induced LTP in WT and AT-1 Tg mice. (G) paired pulse low-frequency facilitation LTD in WT and AT-1 Tg mice. (H–J) Western blot of selected relevant proteins. List and function of targeted proteins (H), quantification (I; n = 5), and representative Western blots (J) are shown. GAPDH served as loading control. *, P < 0.05; **, P < 0.005; #, P < 0.0005. Student’s t test and one-way ANOVA, followed by Tukey-Kramer multiple comparisons test; for LTP and LTD, repeated measures ANOVA, followed by Bonferroni post hoc tests. Bars represent mean ± SD. Alterations in neuronal function were assessed by hippocampal slice electrophysiology. Long-term potentiation (LTP) and its counterpart, long-term depression (LTD), are believed to be correlated with cognitive function, and it is has been demonstrated that a balance between LTP and LTD must be present in order for normal learning and memory formation to occur (Feldman, 2009; Huganir and Nicoll, 2013). We found that AT-1 Tg mice display increased levels of hippocampal LTP elicited by Theta burst stimulation (3xTBS; Fig. 4 F) and significantly reduced LTD induced by paired pulse low-frequency facilitation (Fig. 4 G) when compared with nonTg WT controls. Changes in spine morphology and concomitant altered synaptic plasticity similar to AT-1 Tg mice have been observed in several animal models of neurological disorders (Auffret et al., 2009; Auerbach et al., 2011; Amini et al., 2013; Neuhofer et al., 2015). Importantly, when exposed to Dox during development to repress expression of the transgene, AT-1 Tg mice displayed neither electrophysiological nor morphological changes (unpublished data). To better understand the mechanisms driving the observed changes in neuron structure and activity in AT-1 Tg mice, we selected target proteins known to be involved in neuronal outgrowth and synaptic plasticity, and compared expression levels in the hippocampi of Tg and WT mice (Fig. 4 H). Neurexin1, Neuroligin3, and Rap2a were selected as indicators of dendritic branching and synapse formation (Craig and Kang, 2007; Kawabe et al., 2010). AMPA2/3/4, mGluR5, synaptogyrin1, synapsin3, and Rab12 were selected as markers of pre- and postsynaptic plasticity (Chia et al., 2013; Alvarez-Castelao and Schuman, 2015; Martenson and Tomita, 2015). We found that a majority of these proteins were significantly up-regulated in postsynaptic densities of Tg animals compared with controls (Fig. 4, I and J). Similar results were also obtained by mass spectrometry (Table S2). AT-1 Tg animals display widespread proteomic changes Recognized functions of Nε-lysine acetylation include regulation of expression, activity, molecular stabilization, and conformational assembly of a protein (Kouzarides, 2000). Proteomic studies have shown that a wide range of proteins undergoes Nε-lysine acetylation within the lumen of the ER; they include both ER-resident and -transiting proteins with a variety of biological functions (Choudhary et al., 2009; Pehar et al., 2012b). We have previously shown that changes in expression and activity of AT-1 have profound effects on the ER acetylation profile; specifically, increased activity leads to increased acetylation of ER cargo proteins (Jonas et al., 2010), whereas reduced activity has the opposite effect (Peng et al., 2014). To dissect the molecular mechanisms of the AT-1 Tg phenotype, we first resolved the neuron-specific ER acetylome of the adult AT-1 Tg brain. The analysis identified 395 acetylation sites on 152 proteins (Table S1) involved in different biological pathways, as assessed by the PANTHER classification system (Fig. 5 A). STRING analysis identified six major clusters (Fig. 5 B), all highly relevant to our observed phenotype, suggesting that the increased acetylation of proteins may have consequences for neuronal outgrowth and adhesion, synaptic activity and structure, cell surface transporters, and protein synthesis. Figure 5. Analysis of the brain-specific ER acetylome in AT-1 Tg mice. (A) PANTHER analysis of identified acetylated ER-cargo proteins. Proteins were grouped based on biological functions. (B) Clusters of highly related proteins identified by STRING analysis. Proteins were grouped based on biological function. Analysis was done in triplicate. See Table S1. To further elucidate the impact of increased AT-1 activity on protein expression, we conducted large-scale liquid chromatography coupled to tandem mass spectrometry (MS) analysis to identify global changes in the proteome of AT-1 Tg animals. We found that in hippocampal tissue, 476 proteins were up-regulated in Tg mice compared with CamK2 controls (Table S2). CamK2 mice were selected as the control group to correct for possible nonspecific variations in protein expression. Identified proteins were responsible for a variety of biological functions (Fig. 6 A). STRING analysis at 90% confidence revealed eight clusters that were highly correlated to the observed clusters in the acetylome data (Fig. 6 B). Figure 6. AT-1 Tg animals display widespread proteomic changes. (A) PANTHER analysis of up-regulated proteins. (B) Clusters of related proteins identified by STRING analysis at high confidence (90%). Proteins were grouped based on biological functions. Analysis was done in triplicate. See Table S2. When viewed globally, the acetylome (Fig. 5) and proteome (Fig. 6) data identify four major biological functions: (1) protein translation and quality control; (2) transport of synaptic vesicles, assembly of synaptic connections, and regulation of synaptic activity; (3) neuronal migration, outgrowth and adhesion; (4) mitochondria activity, specifically electron transport chain, tricarboxylate acid cycle, and acetyl-CoA metabolism. Therefore, these data suggest that the increased influx of acetyl-CoA into the ER lumen in AT-1 Tg mice affects proteins related to synaptic plasticity, neuron structure, transporter activity, and protein synthesis. AT-1 Tg mice display increased efficiency of the secretory pathway The proteome data (Fig. 6) indicated that 163 out of 476 proteins up-regulated in AT-1 Tg mice are either membrane or secreted proteins that are normally translated on the ER surface and inserted into the secretory pathway (Table S2). Although they are involved in different biochemical and cellular functions, STRING analysis revealed that the great majority of them are tightly related to synaptic plasticity (unpublished data), thus highlighting the significance of these proteins for our observed phenotype. We recently reported that the ER-based acetyltransferases, ATase1 and ATase2, associate with the oligosaccharyl transferase complex to acetylate correctly folded glycoproteins (Ding et al., 2014). Studies conducted with two type I membrane glycoproteins suggest that Nε-lysine acetylation might regulate efficiency of transport along the secretory pathway (Costantini et al., 2007; Mak et al., 2014). To investigate whether increased influx of acetyl-CoA into the ER lumen can indeed cause global changes in the efficiency of transport in the secretory pathway, we used human neuroglioma (H4) cells overexpressing AT-1. Cells were labeled with azide-modified mannosamine (ManNAz), which is metabolically incorporated into sialic acid–containing glycoproteins while they traffic through the secretory pathway (Laughlin and Bertozzi, 2007). Sialic acid is the last sugar to be attached to the oligosaccharide chain; both the Golgi-membrane cytidine monophosphate-sialic acid transporter and the Golgi-resident sialyltransferase are located in the trans-Golgi (Hirschberg et al., 1998). Only nascent glycoproteins that have successfully reached the trans-Golgi can be sialylated (Hirschberg et al., 1998). Therefore, ManNAz can specifically determine efficiency of trafficking of newly synthesized glycoproteins. The results show a dramatic increase in the levels of nascent glycoproteins as a result of AT-1 overexpression (Fig. 7 A). To account for possible changes in protein translation, we also labeled cells with O-propargyl-puromycin (OPP), an alkyne analogue of puromycin that incorporates into newly translated proteins. The results revealed a significant increase in rate of protein biosynthesis (Fig. 7 B). However, normalization of ManNAz (rate of transport) and OPP (rate of biosynthesis) labeling still showed increased delivery of secretory glycoproteins to the cell surface (Fig. 7 C). Importantly, OPP incorporation does not discriminate between secretory and nonsecretory proteins, whereas ManNAz does. Therefore, we can conclude that although increased translation can partially account for the increased levels of secretory proteins in AT-1 Tg mice, increased efficiency of transport along the secretory pathway is likely to play a major role. To support the studies conducted with ManNAz, we also determined levels of cell-surface biotinylated proteins. Again, we found a significant increase in AT-1–overexpressing cells (Fig. 7 D). Using a cell-impermeable reagent to biotinylate primary amines can be used as a reliable readout of steady-state levels of cell-surface proteins. Biotin labeling does not visualize secreted proteins or proteins that localize to intracellular organelles (i.e.; Golgi apparatus, lysosomes, etc.). Therefore, in contrast to ManAz labeling, it underestimates the efficiency of the secretory pathway. However, the biotin read-out (Fig. 7 D) is consistent with the ManAz results (Fig. 7 A). Finally, to confirm the aforementioned conclusions, we also determined ManNAz incorporation in mouse embryo fibroblasts (MEFs) generated from AT-1S113R/+ mice. AT-1S113R is a mutant version of AT-1 associated with a familial form of spastic paraplegia (Lin et al., 2008) and is deficient in acetyl-CoA transport activity (Peng et al., 2014). As a result, AT-1S113R/+ mice represent a model of AT-1 haploinsufficiency (Peng et al., 2014). As expected, we observed a significant reduction in levels of cell-surface glycoproteins in AT-1S113R/+ MEFs (Fig. 7 E), thus supporting the conclusion that influx of acetyl-CoA into the ER lumen regulates efficiency of trafficking of secretory glycoproteins. Figure 7. AT-1 Tg mice display increased efficiency of the secretory pathway. (A) Incorporation of ManNAz into sialic acid–containing glycoproteins to assess efficiency of trafficking of newly synthesized glycoproteins. H4(-), transfected with empty vector; H4(AT-1wt), transfected with AT-1wt. (H4(-), n = 8; H4(AT-1wt), n = 9). (B) Incorporation of OPP to determine rate of protein biosynthesis. (H4(-), n = 21; H4(AT-1wt), n = 25). (C) Normalization of ManNAz labeling per rate of OPP incorporation. (D) Western blot assessment of cell surface biotinylated proteins. Left, representative images; right, quantitation. (H4(-), n = 2; H4(AT-1wt), n = 2). (E) Incorporation of ManNAz into sialic acid–containing glycoproteins to assess efficiency of trafficking of newly synthesized glycoproteins into MEFs generated from WT and AT-1S113R/+ mice. (MEF(WT), n = 35; MEF(AT-1S113R/+), n = 43). *, P < 0.05; #, P < 0.0005. Student’s t test. Bars represent mean ± SD. When taken together, the above results suggest that AT-1 Tg mice have increased delivery of ER cargo proteins to the cell surface. The same animals also have increased levels of 163 proteins that are translated on the ER and insert into the secretory pathway (Table S2). It is likely that these two findings are functionally related and that they play an important role in the observed mouse phenotype. Increased activity of AT-1 leads to mitochondria adaption In addition to changes in the efficiency of the secretory pathway, the proteomics results (Fig. 6 and Table S2) revealed a marked up-regulation of proteins related to three important mitochondrial functions: electron chain transport, tricarboxylate acid cycle, and acetyl-CoA metabolism (Fig. 8 A). It is worth remembering that the ER imports acetyl-CoA from the cytosol and that the cytosolic pool of acetyl-CoA is largely supplied by the conversion of mitochondria-derived citrate into acetyl-CoA (Pehar and Puglielli, 2013; Pietrocola et al., 2015). Given the increased influx of cytosolic acetyl-CoA into the ER lumen, caused by the overexpression of AT-1, we hypothesized that levels of cytosolic acetyl-CoA would be decreased in AT-1 Tg mice, forcing the mitochondria to compensate by generating more citrate for cytosolic conversion. To test this hypothesis, we initially determined acetyl-CoA levels in the cytosol of AT-1–overexpressing H4 cells. The results showed a significant reduction when compared with control cells (Fig. 8 B), supporting the conclusion that the increased flux into the ER lumen is able to affect the cytosolic pool of acetyl-CoA. To assess whether this conclusion could be extended to the in vivo settings, we also analyzed the cytosol recovered from total hippocampal tissue (Fig. 8 C), as well as isolated adult cortical neurons (Fig. 8 D) of AT-1 Tg mice. Again, we observed a significant reduction of acetyl-CoA levels when compared with WT mice. Importantly, two crucial proteins that allow cross-talk between mitochondria and cytosolic acetyl-CoA were also found to be up-regulated in our proteomics analysis: the mitochondrial carrier citrate transporter (Slc25a1) and Acly (Table S2). Slc25a1 is the mitochondria membrane transporter that translocates citrate to the cytosol, whereas Acly is a cytosolic-based enzyme that converts mitochondria-derived citrate into acetyl-CoA by using cytosolic CoA and ATP (Pehar and Puglielli, 2013; Pietrocola et al., 2015). Interestingly, Acss2 (also known as AceCS), which contributes to the synthesis of cytosolic acetyl-CoA through the condensation of acetate and CoA (Pehar and Puglielli, 2013; Pietrocola et al., 2015), was not found to be up-regulated (Table S2). It is worth stressing that although both Acly and Acss2 can generate acetyl-CoA, they carry out two different biochemical reactions; they also respond to different inputs and elicit different biological functions (Wellen et al., 2009; Xu et al., 2014). Therefore, our results would suggest that Acly, but not Acss2, is activated to ensure mitochondria adaptation in the AT-1 Tg mice. To investigate this further, we also determined mRNA levels of Slc25a1, Acly, and Acss2 in the hippocampus of AT-1 Tg mice. Again, Slc25a1 and Acly showed significant up-regulation, whereas Acss2 did not (Fig. 8 E). Figure 8. Mitochondrial adaptation in AT-1 Tg mice. (A) STRING analysis of mitochondria proteins found to be up-regulated in AT-1 Tg mice. Individual clusters are identified with different colors. Original data are shown in Table S2. (B) Cytosolic levels of acetyl-CoA in H4 cells. H4(-), control/empty vector; H4(AT-1WT), overexpressing WT AT-1. Results are expressed as percent of H4(-). (H4(-), n = 10; H4 (AT-1WT) n = 11). (C) Cytosolic levels of acetyl-CoA in total hippocampal tissue. (WT, n = 6; AT-1 Tg, n = 6). (D) Cytosolic levels of acetyl-CoA in isolated adult neurons. (WT, n = 3; AT-1 Tg, n = 3). (E) mRNA levels of Slc25a1, Acly, and Acss2 in the hippocampus of AT-1 Tg mice. (WT, n = 7; AT-1 Tg, n = 6). (F) Schematic summary of results showing the mitochondrial adaptation that results from the increased influx of acetyl-CoA into the ER in AT-1 Tg mice. *, P < 0.05. Student’s t test. Bars represent mean ± SD. Collectively, the aforementioned data (Fig. 8, B–E) suggest that the increased transport of acetyl-CoA into the lumen of the ER reduces levels of cytosolic acetyl-CoA, leading to increased generation and delivery of citrate from the mitochondria, followed by increased conversion of citrate into acetyl-CoA (Fig. 8 F). Therefore, we interpret the up-regulation of mitochondria-related pathways as a compensatory mechanism (here defined as mitochondrial adaptation) to the increased expression of AT-1. The aforementioned findings raise the question of how mitochondria can sense cytosolic levels of acetyl-CoA. A possible explanation is that changes in the cytosolic pool of acetyl-CoA cause epigenetic changes, leading to transcriptional activation of targeted mitochondria enzymes. To test this possibility, we used a MS-based strategy to assess the posttranslational modification profile of histone proteins within the hippocampus of WT and AT-1 Tg animals. Previous studies have correlated acetyl-CoA availability to global epigenetic changes (Wellen et al., 2009). The MS results showed a significant increase in the acetylation of H3K27 accompanied by a significant decrease in H3K27 methylation, indicative of increased transcriptional activation in our Tg animals (Fig. 9 A and Table S3). Importantly, the MS data were confirmed by direct assessment of H3K27acetyl and H3K27trimethyl levels by immunoblotting (Fig. 9 B), suggesting that a significant portion of chromatin has “sensed” the altered levels of acetyl-CoA. Figure 9. The mitochondrial adaptation in AT-1 Tg mice is driven by a specific epigenetic activation. (A) Mass spectrometry quantification of posttranslational modifications of histone proteins in the hippocampus of AT-1 Tg mice. Global changes on H3K27/K36 are shown on the left and changes in the acetylation and methylation profile of H3K27 are shown on the right (WT, n = 3; AT-1 Tg, n = 3). (B) Western blot assessment showing acetylation and methylation of H3K27 in AT-1–overexpressing H4 cells. Selected images are shown on the left panel and quantification (n = 4) is shown on the right. (C) ChIP analysis of SLC33A1, ACLY, and ACSS2 after immunoprecipitation (IP) with an anti-H3K27ac antibody. IP was done in H4 cells. (left) Sheared DNA; (right) ChIP-amplification. Analysis was performed in quadruplicate. (D) ChIP-qPCR showing H3K27 acetylation of SLC25A1 and ACLY in H4 cells. Results are expressed as fold of H4 (-) (H4(-), n = 5; H4(AT-1WT), n = 5). (E) ChIP-qPCR showing H3K27 acetylation of Slc25a1 and Acly in total hippocampal tissue. Results are expressed as fold of WT (WT, n = 3; AT-1 Tg, n = 3). (F) ChIP-qPCR showing H3K27 acetylation of Slc25a1 and Acly in isolated adult neurons. Results are expressed as fold of WT (WT, n = 3; AT-1 Tg, n = 3). IP in D–F was performed with an anti-H3K27ac antibody. *, P < 0.05; **, P < 0.005; #, P < 0.0005. Student’s t test and one-way ANOVA followed by Tukey-Kramer multiple comparisons test. Bars represent mean ± SD. When taken together, the aforementioned results suggest that the H3K27 modification might serve as a “sensor” to rapidly respond to changes in cytosolic levels of acetyl-CoA by activating key mitochondria genes. To test this hypothesis, we conducted chromatin immunoprecipitation (ChIP) analysis and probed for SLC25A1 and ACLY, which are key to the mitochondrial adaptation observed in AT-1 Tg mice (see Fig. 8 F). Again, we used ACSS2 as negative control. Immunoprecipitation with an anti-H3K27ac antibody resolved SLC25A1 and ACLY, but not ACSS2 (Fig. 9 C). When coupled to quantitative real-time PCR (qPCR), the ChIP analysis revealed increased H3K27 acetylation of SLC25A1 and ACLY in AT-1–overexpressing H4 cells (Fig. 9 D). These results correlate to the increased mRNA levels of both genes shown in Fig. 8 E. To assess whether these findings could be extended to the in vivo settings, we performed the same analysis with total hippocampal tissue (Fig. 9 E) and isolated adult cortical neurons (Fig. 9 F) of AT-1 Tg mice. Again, we found increased H3K27 acetylation of Slc25a1 and Acly, when compared with WT mice. In conclusion, these findings suggest that the enhancement in acetylation at the H3K27 site directly modifies expression of genes related to mitochondrial activity, providing a selective, targeted mitochondrial adaptation to changes in cytosolic levels of acetyl-CoA. DISCUSSION Here, we report that mice that overexpress human AT-1 display cognitive deficits, autistic-like social behavior, aberrations in synaptic plasticity, increased number of dendritic spines and branches, and widespread proteomic changes. We also report that AT-1 activity regulates the efficiency of the secretory pathway as well as cytosolic levels of acetyl-CoA, which in turn leads to epigenetic modulation of the histone epitope H3K27 and mitochondrial adaptation. In conclusion, our results indicate that increased neuronal expression of AT-1 can cause an autistic-like phenotype by affecting key metabolic pathways. Nε-lysine acetylation occurs on both ER-transiting and -resident proteins with a wide array of biological functions (Choudhary et al., 2009; Pehar et al., 2012b; Fig. 5). Lysine acetylation is emerging as a novel mechanism that contributes to ER-based quality control; proteins that are properly folded are acetylated and proceed through the secretory pathway, whereas proteins that are not acetylated are retained and degraded (Costantini et al., 2007; Ding et al., 2014; Mak et al., 2014). The ER-based acetylation machinery requires the membrane transporter AT-1 (Jonas et al., 2010) and at least two acetyltransferases (Ko and Puglielli, 2009). Heterozygous and homozygous mutations in AT-1/SLC33A1 have been identified in patients with familial spastic paraplegia (Lin et al., 2008) and developmental delay/premature death (Huppke et al., 2012), respectively. Chromosomal duplications affecting the 3q25.31 locus harboring AT-1/SLC33A1 have been associated with ASD and intellectual disability (Sanders et al., 2011; Prasad et al., 2012; Krumm et al., 2013). Additionally, a short chromosomal gain that covered only two genes, AT-1/SLC33A1 and GMPS, was found in children with autism, seizure, abnormal electroencephalogram, and facial dysmorphism (Swisshelm, K., et al. 2014. ASHG Annual Meeting, San Diego, CA). Although not definitive, the aforementioned genetic association supports a causative role of AT-1 in ASD. Importantly, chromosomal duplications of 17p13.1 (harboring SLC13A5), 22q11.21 (harboring SLC25A1), and 17q21.2 (harboring ACLY) have also been associated with ASD (Sanders et al., 2011). SLC13A5 is the cell surface citrate transporter that translocates citrate from the extracellular milieu to the cytosol; SLC25A1 is the mitochondria membrane citrate transporter that translocates citrate from the mitochondria to the cytosol; and ACLY is the enzyme responsible for converting cytosolic citrate into acetyl-CoA. In conclusion, these individual genetic associations affecting SLC14A5, SLC25A1, ACLY, and AT-1/SLC33A1 seem to suggest a model where increased supply and movement of acetyl-CoA into the ER lumen is tightly linked to ASD. This model is supported by the phenotype of AT-1 Tg mice reported here. It is also worth mentioning that, like SLC33A1 (Huppke et al., 2012), mutations in SLC13A5 and SLC25A1 have also been associated with developmental delay of the brain (Thevenon et al., 2014; Hardies et al., 2015; Prasun et al., 2015). In addition, single variants on several other genes that are directly or indirectly related to the citrate-acetyl-CoA metabolic pathway have been identified in ASD cohorts. Some of them have also been associated with developmental delay and epileptic encephalopathy. These include SLC16A3, SLC16A7, SLCA1, SLCA2, SLC25A12, SLC25A14, SLC25A24, and SLC25A27 (Sanders et al., 2011). The molecular basis behind the observed AT-1 Tg phenotype involves increased transport of acetyl-CoA into the lumen of the ER, likely causing changes in the ER acetylome, as well as in the efficiency of the secretory pathway. The ability of Nε-lysine acetylation to regulate activity and/or levels of modified proteins is well documented (Kouzarides, 2000; Pehar and Puglielli, 2013; Pietrocola et al., 2015). Both the acetylome (Fig. 5) and the proteome (Fig. 6) identified a large set of proteins that are implicated in cellular functions that could immediately explain the Tg mouse phenotype, specifically: neuronal migration and adhesion, neurite outgrowth, transport of synaptic vesicles, assembly of synaptic connections, and regulation of synaptic activity. Therefore, it is conceivable to assume that the increased influx of acetyl-CoA, and consequent changes in the ER acetylome and in the efficiency of the secretory pathway, is the primary cause of ASD-like phenotype. The characterization of the AT-1 Tg phenotype also revealed that the animals display decreased levels of cytosolic acetyl-CoA and, as a result, mitochondrial adaptation. This adaptive mechanism appears to be driven by epigenetic changes on H3K27. Importantly, we did not detect widespread epigenetic changes (Table S3). This is likely a result of the compensatory activity of Acly; indeed, silencing of ACLY in mammalian cells causes global changes in histone acetylation (Wellen et al., 2009). Whether the mitochondria adaptation participates in the observed phenotype remains to be fully determined. The fact that genes, such as SLC13A5 and ACLY, which can alter cytosolic levels of acetyl-CoA without affecting mitochondria activity directly, might be linked to ASD (discussed above) would suggest that mitochondria have limited impact. However, genes such as SLC25A12 and SLC25A1, which immediately influence acetyl-CoA production or delivery from the mitochondria, have also been linked to ASD. Therefore, the role of mitochondria might be more complex than expected. It is also possible that cross-talk exists between mitochondria, cytosol, and ER so that either increased supply (from mitochondria and/or cytosol) or movement of acetyl-CoA into the ER lumen causes ASD. In conclusion, the specific role of the mitochondria citrate pathway and of the cytosolic citrate/acetyl-CoA pools in ASD remains to be fully dissected. Although increased efficiency of the secretory pathway seems to be logically connected to increased expression of proteins related to synaptic plasticity and learning and memory formation, a majority of proteins identified by mass spectrometry do not insert into the secretory pathway. However, the dramatic increase in dendritic spine and branch formation observed in our animals would require targeted (and secondary) up-regulation of cytosolic and nuclear proteins related to neurite outgrowth. Establishing synaptic densities would also require the activation of cytosolic scaffolding and adaptor proteins that support the structural integrity of the synapse as well as regulate signal transduction pathways. Indeed, our proteomic assessment found significant increase of cytosolic proteins that are involved in different aspects of the aforementioned biological pathways. In conclusion, our studies have revealed that both decreased (Peng et al., 2014) and increased (present study) AT-1 activity affect neuronal biology. In particular, the data shown here have linked increased AT-1 activity to ASD, and elucidated a novel aspect of acetyl-CoA metabolism that affects learning and memory, synaptic plasticity, protein expression, mitochondrial activity, and epigenetics. Further studies are necessary to determine whether changes in AT-1 activity at various points in the lifespan of the mouse will induce significant changes in neuronal morphology and behavior. MATERIALS AND METHODS Generation of AT-1 Tg mice cDNA encoding human AT-1 was isolated by BamHI and EcoRV digestion from Topo-AT-1 constructs and subcloned into pTRE-Tight vector (Takara Bio Inc.). TRE-AT-1 Tg lines were generated by injection of linearized pTRE-Tight vector (Takara Bio Inc.) containing AT-1 cDNA into the pronucleus of fertilized eggs from FVB mice. Monogenic pTRE-AT-1 mice were backcrossed to WT C57BL/6 mice for five generations, and then bred to CamK2a-tTA mice (B6.Cg-Tg(Camk2a-tTA)1Mmay/DboJ; The Jackson Laboratory), generating nonTg WT, Camk2a-tTA monogenic, TRE-AT-1 monogenic, and CamK2a-tTA;TRE-AT-1 (referred to as AT-1 Tg) mice. Genotyping from tail DNA was performed using the following primers: AT-1 forward (5′-AATCTGGGAAACTGGCCTTCT-3′), AT-1 reverse (5′-TATTACCGCCTTTGAGTGAGCTGA-3′), Camk2a-tTA forward (5′-CGCTGTGGGGCATTTTACTTTAG-3′), and Camk2a-tTA reverse (5′-CATGTCCAGATCGAAATCGTC-3′). Animal experiments were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and were approved by the Institutional Animal Care and Use Committee of the University of Wisconsin-Madison and the Madison Veterans Administration Hospital. Unless specified, WT littermates were used as control throughout our study. Cell cultures Mouse embryonic fibroblasts (MEFs) from WT and AT-1S113R/+ mice were previously described (Peng et al., 2014), and AT-1–overexpressing H4 (human neuroglioma) cells were previously described (Jonas et al., 2010; Pehar et al., 2012b). MEFs and H4 cells were maintained in DMEM supplemented with 10% FBS and 1% Penicillin/Streptomycin/Glutamine solution (Mediatech, Inc). G418 was used as selection marker in H4AT-1 cells (Jonas et al., 2010; Pehar et al., 2012b). Adult neurons were isolated as previously described (Cahoy et al., 2008). In brief, mouse brains were quickly removed and digested with Neural Tissue Dissociation kit (Miltenyi Biotec) and gentleMACS Dissociators (Miltenyi Biotec). After enzymatic and mechanical dissociation, the cell suspension was passed through a 40-µm cell strainer and washed with HBSS. The pellets were spun down at 300 g for 10 min and resuspended with buffer from a Neuron Isolation kit (Miltenyi Biotec). Non-neuron cells were removed by using Non-Neuron Cells Biotin-Antibody Cocktail (Miltenyi Biotec), followed by Anti-Biotin MicroBeads (Miltenyi Biotec). After magnetic separation with QuadroMACS Separator (Miltenyi Biotec), the neuronal cells were collected as the unlabeled fraction from the flow-through part. Primary neuronal cultures were generated from E17-E18 mice. The isolated hippocampus was digested in 0.25% trypsin-EDTA (Gibco) for 15 min at 37°C, followed by trituration into a single-cell suspension with fire-polished glass pipettes. The dissociated cells were resuspended in the plating medium (Dulbecco's modified Eagle's medium with 10% FBS). The cells were plated onto coverslips precoated with poly-d-lysine (100 µg/ml; Sigma-Aldrich) at a density of 150 cells/mm2. 4 h after plating, media was changed to a serum-free neurobasal medium supplemented with 2% B27 (Gibco) and 0.5 mM GlutaMAX (Invitrogen). Media was changed every 3 d. Cultures were maintained in 37°C and 5% CO2 incubator for up to 21 d. For immunocytofluorescence analysis, cultures were fixed with 4% paraformaldehyde and immunostained with NeuN antibody (clone A60; 1:100; EMD Millipore) and tetramethyl rhodamine isothiocyanate (TRITC)-labeled phalloidin (0.5 µg/ml; Sigma-Aldrich). Nuclei were counterstained with DAPI. Behavioral analysis All behavioral tests were conducted and analyzed by an experimenter blind to the genotype of the mice. Unless indicated, mice were approximately 6-mo-old when tested. MWM The test consisted of eight blocks of testing over 4 d, followed by a probe trial. The first day, mice were briefly exposed to the pool to test swimming capabilities. Each mouse received two blocks of testing per day for 4 d, and each block was comprised of four sequential trials, for a total of 32 training trials. Each trial, the mouse was placed into the pool at a different start location and allowed to swim until it either located the platform, or reached the end of the 60-s trial. Behavior was monitored by an overhead camera and Noldus Ethovision software. If the mouse did not locate the platform within the 60-s time limit, it was gently placed on the platform and allowed to rest and look around for a few seconds before beginning the next trial. On the final day of testing, after completing the eighth block of testing, mice were exposed to the pool for a probe trial. For this trial, the platform was removed, and swimming behavior was monitored for 60 s. NOR Each mouse received five sequential 6-min trials, with a 3-min inter-trial interval. The first trial allowed the mouse to acclimate to the empty arena, with no objects presented. Trials two to four were training trials with four objects presented within the arena. For the testing trial, one randomly selected object was removed and replaced with a novel object. This trial was videotaped and subsequently analyzed by a trained observer blind to both mouse condition and the identity of the novel object. Time spent investigating all objects was used as total investigative time; the time spent exploring each object was expressed as percentage total investigative time. All objects used were pretested to guarantee no innate preference, and the novel object was rotated through a pool of these objects. FC Mice were tested in the delay FC paradigm to assess contextual memory formation. On day 1, mice were exposed to a 6-min training protocol, which consisted of two 30-s pairings of 87 dB white noise and a 1.5-s 0.7 mA shock, with an inter-trial interval of 2 min. On day 2, 22 h after exposure to the training protocol, mice were tested for conditioning to context. Mice were placed back in the chambers in which they originally received the shock for 5 min, but no stimuli (noise or shock) were presented. Percent freezing during each portion of the FC paradigm was calculated by the FreezeFrame 3 software (Coulbourn) using video feed and a motion index. Marble burying The marble burying paradigm was used to assess repetitive behaviors. Each mouse was placed into a clean cage with extra bedding and 20 black marbles arranged in a grid pattern. After 30 min of exploration in the cage, the number of marbles completely buried in the bedding and the number of marbles buried >50% were counted. Social interaction The social interaction test was used to assess autistic-like social behaviors in the AT-1 Tg mouse model. The test was conducted in a three-chambered arena with a central compartment and two side compartments. The test consisted of three trials, each 10 min in length, run consecutively. In the first trial, the experimental mouse was exposed to an empty arena and allowed to explore freely. In the second trial, a male juvenile mouse was placed in a mesh cup in one side chamber of the box and an identical empty mesh cup was placed in the opposite side chamber of the box to provide a neutral object control. The mesh cup allowed for visual and scent cues, but prevented aggressive behavior between the mice. During the third trial, the juvenile stimulus mouse from trial two was kept in the test arena, and a novel juvenile male mouse was placed in a cup in the opposite chamber. All trials were videotaped and used for subsequent analysis. Investigation time was scored by a trained observer blind to genotype. Assessment of neuronal morphology Confocal images of cultured neurons were uploaded and analyzed in Imaris 8 (Oxford Instruments) using FilamentTracer module. Filaments and spines were traced, segmented, and assigned using semiautomatic and manual detection methods. Two-dimensional rendering of cells was created based on FilamentTracer analysis in Imaris, using cylinder display. Data were extracted into Prism 6 (Graphpad), and statistical significance was determined by performing Student’s t test and ANOVA. Golgi staining Brains were harvested from 6-mo-old mice and Golgi stained using the FD Neurotech Golgi kit. In brief, brains were immersed in equal parts solution A and solution B and allowed to sit for 2 wk. Brains were moved into solution C for 72 h, and then sliced to 100 µM thickness using a sliding microtome. Electrophysiology Extracellular recordings of field excitatory postsynaptic potentials (fEPSP), were measured from acute hippocampal slices (400 µm) prepared as previously described (Pehar et al., 2010). In brief, enameled bipolar platinum–tungsten-stimulating electrodes were placed along Schaffer collaterals, and fEPSPs were recorded with ACSF-filled recording electrodes (2–4 MOhm) from area CA1 stratum radiatum. Baseline responses were set to an intensity that evoked an fEPSP with a slope of 50% that of the maximum evoked response. For LTP, 3x theta burst (3xTBS) consisted of a total of three trains of 10 bursts (each burst consisting of four stimulations at a frequency of 100 Hz) with an interburst interval of 200 ms. For LTD, paired pulse low frequency stimulation consisted of 900 pairs of stimuli with a 50-ms paired pulse interval. Input/output curves were performed before and at the end of LTD recordings to ensure slice health. Data were analyzed by two-way ANOVA (treatment and time) with repeated measures (mixed model) and Bonferroni post-hoc tests. Postsynaptic densities (PSDs) Hippocampal tissue was homogenized in Tris-acetate buffer (50 mM, pH 7.4) containing 100 µM EGTA, 0.32 M sucrose, and protease and phosphatase inhibitors. Homogenate was centrifuged at 1,000 g for 10 min. Supernatants were then centrifuged at 14,000 g for 20 min, and the resulting pellet was defined as the crude synaptosomal fraction. Synaptosomes were resuspended in Tris-acetate buffer. To purify the PSDs, the synaptosomal fraction was diluted with 20 mM Tris-HCl, pH 6.0, and 0.1 mM CaCl2 containing 1% Triton X-100, mixed for 20 min at 4°C on a plate rocker, and then centrifuged at 25 p.s.i. for 20 min using an air-driven ultracentrifuge (Airfuge; Beckman Coulter). The pellet was resuspended in 20 mM Tris-HCl, pH 8.0, and 0.1 mM CaCl2 containing 1% Triton X-100. Samples were mixed again for 20 min at 4°C on the plate rocker, and then centrifuged on the Airfuge (25 p.s.i.) for 20 min. The insoluble pellet containing the PSD fraction was suspended in Tris-acetate buffer (50 mM, pH 7.4, 100 µM EGTA, and 0.32 M sucrose and proteases inhibitors) and stored at −80°C until use. Immunoblotting Western blotting was performed on a 4–12% Bis-Tris SDS-PAGE system (NuPAGE; Invitrogen) as previously described (Costantini et al., 2006; Pehar et al., 2010; Peng et al., 2014). The following primary antibodies were used in this study: Neurexin 1 (1:1,000; Thermo Fisher Scientific; #18730) Neuroligin 3 (1:1,000; Thermo Fisher Scientific; #18849); Rap2a (1:1,000; Thermo Fisher Scientific; #23298); Synapsin 3 (1:1,000; Thermo Fisher Scientific; #25658); AMPA2/3/4 (1:1,000; Cell Signaling Technologies; #2460), mGluR5 (1:3,000; EMD Millipore; #76316); Gapdh (1:5,000; EMD Millipore; #2302); Synaptogyrin 1 (1:1,000; Abcam; #113886); Rab12 (1:1,000; Abcam; #170046); Psd-95 (1:1,000; Cell Signaling Technologies; #2507); S100 (1:1,000; Abcam; #4066); Iba1 (1:1,000; Wako; #016-20001); βIII Tubulin (1:1,000; Abcam; #18207); L1CAM (1:1,000; Abcam; #24345); NFL160 (1:5,000; Abcam; #9034); H3K27ac (1:1,000; Active Motif; #39134); and H3K27met (1:1,000; Active Motif; #39157). HRP-conjugated anti–mouse or anti–rabbit secondary antibodies were used for chemiluminescent detection (ImageQuant LAS4000; GE Healthcare); goat anti–rabbit Alexa Fluor 680–conjugated or anti–mouse Alexa Fluor 800–conjugated secondary antibodies were used for infrared imaging (LICOR Odyssey Infrared Imaging System; LI-COR Biosciences). Proteomics studies and mass spectrometry ER acetylome The neuron-specific acetylome was determined as previously described (Pehar et al., 2012b). In brief, adult neurons were isolated with the gentle MACS Dissociator (Miltenyi Biotec). Proteins were digested with trypsin before high-resolution high-accuracy LC-MS/MS analysis at the Mass Spectrometry facility at the University of Wisconsin-Madison. Peptides and proteins were identified with the Mascot search engine (Matrix Science) via automated database searching of all tandem mass spectra. The output of the MS analysis was further processed to select proteins that insert into the secretory pathway. Proteomics Hippocampi of three pairs of Tg and control mice were dissected and two hippocampi from the same mouse were placed into the same 1.5-ml tube and stored in –80°C until homogenization. A 200 µl aliquot of lysis buffer (8 M urea, 100 mM Tris-HCl, and protease inhibitor cocktail) was added to each tube, the bottom of which was immersed in the ice bath. Next, tissues were immediately sonicated by Sonic Dismembrator (Model 120; Thermo Fisher Scientific) at 120 W output with three bursts that each lasted 45 s each. Tissues were allowed to cool on the ice bath for 30 s after each burst. The protein concentration was determined by 660 nm protein assay and 10 µg protein of each sample was used for tryptic digestion. The protein of each sample was denatured with 8 µl of 8 M urea that was diluted by 50 mM ammonium bicarbonate buffer and reduced by 1 µl of 0.50 M DTT by incubation at 37°C for 1 h. After incubation, the reduced sample was alkylated by 2.7 µl of 0.55 M IAA and kept in the dark for 15 min. 1 µl of 0.50 M DTT was applied to quench the IAA alkylating reaction for 10 min. 70 µl of 50 mM ammonium bicarbonate was used to dilute the urea to a final concentration of 1 M, and 0.5 µg trypsin was subsequently added into the solution. The protein digestion reaction was incubated for 18 h at 37°C, and subsequently quenched by 2.5 µl of 10% formic acid. The solid phase extraction of the tryptic peptides were performed by Varian 100 µl C 18 Omix Tips (Agilent). The peptides were sequentially eluted with 60 µl 50% ACN in 0.1% formic acid, dried by SpeedVac, and reconstituted with 9 µl 0.1% formic acid. 1 µl 1 µM enolase digestion standards were spiked into each sample as internal standards. Online reversed phase liquid chromatography separation of the tryptic peptides was performed on a nanoAcquity UPLC (Waters Corp.), and subsequently analyzed with Q Exactive quadrupole orbitrap mass spectrometer (Thermo Fisher Scientific). The chromatographic separation was performed by mobile phase A that consisted of 0.1% formic acid in water and mobile phase B with 0.1% formic acid in ACN. 1 µl of each sample was injected onto a custom-packed column that was made by silica tubing (360 µm o.d., 75 µm i.d.) with emitter tips pulled by a laser puller. The column was packed with Waters 150 Å, 1.7 µm, BEH C18 material. The LC gradient started with 5% B and increased to 30% B over 120 min at a flow rate of 350 nl/min. Ions were generated under positive electrospray ionization (ESI) at 2.8 KV capillary voltage; 275°C capillary temperature; 30% collision energy via high-energy collision dissociation (HCD). MS1 scans were acquired over m/z 200–2,000 at 70 k resolution and data-dependent tandem MS scan were acquired via selection of the top 10 most abundant precursor ions by HCD fragmentation with an isolation window of m/z 2.0 at a resolution of 17,500. Other parameters include: automatic gain control 1 × 10−5; maximum ion injection time, 100 ms; dynamic exclusion enabled with unassigned, +1, and greater than +8 charges ignored for MS/MS selection. Each sample was analyzed as technical triplicates. Tryptic peptide identification was performed via Proteome Discoverer 1.4 (Thermo Fisher Scientific). FASTA file was downloaded from Uniprot’s reference database of Mus musculus (release 2014_08) with manually added yeast enolase fasta file (SwissProt P00924). Other parameters include: allowed missed cleavage, 1; enzyme, trypsin; fixed modification, carbamidomethylation of cysteine (+57.0215 D); variable modification, oxidation of methionine (+15.9949 D); peptide mass tolerance, 10 ppm; fragment mass tolerance, 0.1 D. q value was set to achieve 1% false discovery rate via Percolator to verify the identified peptides, and the results were filtered by high confidence peptide identification. For label-free quantification of acquired data, SIEVE (Version 2.1; Thermo Fisher Scientific) was applied. The following parameters were used in the peak alignment and frame generation: m/z min = 300; m/z max = 1,500; frame time width = 6.0 min; frame m/z width = 0.02 D; retention time start = 15 min; retention time stop = 90 min; and peak intensity threshold = 10,000. Alignment was validated using the spiked enolase tryptic peptides with accurate m/z and retention time. Proteins with P < 0.05 were selected. Histone extraction and LC-MS/MS analysis Quantification of histone posttranslational modifications was performed as recently described (Krautkramer et al., 2015). Hippocampi were dounced in ice-cold nuclear isolation buffer (10 mM Tris-HCl, 10 mM NaCl, 3 mM MgCl2, and 0.15% NP-40), and nuclei were pelleted by centrifugation. Histones were acid extracted from nuclei, derivatized with propionic anhydride, and trypsinized as previously described (Lin and Garcia, 2012). Histone peptides were injected onto a Dionex Ultimate3000 nanoflow HPLC with a Waters NanoEase C18 column (100 µm × 15 cm, 3 µm) coupled to a Thermo Fisher Scientific Q-Exactive mass spectrometer at 700 nl/min using a 2–35% gradient of acetonitrile. Histone modifications were identified and quantified using the MaxQuant (v 1.4.1.2). Spectra were searched against the human SwissProt database using a 20 ppm mass tolerance for the first search and a 4.5 ppm mass tolerance for the main search. The enzyme was specified as ArgC with zero missed cleavages. Variable modifications were set as follows: acetyl(K), butyryl(K), dimethyl(K), propionlyl(K), trimethyl(K), and propionyl(peptide N terminus). A reverse decoy database was generating within MaxQuant, and the false discovery rate was set to <0.01 for peptide spectrum matches, proteins, and modification sites. Changes in protein expression as identified by mass spectrometry were further analyzed using the Protein ANalysis THrough Evolutionary Relationship (PANTHER) gene list analysis (Mi et al., 2013) and STRING 10 software (Szklarczyk et al., 2015) to highlight protein–protein interactions. Acetyl-CoA determinations Cytosol was isolated from total hippocampal tissue, isolated adult cortical neurons, and H4 cells using differential centrifugation as previously described (Ko and Puglielli, 2007; Pehar et al., 2014). Acetyl-CoA was measured using the PicoProbe kit (Abcam). Trafficking of secretory proteins Nascent glycoproteins were labeled using Click-It ManNAz (Life Technologies) and visualized using the Click-It Cell Reaction kit (Life Technologies). Biotinylated cell surface proteins were isolated using the Cell Surface Protein Isolation kit (Thermo Fisher Scientific) and visualized by Western blot with Streptavidin-HRP (1:10,000). Rate of protein biosynthesis was assessed using the Click-It Plus OPP Alexa Fluor 488 Protein Synthesis Assay kit (Life Technologies). ChIP analysis Control and AT-1–overexpressing H4 cells were cultured in complete medium in 150-mm Petri dishes until ∼70% confluent. The cells were then fixed by the addition of 280 µl of 37% formaldehyde (Sigma-Aldrich) to 10 ml of culture medium for 10 min at 37°C, harvested, and processed for ChIP using a commercially available kit (Active Motif). H3K27-DNA immune complexes were precipitated with a polyclonal antibody against H3K27ac (Active Motif). PCR was performed using primer sets for Acly, Slc25a1, and Acss2 centered on the first exon region. Primer sets used were as follows: Acly: 5′-GCTTAGCCTGTGAGCTGAT-3′ (Sense), 5′-AGGTGGTGCAGATGTACTTG-3′ (AntiSense); Slc25a1: 5′-GAATGGGTCGTGGTCTCAGTAG-3′ (Sense), 5′-CTGCTAGGATTGCCTTCCC-3′ (AntiSense); Acss2: 5′-AAGAGCGGCAGTGGAAG-3′ (Sense), 5′-TTCGATCCAGCACGTTGTAG-3′ (AntiSense). No antibody and normal IgG served as internal negative controls. Real-time PCR Real-time PCR was performed using the Roche 480 lightcycler and Sybr Green Real Time PCR Master Mix (Life Technologies). The cycling parameters were: 95°C, 10 s; 52°C, 20 s; and 72°C, 30 s, for a maximum of 45 cycles. Controls without reverse transcription were included in each assay. PCR primers specific to each gene were: Slc25a1 and Acly (Bio-Rad Laboratories; Prime PCR assay); Acss2: 5′-AAGAGCGGCAGTGGAAG-3′ (sense), 5′-TTCGATCCAGCACGTTGTAG-3′ (antisense); Gapdh: 5′-ACCACAGTCCATGCCATCAC-3′ (sense), 5′-TCCACCACCCTGTTGCTGTA-3′ (antisense). Expression levels were normalized against Gapdh levels and are expressed as percent of control. Statistical analysis Data analysis was performed using InStat 3.06 statistical software (GraphPad). Data are expressed as mean ± SD. Unless otherwise specified, comparison of the means was performed using Student’s t test or one-way ANOVA, followed by Tukey-Kramer multiple comparisons test. Differences were declared statistically significant if P < 0.05. Unless otherwise specified, throughout the paper the following statistical significance is used: *, P < 0.05; **, P < 0.005; #, P < 0.0005. Online supplemental material Table S1 lists all acetylated peptides identified in the adult brain of AT-1 Tg mice. Table S2 provides the entire list of proteins identified by MS in the brain of WT and AT-1 Tg mice. Table S3 lists all posttranslational modifications of histone proteins identified in WT and AT-1 Tg mice. Online supplemental material is available at http://www.jem.org/cgi/content/full/jem.20151776/DC1. Supplementary Material Supplemental Materials Acknowledgments We thank Dr. Qiang Chang and Dr. Albee Messing for critical reading of an early version of this manuscript. This work was supported by a VA Merit Award (BX001638), National Institutes of Health grants (NS094154, GM065386, DK071801, and S10RR029531), and a core grant to the Waisman Center from National Institute of Child Health and Human Development (P30 HD03352). E.R. Chapman is an Investigator of the Howard Hughes Medical Institute. R. Hullinger was supported by a National Science Foundation Graduate Research Fellowship. The authors declare no competing financial interests. Author contributions: R. Hullinger, M. Li, J. Wang, Y. Peng, E. Bomba-Warczak, J.A. Dowell, H.A. Mitchell, J.M. Denu, L. Li, and L. Puglielli conceived and designed experiments. R. Hullinger, M. Li, J. Wang, Y. Peng, E. Bomba-Warczak, J.A. Dowell, and H.A. Mitchell performed experiments and analyzed data. C. Burger, E.R. Chapman, J.M. Denu, L. Li, and L. Puglielli analyzed data. R. Hullinger and L. Puglielli wrote the paper. All authors contributed to writing and revision of the paper. Abbreviations used: AceCS acetyl-CoA synthetase 2 ACLY ATP-citrate lyase ASD autism spectrum disorder AT-1 acetyl-CoA transporter 1 CoA Coenzyme A FC fear conditioning LTD long-term depression LTP long-term potentiation ManNaz azide-modified mannosamine MEF mouse embryo fibroblasts MWM Morris water maze NOR novel object recognition OPP O-propargyl-puromycin TBS theta burst stimulation ==== Refs Alvarez-Castelao, B., and E.M. Schuman. 2015. The regulation of synaptic protein turnover. J. Biol. Chem. 290 :28623–28630. 10.1074/jbc.R115.657130 26453306 Amini, M., C.L. Ma, R. Farazifard, G. Zhu, Y. Zhang, J. Vanderluit, J.S. Zoltewicz, F. Hage, J.M. Savitt, D.C. Lagace, 2013. Conditional disruption of calpain in the CNS alters dendrite morphology, impairs LTP, and promotes neuronal survival following injury. J. Neurosci. 33 :5773–5784. 10.1523/JNEUROSCI.4247-12.2013 23536090 Auerbach, B.D., E.K. Osterweil, and M.F. Bear. 2011. Mutations causing syndromic autism define an axis of synaptic pathophysiology. Nature. 480 :63–68. 10.1038/nature10658 22113615 Auffret, A., V. Gautheron, M. Repici, R. Kraftsik, H.T. Mount, J. Mariani, and C. Rovira. 2009. Age-dependent impairment of spine morphology and synaptic plasticity in hippocampal CA1 neurons of a presenilin 1 transgenic mouse model of Alzheimer’s disease. J. Neurosci. 29 :10144–10152. 10.1523/JNEUROSCI.1856-09.2009 19675248 Cahoy, J.D., B. Emery, A. Kaushal, L.C. Foo, J.L. Zamanian, K.S. Christopherson, Y. Xing, J.L. Lubischer, P.A. Krieg, S.A. Krupenko, 2008. A transcriptome database for astrocytes, neurons, and oligodendrocytes: a new resource for understanding brain development and function. J. Neurosci. 28 :264–278. 10.1523/JNEUROSCI.4178-07.2008 18171944 Chia, P.H., P. Li, and K. Shen. 2013. Cell biology in neuroscience: cellular and molecular mechanisms underlying presynapse formation. J. Cell Biol. 203 :11–22. 10.1083/jcb.201307020 24127213 Choudhary, C., C. Kumar, F. Gnad, M.L. Nielsen, M. Rehman, T.C. Walther, J.V. Olsen, and M. Mann. 2009. Lysine acetylation targets protein complexes and co-regulates major cellular functions. Science. 325 :834–840. 10.1126/science.1175371 19608861 Costantini, C., H. Scrable, and L. Puglielli. 2006. An aging pathway controls the TrkA to p75NTR receptor switch and amyloid beta-peptide generation. EMBO J. 25 :1997–2006. 10.1038/sj.emboj.7601062 16619032 Costantini, C., M.H. Ko, M.C. Jonas, and L. Puglielli. 2007. A reversible form of lysine acetylation in the ER and Golgi lumen controls the molecular stabilization of BACE1. Biochem. J. 407 :383–395. 10.1042/BJ20070040 17425515 Craig, A.M., and Y. Kang. 2007. Neurexin-neuroligin signaling in synapse development. Curr. Opin. Neurobiol. 17 :43–52. 10.1016/j.conb.2007.01.011 17275284 Dere, E., L. Dahm, D. Lu, K. Hammerschmidt, A. Ju, M. Tantra, A. Kästner, K. Chowdhury, and H. Ehrenreich. 2014. Heterozygous ambra1 deficiency in mice: a genetic trait with autism-like behavior restricted to the female gender. Front. Behav. Neurosci. 8 :181. 10.3389/fnbeh.2014.00181 24904333 Ding, Y., C.D. Dellisanti, M.H. Ko, C. Czajkowski, and L. Puglielli. 2014. The endoplasmic reticulum-based acetyltransferases, ATase1 and ATase2, associate with the oligosaccharyltransferase to acetylate correctly folded polypeptides. J. Biol. Chem. 289 :32044–32055. 10.1074/jbc.M114.585547 25301944 Feldman, D.E. 2009. Synaptic mechanisms for plasticity in neocortex. Annu. Rev. Neurosci. 32 :33–55. 10.1146/annurev.neuro.051508.135516 19400721 Gkogkas, C.G., A. Khoutorsky, I. Ran, E. Rampakakis, T. Nevarko, D.B. Weatherill, C. Vasuta, S. Yee, M. Truitt, P. Dallaire, 2013. Autism-related deficits via dysregulated eIF4E-dependent translational control. Nature. 493 :371–377. 10.1038/nature11628 23172145 Hardies, K., C.G. de Kovel, S. Weckhuysen, B. Asselbergh, T. Geuens, T. Deconinck, A. Azmi, P. May, E. Brilstra, F. Becker, autosomal recessive working group of the EuroEPINOMICS RES Consortium. 2015. Recessive mutations in SLC13A5 result in a loss of citrate transport and cause neonatal epilepsy, developmental delay and teeth hypoplasia. Brain. 138 :3238–3250. 10.1093/brain/awv263 26384929 Hirschberg, C.B., P.W. Robbins, and C. Abeijon. 1998. Transporters of nucleotide sugars, ATP, and nucleotide sulfate in the endoplasmic reticulum and Golgi apparatus. Annu. Rev. Biochem. 67 :49–69. 10.1146/annurev.biochem.67.1.49 9759482 Huganir, R.L., and R.A. Nicoll. 2013. AMPARs and synaptic plasticity: the last 25 years. Neuron. 80 :704–717. 10.1016/j.neuron.2013.10.025 24183021 Huppke, P., C. Brendel, V. Kalscheuer, G.C. Korenke, I. Marquardt, P. Freisinger, J. Christodoulou, M. Hillebrand, G. Pitelet, C. Wilson, 2012. Mutations in SLC33A1 cause a lethal autosomal-recessive disorder with congenital cataracts, hearing loss, and low serum copper and ceruloplasmin. Am. J. Hum. Genet. 90 :61–68. 10.1016/j.ajhg.2011.11.030 22243965 Jonas, M.C., M. Pehar, and L. Puglielli. 2010. AT-1 is the ER membrane acetyl-CoA transporter and is essential for cell viability. J. Cell Sci. 123 :3378–3388. 10.1242/jcs.068841 20826464 Kawabe, H., A. Neeb, K. Dimova, S.M. Young Jr., M. Takeda, S. Katsurabayashi, M. Mitkovski, O.A. Malakhova, D.E. Zhang, M. Umikawa, 2010. Regulation of Rap2A by the ubiquitin ligase Nedd4-1 controls neurite development. Neuron. 65 :358–372. 10.1016/j.neuron.2010.01.007 20159449 Ko, M.H., and L. Puglielli. 2007. The sterol carrier protein SCP-x/pro-SCP-2 gene has transcriptional activity and regulates the Alzheimer disease gamma-secretase. J. Biol. Chem. 282 :19742–19752. 10.1074/jbc.M611426200 17485462 Ko, M.H., and L. Puglielli. 2009. Two endoplasmic reticulum (ER)/ER Golgi intermediate compartment-based lysine acetyltransferases post-translationally regulate BACE1 levels. J. Biol. Chem. 284 :2482–2492. 10.1074/jbc.M804901200 19011241 Kouser, M., H.E. Speed, C.M. Dewey, J.M. Reimers, A.J. Widman, N. Gupta, S. Liu, T.C. Jaramillo, M. Bangash, B. Xiao, 2013. Loss of predominant Shank3 isoforms results in hippocampus-dependent impairments in behavior and synaptic transmission. J. Neurosci. 33 :18448–18468. 10.1523/JNEUROSCI.3017-13.2013 24259569 Kouzarides, T. 2000. Acetylation: a regulatory modification to rival phosphorylation? EMBO J. 19 :1176–1179. 10.1093/emboj/19.6.1176 10716917 Krautkramer, K.A., L. Reiter, J.M. Denu, and J.A. Dowell. 2015. Quantification of SAHA-dependent changes in histone modifications using data-independent acquisition mass spectrometry. J. Proteome Res. 14 :3252–3262. 10.1021/acs.jproteome.5b00245 26120868 Krumm, N., B.J. O’Roak, E. Karakoc, K. Mohajeri, B. Nelson, L. Vives, S. Jacquemont, J. Munson, R. Bernier, and E.E. Eichler. 2013. Transmission disequilibrium of small CNVs in simplex autism. Am. J. Hum. Genet. 93 :595–606. 10.1016/j.ajhg.2013.07.024 24035194 Laughlin, S.T., and C.R. Bertozzi. 2007. Metabolic labeling of glycans with azido sugars and subsequent glycan-profiling and visualization via Staudinger ligation. Nat. Protoc. 2 :2930–2944. 10.1038/nprot.2007.422 18007630 Lin, S., and B.A. Garcia. 2012. Examining histone posttranslational modification patterns by high-resolution mass spectrometry. Methods Enzymol. 512 :3–28. 10.1016/B978-0-12-391940-3.00001-9 22910200 Lin, P., J. Li, Q. Liu, F. Mao, J. Li, R. Qiu, H. Hu, Y. Song, Y. Yang, G. Gao, 2008. A missense mutation in SLC33A1, which encodes the acetyl-CoA transporter, causes autosomal-dominant spastic paraplegia (SPG42). Am. J. Hum. Genet. 83 :752–759. 10.1016/j.ajhg.2008.11.003 19061983 Lugo, J.N., G.D. Smith, E.P. Arbuckle, J. White, A.J. Holley, C.M. Floruta, N. Ahmed, M.C. Gomez, and O. Okonkwo. 2014. Deletion of PTEN produces autism-like behavioral deficits and alterations in synaptic proteins. Front. Mol. Neurosci. 7 :27. 10.3389/fnmol.2014.00027 24795561 Mak, A.B., M. Pehar, A.M. Nixon, R.A. Williams, A.C. Uetrecht, L. Puglielli, and J. Moffat. 2014. Post-translational regulation of CD133 by ATase1/ATase2-mediated lysine acetylation. J. Mol. Biol. 426 :2175–2182. 10.1016/j.jmb.2014.02.012 24556617 Martenson, J.S., and S. Tomita. 2015. Synaptic localization of neurotransmitter receptors: comparing mechanisms for AMPA and GABAA receptors. Curr. Opin. Pharmacol. 20 :102–108. 10.1016/j.coph.2014.11.011 25529200 Mi, H., A. Muruganujan, and P.D. Thomas. 2013. PANTHER in 2013: modeling the evolution of gene function, and other gene attributes, in the context of phylogenetic trees. Nucleic Acids Res. 41 :D377–D386. 10.1093/nar/gks1118 23193289 Neuhofer, D., C.M. Henstridge, B. Dudok, M. Sepers, O. Lassalle, I. Katona, and O.J. Manzoni. 2015. Functional and structural deficits at accumbens synapses in a mouse model of Fragile X. Front. Cell. Neurosci. 9 :100. 10.3389/fncel.2015.00100 25859182 Pehar, M., and L. Puglielli. 2013. Lysine acetylation in the lumen of the ER: a novel and essential function under the control of the UPR. Biochim. Biophys. Acta. 1833 :686–697. 10.1016/j.bbamcr.2012.12.004 23247107 Pehar, M., K.J. O’Riordan, M. Burns-Cusato, M.E. Andrzejewski, C.G. del Alcazar, C. Burger, H. Scrable, and L. Puglielli. 2010. Altered longevity-assurance activity of p53:p44 in the mouse causes memory loss, neurodegeneration and premature death. Aging Cell. 9 :174–190. 10.1111/j.1474-9726.2010.00547.x 20409077 Pehar, M., M.C. Jonas, T.M. Hare, and L. Puglielli. 2012 a. SLC33A1/AT-1 protein regulates the induction of autophagy downstream of IRE1/XBP1 pathway. J. Biol. Chem. 287 :29921–29930. 10.1074/jbc.M112.363911 22787145 Pehar, M., M. Lehnus, A. Karst, and L. Puglielli. 2012 b. Proteomic assessment shows that many endoplasmic reticulum (ER)-resident proteins are targeted by N(epsilon)-lysine acetylation in the lumen of the organelle and predicts broad biological impact. J. Biol. Chem. 287 :22436–22440. 10.1074/jbc.C112.362871 22628546 Pehar, M., M.H. Ko, M. Li, H. Scrable, and L. Puglielli. 2014. P44, the ‘longevity-assurance’ isoform of P53, regulates tau phosphorylation and is activated in an age-dependent fashion. Aging Cell. 13 :449–456. 10.1111/acel.12192 24341977 Peng, Y., M. Li, B.D. Clarkson, M. Pehar, P.J. Lao, A.T. Hillmer, T.E. Barnhart, B.T. Christian, H.A. Mitchell, B.B. Bendlin, 2014. Deficient import of acetyl-CoA into the ER lumen causes neurodegeneration and propensity to infections, inflammation, and cancer. J. Neurosci. 34 :6772–6789. 10.1523/JNEUROSCI.0077-14.2014 24828632 Pietrocola, F., L. Galluzzi, J.M. Bravo-San Pedro, F. Madeo, and G. Kroemer. 2015. Acetyl coenzyme A: a central metabolite and second messenger. Cell Metab. 21 :805–821. 10.1016/j.cmet.2015.05.014 26039447 Prasad, A., D. Merico, B. Thiruvahindrapuram, J. Wei, A.C. Lionel, D. Sato, J. Rickaby, C. Lu, P. Szatmari, W. Roberts, 2012. A discovery resource of rare copy number variations in individuals with autism spectrum disorder. G3 (Bethesda). 2 :1665–1685. 10.1534/g3.112.004689 23275889 Prasun, P., S. Young, G. Salomons, A. Werneke, Y.H. Jiang, E. Struys, M. Paige, M.L. Avantaggiati, and M. McDonald. 2015. Expanding the clinical spectrum of mitochondrial citrate carrier (SLC25A1) deficiency: facial dysmorphism in siblings with epileptic encephalopathy and combined D,L-2-hydroxyglutaric aciduria. JIMD Rep. 19 :111–115. 10.1007/8904_2014_378 25614306 Pucilowska, J., J. Vithayathil, E.J. Tavares, C. Kelly, J.C. Karlo, and G.E. Landreth. 2015. The 16p11.2 deletion mouse model of autism exhibits altered cortical progenitor proliferation and brain cytoarchitecture linked to the ERK MAPK pathway. J. Neurosci. 35 :3190–3200. 10.1523/JNEUROSCI.4864-13.2015 25698753 Sanders, S.J., A.G. Ercan-Sencicek, V. Hus, R. Luo, M.T. Murtha, D. Moreno-De-Luca, S.H. Chu, M.P. Moreau, A.R. Gupta, S.A. Thomson, 2011. Multiple recurrent de novo CNVs, including duplications of the 7q11.23 Williams syndrome region, are strongly associated with autism. Neuron. 70 :863–885. 10.1016/j.neuron.2011.05.002 21658581 Shi, L., and B.P. Tu. 2015. Acetyl-CoA and the regulation of metabolism: mechanisms and consequences. Curr. Opin. Cell Biol. 33 :125–131. 10.1016/j.ceb.2015.02.003 25703630 Speed, H.E., M. Kouser, Z. Xuan, J.M. Reimers, C.F. Ochoa, N. Gupta, S. Liu, and C.M. Powell. 2015. Autism-associated insertion mutation (InsG) of Shank3 exon 21 causes impaired synaptic transmission and behavioral deficits. J. Neurosci. 35 :9648–9665. 10.1523/JNEUROSCI.3125-14.2015 26134648 Szklarczyk, D., A. Franceschini, S. Wyder, K. Forslund, D. Heller, J. Huerta-Cepas, M. Simonovic, A. Roth, A. Santos, K.P. Tsafou, 2015. STRING v10: protein-protein interaction networks, integrated over the tree of life. Nucleic Acids Res. 43 (D1 ):D447–D452. 10.1093/nar/gku1003 25352553 Thevenon, J., M. Milh, F. Feillet, J. St-Onge, Y. Duffourd, C. Jugé, A. Roubertie, D. Héron, C. Mignot, E. Raffo, 2014. Mutations in SLC13A5 cause autosomal-recessive epileptic encephalopathy with seizure onset in the first days of life. Am. J. Hum. Genet. 95 :113–120. 10.1016/j.ajhg.2014.06.006 24995870 Wellen, K.E., G. Hatzivassiliou, U.M. Sachdeva, T.V. Bui, J.R. Cross, and C.B. Thompson. 2009. ATP-citrate lyase links cellular metabolism to histone acetylation. Science. 324 :1076–1080. 10.1126/science.1164097 19461003 Xu, M., J.S. Nagati, J. Xie, J. Li, H. Walters, Y.A. Moon, R.D. Gerard, C.L. Huang, S.A. Comerford, R.E. Hammer, 2014. An acetate switch regulates stress erythropoiesis. Nat. Med. 20 :1018–1026. 10.1038/nm.3587 25108527

PMC004xxxxxx/PMC4925026.txt

==== Front J Exp Med J Exp Med jem jem The Journal of Experimental Medicine 0022-1007 1540-9538 The Rockefeller University Press 27353089 20160081 10.1084/jem.20160081 Research Articles Article Mast cells regulate myofilament calcium sensitization and heart function after myocardial infarction The role of mast cells in myocardial infarction http://orcid.org/0000-0002-0016-887X Ngkelo Anta 1 Richart Adèle 1 http://orcid.org/0000-0002-5192-2860 Kirk Jonathan A. 2 http://orcid.org/0000-0003-3184-3740 Bonnin Philippe 3 Vilar Jose 1 Lemitre Mathilde 1 Marck Pauline 4 Branchereau Maxime 4 http://orcid.org/0000-0002-5268-0584 Le Gall Sylvain 1 http://orcid.org/0000-0002-1559-0774 Renault Nisa 1 Guerin Coralie 5 http://orcid.org/0000-0002-4970-8988 Ranek Mark J. 2 http://orcid.org/0000-0002-8081-8296 Kervadec Anaïs 1 http://orcid.org/0000-0003-1222-0732 Danelli Luca 678 Gautier Gregory 67 Blank Ulrich 678 http://orcid.org/0000-0002-3560-1389 Launay Pierre 67 http://orcid.org/0000-0002-6271-7125 Camerer Eric 1 http://orcid.org/0000-0002-2350-2914 Bruneval Patrick 19 http://orcid.org/0000-0002-9845-4064 Menasche Philippe 19 Heymes Christophe 4 Luche Elodie 10 http://orcid.org/0000-0001-9647-3248 Casteilla Louis 10 Cousin Béatrice 10 Rodewald Hans-Reimer 11 http://orcid.org/0000-0003-1596-2299 Kass David A. 2 Silvestre Jean-Sébastien 1 1 Institut National de la Santé et de la Recherche Médicale (INSERM), UMRS-970, Centre de Recherche Cardiovasculaire, Université Paris Descartes, Sorbonne Paris Cité, F-75015 Paris, France 2 Division of Cardiology, Johns Hopkins Medical Institutions, Baltimore, MD 212015 3 INSERM, U965, Hôpital Lariboisière–Fernand-Widal, Assistance Publique Hôpitaux de Paris, F-75010 Paris, France 4 INSERM, UMR-1048, Institut des Maladies Métaboliques et Cardiovasculaires, F-31004 Toulouse, France 5 National Cytometry Platform, Department of Infection and Immunity, Luxembourg Institute of Health, L-4354 Esch-sur-Alzette, Luxembourg 6 Laboratoire d’Excellence INFLAMEX, Université Paris Diderot, Sorbonne Paris Cité, F-75018 Paris, France 7 INSERM, U1149, F-75018 Paris, France 8 Centre National de la Recherche Scientifique (CNRS) ERL 8252, F-75018 Paris, France 9 Hôpital European George Pompidou, Assistance Publique Hôpitaux de Paris, F-75015 Paris, France 10 STROMALab, Etablissement Français du Sang, INSERM U1031, CNRS ERL 5311, Université de Toulouse, F-31004 Toulouse, France 11 Division of Cellular Immunology, German Cancer Research Center, D-69120 Heidelberg, Germany Correspondence to Jean-Sébastien Silvestre: jean-sebastien.silvestre@inserm.fr 27 6 2016 213 7 13531374 16 1 2016 12 5 2016 ©2016 Ngkelo et al. 2016 https://creativecommons.org/licenses/by-nc-sa/3.0/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 3.0 Unported license, as described at http://creativecommons.org/licenses/by-nc-sa/3.0/). Ngkelo et al. use a mast cell–deficient mouse model to reveal a protective role of mast cells in myocardial infarction, through regulation of the cardiac contractile machinery. Acute myocardial infarction (MI) is a severe ischemic disease responsible for heart failure and sudden death. Inflammatory cells orchestrate postischemic cardiac remodeling after MI. Studies using mice with defective mast/stem cell growth factor receptor c-Kit have suggested key roles for mast cells (MCs) in postischemic cardiac remodeling. Because c-Kit mutations affect multiple cell types of both immune and nonimmune origin, we addressed the impact of MCs on cardiac function after MI, using the c-Kit–independent MC-deficient (Cpa3Cre/+) mice. In response to MI, MC progenitors originated primarily from white adipose tissue, infiltrated the heart, and differentiated into mature MCs. MC deficiency led to reduced postischemic cardiac function and depressed cardiomyocyte contractility caused by myofilament Ca2+ desensitization. This effect correlated with increased protein kinase A (PKA) activity and hyperphosphorylation of its targets, troponin I and myosin-binding protein C. MC-specific tryptase was identified to regulate PKA activity in cardiomyocytes via protease-activated receptor 2 proteolysis. This work reveals a novel function for cardiac MCs modulating cardiomyocyte contractility via alteration of PKA-regulated force–Ca2+ interactions in response to MI. Identification of this MC-cardiomyocyte cross-talk provides new insights on the cellular and molecular mechanisms regulating the cardiac contractile machinery and a novel platform for therapeutically addressable regulators. ==== Body pmcAcute myocardial infarction (MI) is a severe ischemic disease responsible for sudden death and heart failure with prevalence rates rapidly increasing worldwide (White et al., 2014). The evolution in clinical practice has substantially reduced mortality and morbidity associated with this condition. However, given the adverse hemorrhagic effects of the integration of antithrombotic therapy and the high socioeconomic burden of ischemic heart disease, a need for novel effective targets is emerging (White and Chew, 2008). Hence, efforts are directed toward pivotal pathways shaping cardiac homeostasis such as the inflammatory cellular responses (Zouggari et al., 2013; Boag et al., 2015; de Couto et al., 2015) as well as the molecular mechanisms that drive cardiac contractile function (Gorski et al., 2015; Movsesian, 2015). Substantial interest has been drawn on the role of cardiac mast cells (MCs) in mediating postischemic adverse myocardial remodeling (Kritikou et al., 2016). MCs are innate immune cells, characterized morphologically by numerous cytoplasmic granules that contain a variety of mediators such as proteoglycans, histamine, proteases (chymase and tryptase), and proinflammatory cytokines that are released upon MC activation to influence the local tissue microenvironment (Wernersson and Pejler, 2014). To date, several studies investigating the role of MCs in cardiac function and remodeling have been contradicting (Janicki et al., 2015), which may relate to the use of c-Kit mutant mice (the c-kit W/Wv [Kitamura et al., 1978]) and the more recent Kit W-sh/W-sh mice (Kitamura et al., 1978; Grimbaldeston et al., 2005) with mutations in the gene encoding the receptor tyrosine kinase c-Kit with subsequent MC deficiency. Because deficient c-Kit signaling affects other lineages, including hematopoietic stem cells, progenitor cells, red blood cells, neutrophils, cardiomyocytes, melanocytes, and germ cells (Katz and Austen, 2011), it remains ambiguous to what extent MC absence is responsible for the observed phenotypes. Therefore, the distinct role of MCs, independently of c-Kit functions, on regulating postischemic cardiac remodeling and function is unknown. Here we addressed the role of MCs in regulating cardiac function and contractility in response to acute MI by using the recently developed “Cre-mediated MC eradication” (Cre-Master or Cpa3cre/+) mouse model, which yields constitutive and c-Kit–independent MC deficiency (Feyerabend et al., 2011). We show that MCs play a key role in regulating cardiomyocyte contractility and subsequently cardiac function after MI. We describe an MC-dependent mechanism of protein kinase A (PKA) activity and myofilament protein phosphorylation through MC-released tryptase. RESULTS MCs accumulate in the heart at day 7 after MI To investigate the kinetics of mature MC infiltration after MI, digested infarcted tissue underwent flow cytometry/imaging analysis. Mature MCs were identified as c-kit+FcεRI+ by flow cytometry (Fig. 1 A), and the combination of these markers’ expression was verified as corresponding to the typical granulated morphology of MCs by the side scatter light imaging on ImageStream (Fig. 1 B). MC numbers in the sham-operated hearts were very low, but a significant accumulation of MCs was observed at day 7 after MI (infarct: 30,341 ± 2,600 cells/g of tissue vs. sham: 628 ± 218 cells/g of tissue, P = 0.0025; Fig. 1 C). This was followed by a progressive decrease in MC numbers from day 10 until day 21 (Fig. 1 C). Based on metachromatic toluidine blue (TB) staining (Tallini et al., 2009), 91.3 ± 4.1% of cardiac MCs were degranulated at day 7 after MI (Fig. 1, D and E). In addition, there was a significant increase in the mRNA expression of mouse MC chymase (mMCP4) and tryptase (mMCP6) starting and/or peaking at day 7 in the infarcted myocardium (vs. sham-operated myocardium; Fig. 1, F and G), consistent with the connective tissue MC phenotype (CTMC; Forman et al., 2006). c-Kit+tryptase+ cells were also identified in human biopsies from coronary artery bypass surgery (not depicted). Figure 1. Characterization of cardiac mature MCs after MI. (A and B) Representative fluorescence minus one control (FMO) and flow cytometry gating for mature cardiac MCs (c-kit+FcεRI+) at day 7 in sham-operated and infarcted myocardium (A) and representative image of ImageStream flow cytometer showing the morphology of Vybrant+c-kit+FcεRI+ cells under brightfield and side scatter (SSC) imaging with corresponding negative controls (B). (C) Time-dependent monitoring cardiac mature MCs/gram of cardiac tissue in response to infarction (n = 4–8, two independent experiments). *, P < 0.05; **, P < 0.01. Kruskal–Wallis and Dunn’s post hoc test for comparisons for sham versus MI at different time points. (D) Representative cardiac TB staining of mature MCs at granulated (above) and degranulated (below) states found both in the myocardium (left) and at the periphery (right). Arrows point to representative cardiac MCs. (E) Cardiac degranulated cells as the percentage of total cells counted by TB staining (n = 6, two independent experiments). *, P < 0.05. Kruskal–Wallis and Dunn’s post hoc test for comparisons at different time points. (F and G) Cardiac mRNA expression of chymase mMCP4 and tryptase mMCP6 at different time points after the sham operation or infarction (n = 6–8). *, P < 0.05; **, P < 0.01. Kruskal–Wallis and Dunn’s post hoc test for comparisons for sham versus MI at different time points. All data shown are representative of at least three independent experiments. All values are presented as mean ± SEM. Sham, sham-operated animals. MC progenitors (MCPs) are recruited into the heart and give rise to mature MCs in a stem cell factor (SCF)–dependent manner Mature MCs increase at tissue sites by infiltration of precursors from the peripheral blood and reach differentiation/maturation by locally secreted growth factors (Welker et al., 2000). To assess whether this is the case in response to MI, we identified MCPs as Lin−CD45+CD34+β7-integrin+FcγRII/III+ cells (Fig. 2 A; Chen et al., 2005; Schmetzer et al., 2016) and monitored their numbers in the BM, white adipose tissue (WAT), heart (Fig. 2, B–D), blood, and spleen (Fig. 2, J and K). MCPs increased significantly in numbers in the BM at days 3–5 and the WAT at day 3 in response to infarction (Fig. 2, B and C). In the infarcted heart, an accumulation of MCPs was observed at day 3 (infarct: 2,600 ± 600 cells/g vs. sham: 506 ± 215 cells/g of tissue, P = 0.017) that persisted until day 5 (Fig. 2 D). To evaluate whether the increase in MCP density in the heart is a result of local cardiac proliferation, BrdU incorporation was measured at days 3, 5, and 7 after infarction. A significant increase in BrdU-labeled MCPs was observed at day 5, 24 h after cumulative BrdU injections, with 2.8 ± 0.3% of the MCP cells being in a proliferative state (P = 0.007 vs. 0.27 ± 0.07% of BrdU+ MCPs on day 3; Fig. 2, E and F). The c-kit ligand SCF is responsible for the proliferation of MCPs both in vitro (Kirshenbaum et al., 1992) and in vivo (Matsuzawa et al., 2003) and can be produced by several tissue-resident cells, such as fibroblasts, endothelial cells, and cardiac stem cells (Guo et al., 2009; Xiang et al., 2013). SCF mRNA levels significantly increased in the peri-infarcted tissue at day 5 after MI compared with sham-operated mice (Fig. 2 G). Neutralization of SCF by the systemic administration of an anti-SCF antibody (Oliveira et al., 2002) reduced proliferation and numbers of MCPs at day 5 after MI (Fig. 2 H) as well as numbers of mature MCs at day 7 after MI (SCF antibody: 0.01 ± 0.001% vs. rabbit serum: 0.03 ± 0.005% of c-kit+FcεRI+ cells, P = 0.026; Fig. 2 I). MCPs were also detectable in the blood circulation, but numbers were too low to draw conclusions on their kinetics of circulation (Fig. 2 J). Similarly, no significant differences were obtained in the splenic MCP numbers in response to MI (Fig. 2 K). Figure 2. SCF-dependent accumulation of cardiac mature MCs. (A) Representative flow cytometry gating for MCPs (Lin−CD45+CD34+β7-integrin+FCγRII/III+) in the BM of sham-operated mice versus mice with MI at day 3. (B and C) Time-dependent increase of MCPs at days 3 and 5 after MI in the BM (B) and at day 3 in WAT (C) compared with sham control mice (n = 4–9, two independent experiments). *, P < 0.05, Kruskal–Wallis and Dunn’s post hoc test for comparisons of sham versus MI at different time points. (D) Numbers of MCPs/gram of cardiac tissue after infarct versus sham-operated mice (n = 4–9, two independent experiments). *, P < 0.05, Kruskal–Wallis and Dunn’s post hoc test for comparisons of sham versus MI at different time points. (E) Increased proliferation of cardiac MCPs at day 5 of infarct, 24 h after BrdU administration (n = 8, two independent experiments). **, P < 0.01, Kruskal–Wallis and Dunn’s post hoc test for comparisons at different time points. (F) Representative flow cytometry analysis of BrdU+ MCPs at days 3, 5, and 7 after infarct. (G) mRNA expression of SCF in the cardiac tissue in response to infarction (vs. sham) peaking at day 5 (n = 5–10, two independent experiments). *, P < 0.05, Kruskal–Wallis and Dunn’s post hoc test for comparisons of sham versus MI at different time points. (H) No significant effect of SCF antibody treatment on proliferation (left) and number (right) of CD45+β7-integrin+FCγRII/III+ cells at day 5 after MI (n = 6–8, two independent experiments). Mann–Whitney test for comparisons between groups. (I) SCF antibody reduced total numbers of mature MCs at day 7 after MI (n = 6–8, two independent experiments). *, P < 0.05, Mann–Whitney test for comparisons between groups. (J and K) MCPs (Lin−CD45+CD34+FCγRII/III+β7+) were identified circulating in the blood (J) at low percentages and in the spleen (K) with no statistically significant changes in response to MI. Kruskal–Wallis and Dunn’s post hoc test for comparisons of sham versus MI at different time points. All values are presented as mean ± SEM. anti, antibody against; D, day; Rb, rabbit; Sham, sham-operated animals. MCs preserve cardiac function after MI We next evaluated the functional effect of MC accumulation in the infarcted myocardium by evaluating cardiac function 2 wk after infarction in the MC-eradicated Cpa3cre/+ mice, which are selectively deficient for mature MCs (Feyerabend et al., 2011). c-kit+FcεRI+ cells were absent in the heart of Cpa3cre/+ mice at day 7 after the infarct, as confirmed both by flow cytometry and TB staining–assisted counting (Fig. 3, A and B). Heart function of Cpa3cre/+ mice was comparable with WT mice when sham operated. However, postinfarcted hearts lacking MCs displayed a significantly lower shortening fraction (SF [SF%]; Erdei et al., 2004; WT 28.15 ± 1.7% SF vs. Cpa3cre/+: 17.27 ± 1.9% SF, P = 0.0043; Fig. 3 C). In addition, the left ventricular internal postsystolic diameter was significantly increased in Cpa3cre/+ compared with WT mice after infarction. No significant differences were observed between the two groups in left ventricular end-diastolic internal diameter (Fig. 3 C) or heart rate (not depicted). Left ventricular posterior wall end-diastolic thickness and interventricular end-diastolic septal diameter were also decreased in Cpa3cre/+ compared with WT mice (not depicted). Figure 3. Depressed cardiac function after MI in Cpa3cre/+ MC-deficient mice. (A) Representative flow cytometry gating of c-kit+FcεRI+ cells in WT that are absent in Cpa3cre/+ mice at day 7 after MI (top); bar graphs show the numbers of mature MCs/gram of cardiac tissue (bottom; n = 5–8, two independent experiments). (B) Representative cardiac TB staining (top) and quantitative evaluation of mature MCs in infarcted heart (n = 5–6, two independent experiments). Arrows point to representative cardiac MCs. (C) Left ventricular %SF, LVIDd (left ventricular internal end-diastolic diameter), and LVIDs (left ventricular internal end-systolic diameter) measurements showing significant reduction of cardiac function (day 14) in Cpa3cre/+ infarcted mice compared with WT (Cpa3+/+) infarcted mice, with no differences on basal heart function (n = 6–7, two independent experiments). *, P < 0.05; **, P < 0.01, Kruskal–Wallis and Dunn’s post hoc test for comparisons between groups. (D) Cardiac fibrosis, infarct size, capillaries’ density, percentage of cell apoptosis, cardiomyocyte size, and number after infarction (day 14) in WT and Cpa3cre/+ mice (n = 6–7, two independent experiments). (E) Representative images used for quantification of fibrosis, capillary density, infarct size, and density of apoptotic cardiac cells. Arrows point to representative apoptotic cells. All values are presented as mean ± SEM. Sham, sham-operated animals; WT, Cpa+/+ littermates. To identify the cause of reduced heart function in the MC-deficient mice, we evaluated inflammation and cardiac remodeling after infarction. There was no difference in the levels of collagen deposition, infarct size, capillary density, cardiomyocyte size, and number, as well as in the amount of apoptotic resident cells in the Cpa3cre/+ mice compared with WT mice in response to infarction (Fig. 3, D and E). In line with these results, we found no differences in the numbers of recruited monocytes, macrophages, B lymphocytes, and CD8+ and CD3+ T lymphocytes to the infarcted heart in Cpa3cre/+ in response to MI compared with WT mice (Fig. 4 A). Accordingly, we assessed the release of key inflammatory mediators in the infarcted heart, including CCL2, CCL7, IL-6, IL-1β, and TNF, and found no difference in the Cpa3cre/+ mice compared with WT mice at all time points tested (Fig. 4 B). Figure 4. Cpa3cre/+ mice have a normal inflammatory response after MI. (A) Inflammatory cell density (monocytes, macrophages, and B and T lymphocytes) in Cpa3cre/+ mice was comparable with that of WT mice at days 3, 5, and 7 after MI (n = 7–8, two independent experiments). (B) Inflammatory mediators’ concentration (IL3, CCL7, CCL2, IL-6, IL-10, IL-1β, IL12p70, and TNF; pg/µg of protein) was analyzed by FlowCytomix, and no differences were observed between WT and Cpa3cre/+ mice at days 3, 5, 7, and 14 after MI (n = 6–8, 2 independent experiments). All values are presented as mean ± SEM. WT, Cpa+/+ littermates. Cardiac remodeling in c-Kit W/Wv and disodium cromoglycate (DSCG)–treated mice after MI Accumulation of cardiac MCs after infarction was SCF dependent, and a high percentage of cardiac MCs were degranulated in the infarcted heart. To assess whether these mechanisms are involved in the MC-dependent decline in cardiac function after the infarct, we used two widely studied animal models: the c-Kit W/Wv mice and systemic treatment with the degranulation inhibitor DSCG (Zhang et al., 2016). Both animal models are associated with impaired MC function and have been extensively used to investigate the biological role of MCs in a broad range of disease pathologies (Kovanen, 2009; He and Shi, 2013). We first assessed the number of cardiac MCs by TB staining after infarct in both mouse models. c-Kit W/Wv mice were MC poor but not completely deficient after MI, consistent with another report where inflammatory signaling can reverse MC absence in these mice (Feyerabend et al., 2011), and as expected, DSCG treatment did not affect cardiac MC density (Fig. 5 A). We next evaluated cardiac function at 2 wk after MI and found a significant reduction in left ventricular ejection fraction (EF) in both DSCG-treated (16.3 ± 1.6% EF) and c-Kit W/Wv mice (20 ± 2.5% EF) compared with their respective controls and littermates (PBS: 30.2 ± 2.8% EF, c-Kit+/+: 25.5 ± 4.6% EF; Fig. 5 B). However, in contrast with the Cpa3cre/+ mice, there were significant differences in cardiac remodeling of these animal models of MC-impaired function. DSCG treatment had no effect on the infarct size, capillary density, or cardiomyocyte size but significantly increased levels of cardiac interstitial fibrosis and cell apoptosis (Fig. 5, C and D). In contrast, the c-Kit W/Wv mice had a significant increase in the number of cardiac apoptotic cells after MI but no difference in the rest of the remodeling parameters evaluated (Fig. 5, C and D). Because remodeling parameters were unaffected in the MC-eradicated Cpa3cre/+ mice, we conclude that MCs preserve heart function after infarction but do not directly regulate cardiac remodeling. Figure 5. Cardiac function and remodeling in c-KitW/Wv and DSCG-treated mice. (A) TB-assisted MC counting in c-Kit W/Wv and DSCG-treated mice compared with their controls (c-Kit+/+ and PBS treated, respectively) on 5-µm slide sections (n = 6). (B) Reduced left ventricular (LV) cardiac function (EF%) at day 14 after MI in Cpa3Cre/+, DSCG-treated animals and c-Kit W/Wv compared with their respective controls (n = 6). (C) Evaluation of cardiac remodeling parameters in c-Kit W/Wv and DSCG-treated mice at day 14 after infarction, showing significantly increased levels of fibrosis in DSCG-treated mice and increased number of apoptotic cells in c-Kit W/Wv mice and DSCG-treated mice, with no effect on other parameters (n = 6). (D) Representative images of fibrosis, capillary density, infarct size, and apoptotic cardiac cells. Arrows point to representative apoptotic cells. All data represent two independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001, Mann–Whitney nonparametric test. All values are presented as mean ± SEM. PBS, PBS-treated animals; WT, Cpa+/+ littermates. Post-MI cardiac MCs originate primarily from the WAT Most studies on MC origin, circulation, and maturation are focused on BM-derived MCPs (Chen et al., 2005; Franco et al., 2010), but a recent study by Poglio et al. (2010) identified the WAT as a reservoir of progenitor MCs that do not originate from the BM and are able to home to peripheral tissues. Because we observed increased numbers of MCPs in both the WAT and the BM at days 3 and 5 (Fig. 2), we next identified the primary source of origin of cardiac MCs after infarction. We transplanted lethally irradiated C57BL/6 mice with BM-derived cells from WT and Cpa3cre/+ mice. SF% in WT mice reconstituted with Cpa3cre/+-derived BM cells was comparable with the SF% of WT animals reconstituted with BM-WT as measured 2 wk after the infarct (Fig. 6 A). In this line, reconstitution of WT mice with Cpa3cre/+-derived BM cells did not lead to absence of mature cardiac MCs after infarction, and conversely, reconstitution of Cpa3cre/+ mice with WT-derived BM cells did not reverse the absence of mature MCs in the Cpa3cre/+ mice after infarction (Fig. 6 B). Poglio et al. (2010) developed a competitive repopulation assay for assessing the function of WAT hematopoietic stem/progenitor cells (HSPCs) in comparison with BM competitor cells. WAT HSPCs (previously described as c-Kit+Lin−Sca+; Poglio et al., 2012) were FACS sorted from CD45.2 Cpa3cre/+ mice or CD45.2 red MC and basophil mice (RMB), where the 3′-UTR of the Ms4a2 gene encoding the FcεRIβ chain includes a cassette composed of a sequence coding for the bright td-Tomato (tdT) fluorescence protein (Dahdah et al., 2014). Lethally irradiated CD45.1 mice were cotransplanted with 2 × 103 WAT HSPC CD45.2+ and 2 × 105 BM CD45.1+ cells. WAT-HSPCs do not contain mature MCs (Fig. 6 C). Chimerism in the WAT was confirmed at 8 wk after transplantation by flow cytometry with a percentage of 20–30% of CD45.2+ cells present (Fig. 6 D). Because WAT-HSPCs do not reconstitute hematopoietic organs (Poglio et al., 2012), chimerism in the BM was very low (Fig. 6 D). Transplantation of WT mice with WAT HSPCs from Cpa3cre/+ mice significantly reduced SF% after MI (WT WAT donor: 33.3 ± 3.2% SF vs. Cpa3cre/+ WAT donor: 21.7 ± 1.8% SF, P = 0.026; Fig. 6 E). In addition, no significant differences were found in BM and WAT transplanted mice on infarct size, capillary density, or fibrosis at day 14 after infarction (Fig. 6, A and E). To further investigate the origin of MCs, we compared the levels of tdT+c-Kit+ cells in WT mice transplanted either with both WAT-HSPCs from WT mice and BM cells from RMB mice or with both WAT-derived HSPCs from RMB mice and WT mice–derived BM cells. Similar numbers of c-Kit+FcεRI+ cells were observed in the heart at day 7 after infarction, regardless of the origin of transplanted cells. However, when tdT staining was used to monitor their origin, the mice transplanted with RMB-BM cells had significantly less tdT+c-Kit+ cells in the infarcted myocardium at day 7 after MI compared with WT mice transplanted with RMB-WAT cells (P = 0.0317; Fig. 6, F–H). Collectively, although MCPs were detectable both in the BM and the WAT (Fig. 2), WAT-derived MCPs appeared to be more efficient in homing toward the cardiac tissue after MI. Figure 6. Cardiac MCs derive primarily from WAT HSPCs. (A) Transplantation of WT mice with Cpa3cre/+-derived BM cells had no effect on left ventricular SF%, infarct size, fibrosis, and capillary density in reference to WT animals receiving WT-derived BM cells (n = 16, from three independent experiments). (B) Transplantation of lethally irradiated WT or Cpa3cre/+ mice with WT or Cpa3cre/+-derived BM cells showed no reconstitution or inhibition of cardiac MCs, respectively, as counted on TB-stained heart sections (n = 6–12, two independent experiments). Kruskal–Wallis and Dunn’s post hoc test for comparisons between groups. (C) Representative flow cytometry gating strategy for isolation of HSPCs (CD45.2+c-kit+Sca-1+) in WAT showing the lack of mature FcεRI+ MCs. (D, left) Schematic overview of co-transplantation of lethally irradiated CD45.1+ WT mice with both FACS-sorted WAT-HSPCs (c-kit+Lin−Sca+) from CD45.2+ mice and BM cells from CD45.1+ cells. WAT-HSPCs were isolated from WT, Cpa3cre/+, or RMB mice. (right) Representative example of chimerism evaluation in the WAT and the BM by flow cytometry–assisted counting of WAT-derived CD45.2+ cells in CD45.1+ recipient mice transplanted with both FACS-sorted WAT-HSPCs from WT CD45.2+ mice and BM cells from WT CD45.1+ cells. (E) Transplantation of WT mice with WAT-HSPCs from Cpa3cre/+ led to depressed left ventricular SF% at day 14 after infarction (n = 7, two independent experiments) without any changes in infarct size, fibrosis, and capillary density. *, P < 0.05, Mann–Whitney nonparametric test. (F) WT mice transplanted with either WAT-HSPCs and RMB-derived BM cells or RMB-derived WAT HSPCs and BM cells have equal amounts of mature cardiac MCs at day 7 after infarction (n = 4–5, two independent experiments). (G) Cardiac tdT+c-kit+ cell numbers were higher in the WT mice transplanted with RMB-derived WAT HSPCs and BM cells compared with those transplanted with WAT-HSPC and RMB-derived BM cells (day 7 after MI; n = 8–10, two independent experiments). *, P < 0.05, Mann–Whitney nonparametric test. (H) Representative image of tdT+ cells in the heart 7 d after infarction. All values are presented as mean ± SEM. ns, not significant; WT, Cpa+/+ littermates. Cardiac MCs regulate myofilament sensitization to Ca2+ MC eradication had negative effects on the cardiac function after infarction with decreased systolic left ventricular diameter. There was no difference in infarct size or cardiac remodeling in the hearts of Cpa3cre/+ mice, suggesting a more direct impact on the myocardial contractile function. To further elucidate this MC-specific cardiac response, we examined the function and contractility properties of cardiomyocytes in the MC-depleted myocardial environment 14 d after the infarct. First, we assessed cardiomyocyte shortening (%) in response to field stimulation (1, 2, and 4 Hz). As shown in Fig. 7 A, MC deficiency significantly depressed the frequency response of left ventricular cardiomyocytes (Cpa3cre/+: 6.08 ± 0.5% vs. WT: 8.8 ± 0.8% 2Hz; P = 0.009). Absence of MCs in the infarcted myocardium led to abnormal contraction and relaxation kinetics of the left ventricular cardiomyocytes (Fig. 7 B), but there was no effect on the amplitude of Ca2+ transient peak in electrically stimulated cardiomyocytes (Fig. 7 C). In addition, sarcoplasmic reticulum load was maintained, indicating that Ca2+ influx and efflux were unaffected by MC eradication (Fig. 7 D). To further investigate contractile function, we determined myofilament force–Ca2+ dependence of left ventricular endocardial skinned myocytes at a steady-state (at day 14 after infarct). This dependence was markedly reduced in WT infarcted hearts compared with sham-operated hearts with maximum Ca2+-activated force (Fmax) declining nearly 50% (sham: 42.3 ± 0.5 mN/mm2 vs. infarct: 19.8 ± 3.5 mN/mm2, P = 0.0062; Fig. 7, E and F). Similar reduction in Fmax was exhibited by the myocytes from Cpa3cre/+ mice (sham: 35.9 ± 2.4 mN/mm2 vs. infarct: 19.2 ± 4.4 mN/mm2, P = 0.0186; Fig. 7, E and F). Although Fmax and Hill coefficient remained unchanged between WT and Cpa3cre/+ infarcted hearts, skinned myocytes from the MC-deficient ventricular environment displayed Ca2+ desensitization, as quantified by an increase in EC50 (Ca2+ required to achieve 50% maximal force; WT: 2.8 ± 0.09 µM vs. Cpa3cre+/−: 3.5 ± 0.11 µM, P = 0.007; Fig. 7, E–H). Figure 7. Reduced contractility and myofilament Ca2+ sensitization in Cpa3cre/+ mice after infarction. (A) Depressed left ventricular cardiomyocyte cell shortening (%) of Cpa3cre/+-derived intact cardiomyocytes versus WT cardiomyocytes in response to field stimulation (1, 2, and 4 Hz; n = 30–50, two independent experiments). *, P < 0.05; **, P < 0.01, Kruskal–Wallis and Dunn’s post hoc test for comparisons between groups. (B) Both contraction and relaxation kinetics were significantly reduced in Cpa3cre/+-derived cardiomyocytes versus WT cardiomyocytes (n = 30–50, two independent experiments). *, P < 0.05, Kruskal–Wallis and Dunn’s post hoc test for comparisons between groups. (C) MC deficiency had no effect on Ca2+ transient peak (n = 30–50, two independent experiments). (D) SR Ca2+ content in response to 1-Hz electric stimulation and after caffeine (10 mmol/l) was similar in both WT and Cpa3cre/+-derived cardiomyocytes (n = 30–50, two independent experiments). (E) Force–calcium fitted curves reflecting Ca2+ responsiveness of left ventricle peri-infarcted skinned myocytes from WT and Cpa3cre/+ mice after sham operation or MI. Reduced Ca2+ sensitivity in both WT and Cpa3cre/+-derived skinned myocytes in response to MI with a shift to the right in Cpa3cre/+-derived myocytes versus WT (n = 5–8, two independent experiments). (F and G) Fmax was not altered between WT and Cpa3cre/+-derived myocytes (F), but MC deficiency caused Ca2+ desensitization with a significant increase in EC50 (G; n = 5–8, two independent experiments). **, P < 0.01, Kruskal–Wallis and Dunn’s post hoc test for comparisons between groups. (H) No difference in Hill coefficient as a measure of Ca2+ cooperativity (n = 5–8, two independent experiments). All values are presented as mean ± SEM. ns, not significant; Sham, sham-operated animals; SR, sarcoplasmic reticulum. MC-dependent Ca2+ sensitization is mediated by myofilament phosphorylation Ca2+ sensitivity is regulated by phosphorylation of several sarcomeric proteins involved in the regulation of actomyosin interactions, including cardiac troponin I (cTnI) and myosin-binding protein C (MyBPC), both catalyzed by PKA (Layland et al., 2005; Stelzer et al., 2007). We evaluated the phosphorylation levels of cTnI and MyBPC at residues known to confer a negative effect on Ca2+ sensitization in the mouse myocardium (Chen et al., 2010). Myofilament phosphorylation in response to infarction was significantly higher in the peri-infarcted cardiac tissue of Cpa3cre/+ mice compared with WT mice both on cTnI (Ser22/23) and on MyBPC (Ser282 and Ser273) in a time-dependent manner, being observed 14 d after MI (but not at day 7; Fig. 8, A–C). This hyperphosphorylation event correlated with an increased activity of PKA in the Cpa3cre/+ peri-infarcted area compared with WT mice at day 14 after infarction (WT infarct: 1.2 ± 0.07% vs. Cpa3cre/+ infarct: 2.3 ± 0.2%, P = 0.0005; Fig. 8 D) that was not exhibited by the sham-operated cardiac tissue. Figure 8. MC-dependent myofilament phosphorylation via tryptase-induced PAR2 activation. (A and B) Representative Western blots for phosphorylated cTnI (Ser22/23) and MyBPC (Ser273, Ser302, and Ser282) and total proteins levels from peri-infarcted cardiac tissue of WT and Cpa3cre/+ at day 7 after MI (A) and at day 14 after infarction (B). (C) Quantified levels of p273-, p282-, and p302-MyBPC and p22/23-cTnI normalized to total protein levels showing increased myofilament phosphorylation in Cpa3cre/+ mice at day 14 after infarction (n = 7–13, three independent experiments). *, P < 0.05; **, P < 0.01, Kruskal–Wallis and Dunn’s post hoc test for comparisons between groups. (D) Increased PKA activity (A.U.) in peri-infarcted area of Cpa3cre/+ cardiac tissue at day 14 after infarction (n = 4 for sham, n = 10 for MI from four independent experiments). *, P < 0.05; **, P < 0.01, Kruskal–Wallis and Dunn’s post hoc test for comparisons between groups. (E) Treatment with recombinant MC chymase cleaved PAR2 in a non–R36-specific/canonical manner (n = 3, two independent experiments). (F) Recombinant mouse tryptase cleaved PAR2 only at canonical R36 site (n = 4, two independent experiments). **, P < 0.01, Kruskal–Wallis and Dunn’s post hoc test for comparisons between groups. (G) IBMX-induced activation of PKA in H9C2 cardiomyocytes is inhibited in the presence of tryptase or PAR2-AP; 6-Bz-cAMP was used as a positive control (n = 4, two independent experiments). *, P < 0.05, Kruskal–Wallis and Dunn’s post hoc test for comparisons between groups. (H) siRNA-induced mRNA knockdown of tryptase in cardiac MCs at 16 h after transfection (n = 4, two independent experiments). **, P < 0.01, Mann–Whitney nonparametric test. (I) Depressed SF% (day 14) is restored in Cpa3cre/+ mice by trans-cutaneous (echo-guided) injection of cardiac FACS-sorted Vybrant+ MCs (sorted at day 6 after infarction; n = 7) but not restored by MCs treated with tryptase siRNA (n = 3). All data are from two independent experiments. *, P < 0.05; **, P < 0.01, Kruskal–Wallis and Dunn’s post hoc test for comparisons between groups. All values are presented as mean ± SEM. ns, not significant. Identification of tryptase-induced protease-activated receptor 2 (PAR2) activation as a mechanism of MC-regulated PKA activity Because MC absence leads to increased PKA activity, we hypothesized that activation of a Gi-coupled protein receptor by an MC-released mediator could be involved. PAR2 can be cleaved by trypsin-like proteases such as MC tryptase (McLarty et al., 2011; Weithauser and Rauch, 2014), leading to the activation of Gi and consequential inhibition of the adenylyl cyclase–c-AMP–PKA axis (Sriwai et al., 2013). We analyzed the PAR2 cleaving capacity of both MC-specific tryptase and chymase that are all absent in the hearts of the Cpa3cre/+ mice after infarct (not depicted). Although MC chymase (mMCP4) cleaved PAR2, cleavage was not sensitive to replacement of arginine 36 of the canonical cleavage site with glycine (Fig. 8 E) and therefore unlikely to be signaling productive. Moreover, cardiac function at day 14 after infarction in mMCP4−/− mice was comparable with WT mice as analyzed by echocardiography (not depicted). In contrast, tryptase showed canonical PAR2 cleavage consistent with receptor activation (Fig. 8 F). We next investigated whether tryptase-induced PAR2 cleavage could regulate intracellular PKA activity in cardiomyocytes. As shown in Fig. 8 G, IBMX-induced PKA activation in H9C2 cells was attenuated by a PAR2-activating peptide (PAR2-AP; P = 0.04), which is known to activate Gi (Sriwai et al., 2013). Consistent with PAR2 cleavage, treatment with MC tryptase also significantly inhibited IBMX-induced PKA activity (P = 0.03). Because Cpa3cre/+ mice express very low levels of tryptase mRNA (not depicted), we assessed whether the tryptase signaling is responsible for reduced cardiac function in the MC-eradicated mice. We sorted live (Vybrant+) mature cardiac MCs (c-Kit+FcεRI+) at day 6 after MI, which expressed both connective tissue MC markers chymase (mMCP4) and tryptase (mMCP6; compared with nondetectable expression in cells sorted from the intestine; not depicted). Cardiac sorted MCs were transfected with tryptase siRNA or scramble siRNA (Fig. 8 H) for 16 h; viable cells were counted and were injected transcutaneously under echo-guidance into the myocardium of Cpa3cre/+ mice at day 6 after infarction, 1 d before their physiological peak in the heart. Reconstitution of cardiac MCs transfected with scramble siRNA restored SF% of Cpa3cre/+ mice as assessed at day 14 after infarction (WT: 28.15 ± 1.7% SF, Cpa3cre/+: 18.5 ± 2.08% SF, Cpa3cre/+ receiving MCs transfected with scramble SiRNA: 26.82 ± 2.9% SF; Fig. 8 I). However, injections with MCs transfected with MC tryptase siRNA failed to restored SF% of Cpa3cre/+ (Fig. 8 I). DISCUSSION Here we report a novel role for cardiac MCs in the regulation of the myofilament force–Ca2+ relationship. Mature cardiac MCs respond functionally to MI and regulate myofilament cTnI and MyBPC phosphorylation. This integral MC-dependent effect preserves the contractile reserve in concert with Ca2+ flux to the sarcomeres (Arteaga et al., 2005; Solaro and Arteaga, 2007). Mutations in c-Kit, as well as MC stabilization, have served as standard models to decipher MC functions. However, studies using c-kit mutations do not provide an exclusive functional role of MCs. As a prototypic example, kit mutant animals are protective against antibody-induced arthritis in contrast to Cpa3cre/+ mice that are susceptible, proving that c-kit effects are not MC exclusive. Similarly, MC deficiency, in the absence of c-kit mutations, plays neither a role in the regulation of weight gain or insulin resistance (Gutierrez et al., 2015) nor in wound healing and skin carcinogenesis (Antsiferova et al., 2013). Similarly, the selectivity and the efficacy of using an MC stabilizer treatment, such as DSCG, have been questioned. DSCG failed to inhibit IgE-dependent MC degranulation in mice, and it had nonspecific inhibitory effects on LPS-induced TNF release (Oka et al., 2012). In addition, several studies have identified MC-independent effects of DSCG treatment, including regulation of heat shock protein 90 (Okada et al., 2003), S100 proteins (Okada et al., 2002; Arumugam et al., 2006), and G protein–coupled receptor activation (Yang et al., 2010). Hence, in an attempt to reevaluate the role of cardiac MCs, we assessed the functional effect of MC deficiency (Cpa3cre/+ mice), c-kit deficiency (c-kitW/Wv mice), and inhibition of degranulation (DSCG administration) on cardiac function after infarction. In agreement with previous studies, we found increased MC density in the cardiac tissue after MI (Frangogiannis et al., 1998; Reid et al., 2011). Nevertheless, we show that, independently of c-Kit signaling, MCs preserve heart function via a mechanism that does not involve changes in cardiac remodeling after MI. Furthermore, we analyzed the recruitment of inflammatory cells and release of their mediators in the Cpa3cre/+ mice. Our data showed no effects of MCs deficiency in the regulation of local inflammation at the times studied. However, besides the absence of MCs, a reduced basophil compartment has previously been observed in Cpa3cre/+ animals, and this decrease should be considered when immunological functions are assessed in this model (Feyerabend et al., 2011). Depressed cardiac function in response to MC deficiency is mediated by a reduction in cardiomyocyte cell shortening, the kinetics of contraction/relaxation, and Ca2+ sensitization, as reflected by a right shift on the Ca2+–force response curve and an increase in EC50. These effects correlated both with an increase in p-Ser22/23 cTnI, p-Ser273, and p-Ser282 MyBPC and increased PKA activity. Phosphorylation of cTnI at its N-terminal Ser22/23 residues leads to its dissociation (Finley et al., 1999) or low-affinity binding (Ward et al., 2003) to the N-terminal domain of cTnC (NTnC), blocking the transfer of the Ca2+ signal through NTnC to cTnI, then to actin and further along the thin filament (Sykes, 2003; Li et al., 2004). This PKA-dependent mechanism leads to decreased Ca2+ sensitivity (Ramirez-Correa et al., 2010), increased cross-bridge cycling, and accelerated relaxation of the muscle (Herron et al., 2001). In agreement with our findings, such posttranslational modifications on cTnI have been previously described in postinfarcted stunned myocardium and heart failure (Dong et al., 2012; Nixon et al., 2014; Thoemmes et al., 2014). PKA also phosphorylates Ser273, Ser282, and Ser302 of the N terminus of MyBPC, the region that binds to the S2 segment of myosin, close to its arm domain (Gruen et al., 1999). PKA-induced phosphorylation of MyBPC loosens the thick filament structure, preventing its binding to myosin, thereby changing force–Ca2+ response (Levine et al., 2001; Kulikovskaya et al., 2003) and altering contractile function (Stelzer et al., 2007). MyBPC phosphorylation is necessary for basal myocardial function, and both cardiac MyBPC-null mouse hearts and fully phosphorylated MyBPC show accelerated myofilament kinetics (Barefield and Sadayappan, 2010). Although increased PKA activity generally correlates with positive inotropic effects via L-type Ca+2 channel and phospholamban (PLB) phosphorylation, we saw no difference in Ca+2 transients, but rather a reduction in contractility via desensitization. The mechanisms by which such modulation of PKA is strictly affecting myofilament proteins and not calcium handling proteins are not clear. Specificity and efficiency of PKA substrate phosphorylation require spatial-temporal regulation of cAMP/PKA by A kinase–anchoring proteins (AKAPs). The latter possess the ability to coordinate signaling pathways by scaffolding different proteins and facilitating a unique PKA-regulated signal transduction (Manni et al., 2008). Possible changes that disrupt AKAP/PKA complexes with PLB or L-type Ca+2 channel or that regulate AKAPs docking PKA in proximity of its sarcomeric substrates (Rababa’h et al., 2015), such as the recently identified cardiac troponin C (Sumandea et al., 2011), could be possible mechanisms involved in the MC-dependent Ca+2 desensitization. Extensive further investigation will be needed to fully uncover how MCs regulate the contractile machinery in a PKA-dependent mechanism. Nevertheless, this is the first time that an MC to myofilament signal exchange is documented. To further elucidate the mechanism of MC-dependent PKA inactivation, we sought to identify how MC signaling interferes with intracellular cardiomyocyte PKA. The PAR2 receptor is known to be cleaved/activated by MC-derived tryptases, and such cross-talk has already been identified between MCs and fibroblasts (McLarty et al., 2011; Murray et al., 2012). Indeed, MC tryptase was able to cleave PAR2 in vitro at the canonical site. Functionally, treatment of H9C2 cells with tryptase inhibited IBMX-induced PKA activation, and this effect was reproduced by treatment with the PAR2-AP, which is known to lead to Gi activation. Tryptase-induced PAR2 activation has been previously suggested to be pertussis toxin sensitive in airway smooth muscle cells (Berger et al., 2001). Although PAR2 deficiency has been shown to be protective against infarct size and cardiac remodeling (Antoniak et al., 2010), studies with PAR2-AP in WT mice showed that PAR-2 mediates protective effects in cardiac ischemia/reperfusion injury (Napoli et al., 2000; McLean et al., 2002; Jiang et al., 2007; Zhong and Wang, 2009). Such difference in the functional outcome between PAR2-deficient mice and PAR2-AP rely on the fact that the activating peptide changes the receptor’s three-dimensional structure and subsequently its “biased agonism” in ways that cannot be compared with its activation by proteases such as tryptase (Rajagopal et al., 2010). In agreement with our findings, however, Somasundaram et al. (2005) identified PAR2-AP mimicking the signaling transduction of tryptase in canine venous endothelial cells after ischemia/reperfusion, and others have shown the same effects in a variety of pathological models (Corvera et al., 1997; Akers et al., 2000; Berger et al., 2001; Shpacovitch et al., 2002). In conclusion, our data propose a novel function of tryptase-induced PAR2 activation, similar to that of PAR2-AP, which regulates the intracellular PKA activity of cardiomyocytes. MC trafficking in response to inflammatory or allergic triggers has been mainly attributed to BM-derived MCPs that circulate in the blood and home peripheral tissues before reaching terminally differentiated mature MCs. WAT has been described to contain stem cells with multilineage properties (Cousin et al., 2003; Han et al., 2010; Poglio et al., 2010, 2012) that may be of clinical value in repair or replacement of various cell lineages (Tran and Kahn, 2010). Recent studies have described both MCs (Liu et al., 2009) and MCPs (Poglio et al., 2010) in WAT. Poglio et al. (2010) documented that the WAT stromal vascular fraction hosts an MC lineage that does not originate from the BM and that can home to peripheral tissues including the skin and the intestine. MCs of WAT origin are infiltrating the cardiac tissue more efficiently than the BM competitor cells. Similarly, in chimeric mice whose BM cells were donated from GFP transgenic animals, cardiac MCs populating the ischemic milieu did not carry the GFP transgene, suggesting that MCs that home to the infarcted heart do not arise from BM progenitors (Fazel et al., 2006). Nevertheless, the density of MCPs was increased in the BM, and a small proportion of the mature cardiac MCs after MI did originate from the BM progenitors. Therefore, although we cannot exclude the BM as a source of cardiac MCs, MCPs originating primarily from the WAT dominate in the infarcted heart in our experimental conditions. Interestingly, in metabolic disorders and obesity, MC density in WAT is increased, suggesting a defect in their homing and/or production. Further studies are required to understand how homing of WAT MCs is regulated in response to acute MI or other related morbidities such as diabetes and obesity (Liu et al., 2009; Divoux et al., 2012) and would be of great therapeutic interest. We identified a novel role for cardiac MCs in preserving postischemic cardiac function. In response to MI, WAT-derived MCPs are mainly recruited to the cardiac tissue, proliferate, and differentiate into mature cells in an SCF-dependent manner. MC deficiency reduces PKA-mediated myofilament phosphorylation and Ca2+ sensitization (Fig. 9). In conclusion, MCs can directly regulate the contractile machinery of cardiomyocytes, and their presence in the infarcted myocardium is indispensable. Identification of the mechanisms that regulate MC activation and its signaling on PKA-dependent sarcomere function will provide novel insights in the regulation of contractile function after acute MI. Figure 9. Schematic diagram showing the proposed mechanism of MC-dependent myofilament Ca2+ sensitization after MI. MCs, originating primarily from WAT, infiltrate the heart after MI and regulate cardiac function via regulation of myofilament protein phosphorylation, cTnI, and MyBPC. The mechanism proposed involves tryptase-regulated PAR2 activation with subsequent Gi activation inhibiting cAMP/PKA activity. AC, adenylyl cyclase; p, phosphoryl group. MATERIALS AND METHODS MI All experiments were conducted according to the ethical committee for animal experimentation (Paris Descartes University, CEEA 34) and the National Charter on the ethics on animal experimentation from the French Minister of High Education and Research under the following reference: Molecular and cellular mechanisms involved in post-ischemic tissue remodelling (project n° 13-06/reference MESR: n° 01373.01). C57BL/6J mice were obtained from Janvier (backcrossed at least 12 times). MC-sufficient WBB6F1/J-Kit+/+ and MC-deficient WBB6F1/J-KitW/KitWv/J mice were obtained from The Jackson Laboratory (backcrossed at least 10 times). C57BL/6J mMCP4−/− and their WT littermates (mMCP4+/+, WT) mice were provided by U. Blank (backcrossed at least seven times). C57BL/6J Cpa3cre/+ mice and their WT littermates (Cpa3+/+, WT) were a gift from H.-R. Rodewald (backcrossed at least eight times). All mice were studied at the age of 8 wk. MI was induced by left coronary ligation as previously described (Kumar et al., 2005). Mice were anesthetized using ketamine (100 mg/kg body weight) and xylazine (10 mg/kg body weight) via intraperitoneal injection, intubated, and ventilated using a small-animal respirator. The chest wall was shaved, and a thoracotomy was performed in the fourth left intercostal space. The pericardial sac was removed, and the left anterior descending artery was permanently ligated using a 7/0 monofilament suture (Peters Surgical) at the site of its emergence close to the left atrium. The thoracotomy was closed with 6/0 monofilament sutures. The exact same procedure was performed for the sham-operated mice except that the ligation was not tied. The endotracheal tube was removed once spontaneous respiration resumed, and mice were placed on a warm pad at 37°C until awakened. For the DSCG treatment experiments, mice were treated with DSCG (50 µg/g/mouse) or PBS (150 µl) 1 h after the operation and every day for 7 d. For evaluation of SCF effect on MC proliferation and density, mice were treated with the anti-SCF antibody provided by N.W. Lukacs and S. Morris (Medical School, University of Michigan, Ann Arbor, MI; Dolgachev et al., 2009) or rabbit serum (400 µl/mouse; Sigma-Aldrich) at days 1, 3, and 5 after infarction. For assessment of cell proliferation with the APC BrdU Flow kit (BD), 100 µl BrdU was intraperitoneally injected into mice 24 h before tissue isolation and digestion. BM and WAT transplantation BM cells were flushed from the femurs of C57BL/6J under sterile conditions, and 8 × 106 cells were injected into the retroorbital sinuses of C57BL/6J mice irradiated with 10 Gy (one dose). Competitive repopulation assays were conducted as described previously (Poglio et al., 2010, 2012). In brief, 2 × 103 c-Kit+/Lin−/Sca-1+ cells sorted from the WAT of donor mice were mixed with 2 × 105 competitor BM total cells. The mixed population was intravenously injected into lethally irradiated (10 Gy, 137Cs source) recipient mice. Reconstituted mice were then allowed to recover for 2 mo before MI. Echocardiographic measurements Transthoracic echocardiography was performed 14 d after surgery using an echocardiograph (ACUSON S3000 ultrasound; Siemens AG) equipped with a 14-MHz linear transducer (1415SP). The investigator was blinded to group assignment. Mice were anesthetized by isoflurane inhalation. Percentages of SF (%SF; Zouggari et al., 2013) or EF (%EF) were calculated as previously described (Dormishian et al., 2013). Immunohistochemistry Cardiac remodeling after MI was assessed at day 14. Hearts were excised, rinsed in PBS, and frozen in liquid nitrogen. Hearts were cut by a cryostat (CM 3050S; Leica) into 5–7-µm-thick sections. TB was used to assess MC degranulation as previously described (Xaubet et al., 1991). Masson’s trichrome and Sirius red stains were performed for infarct size and myocardial fibrosis evaluation. Infarct size (in %) was calculated as total infarct circumference divided by total left ventricle circumference. The collagen volume fraction was calculated as the ratio of the total area of interstitial fibrosis to the myocytes area in the entire visual field of the section. Endothelial cells within capillaries were visualized after BS-1 lectin staining (1:100, FITC-conjugated Griffonia simplicifolia; Sigma-Aldrich) and cardiomyocytes with wheat germ agglutinin (1:200, Texas red conjugated; Thermo Fisher Scientific). Image analysis and measurements were conducted with ImageJ processing program (National Institutes of Health). Cleavage assay for PAR2 receptor Secreted epithelial alkaline phosphatase (SEAP)–tagged versions of human PAR2 (UniProtKB accession P55085; fused at Q27) protein as well as the canonical cleavage site mutant R36G were generated to allow quantification of receptor cleavage (Ludeman et al., 2004). PAR1 (F2r) knockout mouse lung fibroblasts (KOLFs; Trejo et al., 1996) were grown in DMEM supplemented with antibiotics and 5% serum. Transfection was performed with TransitLT1 transfection (Euromedex) in complete medium as recommended. 48 h after transfection, cells were washed with Opti-MEM for 1 h and incubated for one additional hour in Opti-MEM with or without the respective MC-specific proteases (500 ng/ml), and supernatants were collected. A second 20-min incubation with 10 nM trypsin stripped all remaining SEAP moiety (as verified in separate experiments) from the cell surface. SEAP activity in the conditioned media was determined at OD405 after hydrolysis of para-nitrophenyl phosphate (pNPP; Sigma-Aldrich). This permitted calculation of percentage of surface receptors cleaved during the initial 60-min incubation. Membrane-permeabilized myocytes Membrane-permeabilized or “skinned” myocyte experiments were performed as previously described (Kirk et al., 2014). Tissue from the peri-infarct zone was flash frozen in liquid nitrogen and stored at −80°C. At the time of the experiment, the tissue was homogenized in isolation solution (5.5 mM Na2ATP, 7.11 mM MgCl2, 2.0 mM EGTA, 108.01 mM KCl, 8.91 mM KOH, 10 mM imidazole, 10 mM DTT, protease inhibitors [Sigma-Aldrich], and phosphatase inhibitors [Roche] with 0.3% Triton X-100). Triton X-100 is a detergent that permeabilizes the myocyte membrane, allowing the free movement of calcium into the cell. Myocytes were then washed in isolation solution without Triton X-100 to remove the detergent. Using silicone, a single myocyte was glued to the tips of 150-µm diameter minutia pins. One pin was attached to a force transducer (Aurora Scientific Inc.). A video camera (Imperx) and the high-speed video sarcomere length program (Aurora Scientific Inc.) were used to continuously monitor sarcomere length. Studies were conducted at a sarcomere length of 2.0 µm. The mounted myocytes were kept in a relax buffer (5.95 mM Na2ATP, 6.41 mM MgCl2, 10 mM EGTA, 100 mM BES, 10 mM phosphocreatine, 50.25 mM potassium propionate, 10 mM DTT, and protease and phosphatase inhibitors) with no calcium. Force was measured as the myocyte was exposed to increasing calcium concentrations, by moving the myocyte from baths containing different ratios of relax and activating solutions. The activating solution contained 5.95 mM Na2ATP, 6.2 mM MgCl2, 10 mM Ca2+EGTA, 100 mM BES, 10 mM phosphocreatine, 29.98 mM potassium propionate, 10 mM DTT, and protease and phosphatase inhibitors. All buffers were adjusted to a pH of 7.0. A complete activation of the myocyte occurred at the beginning and end of the experiment, and the myocyte was discarded if there was <10% rundown. Force–calcium data were fit to the Hill equation: F = Fmax × Cah/(ECh50 + Cah), yielding Fmax, calcium sensitivity (Ca2+ required to achieve 50% maximal force, EC50), and cooperativity (Hill coefficient, h). Organ digestion/cell isolation Peripheral blood was isolated from the inferior vena cava puncture with heparin solution. Whole blood was lysed after immunofluorescence staining using the FACS lysing solution (BD), and total blood leukocyte numbers were counted using trypan blue (Sigma-Aldrich). BM cells were washed through the femur and the tibia and filtered through a 40-µm nylon mesh (BD). Spleens were collected, minced with fine scissors, and filtered through a 40-µm nylon mesh (BD). For both splenocytes and BM-derived cells, the cell suspension was centrifuged at 400 g for 10 min at 4°C. Red blood cells were lysed using red blood cell lysing buffer (Sigma-Aldrich), and splenocytes and BM cells were washed with PBS supplemented with 1% vol/vol fetal bovine serum. Hearts were collected, and the left ventricle was isolated, minced with fine scissors, and gently passed through the Bel-Art Scienceware 12-well tissue disaggregator (Thermo Fisher Scientific). Cells were collected, filtered through 40-µm nylon mesh, and washed with PBS with 1% vol/vol fetal bovine serum. Subcutaneous WAT was dissected and mechanically dissociated. WAT fragments were digested with collagenase, and stroma-vascular cells were collected by centrifugation after elimination of undigested fragments by filtration as described previously (Poglio et al., 2010). Red blood cells were removed by incubation in hemolysis buffer (STEMCELL Technologies). Cells were counted and used for HSPC sorting. Flow cytometry Cells isolated from the tissue of interest were incubated in the dark at 4°C for 30 min with the following antibody mix. For mature MCs: APC-conjugated ckit/CD117 (2B8; BD) and PE-Cy7–conjugated Ly-6A/E (Sca-1; D7; BD). For MCPs: DAPI-conjugated Lineage (eBioscience), FITC-conjugated CD45.2 (BD), CD45.1 (BD), APC-conjugated CD34 (HM34; BioLegend), PerCP/Cy5.5-conjugated CD16/32 (FcγII/III; Thermo Fisher Scientific), and PE-conjugated Integrin-β7 (FIB504; eBioscience). When cardiac-derived cells were analyzed for marker expression, Vybrant DyeCycle violet stain (Thermo Fisher Scientific) was used to stain live cells and eliminate both debris subsequently nonspecific autofluorescence. For detection of cardiac MCs, granulated cells were first gated on SSC/FSC. The total number of cells was normalized to tissue weight. Cells were analyzed using a flow cytometer (LSR II; BD) or were sorted with FACSAria II (BD). The APC BrdU Flow kit (BD) was used for analysis of proliferation according to the manufacturer’s instructions. HSPC isolation was performed as previously described (Poglio et al., 2010). In brief, freshly isolated WAT–stromal vascular fraction (SVF) cells were stained in PBS containing FcR Block reagent and CD117 (2B8; BD), Lineage (BD), Sca-1 (E13-161.7; BD), and CD45.2 (BD) antibodies. Cells were washed in PBS and sorted with FACSAria II. Data acquisition and analysis were performed using FACSDiva software (BD) or FlowJo 7.5. For detection of inflammatory cells, total cardiac cells were gated on PerCP-conjugated CD45 (BD), and the following antibodies were used: DAPI-conjugated anti-CD11b (BD), PE-conjugated anti-Ly6G (1A8; BD), APC-conjugated anti-F4/80 (BioLegend), FITC-conjugated anti-Ly6C (BioLegend), APC-conjugated anti-CD8a (BD), PE-conjugated anti-CD45R/B220 (RA3-6B2; eBioscience), PE–Alexa Fluor 700–conjugated anti-CD3 (eBioscience), and FITC-conjugated anti-CD4 (eBioscience). All antibodies were used at a dilution of 1:100. Monocytes were identified as CD11bhiLy6G−7/4hi/lo. Neutrophils were identified as CD11b+Ly6Ghi. Macrophages were identified as CD11b+Ly6G−F4/80+. Mature B lymphocytes were identified as B220+CD3− and T cells as CD4+CD3+ or CD4+CD8+. The total number of cells was then normalized to heart weight. Cells were analyzed using a flow cytometer (LSR II). Transcutaneous echo-guided intramyocardial injections of cardiac MCs For each experiment, cardiac MCs were FACS sorted as Vybrant+c-kit+FcεRI+ cells from a total of 30 C57BL/6J mice at day 6 of MI. Hearts were collected, and the left ventricle was isolated, minced with fine scissors, and gently passed through the Bel-Art Scienceware 12-well tissue disaggregator (Thermo Fisher Scientific). Cells were then collected and filtered through 40-µm nylon mesh. During sorting, cells were kept always in PBS complemented with 5% fetal calf serum at 4°C under sterile conditions. FACS-sorted MCs were counted, plated at a density of 2.5 × 105 cells/ml, and cultured in DMEM complemented with 20 mM l-glutamine, 10% fetal calf serum (inactivated at 56°C), 2 mM l-glutamine, 1 mM pyruvate, 100 U/ml penicillin/streptomycin, and 1% MEM nonessential amino acid solution (Sigma-Aldrich) in the presence of 0.1 mg/ml recombinant mouse SCF for MC survival overnight. Cells were then transfected with the Nucleofector 2B device (program C-005) in 100 µl Nucleofector solution T (Lonza) with ON-TARGETplus Mouse Tpsb2 (mMCP6) siRNA (L-064669-02-0005; GE Healthcare) or non-targeting (scramble) siRNA (GE Healthcare) for 16 h according to the manufacturer’s instructions. Cells were washed, their viability was assessed with Trypan blue, and they were counted and resuspended in PBS to give a total of 80,000 cells/60 µl of suspension for trans-cutaneous injections. Closed-chest, trans-cutaneous, echo-guided injections were then used to deliver cells or PBS directly to the peri-infarcted myocardium as previously described (Toeg et al., 2013). In brief, Cpa3cre/+ mice with 6-d-old infarcts were anesthetized by isoflurane inhalation and fixed in a supine position on a heating pad. The infarct region was located by echocardiography using a Vevo 2100 imaging system and a MS400 probe appropriate for mouse cardiovascular imaging (18–38 MHz; VisualSonics Inc.). A micromanipulator (VisualSonics) was used to guide a 1.5-in-long 27-G needle (Dominic Dutscher) into the hypokinetic anterior wall of the peri-infarcted myocardium under continual echo guidance. Four 15-µl injections were used to deliver a total of 80,000 cells resuspended in sterile PBS. FlowCytomix The bead-based multiplex immunoassay for the flow cytometer (eBioscience) was used to measure the levels of inflammatory mediators in the cardiac tissue. Peri-infarcted left ventricular tissue was homogenized using a glass Dounce homogenizer in ice-cold Tris-HCl (50 mmol/l, pH 7.5), NaCl (150 mmol/l), EDTA (1 mmol/l), NP-40 (1%), Na-orthovanadate (1 mM), β-glycerophosphate (40 mM; Sigma-Aldrich), and protease cocktail inhibitors (Roche). 500 µg protein homogenates was incubated with antibody-coated beads recognizing IL3, Mcp3, Mcp1, IL-6, IL-10, IL1β, IL12p70, and TNF. The target analytes were captured by the specific antibodies, and the samples were incubated with biotin-conjugated secondary antibodies. Streptavidin-PE was added for detection, and flow cytometry was used to differentiate bead populations according to size and fluorescent signature. Analyte concentration was measured using standard curves on the FlowCytomix Pro Software (eBioscience). Data were expressed as picogram of analyte per microgram of cardiac protein homogenate. Quantitative real-time PCR Quantitative real-time PCR was performed on a Step-One Plus (Applied Biosystems). GAPDH was used to normalize gene expression. The following primer sequences were used: SCF forward, 5′-GGAGATCTGGAATCCTGTGA-3′; and reverse, 5′-CCCGGCGACATAGTTGAGGGTTAT-3′; mMCP6 forward, 5′-CTGGGGCGACATTGATAATGACGAGCCTCT-3′; and reverse, 5′-CCCCCTGAATCGCCCTGGCAGGAGT-3′; and mMCP4 forward, 5′-CTGGGGCTGGAGCTGAGGAGATTA-3′; and reverse, 5′-CAACACAAATTGGCGGGTTATGAGAA-3′. Tissue homogenization and Western blot Cardiac tissue was homogenized using a glass Dounce homogenizer in ice-cold Tris-HCl (50 mmol/l, pH 7.5), NaCl (150 mmol/l), EDTA (1 mmol/l), NP-40 (1%), Na-orthovanadate (1 mM), β-glycerophosphate (40 mM; Sigma-Aldrich), and protease cocktail inhibitors (Roche). After homogenization, protein concentration was measured by the SMART Micro BCA Protein Assay kit (Intron). Total tissue homogenate was denatured in loading buffer and 0.1 M DTT at 95°C for 5 min, and 30 µg was loaded on freshly prepared 9% Bis-Acrylamide gels (ratio 29:1), transferred onto nitrocellulose membranes (Bio-Rad Laboratories), and blotted using custom MyBPC antibodies against phospho Ser-273 (1:2,500), Ser-282 (1:2,500), and Ser-302 (1:10,000) and total (1:5,000; Kirk et al., 2014); phospho TnI (1:1,000; Cell Signaling Technology); and total TnI (1:10,000; Spectral Diagnostics). In some cases, blots were stripped with freshly prepared stripping buffer (2 M Tris-HCL, pH 6.8, 10% SDS, and 100 mM β-mercaptoethanol) and reprobed. Blots were scanned before being reprobed to ensure efficient stripping. All phosphorylated protein levels were normalized to total protein levels. Cardiomyocyte contractility and Ca2+-transient measurements Left ventricle myocytes were freshly isolated from the noninfarcted free wall and recorded as previously described (Fauconnier et al., 2010). Real-time Ca2+ measurements and cell shortening were performed on freshly isolated myocytes incubated in a physiological Tyrode solution (140 mM NaCl, 5 mM KCl, 8 mM NaH2PO4, 1 mM MgSO4, 10 mM HEPES, 5 mM Taurine, and 10 mM glucose). Cardiomyocytes were loaded with the ratiometric Ca2+ dye Fura-2AM at room temperature for 20 min (2 µM, Invitrogen), and cell shortening/Ca2+ transients were recorded using electrical-field stimulation (1, 2, and 4 Hz). Fluorescence wavelengths emitted at 340 nm (F340) and 380 nm (F380) were simultaneously recorded using the IonOptix system coupled to a microscope (×40; ZEISS). SR Ca2+ content was evaluated by measuring the peak amplitude of the cytosolic Ca2+ transient induced by rapid perfusion of caffeine (10 mM) after a 20-s pacing period (1 Hz) followed by a 10-s rest period with a calcium-sodium–free solution containing 140 mM LiCl, 6 mM KCl, 1 mM MgCl, 1 mM EGTA, 10 mM glucose, and 10 HEPES, pH 7.4. Cell culture H9c2 cells, derived from the embryonic rat ventricle (ATCC CRL1446), were seeded at a density of 5 × 105 cells/100-mm plate and cultured at 37°C in a 5% CO2 humidified atmosphere in DMEM supplemented with 0.2 mM glutamine, 100 U/ml penicillin, 100 µg/ml streptomycin, and 10% FBS. When cells reached a subconfluence state (80%), they were treated with IBMX (3-isobutyl-1-methylxanthine; Sigma-Aldrich) at 100 µM for 5 and 10 min with or without co-treatment with 500 ng/ml mouse recombinant tryptase (R&D Systems) or the 100 µM PAR2-AP SLIGRL-NH2 (Abcam). Cells were homogenized and sonicated in cell lysis buffer (#9803; Cell Signaling Technology), and protein concentration was measured by the SMART Micro BCA Protein Assay kit (Intron). PKA activity assay PKA activity was measured by the commercial kit (Enzo Life Sciences) according to manufacturer’s instructions. In essence, active PKA was used to generate a standard curve, and 15 µg of cardiac peri-infarcted tissue homogenate or 10 µg of cell lysate was used for the assay. Statistical analysis Group comparisons were made using the Kruskal–Wallis test by ranks (nonparametric one-way analysis of variance), with Dunn’s post hoc comparisons. Normal distribution was evaluated with the Shapiro–Wilk test. For individual comparisons, Mann–Whitney’s U test was used as described in the figure legends. All values are presented as mean ± SEM, and a p-value <0.05 was considered significant. Acknowledgments We would like to thank Nicholas W. Lukacs and Susan Morris for kindly providing the SCF antibody. We are also grateful to the imaging platform for small animals of Paris Descartes University for echocardiographic analysis. The authors declare no competing financial interests. Author contributions: A. Ngkelo designed and performed the study, interpreted the data, and drafted the manuscript. A. Richart performed experiments and interpreted the data. J.A. Kirk and M.J. Ranek contributed to study design and performed the myofilament calcium sensitization and PKA activity experiments. P. Bonnin performed and interpreted the ultrasound studies. C. Guerin performed and interpreted the ImageStream analysis. J. Vilar contributed to study design and performed experimental optimization and data acquisition. M. Lemitre performed and analyzed cardiac remodeling parameters. S. Le Gall performed the PAR2 cleavage experiments and interpretation. N. Renault and A. Kervadec performed the trans-cutaneous echo-guided intramyocardial injections. P. Menasche, M. Branchereau, and C. Heymes performed the cardiomyocyte contractility measurements and interpretation. L. Danelli and U. Blank provided the mcp4−/− mice. G. Gautier and P. Launay provided the RBM mice. E. Camerer provided the KOLF cells, PAR2-AP, and the system of PAR2 cleavage measurement. P. Bruneval and P. Marck provided the Vevo 2100 imaging system for trans-cutaneous echo-guided intramyocardial injections and the human patient MC staining data. E. Luche, L. Casteilla, and B. Cousin performed the WAT transplantation experiments. H.-R. Rodewald provided the cpa3cre/+ mice and drafted the manuscript. D.A. Kass contributed to study design and data interpretation. J.-S. Silvestre designed the study, analyzed and interpreted the data, and drafted the manuscript. Abbreviations used: AKAP A kinase–anchoring protein cTnI cardiac troponin I DSCG disodium cromoglycate EF ejection fraction HSPC hematopoietic stem/progenitor cell MC mast cell MCP MC progenitor MI myocardial infarction MyBPC myosin-binding protein C PAR2-AP PAR2-activating peptide PKA protein kinase A RMB red MC and basophil mice SCF stem cell factor SEAP secreted epithelial alkaline phosphatase SF shortening fraction TB toluidine blue WAT white adipose tissue ==== Refs Akers, I.A., M. Parsons, M.R. Hill, M.D. Hollenberg, S. Sanjar, G.J. Laurent, and R.J. McAnulty. 2000. Mast cell tryptase stimulates human lung fibroblast proliferation via protease-activated receptor-2. Am. J. Physiol. Lung Cell. Mol. Physiol. 278 :L193–L201.10645907 Antoniak, S., M. Rojas, D. Spring, T.A. Bullard, E.D. Verrier, B.C. Blaxall, N. Mackman, and R. Pawlinski. 2010. Protease-activated receptor 2 deficiency reduces cardiac ischemia/reperfusion injury. Arterioscler. Thromb. Vasc. Biol. 30 :2136–2142. 10.1161/ATVBAHA.110.213280 20724699 Antsiferova, M., C. Martin, M. Huber, T.B. Feyerabend, A. Förster, K. Hartmann, H.R. Rodewald, D. Hohl, and S. Werner. 2013. Mast cells are dispensable for normal and activin-promoted wound healing and skin carcinogenesis. J. Immunol. 191 :6147–6155. 10.4049/jimmunol.1301350 24227781 Arteaga, G.M., C.M. Warren, S. Milutinovic, A.F. Martin, and R.J. Solaro. 2005. Specific enhancement of sarcomeric response to Ca2+ protects murine myocardium against ischemia-reperfusion dysfunction. Am. J. Physiol. Heart Circ. Physiol. 289 :H2183–H2192. 10.1152/ajpheart.00520.2005 16024565 Arumugam, T., V. Ramachandran, and C.D. Logsdon. 2006. Effect of cromolyn on S100P interactions with RAGE and pancreatic cancer growth and invasion in mouse models. J. Natl. Cancer Inst. 98 :1806–1818. 10.1093/jnci/djj498 17179482 Barefield, D., and S. Sadayappan. 2010. Phosphorylation and function of cardiac myosin binding protein-C in health and disease. J. Mol. Cell. Cardiol. 48 :866–875. 10.1016/j.yjmcc.2009.11.014 19962384 Berger, P., D.W. Perng, H. Thabrew, S.J. Compton, J.A. Cairns, A.R. McEuen, R. Marthan, J.M. Tunon De Lara, and A.F. Walls. 2001. Tryptase and agonists of PAR-2 induce the proliferation of human airway smooth muscle cells. J. Appl. Physiol. 91 :1372–1379.11509538 Boag, S.E., R. Das, E.V. Shmeleva, A. Bagnall, M. Egred, N. Howard, K. Bennaceur, A. Zaman, B. Keavney, and I. Spyridopoulos. 2015. T lymphocytes and fractalkine contribute to myocardial ischemia/reperfusion injury in patients. J. Clin. Invest. 125 :3063–3076. 10.1172/JCI80055 26168217 Chen, C.C., M.A. Grimbaldeston, M. Tsai, I.L. Weissman, and S.J. Galli. 2005. Identification of mast cell progenitors in adult mice. Proc. Natl. Acad. Sci. USA. 102 :11408–11413. 10.1073/pnas.0504197102 16006518 Chen, P.P., J.R. Patel, I.N. Rybakova, J.W. Walker, and R.L. Moss. 2010. Protein kinase A-induced myofilament desensitization to Ca2+ as a result of phosphorylation of cardiac myosin-binding protein C. J. Gen. Physiol. 136 :615–627. 10.1085/jgp.201010448 21115695 Corvera, C.U., O. Déry, K. McConalogue, S.K. Böhm, L.M. Khitin, G.H. Caughey, D.G. Payan, and N.W. Bunnett. 1997. Mast cell tryptase regulates rat colonic myocytes through proteinase-activated receptor 2. J. Clin. Invest. 100 :1383–1393. 10.1172/JCI119658 9294103 Cousin, B., M. André, E. Arnaud, L. Pénicaud, and L. Casteilla. 2003. Reconstitution of lethally irradiated mice by cells isolated from adipose tissue. Biochem. Biophys. Res. Commun. 301 :1016–1022. 10.1016/S0006-291X(03)00061-5 12589814 Dahdah, A., G. Gautier, T. Attout, F. Fiore, E. Lebourdais, R. Msallam, M. Daëron, R.C. Monteiro, M. Benhamou, N. Charles, 2014. Mast cells aggravate sepsis by inhibiting peritoneal macrophage phagocytosis. J. Clin. Invest. 124 :4577–4589. 10.1172/JCI75212 25180604 de Couto, G., W. Liu, E. Tseliou, B. Sun, N. Makkar, H. Kanazawa, M. Arditi, and E. Marbán. 2015. Macrophages mediate cardioprotective cellular postconditioning in acute myocardial infarction. J. Clin. Invest. 125 :3147–3162. 10.1172/JCI81321 26214527 Divoux, A., S. Moutel, C. Poitou, D. Lacasa, N. Veyrie, A. Aissat, M. Arock, M. Guerre-Millo, and K. Clément. 2012. Mast cells in human adipose tissue: link with morbid obesity, inflammatory status, and diabetes. J. Clin. Endocrinol. Metab. 97 :E1677–E1685. 10.1210/jc.2012-1532 22745246 Dolgachev, V.A., M.R. Ullenbruch, N.W. Lukacs, and S.H. Phan. 2009. Role of stem cell factor and bone marrow-derived fibroblasts in airway remodeling. Am. J. Pathol. 174 :390–400. 10.2353/ajpath.2009.080513 19147822 Dong, X., C.A. Sumandea, Y.C. Chen, M.L. Garcia-Cazarin, J. Zhang, C.W. Balke, M.P. Sumandea, and Y. Ge. 2012. Augmented phosphorylation of cardiac troponin I in hypertensive heart failure. J. Biol. Chem. 287 :848–857. 10.1074/jbc.M111.293258 22052912 Dormishian, M., G. Turkeri, K. Urayama, T.L. Nguyen, M. Boulberdaa, N. Messaddeq, G. Renault, D. Henrion, and C.G. Nebigil. 2013. Prokineticin receptor-1 is a new regulator of endothelial insulin uptake and capillary formation to control insulin sensitivity and cardiovascular and kidney functions. J. Am. Heart Assoc. 2 :e000411. 10.1161/JAHA.113.000411 24152983 Erdei, A., M. Andrásfalvy, H. Péterfy, G. Tóth, and I. Pecht. 2004. Regulation of mast cell activation by complement-derived peptides. Immunol. Lett. 92 :39–42. 10.1016/j.imlet.2003.11.019 15081525 Fauconnier, J., J. Thireau, S. Reiken, C. Cassan, S. Richard, S. Matecki, A.R. Marks, and A. Lacampagne. 2010. Leaky RyR2 trigger ventricular arrhythmias in Duchenne muscular dystrophy. Proc. Natl. Acad. Sci. USA. 107 :1559–1564. 10.1073/pnas.0908540107 20080623 Fazel, S., M. Cimini, L. Chen, S. Li, D. Angoulvant, P. Fedak, S. Verma, R.D. Weisel, A. Keating, and R.K. Li. 2006. Cardioprotective c-kit+ cells are from the bone marrow and regulate the myocardial balance of angiogenic cytokines. J. Clin. Invest. 116 :1865–1877. 10.1172/JCI27019 16823487 Feyerabend, T.B., A. Weiser, A. Tietz, M. Stassen, N. Harris, M. Kopf, P. Radermacher, P. Möller, C. Benoist, D. Mathis, 2011. Cre-mediated cell ablation contests mast cell contribution in models of antibody- and T cell-mediated autoimmunity. Immunity. 35 :832–844. 10.1016/j.immuni.2011.09.015 22101159 Finley, N., M.B. Abbott, E. Abusamhadneh, V. Gaponenko, W. Dong, G. Gasmi-Seabrook, J.W. Howarth, M. Rance, R.J. Solaro, H.C. Cheung, and P.R. Rosevear. 1999. NMR analysis of cardiac troponin C-troponin I complexes: effects of phosphorylation. FEBS Lett. 453 :107–112. 10.1016/S0014-5793(99)00693-6 10403385 Forman, M.F., G.L. Brower, and J.S. Janicki. 2006. Rat cardiac mast cell maturation and differentiation following acute ventricular volume overload. Inflamm. Res. 55 :408–415. 10.1007/s00011-006-6016-z 17109067 Franco, C.B., C.C. Chen, M. Drukker, I.L. Weissman, and S.J. Galli. 2010. Distinguishing mast cell and granulocyte differentiation at the single-cell level. Cell Stem Cell. 6 :361–368. 10.1016/j.stem.2010.02.013 20362540 Frangogiannis, N.G., J.L. Perrard, L.H. Mendoza, A.R. Burns, M.L. Lindsey, C.M. Ballantyne, L.H. Michael, C.W. Smith, and M.L. Entman. 1998. Stem cell factor induction is associated with mast cell accumulation after canine myocardial ischemia and reperfusion. Circulation. 98 :687–698. 10.1161/01.CIR.98.7.687 9715862 Gorski, P.A., D.K. Ceholski, and R.J. Hajjar. 2015. Altered myocardial calcium cycling and energetics in heart failure--a rational approach for disease treatment. Cell Metab. 21 :183–194. 10.1016/j.cmet.2015.01.005 25651173 Grimbaldeston, M.A., C.C. Chen, A.M. Piliponsky, M. Tsai, S.Y. Tam, and S.J. Galli. 2005. Mast cell-deficient W-sash c-kit mutant Kit W-sh/W-sh mice as a model for investigating mast cell biology in vivo. Am. J. Pathol. 167 :835–848. 10.1016/S0002-9440(10)62055-X 16127161 Gruen, M., H. Prinz, and M. Gautel. 1999. cAPK-phosphorylation controls the interaction of the regulatory domain of cardiac myosin binding protein C with myosin-S2 in an on-off fashion. FEBS Lett. 453 :254–259. 10.1016/S0014-5793(99)00727-9 10405155 Guo, J., W. Jie, D. Kuang, J. Ni, D. Chen, Q. Ao, and G. Wang. 2009. Ischaemia/reperfusion induced cardiac stem cell homing to the injured myocardium by stimulating stem cell factor expression via NF-κB pathway. Int. J. Exp. Pathol. 90 :355–364. 10.1111/j.1365-2613.2009.00659.x 19563618 Gutierrez, D.A., S. Muralidhar, T.B. Feyerabend, S. Herzig, and H.R. Rodewald. 2015. Hematopoietic Kit deficiency, rather than lack of mast cells, protects mice from obesity and insulin resistance. Cell Metab. 21 :678–691.25955205 Han, J., Y.J. Koh, H.R. Moon, H.G. Ryoo, C.H. Cho, I. Kim, and G.Y. Koh. 2010. Adipose tissue is an extramedullary reservoir for functional hematopoietic stem and progenitor cells. Blood. 115 :957–964. 10.1182/blood-2009-05-219923 19897586 He, A., and G.P. Shi. 2013. Mast cell chymase and tryptase as targets for cardiovascular and metabolic diseases. Curr. Pharm. Des. 19 :1114–1125. 10.2174/1381612811319060012 23016684 Herron, T.J., F.S. Korte, and K.S. McDonald. 2001. Power output is increased after phosphorylation of myofibrillar proteins in rat skinned cardiac myocytes. Circ. Res. 89 :1184–1190. 10.1161/hh2401.101908 11739284 Janicki, J.S., G.L. Brower, and S.P. Levick. 2015. The emerging prominence of the cardiac mast cell as a potent mediator of adverse myocardial remodeling. Methods Mol. Biol. 1220 :121–139. 10.1007/978-1-4939-1568-2_8 25388248 Jiang, R., A. Zatta, H. Kin, N. Wang, J.G. Reeves, J. Mykytenko, J. Deneve, Z.Q. Zhao, R.A. Guyton, and J. Vinten-Johansen. 2007. PAR-2 activation at the time of reperfusion salvages myocardium via an ERK1/2 pathway in in vivo rat hearts. Am. J. Physiol. Heart Circ. Physiol. 293 :H2845–H2852. 10.1152/ajpheart.00209.2007 17720772 Katz, H.R., and K.F. Austen. 2011. Mast cell deficiency, a game of kit and mouse. Immunity. 35 :668–670. 10.1016/j.immuni.2011.11.004 22118523 Kirk, J.A., R.J. Holewinski, V. Kooij, G. Agnetti, R.S. Tunin, N. Witayavanitkul, P.P. de Tombe, W.D. Gao, J. Van Eyk, and D.A. Kass. 2014. Cardiac resynchronization sensitizes the sarcomere to calcium by reactivating GSK-3β. J. Clin. Invest. 124 :129–139. 10.1172/JCI69253 24292707 Kirshenbaum, A.S., J.P. Goff, S.W. Kessler, J.M. Mican, K.M. Zsebo, and D.D. Metcalfe. 1992. Effect of IL-3 and stem cell factor on the appearance of human basophils and mast cells from CD34+ pluripotent progenitor cells. J. Immunol. 148 :772–777.1370517 Kitamura, Y., S. Go, and K. Hatanaka. 1978. Decrease of mast cells in W/Wv mice and their increase by bone marrow transplantation. Blood. 52 :447–452.352443 Kovanen, P.T. 2009. Mast cells in atherogenesis: actions and reactions. Curr. Atheroscler. Rep. 11 :214–219. 10.1007/s11883-009-0033-7 19361353 Kritikou, E., J. Kuiper, P.T. Kovanen, and I. Bot. 2016. The impact of mast cells on cardiovascular diseases. Eur. J. Pharmacol. 778 :103–115. 10.1016/j.ejphar.2015.04.050 25959384 Kulikovskaya, I., G. McClellan, R. Levine, and S. Winegrad. 2003. Effect of extraction of myosin binding protein C on contractility of rat heart. Am. J. Physiol. Heart Circ. Physiol. 285 :H857–H865. 10.1152/ajpheart.00841.2002 12860568 Kumar, D., T.A. Hacker, J. Buck, L.F. Whitesell, E.H. Kaji, P.S. Douglas, and T.J. Kamp. 2005. Distinct mouse coronary anatomy and myocardial infarction consequent to ligation. Coron. Artery Dis. 16 :41–44. 10.1097/00019501-200502000-00008 15654199 Layland, J., R.J. Solaro, and A.M. Shah. 2005. Regulation of cardiac contractile function by troponin I phosphorylation. Cardiovasc. Res. 66 :12–21. 10.1016/j.cardiores.2004.12.022 15769444 Levine, R., A. Weisberg, I. Kulikovskaya, G. McClellan, and S. Winegrad. 2001. Multiple structures of thick filaments in resting cardiac muscle and their influence on cross-bridge interactions. Biophys. J. 81 :1070–1082. 10.1016/S0006-3495(01)75764-5 11463648 Li, M.X., X. Wang, and B.D. Sykes. 2004. Structural based insights into the role of troponin in cardiac muscle pathophysiology. J. Muscle Res. Cell Motil. 25 :559–579. 10.1007/s10974-004-5879-2 15711886 Liu, J., A. Divoux, J. Sun, J. Zhang, K. Clément, J.N. Glickman, G.K. Sukhova, P.J. Wolters, J. Du, C.Z. Gorgun, 2009. Genetic deficiency and pharmacological stabilization of mast cells reduce diet-induced obesity and diabetes in mice. Nat. Med. 15 :940–945. 10.1038/nm.1994 19633655 Ludeman, M.J., Y.W. Zheng, K. Ishii, and S.R. Coughlin. 2004. Regulated shedding of PAR1 N-terminal exodomain from endothelial cells. J. Biol. Chem. 279 :18592–18599. 10.1074/jbc.M310836200 14982936 Manni, S., J.H. Mauban, C.W. Ward, and M. Bond. 2008. Phosphorylation of the cAMP-dependent protein kinase (PKA) regulatory subunit modulates PKA-AKAP interaction, substrate phosphorylation, and calcium signaling in cardiac cells. J. Biol. Chem. 283 :24145–24154. 10.1074/jbc.M802278200 18550536 Matsuzawa, S., K. Sakashita, T. Kinoshita, S. Ito, T. Yamashita, and K. Koike. 2003. IL-9 enhances the growth of human mast cell progenitors under stimulation with stem cell factor. J. Immunol. 170 :3461–3467. 10.4049/jimmunol.170.7.3461 12646606 McLarty, J.L., G.C. Meléndez, G.L. Brower, J.S. Janicki, and S.P. Levick. 2011. Tryptase/protease-activated receptor 2 interactions induce selective mitogen-activated protein kinase signaling and collagen synthesis by cardiac fibroblasts. Hypertension. 58 :264–270. 10.1161/HYPERTENSIONAHA.111.169417 21730297 McLean, P.G., D. Aston, D. Sarkar, and A. Ahluwalia. 2002. Protease-activated receptor-2 activation causes EDHF-like coronary vasodilation: selective preservation in ischemia/reperfusion injury: involvement of lipoxygenase products, VR1 receptors, and C-fibers. Circ. Res. 90 :465–472. 10.1161/hh0402.105372 11884377 Movsesian, M. 2015. New pharmacologic interventions to increase cardiac contractility: challenges and opportunities. Curr. Opin. Cardiol. 30 :285–291. 10.1097/HCO.0000000000000165 25807221 Murray, D.B., J. McLarty-Williams, K.T. Nagalla, and J.S. Janicki. 2012. Tryptase activates isolated adult cardiac fibroblasts via protease activated receptor-2 (PAR-2). J. Cell Commun. Signal. 6 :45–51. 10.1007/s12079-011-0146-y 21786087 Napoli, C., C. Cicala, J.L. Wallace, F. de Nigris, V. Santagada, G. Caliendo, F. Franconi, L.J. Ignarro, and G. Cirino. 2000. Protease-activated receptor-2 modulates myocardial ischemia-reperfusion injury in the rat heart. Proc. Natl. Acad. Sci. USA. 97 :3678–3683. 10.1073/pnas.97.7.3678 10737808 Nixon, B.R., S.D. Walton, B. Zhang, E.A. Brundage, S.C. Little, M.T. Ziolo, J.P. Davis, and B.J. Biesiadecki. 2014. Combined troponin I Ser-150 and Ser-23/24 phosphorylation sustains thin filament Ca2+ sensitivity and accelerates deactivation in an acidic environment. J. Mol. Cell. Cardiol. 72 :177–185. 10.1016/j.yjmcc.2014.03.010 24657721 Oka, T., J. Kalesnikoff, P. Starkl, M. Tsai, and S.J. Galli. 2012. Evidence questioning cromolyn’s effectiveness and selectivity as a ‘mast cell stabilizer’ in mice. Lab. Invest. 92 :1472–1482. 10.1038/labinvest.2012.116 22906983 Okada, M., H. Tokumitsu, Y. Kubota, and R. Kobayashi. 2002. Interaction of S100 proteins with the antiallergic drugs, olopatadine, amlexanox, and cromolyn: identification of putative drug binding sites on S100A1 protein. Biochem. Biophys. Res. Commun. 292 :1023–1030. 10.1006/bbrc.2002.6761 11944917 Okada, M., H. Itoh, T. Hatakeyama, H. Tokumitsu, and R. Kobayashi. 2003. Hsp90 is a direct target of the anti-allergic drugs disodium cromoglycate and amlexanox. Biochem. J. 374 :433–441. 10.1042/bj20030351 12803546 Oliveira, S.H., D.D. Taub, J. Nagel, R. Smith, C.M. Hogaboam, A. Berlin, and N.W. Lukacs. 2002. Stem cell factor induces eosinophil activation and degranulation: mediator release and gene array analysis. Blood. 100 :4291–4297. 10.1182/blood.V100.13.4291 12453875 Poglio, S., F. De Toni-Costes, E. Arnaud, P. Laharrague, E. Espinosa, L. Casteilla, and B. Cousin. 2010. Adipose tissue as a dedicated reservoir of functional mast cell progenitors. Stem Cells. 28 :2065–2072. 10.1002/stem.523 20845475 Poglio, S., F. De Toni, D. Lewandowski, A. Minot, E. Arnaud, V. Barroca, P. Laharrague, L. Casteilla, and B. Cousin. 2012. In situ production of innate immune cells in murine white adipose tissue. Blood. 120 :4952–4962. 10.1182/blood-2012-01-406959 23071275 Rababa’h, A., S. Singh, S.V. Suryavanshi, S.E. Altarabsheh, S.V. Deo, and B.K. McConnell. 2015. Compartmentalization role of A-kinase anchoring proteins (AKAPs) in mediating protein kinase A (PKA) signaling and cardiomyocyte hypertrophy. Int. J. Mol. Sci. 16 :218–229. 10.3390/ijms16010218 Rajagopal, S., K. Rajagopal, and R.J. Lefkowitz. 2010. Teaching old receptors new tricks: biasing seven-transmembrane receptors. Nat. Rev. Drug Discov. 9 :373–386. 10.1038/nrd3024 20431569 Ramirez-Correa, G.A., S. Cortassa, B. Stanley, W.D. Gao, and A.M. Murphy. 2010. Calcium sensitivity, force frequency relationship and cardiac troponin I: critical role of PKA and PKC phosphorylation sites. J. Mol. Cell. Cardiol. 48 :943–953. 10.1016/j.yjmcc.2010.01.004 20083117 Reid, A.C., J.A. Brazin, C. Morrey, R.B. Silver, and R. Levi. 2011. Targeting cardiac mast cells: pharmacological modulation of the local renin-angiotensin system. Curr. Pharm. Des. 17 :3744–3752. 10.2174/138161211798357908 22103845 Schmetzer, O., P. Valentin, M.K. Church, M. Maurer, and F. Siebenhaar. 2016. Murine and human mast cell progenitors. Eur. J. Pharmacol. 778 :2–10. 10.1016/j.ejphar.2015.07.016 26164789 Shpacovitch, V.M., T. Brzoska, J. Buddenkotte, C. Stroh, C.P. Sommerhoff, J.C. Ansel, K. Schulze-Osthoff, N.W. Bunnett, T.A. Luger, and M. Steinhoff. 2002. Agonists of proteinase-activated receptor 2 induce cytokine release and activation of nuclear transcription factor kappaB in human dermal microvascular endothelial cells. J. Invest. Dermatol. 118 :380–385. 10.1046/j.0022-202x.2001.01658.x 11841560 Solaro, R.J., and G.M. Arteaga. 2007. Heart failure, ischemia/reperfusion injury and cardiac troponin. Adv. Exp. Med. Biol. 592 :191–200. 10.1007/978-4-431-38453-3_17 17278366 Somasundaram, P., G. Ren, H. Nagar, D. Kraemer, L. Mendoza, L.H. Michael, G.H. Caughey, M.L. Entman, and N.G. Frangogiannis. 2005. Mast cell tryptase may modulate endothelial cell phenotype in healing myocardial infarcts. J. Pathol. 205 :102–111. 10.1002/path.1690 15586361 Sriwai, W., S. Mahavadi, O. Al-Shboul, J.R. Grider, and K.S. Murthy. 2013. Distinctive G protein-dependent signaling by protease-activated receptor 2 (PAR2) in smooth muscle: Feedback inhibition of RhoA by cAMP-independent PKA. PLoS One. 8 :e66743. 10.1371/journal.pone.0066743 23825105 Stelzer, J.E., J.R. Patel, J.W. Walker, and R.L. Moss. 2007. Differential roles of cardiac myosin-binding protein C and cardiac troponin I in the myofibrillar force responses to protein kinase A phosphorylation. Circ. Res. 101 :503–511. 10.1161/CIRCRESAHA.107.153650 17641226 Sumandea, C.A., M.L. Garcia-Cazarin, C.H. Bozio, G.A. Sievert, C.W. Balke, and M.P. Sumandea. 2011. Cardiac troponin T, a sarcomeric AKAP, tethers protein kinase A at the myofilaments. J. Biol. Chem. 286 :530–541. 10.1074/jbc.M110.148684 21056973 Sykes, B.D. 2003. Pulling the calcium trigger. Nat. Struct. Biol. 10 :588–589. 10.1038/nsb0803-588 12886291 Tallini, Y.N., K.S. Greene, M. Craven, A. Spealman, M. Breitbach, J. Smith, P.J. Fisher, M. Steffey, M. Hesse, R.M. Doran, 2009. c-kit expression identifies cardiovascular precursors in the neonatal heart. Proc. Natl. Acad. Sci. USA. 106 :1808–1813. 10.1073/pnas.0808920106 19193854 Thoemmes, S.F., C.A. Stutzke, Y. Du, M.D. Browning, P.M. Buttrick, and L.A. Walker. 2014. Characterization and validation of new tools for measuring site-specific cardiac troponin I phosphorylation. J. Immunol. Methods. 403 :66–71. 10.1016/j.jim.2013.11.020 24291343 Toeg, H.D., R. Tiwari-Pandey, R. Seymour, A. Ahmadi, S. Crowe, B. Vulesevic, E.J. Suuronen, and M. Ruel. 2013. Injectable small intestine submucosal extracellular matrix in an acute myocardial infarction model. Ann. Thorac. Surg. 96 :1686–1694. 10.1016/j.athoracsur.2013.06.063 24083799 Tran, T.T., and C.R. Kahn. 2010. Transplantation of adipose tissue and stem cells: role in metabolism and disease. Nat. Rev. Endocrinol. 6 :195–213. 10.1038/nrendo.2010.20 20195269 Trejo, J., A.J. Connolly, and S.R. Coughlin. 1996. The cloned thrombin receptor is necessary and sufficient for activation of mitogen-activated protein kinase and mitogenesis in mouse lung fibroblasts. Loss of responses in fibroblasts from receptor knockout mice. J. Biol. Chem. 271 :21536–21541. 10.1074/jbc.271.35.21536 8702939 Ward, D.G., S.M. Brewer, M.P. Cornes, and I.P. Trayer. 2003. A cross-linking study of the N-terminal extension of human cardiac troponin I. Biochemistry. 42 :10324–10332. 10.1021/bi034495r 12939162 Weithauser, A., and U. Rauch. 2014. Role of protease-activated receptors for the innate immune response of the heart. Trends Cardiovasc. Med. 24 :249–255. 10.1016/j.tcm.2014.06.004 25066486 Welker, P., J. Grabbe, T. Zuberbier, S. Guhl, and B.M. Henz. 2000. Mast cell and myeloid marker expression during early in vitro mast cell differentiation from human peripheral blood mononuclear cells. J. Invest. Dermatol. 114 :44–50. 10.1046/j.1523-1747.2000.00827.x 10620114 Wernersson, S., and G. Pejler. 2014. Mast cell secretory granules: armed for battle. Nat. Rev. Immunol. 14 :478–494. 10.1038/nri3690 24903914 White, H.D., and D.P. Chew. 2008. Acute myocardial infarction. Lancet. 372 :570–584. 10.1016/S0140-6736(08)61237-4 18707987 White, H.D., K. Thygesen, J.S. Alpert, and A.S. Jaffe. 2014. Republished: clinical implications of the third universal definition of myocardial infarction. Postgrad. Med. J. 90 :502–510. 10.1136/postgradmedj-2012-302976rep 25140007 Xaubet, A., J.A. Moisés, C. Agustí, J.A. Martos, and C. Picado. 1991. Identification of mast cells in bronchoalveolar lavage fluid. Comparison between different fixation and staining methods. Allergy. 46 :222–227. 10.1111/j.1398-9995.1991.tb00575.x 1711798 Xiang, F.L., X. Lu, Y. Liu, and Q. Feng. 2013. Cardiomyocyte-specific overexpression of human stem cell factor protects against myocardial ischemia and reperfusion injury. Int. J. Cardiol. 168 :3486–3494. 10.1016/j.ijcard.2013.04.165 23680593 Yang, Y., J.Y. Lu, X. Wu, S. Summer, J. Whoriskey, C. Saris, and J.D. Reagan. 2010. G-protein-coupled receptor 35 is a target of the asthma drugs cromolyn disodium and nedocromil sodium. Pharmacology. 86 :1–5. 10.1159/000314164 20559017 Zhang, T., D.F. Finn, J.W. Barlow, and J.J. Walsh. 2016. Mast cell stabilisers. Eur. J. Pharmacol. 778 :158–168. 10.1016/j.ejphar.2015.05.071 26130122 Zhong, B., and D.H. Wang. 2009. Protease-activated receptor 2-mediated protection of myocardial ischemia-reperfusion injury: role of transient receptor potential vanilloid receptors. Am. J. Physiol. Regul. Integr. Comp. Physiol. 297 :R1681–R1690. 10.1152/ajpregu.90746.2008 19812353 Zouggari, Y., H. Ait-Oufella, P. Bonnin, T. Simon, A.P. Sage, C. Guérin, J. Vilar, G. Caligiuri, D. Tsiantoulas, L. Laurans, 2013. B lymphocytes trigger monocyte mobilization and impair heart function after acute myocardial infarction. Nat. Med. 19 :1273–1280. 10.1038/nm.3284 24037091

PMC004xxxxxx/PMC4969795.txt

==== Front J Gen Physiol J Gen Physiol jgp jgp The Journal of General Physiology 0022-1295 1540-7748 The Rockefeller University Press 27481713 201611586 10.1085/jgp.201611586 Research Articles Research Article The α2δ-1 subunit remodels CaV1.2 voltage sensors and allows Ca2+ influx at physiological membrane potentials The α2δ-1 subunit remodels CaV1.2 voltage sensors http://orcid.org/0000-0001-7267-9655 Savalli Nicoletta 1 Pantazis Antonios 1 Sigg Daniel 5 Weiss James N. 234 Neely Alan 16 http://orcid.org/0000-0003-4225-9814 Olcese Riccardo 134 1 Department of Anesthesiology, Division of Molecular Medicine, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA 90095 2 Department of Medicine (Cardiology), David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA 90095 3 Department of Physiology, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA 90095 4 Cardiovascular Research Laboratories, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA 90095 5 dPET, Spokane, WA 99223 6 Centro Interdisciplinario de Neurociencias de Valparaíso, Facultad de Ciencias, Universidad de Valparaíso, Valparaíso 2360102, Chile Correspondence to Riccardo Olcese: rolcese@ucla.edu 8 2016 148 2 147159 17 2 2016 30 6 2016 © 2016 Savalli et al. 2016 https://creativecommons.org/licenses/by-nc-sa/3.0/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 3.0 Unported license, as described at http://creativecommons.org/licenses/by-nc-sa/3.0/). Voltage-sensing domains (VSDs) in voltage-gated calcium channels sense the potential difference across membranes and interact with the pore to open it. Savalli et al. find that the accessory subunit α2δ-1 increases the sensitivity of VSDs I–III and also their efficiency of coupling to the pore. Excitation-evoked calcium influx across cellular membranes is strictly controlled by voltage-gated calcium channels (CaV), which possess four distinct voltage-sensing domains (VSDs) that direct the opening of a central pore. The energetic interactions between the VSDs and the pore are critical for tuning the channel’s voltage dependence. The accessory α2δ-1 subunit is known to facilitate CaV1.2 voltage-dependent activation, but the underlying mechanism is unknown. In this study, using voltage clamp fluorometry, we track the activation of the four individual VSDs in a human L-type CaV1.2 channel consisting of α1C and β3 subunits. We find that, without α2δ-1, the channel complex displays a right-shifted voltage dependence such that currents mainly develop at nonphysiological membrane potentials because of very weak VSD–pore interactions. The presence of α2δ-1 facilitates channel activation by increasing the voltage sensitivity (i.e., the effective charge) of VSDs I–III. Moreover, the α2δ-1 subunit also makes VSDs I–III more efficient at opening the channel by increasing the coupling energy between VSDs II and III and the pore, thus allowing Ca influx within the range of physiological membrane potentials. National Heart, Lung, and Blood Institute 10.13039/100000050 P01HL078931 National Institute of General Medical Sciences 10.13039/100000057 R01GM110276 American Heart Association 10.13039/100000968 14SDG20300018 ==== Body pmcINTRODUCTION Calcium influx through voltage-activated calcium (CaV) channels translates electrical signals into a variety of physiological outcomes such as cell contraction, neurotransmitter or hormonal release, and gene expression (Catterall, 2011; Zamponi et al., 2015). The specificity of the Ca2+ signal relies on the activity of the CaV channel complex being perfectly tuned to voltage signals. CaV channels are multimeric proteins formed by the pore-forming α1 subunit and at least three auxiliary subunits, β, α2δ, and calmodulin, in a 1:1:1:1 stoichiometry, resulting in an asymmetric structural architecture (Fig. 1; Findeisen and Minor, 2010; Catterall, 2011; Dolphin, 2013; Ben-Johny and Yue, 2014; Neely and Hidalgo, 2014; Campiglio and Flucher, 2015; Wu et al., 2015). The α2δ auxiliary subunit is a large (∼170 kD), mostly extracellular protein with a single membrane-anchoring segment (Davies et al., 2010) that binds to the α1 subunit from the extracellular side (Cassidy et al., 2014). α2 and δ proteins are the products of the same gene as a preprotein that is posttranslationally proteolysed and then linked by a disulfide-bond to form the mature α2δ protein (Calderón-Rivera et al., 2012). Four genes (CACNA2D1–4) encode for distinct α2δ isoforms (α2δ-1–4), which are all expressed in the brain (Dolphin, 2013). In addition to brain tissue, α2δ-1 is strongly expressed in cardiac, smooth, and skeletal muscles, whereas α2δ-4 is found in endocrine tissues and the retina. Mutations in the α2δ-1 gene can lead to Brugada (Burashnikov et al., 2010) and short QT (Templin et al., 2011; Bourdin et al., 2015) syndromes and are associated with epilepsy and mental disability (Vergult et al., 2015). In mice, naturally occurring mutations in the α2δ-2 gene lead to ataxia and epilepsy (Barclay et al., 2001), whereas the α2δ-3 protein is important for synaptic morphogenesis (Kurshan et al., 2009) and nociception (Neely et al., 2010). Mutations in α2δ-4 can result in night blindness (Wycisk et al., 2006). Moreover, α2δ-1 and -2 have been identified as the molecular targets of gabapentinoid drugs (such as gabapentin and pregabalin), mediating their analgesic action in neuropathic pain (Field et al., 2006; Hendrich et al., 2008; Uchitel et al., 2010). Finally, it has been shown that α2δ proteins also play an important role in synapse formation (Eroglu et al., 2009). Figure 1. CaV1.2 channel topology and subunit composition. (A) In CaV1.2, the pore-forming α1C subunit consists of four tandem repeats (I–IV), each crossing the membrane six times (S1–S6). Helices S1–S4 form the VSD, and helices S5 and S6 form the pore domain (PD). The intracellular loop between repeats I and II encompasses the binding site for the auxiliary β subunit (α-interacting domain [AID]), whereas the C terminus includes the IQ region, where calmodulin binds. (B) CaV1.2 channels are multimeric proteins composed of the α1C pore-forming subunit, a mostly extracellular α2δ subunit, and an intracellular β subunit. In this cartoon representation, the four repeats that constitute the α1C subunit are arranged clockwise, as in a recent cryo-EM structure of related α1S (CaV1.1) channel (Wu et al., 2015): repeats III (green) and IV (yellow) are in the front, whereas I (blue) and II (red) are in the back. (C) Side and top views of the atomic structure of a voltage-gated Na+ channel (NaVAb; PDB accession no. 4EKW; Payandeh et al., 2012), shown as putative structural representation of a CaV α1 channel. S4 helices were fluorescently labeled at their extracellular flank, one VSD at a time (pink triangles). Several studies report that the interaction of α2δ-1 with the pore-forming α1C subunits (L-type CaV1.2) favors channel activation, as manifested by a hyperpolarizing shift of channel opening (Felix et al., 1997; Platano et al., 2000; Bourdin et al., 2015) and an accelerated time course of activation (Bangalore et al., 1996). Accordingly, channel activation is diminished with α2δ-1 down-regulation (Tuluc et al., 2007; Fuller-Bicer et al., 2009) and enhanced with α2δ-1 up-regulation (Li et al., 2006). Thus, by facilitating channel opening, α2δ-1 allows CaV1.2 channels to operate at physiological membrane potentials. However, the molecular mechanism by which α2δ-1 facilitates CaV1.2 activation is as yet poorly understood. Because α2δ-1 modulates the voltage-dependent properties of CaV1.2 channels (Felix et al., 1997; Platano et al., 2000; Bourdin et al., 2015) and associates with α1C subunits asymmetrically (Walsh et al., 2009a), we hypothesized that α2δ-1 differentially modulates each of the four voltage-sensing domains (VSDs), as well as their contribution to channel opening. In fact, the pore-forming α1C subunit consists of four homologous, but nonidentical, concatenated repeats (I–IV), each composed of a VSD (transmembrane helices S1–S4) and a quarter of the pore domain (S5–S6; Catterall, 2011). Using voltage clamp fluorometry (VCF), we have recently revealed the functional heterogeneity of the four CaV1.2 VSDs, whereby each undergoes structural changes during channel activation, with unique voltage- and time-dependent properties, such that the activation of VSDs II and III, and to a lesser extent VSD I, energetically contributes to channel opening (Pantazis et al., 2014). VCF is a powerful investigative tool that allows for simultaneous measurements of ionic current kinetics and structural rearrangements occurring within specific protein domains, the latter tracked using environmentally sensitive fluorophores (Claydon and Fedida, 2007; Gandhi and Olcese, 2008; Talwar and Lynch, 2015; Zhu et al., 2016). VCF has been a successful approach in the study of numerous voltage-sensitive proteins (Mannuzzu et al., 1996; Cha et al., 1999; Smith and Yellen, 2002; Savalli et al., 2006; Kohout et al., 2008; Osteen et al., 2010; Tombola et al., 2010), transporters (Meinild et al., 2002; Larsson et al., 2004; Ghezzi et al., 2009), and receptors (Dahan et al., 2004; Lörinczi et al., 2012). Using VCF, auxiliary subunit modulation of VSDs has also been detected in ion channels, such as BK channels (Savalli et al., 2007) or KV7.1 (Ruscic et al., 2013). However, VCF has only recently been adapted to investigate CaV channels (Pantazis et al., 2014). In this study, by using VCF and a structurally relevant allosteric model of CaV1.2 activation, we show that the α2δ-1 auxiliary subunit (a) facilitates the voltage-dependent activation of CaV1.2 VSDs I–III; (b) accelerates VSD I kinetics; and (c) increases the energetic contribution of VSDs I–III to pore opening. These results unravel the molecular mechanisms by which α2δ-1 exerts its modulation on CaV1.2 channel activation, allowing for Ca2+ influx to occur in excitable cells at physiological membrane potentials. MATERIALS AND METHODS Molecular biology Human α1C-77 subunits (GenBank accession no. CAA84346; Soldatov, 1992) of CaV1.2 channels, with a Cys substituted at an extracellular position in the S3–S4 linker of each VSD at a time, were used (F231C, L614C, V994C, or S1324C for VSDs I–IV, respectively) as previously described (Pantazis et al., 2014). Single-point mutations were generated using the QuikChange Site-Directed Mutagenesis kit (Agilent Technologies) and confirmed by sequencing. Auxiliary subunits α2δ-1 (UniProt accession no. P13806) and β3 (UniProt accession no. P54286) were also coexpressed. The cRNA of the different subunits was transcribed in vitro (mMESSAGE MACHINE; Ambion) and injected into stage VI Xenopus laevis oocytes (50 nl at 0.1–0.5 µg/µl). VCF 3–4 d after injection, oocytes were incubated with thiol-reactive fluorophores sensitive to environmental changes (10 µM tetramethylrhodamine-5-maleimide [TMRM] for VSD II and 20 µM 2-((5(6)-tetramethyl-rhodamine)carboxylamino)ethyl methanethiosulfonate [MTS-TAMRA] for VSD I, III, or IV) in a depolarizing solution (120 mM K-methanesulfonate [MES], 2 mM Ca(MES)2, and 10 mM HEPES, pH 7.0). Subsequently, oocytes were voltage clamped using the cut-open oocyte technique (Stefani and Bezanilla, 1998; Pantazis and Olcese, 2013). Fluorescence changes and ionic currents were acquired simultaneously from the same membrane area (Gandhi and Olcese, 2008; Pantazis and Olcese, 2013). External solution was 2 mM Ba(MES)2, 120 mM NaMES, and 10 HEPES, pH 7.0, supplemented with 0.1 ouabain to eliminate charge movement from Na/K ATPase (Neely et al., 1994). Internal solution was 120 mM K-glutamate and 10 mM HEPES, pH 7.0. Pipette solution was 2.7 M Na-MES, 10 mM NaCl, and 10 mM Na-HEPES, pH 7.0. Before experiments, oocytes were injected with 10 mM BAPTA•4K, pH 7.0, to prevent activation of native Ca2+- and Ba2+-dependent Cl− channels (Barish, 1983). Data analysis The voltage dependence of ionic conductance (G(V), estimated from the peaks of the tail currents) and fluorescence changes (F(V)) were empirically characterized by fitting to one or two Boltzmann functions as f(V) = {1 + exp([q · (Vhalf − Vm)(F/RT)])}−1, where q is the effective charge, Vhalf is the half-activation potential, Vm is the membrane potential, T is the absolute temperature, and F and R are the Faraday and Gas constants, respectively. F(V) curves can be satisfactorily described by single Boltzmann functions both the absence and the presence of the α2δ-1 subunit (see Figs. 4 and 5). The time course of fluorescence onset (VSD activation) was fitted in background-subtracted fluorescence traces to the sum of two exponential components:f(t)=B+∑i=12Ai⋅exp(−t/τi), where B is the baseline, A is the amplitude, t is time, and τ is the time constant. Fitting was performed in Matlab (MathWorks) by least squares (Optimization Toolbox). Only traces with sufficient signal-to-noise ratio (S:N >2) were included in the kinetics statistics. S:N is defined as mean signal amplitude divided by the root mean square. Fractional amplitude–weighted time constants (τavg) were calculated usingτavg=∑i=12αi⋅τi, whereαi=AiA1+A2. Data are reported as mean ± SEM; statistical analysis was performed using Excel (Microsoft). Allosteric model Modeling CaV1.2 kinetics and activation curves through a five-particle allosteric scheme was performed as described previously (Pantazis et al., 2014).In brief, equilibrium states were determined from the values of five particle equilibrium constants (L, J1–4) and four VSD–pore coupling constants D1–4 using the channel partition function:Z=(1+J1)(1+J2)(1+J3)(1+J4)+L(1+J1D1)(1+J2D2)(1+J3D3)(1+J4D4). The pore particle (L) derives its voltage dependence through a gating charge displacement ΔqL and a characteristic midpoint voltage VL; for example, L = exp(ΔqL(V − VL)/kT). Similar expressions applied to J1–4 were used to describe intrinsic VSD activation. The four allosteric factors D1–4 are related to VSD–pore interaction energies W1–4 through Di = exp(–Wi/kT). The equilibrium curves 〈k=l,j1–4〉 for the five gating particles are easily derived from the partition function through the relations 〈k〉=∂lnZ/∂lnK, which were used to fit the experimental conductance (G(V)) and fluorescence (F(V)) curves. A kinetic model of channel activation that reduces to the thermodynamic model under equilibrium conditions was obtained by assigning two additional variables for each particle transition: a frequency factor ν and a fractional position x of the transition barrier between resting and active states (Sigg, 2014). The forward and backward rate constants for the transition between a configuration i and any of the accessible configurations j after the activation of one of the five gating particles were expressed asαi−j=νk(ZjZi)xk andβj−i=νk(ZiZj)1−xk, where k refers to the transitioning particle and Zi and Zj are the configuration-specific contributions to the overall partition function Z (obtained by expanding the earlier expression of Z into its 32 terms). The channel kinetics were solved by integrating p→(t)=p→(0)⋅exp(Qt), which describes state probabilities of all states (p). Q is the standard rate matrix as described in Colquhoun and Hawkes (1981). The initial condition p(0) for the holding potential was obtained from p→(t→∞). The time dependence of a quantity of interest A (ionic current or fluorescence) was obtained from〈A(t)〉=N∑ipi(t)ai, where ai is the value of the desired quantity at configuration i. To find the set of parameters that best described the data, several approaches were used such as Marquadt–Levenberg, implemented in Berkeley Madonna, and Nedler–Mead, developed in Matlab, minimizing the error function “ssq” generated by the weighed sum of 10 error functions, five for steady-state voltage dependencies and five for time-dependent signals. Each individual error function corresponds to the sum of the squares of the difference between the experimental and simulated datasets normalized by n and the square of the maximum value. To test for the uniqueness of the solution and estimating the 95% credible interval of each parameter, we used a Bayesian approach using Markov chain Monte Carlo (MCMC) sampling as in Hines et al. (2014), but instead of using likelihood ratio to test for high posterior probabilities, we useda=min[1,exp(ssqi–ssqi+1T)]. Then, transitions of the Markov chain were accepted with probability a, as described in Li (2012). To further constrain the solution space and take better advantage of the time-dependent data, we added a set of penalty functions as further explained in the supplementary figure legends. Online supplemental material Fig. S1 shows representative Ba2+ current traces from Xenopus oocytes expressing CaV1.2 channel complexes formed by α1C + β3 subunits. Fig. S2 shows histograms of posterior distributions of 14 parameters obtained from a 100,000-trial MCMC run. Fig. S3 shows histograms of posterior distributions of parameters xi and ni (i = L,1,2,3,4). Online supplemental material is available at http://www.jgp.org/cgi/content/full/jgp.201611586/DC1. RESULTS α2δ-1 facilitates the voltage-dependent activation of human CaV1.2 channels To understand the mechanism of α2δ-1 subunits modulation of human CaV1.2 channels, we expressed in Xenopus oocytes CaV1.2 channels consisting of α1C and β3 subunits, in the presence or the absence of α2δ-1 proteins. We voltage clamped the cells using the cut-open oocyte voltage clamp technique (Stefani and Bezanilla, 1998; Pantazis and Olcese, 2013) and recorded ionic currents (Fig. 2, A and B) using Ba2+ as the charge carrier to prevent calcium-dependent inactivation (Peterson et al., 1999; Qin et al., 1999). Because CaV1.2 channels lacking α2δ-1 subunits activate slowly (Fig. 2 A), relatively longer depolarizations were necessary for ionic current to reach quasi–steady state (Fig. S1), a condition necessary to construct conductance versus voltage relationships (G(V)) from tail currents. We observed that α2δ-1 coexpression strongly facilitated channel opening in human CaV1.2 channels by shifting the CaV1.2 half-activation potential (Vhalf) of the G(V) curves by ∼50 mV toward more hyperpolarized potentials (Fig. 2 C), in agreement with data previously obtained from the rabbit isoforms (Felix et al., 1997; Platano et al., 2000; Bourdin et al., 2015). G(V) curves obtained from channels expressed with the full complement of auxiliary subunits were well described by the sum of two Boltzmann functions with distinct voltage-dependent properties (Vhalf1 = −4.02 ± 0.38 mV, z1 = 3.46 ± 0.11 e0, G1 = 57.56 ± 4.82%, Vhalf2 = 42.56 ± 1.87 mV, z2 = 1.27 ± 0.04 e0, G2 = 42.44 ± 4.82%; n = 4; Fig. 2 C), alluding to a complex voltage-dependent activation mechanism with more than one voltage-dependent opening transitions; in contrast, G(V) curves for channels lacking α2δ-1 were well accounted for by a single Boltzmann distribution (Vhalf = 68.01 ± 1.32 mV and z = 1.19 ± 0.01 e0; n = 7), which is a tentative (yet tantalizing) indication that these channels gate in a two-state process. We further mechanistically evaluated this premise using an allosteric model of voltage-dependent CaV activation. Figure 2. The α2δ-1 subunit facilitates CaV1.2 channel opening. (A and B) Representative Ba2+ current traces from Xenopus oocytes expressing human CaV1.2 channel complexes with different subunit composition: α1C + β3 subunits (A) or α1C + β3 + α2δ-1 subunits (B). The voltage protocol is reported above the current traces. (C) Mean conductance versus voltage (G(V)) relationships were constructed from tail currents as in B and Fig. S1 (mean ± SEM; error bars are within the symbols; n = 7 for α1C + β3 and n = 4 for α1C + β3 + α2δ-1). The α2δ-1 auxiliary subunit facilitates channel opening, as manifested by a hyperpolarizing shift of the conductance voltage dependence by ∼50 mV. α2δ-1 increases the rate of VSD I activation Given the large difference (>50 mV) in the voltage dependence of CaV1.2 activation in the presence or absence of α2δ-1 subunits (Fig. 2), we tested the hypothesis that α2δ-1 association with α1C induces a functional remodeling of one or more VSDs, altering their gating properties. We used the VCF technique (Mannuzzu et al., 1996; Cha and Bezanilla, 1997; Gandhi and Olcese, 2008) to track the molecular rearrangements of the individual VSDs of human CaV1.2 channels in the presence or absence of α2δ-1. Briefly, this involves the introduction of a cysteine residue one at a time at a strategic and specific position extracellular to the S4 helix in each CaV1.2 VSD, as shown in our previous study (Pantazis et al., 2014). In voltage-gated ion channels, the S4 segment typically contains the positively charged amino acids effectively responsible for voltage sensing and undergoes structural rearrangements during depolarizations (Tombola et al., 2006; Bezanilla, 2008; Chanda and Bezanilla, 2008; Swartz, 2008; Catterall, 2010; Palovcak et al., 2014). CaV1.2 channels (α1C + β3 subunits) were expressed with or without α2δ-1 subunits in Xenopus oocytes. After labeling of the cysteines with thiol-reactive fluorophores that are sensitive to the environment, we used VCF to simultaneously study the voltage-dependent activation of the pore (ionic current) and each of the four VSDs (fluorescence) in conducting CaV1.2 channels. The labeling positions and fluorophores used here were the same as in our previous work (Pantazis et al., 2014). The effect of α2δ-1 subunits on the kinetics of VSD activation was quantified for 100-ms depolarizations to 20 mV (Table 1). The activation of VSD I was accelerated by approximately twofold by the α2δ-1 subunit, increasing the fractional amplitude of the fast component of activation (Fig. 3 and Table 1). This result suggests that the acceleration of ionic current by the α2δ-1 subunit (Fig. 2, A and B; Felix et al., 1997; Platano et al., 2000; Tuluc et al., 2007) may result from a faster VSD I, consistent with the findings of Nakai et al. (1994), who demonstrated a relevant role for VSD I in controlling CaV1.2 current kinetics by transferring CaV1.1 VSD I sequences into the corresponding location in CaV1.2. In contrast, the kinetics of VSDs II–IV were practically unaffected by α2δ-1 (Fig. 3 and Table 1). Table 1. Effect of the α2δ-1 subunit on CaV1.2 (α1C/β3) VSD activation kinetics (100-ms depolarizations to 20 mV) VSD Parameter No α2δ With α2δ-1 VSD I τ1 (ms) 5.48 ± 1.57 (n = 4) 3.8 ± 0.37 (n = 3) α1 (%) 47.2 ± 3.7 63.9 ± 7.7 τ2 (ms) 56.3 ± 7.9 29.6 ± 6.0 τavg (ms) 31.4 ± 2.3 13.6 ± 3.7 VSD II τ1 (ms) 0.83 ± 0.32 (n = 3) 1.04 ± 0.39 (n = 3) α1 (%) 43.1 ± 2.0 66.2 ± 8.6 τ2 (ms) 28.3 ± 12.5 30.7 ± 5.7 τavg (ms) 16.5 ± 7.2 11.9 ± 3.6 VSD III τ1 (ms) 2.39 ± 0.44 (n = 6) 2.0 ± 0.17 (n = 4) α1 (%) 91.8 ± 5.5 83.5 ± 9.7 τ2 (ms) 37.9 ± 7.9 32.0 ± 9.0 τavg (ms) 4.6 ± 1.1 5.09 ± 0.98 VSD IV τ1 (ms) 21.5 ± 1.9 (n = 6) 16.5 ± 2.1 (n = 3) α1 (%) 100 100 τ2 (ms) NA NA τavg (ms) 21.5 ± 1.9 16.5 ± 2.1 NA, not applicable. Figure 3. Voltage-dependent rearrangements of the individual CaV1.2 VSDs in the presence and absence of the α2δ subunit. Fluorescence traces reporting the structural rearrangements of each CaV1.2 VSD in the absence (top) or presence (bottom) of the α2δ-1 subunit are shown. The channels were expressed in Xenopus oocytes and fluorescently labeled at extracellular S4 positions in VSD I (F231C), VSD II (L614C), VSD III (V994C), or VSD IV (S1324C). The voltage protocol is reported above the traces. The holding potential was –90 mV. Because VSD IV was partially activated at –90 mV (Fig. 4 D), oocytes expressing S1324C channels underwent a prepulse to –160 mV to allow VSD IV to return to its resting state before each depolarization. α2δ-1 facilitates the voltage-dependent activation of VSDs I, II, and III To assess how the voltage dependence of the individual VSDs was affected by α2δ-1 subunit association, we constructed activation curves (F(V)) from the corresponding fluorescence intensity at the end of 100-ms pulses over a wide range of membrane potentials (Fig. 4). The F(V) values of VSDs I–III were shifted to more negative potentials in the presence of the α2δ-1 subunit, whereas VSD IV activation was unaffected (Fig. 4, A–D). Moreover, VSDs I–III exhibited a steeper slope of voltage-dependent activation: the effective charge q increased by approximately twofold (Fig. 4, A–C; and Table 2). Overall, in the absence of the α2δ-1 subunit, the activation curves of the four VSDs were highly disparate, spread over a voltage range spanning ∼90 mV (Fig. 5 A and Table 2), while the variance of the Vhalf values was 1,300 mV2. The association of α2δ-1 subunit narrowed the range of membrane potentials at which VSDs activated (∼50 mV; Table 2, Vhalf variance: 520 mV2) and shifted the G(V) such that VSD voltage dependence was closer to pore opening (Fig. 5 B). A separation of the half activation potential of VSD activation and channel opening can be interpreted as decreased coupling between VSD activation and pore gating (Sigg, 2014). Taken together, these results are consistent with the view that α2δ-1 is required to increase the coupling between VSDs I–III and the CaV1.2 pore. In addition, because the effective charge q is the summed displacements of residue charges across the membrane potential profile, we cannot exclude that α2δ-1 association also increases CaV1.2 voltage sensitivity by altering the shape of the profile (for example, making it steeper in the region of charge translation). To discriminate among these mechanisms (increased coupling, increased effective charge, or both), we modeled CaV1.2 activation with an allosteric model used previously (Pantazis et al., 2014). Figure 4. The α2δ-1 subunit facilitates the voltage-dependent activation of CaV1.2 VSDs I–III, whereas VSD IV is unperturbed. (A–D) Mean voltage dependence of VSD activation constructed from experiments as in Fig. 3. The α2δ-1 subunit facilitated the activation of VSDs I–III, as revealed by a more hyperpolarized voltage dependence of VSD activation, although to a different extent, whereas VSD IV was unaffected. Boltzmann fitting parameters are reported in Table 2. Table 2. Fitting parameters for the Boltzmann functions fitting the fluorescence data from each VSD ( Fig. 4 ) VSD Parameter No α2δ With α2δ−1 VSD I q (e0) 1.6 ± 0.1 (n = 5) 2.8 ± 0.1 (n = 5) Vhalf (mV) 36.5 ± 3.1 6.1 ± 1.3 VSD II q (e0) 1.2 ± 0.1 (n = 6) 2.7 ± 0.2 (n = 3) Vhalf (mV) −6.7 ± 1.8 −30.8 ± 3.9 VSD III q (e0) 0.9 ± 0.1 (n = 6) 1.5 ± 0.09 (n = 5) Vhalf (mV) 0.9 ± 4.1 −22.0 ± 1.7 VSD IV q (e0) 0.9 ± 0.04 (n = 7) 1.1 ± 0.1 (n = 4) Vhalf (mV) −51.4 ± 4 −48.5 ± 2.5 Figure 5. The α2δ-1 subunit remodels VSDs I–III. (A and B) Mean normalized G(V) and F(V) data points from α1C + β3 (A) or α1C + β3 + α2δ-1 (B) channels and the corresponding Boltzmann fits are shown superimposed (mean ± SEM). These results suggest that the facilitation of voltage-dependent pore activation by α2δ-1 subunits is likely mediated by the remodeling of VSDs I–III. α2δ-1 facilitates CaV1.2 activation by increasing the energetic contribution of VSDs I–III to pore opening We analyzed the VCF data with the 32-state allosteric model for CaV1.2 activation (Pantazis et al., 2014), consisting of five gating elements (one pore, four VSDs; Fig. 6 A) and therefore relevant to CaV1.2 architecture. Pore and VSDs can exist in two states: closed-open and resting-active, each undergoing voltage-dependent transitions. Thus, in this model, the pore as well as the VSDs are intrinsically voltage dependent (half-activation potential V and charge displacement q). The activation of one or more VSDs stabilizes the open state of the pore through energy coupling with magnitude Wi (i = 1–4). Figure 6. An allosteric structure-based model of voltage-dependent CaV1.2 activation accounts for time- and voltage-dependent properties of CaV1.2 channels lacking the α2δ-1 subunit. (A) Scheme of the model used to simultaneously fit current and optically tracked conformational changes from each of the four CaV1.2 VSDs. Each VSD and the pore are modeled as two-state particles that can undergo resting↔active or closed↔open voltage-dependent transitions, respectively. Activation of a VSD allosterically stabilizes the open pore state by energy W. (B) Mean normalized G(V) and F(V) data points from α1C + β3 channels are shown superimposed with the model predictions (curves). (C) Ionic currents (top) and fluorescence traces from each VSD (normalized to VSD activation; bottom) from α1C + β3 channels. The simultaneous model fits are shown superimposed as black lines. Fitting parameters are reported in Table 3. Kinetic and quasi-equilibrium data from the pore (ionic currents) and each VSD (fluorescence) were simultaneously fitted in the absence of α2δ-1 with no assumption or constraint. The model accurately accounts for the voltage- and time-dependent properties of channels composed of α1C + β3 subunits (Fig. 6, B and C). The most salient feature of the fitted quantities is that the energetic contribution to pore opening (W) of each VSD is small (<1 kT or 25 meV). A comparison of W values with and without the α2δ-1 subunit demonstrates a doubling of the energetic contribution to pore opening by VSDs I and III (W1 and W3) and an approximately threefold increase of W2 in the presence of α2δ-1, whereas the contribution of VSD IV (W4) was practically unchanged (Table 3; parameters with α2δ-1 are from Pantazis et al., 2014 and reported here for clarity). In addition to enhancing VSD I–III energetic contributions to pore opening, α2δ-1 increased the charge displacement (q) of VSDs I and II by ∼140% and ∼200%, respectively (Table 3). In contrast, the intrinsic pore parameters qL and VL varied minimally with the addition of α2δ-1 to the channel. Thus, α2δ-1 modulates the CaV1.2 channel by exerting its effect on the VSDs (VSDs I–III) rather than on the pore. Table 3. Fitting parameters for the model predictions in Fig. 6 Parameter No α2δ With α2δ-1a Pore qL (e0) 0.99 0.76 VL (mV) 103 140 x L 1.0 0.49 νL (s−1) 1,356 670 VSD I q1 (e0) 1.4 2.0 V1 (mV) 37 8.5 x 1 0.72 0.89 ν1 (s−1) 17 110 VSD II q2 (e0) 1.2 2.5 V2 (mV) −6.9 −27 x 2 0.78 0.99 ν2 (s−1) 35 44 VSD III q3 (e0) 0.83 1.0 V3 (mV) 3.7 −11 x 3 1.0 0.93 ν3 (s−1) 145 160 VSD IV q4 (e0) 0.92 1.1 V4 (mV) −54 −52 x 4 0.35 0.55 ν4 (s−1) 16 11 Energetic interaction W1 (meV) −8.0 −16 W2 (meV) −16 −50 W3 (meV) −19 −45 W4 (meV) 4.1 −0.87 a Parameters in this column are from Pantazis et al., 2014. The uniqueness of the solution and the 95% credible interval of each parameter were obtained with a Bayesian approach using MCMC sampling as in Hines et al. (2014). The results are reported in Figs. S2 and S3. DISCUSSION The α1 subunit of voltage-gated CaV channels is a modular, pseudotetrameric protein consisting of a central pore domain coupled to four homologous but not identical VSDs. Several auxiliary subunits, including α2δ and β, associate with the α1 subunit in a 1:1:1 ratio (Catterall, 2011; Wu et al., 2015) to regulate channel trafficking and biophysical properties (Fang and Colecraft, 2011; Dolphin, 2012; Neely and Hidalgo, 2014; Campiglio and Flucher, 2015). Using VCF to optically track the individual VSDs in a human CaV1.2 α1 + β3 complex with and without the auxiliary α2δ-1 subunits, we gained a mechanistic understanding of α2δ-1–mediated facilitation of CaV1.2 activation. We found that α2δ-1 alters the biophysical properties of three VSDs (I–III). The association of α2δ-1 with α1C increases the coupling of VSDs I–III to the channel pore, allowing the CaV1.2 channel to operate in the range of physiological membrane potentials found in excitable cells. The physical nature of α1C/α2δ-1 association The association of α2δ-1 with α1C resulted in a substantial change in the intrinsic voltage-sensing properties of VSDs I–III (Fig. 4 and Table 3). This effect suggests either a direct physical or long-range allosteric interaction of α2δ-1 with these VSDs, causing their structural rearrangement. Our results are consistent with coimmunoprecipitation experiments showing that α2δ subunits bind to extracellular loops of repeat III and possibly to other extracellular loops of α1 (Gurnett et al., 1997). On the other hand, VSD IV remained completely unaffected, suggesting that α2δ-1 and VSD IV do not physically interact and consistent with the findings by Tuluc et al. (2009), who showed that the deletion of S3-S4 loop in VSD IV affects channel gating in CaV1.1 but not its modulation by α2δ-1 subunits. The findings that three VSDs out of four are remodeled by α2δ-1 is in agreement with a low-resolution structure of CaV1.2 channel complexes, which demonstrates that the α2δ-1 subunit forms a cap that embraces ~3/4 of the extracellular surface of the α1C subunit (Walsh et al., 2009a,b). Taking these data together, we propose that the α2δ-1 encapsulates a part of α1C that comprises VSDs I–III, whereas the exposed quarter of the α1C subunit is VSD IV (Fig. 7). Finally, we favor a physical interaction between α2δ-1 and VSDs I–III because of a recent cryo-electron microscopy (cryo-EM) structure of CaV1.1 channels showing that the α2δ-1 subunit interacts with the extracellular loops of repeats I–III (Wu et al., 2015), suggesting a common α1/α2δ topology in CaV1.1 (α1S) and CaV1.2 (α1C) channels. However, contribution of long-range allosteric interactions cannot be ruled out. Figure 7. The α2δ subunit covers ∼3/4 of the CaV1.2 pore-forming subunit, excluding VSD IV. Side and top views of the CaV1.2 (α1C/α2/β) channel volume (pink and light blue) modified from cryo-EM data from Walsh et al. (2009a) are shown. The atomic structures of NaVAb (PDB accession no. 4EKW; Payandeh et al., 2012), with subunits arranged clockwise, and the atomic structure of AID-associated β3 subunit (PDB accession no. 1VYT; Chen et al., 2004) were manually positioned in the cryo-EM volume. Because VSD IV is not perturbed by α2δ-1, we propose that the resolved α1C volume not covered by α2 is occupied by VSD IV, whereas VSDs I–III are encompassed by the α2δ-1 subunit, which alters their biophysical properties and, in the case of VSDs II and III, enhances their coupling to the channel pore. α2δ-1 association alters the intrinsic voltage-sensing properties of VSDs I–III The hyperpolarizing shifts in the F(V) values of VSDs I–III (Fig. 4) indicates that the active state of these voltage sensors is favored in the presence of α2δ-1 subunits. In VSD-gated channels, the activation of charge-bearing S4 segments is facilitated by the formation of salt bridges between positively charged S4 residues and negatively charged residues in adjacent VSD helices (Papazian et al., 1995; Wu et al., 2010; DeCaen et al., 2011; Tuluc et al., 2016). Association of the α2δ-1 subunit may facilitate the formation of such bonds by physically remodeling the spatial organization of the transmembrane helices of VSDs, altering their relative positions. Interestingly, we found that the VSDs’ sensitivity to changes in the membrane potential, i.e., the effective charge or slope of the F(V) curves (q), is almost equal among the four VSDs in the absence of the α2δ-1 subunit (q ≈ 1 e0), whereas q increases by approximately twofold for VSDs I–III in the presence of α2δ-1. The effective charge q of a voltage-sensing residue is given by the product zδ, where z is the valence number and δ is the electrical distance or fraction of membrane potential traversed by the residue. Because it is very unlikely that α2δ-1 association adds voltage-sensing charges, the increased apparent charge observed for VSD I–III suggests that their charged S4 helices move across a greater electrical distance, through either a larger spatial translation (e.g., moving at a steeper angle) or more concentrated electric field lines. Indeed, in VSD-endowed proteins, the local electric field can be tremendously enhanced by the existence of aqueous crevices separated by hydrophobic gaskets comprised of aromatic residue side chains (Asamoah et al., 2003; Starace and Bezanilla, 2004; Ahern and Horn, 2005; Chanda et al., 2005; Long et al., 2007; Tao et al., 2010; Lacroix and Bezanilla, 2011). The α2δ-1 subunit enhances the coupling of VSDs I–III to the pore Do the observed changes in VSD voltage-sensing properties account for the facilitation of CaV1.2 activation by α2δ-1 subunit? To answer this question, we used our allosteric model for CaV1.2 channel activation, which predicts the time- and voltage-dependent properties of each VSD and the pore (Pantazis et al., 2014). This model successfully accounted for the effects of α2δ-1 binding by both increasing the energetic contributions of VSDs I–III to pore opening and increasing the effective charges of VSDs I and II (Fig. 6 and Table 3). Specifically, in channels lacking the α2δ-1 subunit, VSDs I–III make a weak contribution to channel opening (W > −20 meV, equivalent to 0.8 kT or an allosteric factor of 2.2). The striking outcome of this study is that the energetic contribution of the activation of VSDs II and III to pore opening in channels lacking α2δ is greatly reduced. This is in contrast to channels containing α2δ-1, where VSDs II and III contribute two to three times as much energy toward channel opening (~−95 meV, ∼3.7 kT) or, in an alternative interpretation, their activation is obligatory for pore opening (Pantazis et al., 2014). The diminished VSD–pore conformational coupling in channels lacking α2δ is also supported by the good approximation of the G(V) by a Boltzmann distribution (Fig. 2 C), which implies a single voltage-dependent opening transition without significant input from VSDs. Our previous work on α2δ-containing CaV1.2 channels revealed a surprising disparity in the VSD voltage dependencies, greater than that observed in related pseudotetrameric NaV channels: VSD activations (Vhalf values) spanned 50 mV. The functional heterogeneity of the four VSDs was attributed to (a) the different amino acid composition of each VSD and (b) the structural asymmetry of the channel complex arising from its 1:1:1 α1/β/α2δ subunit stoichiometry. Interestingly, in this work, we found that increasing CaV1.2 structural symmetry (by excluding α2δ-1 subunits) in fact made the VSD voltage dependencies even more disparate, spanning ∼90 mV (Fig. 5). Our current model interpreted this finding as a result of direct modification of VSD voltage-sensing properties and reduced coupling of VSDs I–III to the pore by α2δ-1. Another possible explanation is that α2δ-1 acts as an allosteric center (in addition to the pore), increasing the coupling between voltage sensors. However, this possibility implies that α2δ-1 also undergoes conformational changes, for which there is yet no experimental evidence. Perhaps future studies could explore the possibility. Conclusions In summary, we have used VCF to optically track the molecular rearrangements of the individual VSDs of a human CaV1.2 channel in the presence or absence of α2δ-1. VCF is now a well-established method to assess voltage-dependent conformational changes, allowing us to track the movement of individual VSDs and to resolve slow conformational changes (as those observed in VSD I) that are extremely difficult to capture by gating current measurements. In this work, we have not systematically recorded gating currents, as they could not reveal the individual contributions of each VSD to CaV1.2 activation. Perhaps the most important advantage of VCF is that all recordings could be performed in conducting channels, whereby ionic currents and VSD movements were sampled simultaneously, without the use of pore-blockers. We found that the α2δ-1 auxiliary subunit significantly alters the voltage dependence of VSDs I–III, facilitating their activation, but not that of VSD IV. A 32-state allosteric model, consistent with the CaV1.2 molecular architecture, predicts the major kinetic and steady-state features of the experimental data, revealing that the association of α2δ-1 with α1C (in the presence of β3) specifically increased the coupling energy of VSDs I–III to the pore, as well as effective gating charge in segments I and II. Without the enhanced gating properties brought about by α2δ-1 association, CaV1.2 channels could not operate at physiological membrane potentials. Supplementary Material Supplementary Figures S1-S3 ACKNOWLEDGMENTS We are grateful to Ashraf Kitmitto for sharing the cryo-EM volumes of CaV α1C/α2δ (Walsh et al., 2009a). The human α1C-77 clone was a gift from Nicolaj Soldatov. We thank the members of the Olcese laboratory for insightful discussion and Jing Gao for the weekly preparation of the Xenopus oocytes. This work was supported by the National Institutes of Health (NIH)/National Heart, Lung, and Blood Institute grant P01HL078931 to J.N. Weiss and R. Olcese; NIH/National Institute of General Medical Sciences grant R01GM110276 to R. Olcese; American Heart Association Scientist Development grant 14SDG20300018 to A. Pantazis and postdoctoral fellowship 16POST27250284 to N. Savalli; and Chilean government grants FONDECYT 1161672 and ACT1104 to A. Neely. The Centro Interdisciplinario de Neurociencia de Valparaíso is a Millennium Institute supported by the Millennium Scientific Initiative of the Chilean Ministry of Economy. The authors declare no competing financial interests. Eduardo Rios served as editor. Abbreviations used: cryo-EM cryo-electron microscopy MES methanesulfonate VCF voltage clamp fluorometry VSD voltage-sensing domain ==== Refs Ahern, C.A., and R. Horn. 2005. Focused electric field across the voltage sensor of potassium channels. Neuron. 48 :25–29. 10.1016/j.neuron.2005.08.020 16202706 Asamoah, O.K., J.P. Wuskell, L.M. Loew, and F. Bezanilla. 2003. A fluorometric approach to local electric field measurements in a voltage-gated ion channel. Neuron. 37 :85–98. 10.1016/S0896-6273(02)01126-1 12526775 Bangalore, R., G. Mehrke, K. Gingrich, F. Hofmann, and R.S. Kass. 1996. Influence of L-type Ca channel alpha 2/delta-subunit on ionic and gating current in transiently transfected HEK 293 cells. Am. J. Physiol. 270 :H1521–H1528.8928856 Barclay, J., N. Balaguero, M. Mione, S.L. Ackerman, V.A. Letts, J. Brodbeck, C. Canti, A. Meir, K.M. Page, K. Kusumi, 2001. Ducky mouse phenotype of epilepsy and ataxia is associated with mutations in the Cacna2d2 gene and decreased calcium channel current in cerebellar Purkinje cells. J. Neurosci. 21 :6095–6104.11487633 Barish, M.E. 1983. A transient calcium-dependent chloride current in the immature Xenopus oocyte. J. Physiol. 342 :309–325. 10.1113/jphysiol.1983.sp014852 6313909 Ben-Johny, M., and D.T. Yue. 2014. Calmodulin regulation (calmodulation) of voltage-gated calcium channels. J. Gen. Physiol. 143 :679–692. 10.1085/jgp.201311153 24863929 Bezanilla, F. 2008. How membrane proteins sense voltage. Nat. Rev. Mol. Cell Biol. 9 :323–332. 10.1038/nrm2376 18354422 Bourdin, B., B. Shakeri, M.P. Tétreault, R. Sauvé, S. Lesage, and L. Parent. 2015. Functional characterization of CaVα2δ mutations associated with sudden cardiac death. J. Biol. Chem. 290 :2854–2869. 10.1074/jbc.M114.597930 25527503 Burashnikov, E., R. Pfeiffer, H. Barajas-Martinez, E. Delpón, D. Hu, M. Desai, M. Borggrefe, M. Häissaguerre, R. Kanter, G.D. Pollevick, 2010. Mutations in the cardiac L-type calcium channel associated with inherited J-wave syndromes and sudden cardiac death. Heart Rhythm. 7 :1872–1882. 10.1016/j.hrthm.2010.08.026 20817017 Calderón-Rivera, A., A. Andrade, O. Hernández-Hernández, R. González-Ramírez, A. Sandoval, M. Rivera, J.C. Gomora, and R. Felix. 2012. Identification of a disulfide bridge essential for structure and function of the voltage-gated Ca(2+) channel α(2)δ-1 auxiliary subunit. Cell Calcium. 51 :22–30. 10.1016/j.ceca.2011.10.002 22054663 Campiglio, M., and B.E. Flucher. 2015. The role of auxiliary subunits for the functional diversity of voltage-gated calcium channels. J. Cell. Physiol. 230 :2019–2031. 10.1002/jcp.24998 25820299 Cassidy, J.S., L. Ferron, I. Kadurin, W.S. Pratt, and A.C. Dolphin. 2014. Functional exofacially tagged N-type calcium channels elucidate the interaction with auxiliary α2δ-1 subunits. Proc. Natl. Acad. Sci. USA. 111 :8979–8984. 10.1073/pnas.1403731111 24889613 Catterall, W.A. 2010. Ion channel voltage sensors: Structure, function, and pathophysiology. Neuron. 67 :915–928. 10.1016/j.neuron.2010.08.021 20869590 Catterall, W.A. 2011. Voltage-gated calcium channels. Cold Spring Harb. Perspect. Biol. 3 :a003947. 10.1101/cshperspect.a003947 21746798 Cha, A., and F. Bezanilla. 1997. Characterizing voltage-dependent conformational changes in the Shaker K+ channel with fluorescence. Neuron. 19 :1127–1140. 10.1016/S0896-6273(00)80403-1 9390525 Cha, A., P.C. Ruben, A.L. George Jr., E. Fujimoto, and F. Bezanilla. 1999. Voltage sensors in domains III and IV, but not I and II, are immobilized by Na+ channel fast inactivation. Neuron. 22 :73–87. 10.1016/S0896-6273(00)80680-7 10027291 Chanda, B., and F. Bezanilla. 2008. A common pathway for charge transport through voltage-sensing domains. Neuron. 57 :345–351. 10.1016/j.neuron.2008.01.015 18255028 Chanda, B., O.K. Asamoah, R. Blunck, B. Roux, and F. Bezanilla. 2005. Gating charge displacement in voltage-gated ion channels involves limited transmembrane movement. Nature. 436 :852–856. 10.1038/nature03888 16094369 Chen, Y.H., M.H. Li, Y. Zhang, L.L. He, Y. Yamada, A. Fitzmaurice, Y. Shen, H. Zhang, L. Tong, and J. Yang. 2004. Structural basis of the alpha1-beta subunit interaction of voltage-gated Ca2+ channels. Nature. 429 :675–680. 10.1038/nature02641 15170217 Claydon, T.W., and D. Fedida. 2007. Voltage clamp fluorimetry studies of mammalian voltage-gated K(+) channel gating. Biochem. Soc. Trans. 35 :1080–1082. 10.1042/BST0351080 17956284 Colquhoun, D., and A.G. Hawkes. 1981. On the stochastic properties of single ion channels. Proc. R. Soc. Lond. B Biol. Sci. 211 :205–235. 10.1098/rspb.1981.0003 6111797 Dahan, D.S., M.I. Dibas, E.J. Petersson, V.C. Auyeung, B. Chanda, F. Bezanilla, D.A. Dougherty, and H.A. Lester. 2004. A fluorophore attached to nicotinic acetylcholine receptor βM2 detects productive binding of agonist to the αδ site. Proc. Natl. Acad. Sci. USA. 101 :10195–10200. 10.1073/pnas.0301885101 15218096 Davies, A., I. Kadurin, A. Alvarez-Laviada, L. Douglas, M. Nieto-Rostro, C.S. Bauer, W.S. Pratt, and A.C. Dolphin. 2010. The α2δ subunits of voltage-gated calcium channels form GPI-anchored proteins, a posttranslational modification essential for function. Proc. Natl. Acad. Sci. USA. 107 :1654–1659. 10.1073/pnas.0908735107 20080692 DeCaen, P.G., V. Yarov-Yarovoy, T. Scheuer, and W.A. Catterall. 2011. Gating charge interactions with the S1 segment during activation of a Na+ channel voltage sensor. Proc. Natl. Acad. Sci. USA. 108 :18825–18830. 10.1073/pnas.1116449108 22042870 Dolphin, A.C. 2012. Calcium channel auxiliary α2δ and β subunits: Trafficking and one step beyond. Nat. Rev. Neurosci. 13 :542–555. 10.1038/nrn3317 22805911 Dolphin, A.C. 2013. The α2δ subunits of voltage-gated calcium channels. Biochim. Biophys. Acta. 1828 :1541–1549. 10.1016/j.bbamem.2012.11.019 23196350 Eroglu, C., N.J. Allen, M.W. Susman, N.A. O’Rourke, C.Y. Park, E. Ozkan, C. Chakraborty, S.B. Mulinyawe, D.S. Annis, A.D. Huberman, 2009. Gabapentin receptor alpha2delta-1 is a neuronal thrombospondin receptor responsible for excitatory CNS synaptogenesis. Cell. 139 :380–392. 10.1016/j.cell.2009.09.025 19818485 Fang, K., and H.M. Colecraft. 2011. Mechanism of auxiliary β-subunit-mediated membrane targeting of L-type (Ca(V)1.2) channels. J. Physiol. 589 :4437–4455. 10.1113/jphysiol.2011.214247 21746784 Felix, R., C.A. Gurnett, M. De Waard, and K.P. Campbell. 1997. Dissection of functional domains of the voltage-dependent Ca2+ channel alpha2delta subunit. J. Neurosci. 17 :6884–6891.9278523 Field, M.J., P.J. Cox, E. Stott, H. Melrose, J. Offord, T.Z. Su, S. Bramwell, L. Corradini, S. England, J. Winks, 2006. Identification of the α2-δ-1 subunit of voltage-dependent calcium channels as a molecular target for pain mediating the analgesic actions of pregabalin. Proc. Natl. Acad. Sci. USA. 103 :17537–17542. 10.1073/pnas.0409066103 17088553 Findeisen, F., and D.L. Minor Jr. 2010. Progress in the structural understanding of voltage-gated calcium channel (CaV) function and modulation. Channels (Austin). 4 :459–474. 10.4161/chan.4.6.12867 21139419 Fuller-Bicer, G.A., G. Varadi, S.E. Koch, M. Ishii, I. Bodi, N. Kadeer, J.N. Muth, G. Mikala, N.N. Petrashevskaya, M.A. Jordan, 2009. Targeted disruption of the voltage-dependent calcium channel α2/δ-1-subunit. Am. J. Physiol. Heart Circ. Physiol. 297 :H117–H124. 10.1152/ajpheart.00122.2009 19429829 Gandhi, C.S., and R. Olcese. 2008. The voltage-clamp fluorometry technique. Methods Mol. Biol. 491 :213–231. 10.1007/978-1-59745-526-8_17 18998096 Ghezzi, C., H. Murer, and I.C. Forster. 2009. Substrate interactions of the electroneutral Na+-coupled inorganic phosphate cotransporter (NaPi-IIc). J. Physiol. 587 :4293–4307. 10.1113/jphysiol.2009.175596 19596895 Gurnett, C.A., R. Felix, and K.P. Campbell. 1997. Extracellular interaction of the voltage-dependent Ca2+ channel alpha2delta and alpha1 subunits. J. Biol. Chem. 272 :18508–18512. 10.1074/jbc.272.29.18508 9218497 Hendrich, J., A.T. Van Minh, F. Heblich, M. Nieto-Rostro, K. Watschinger, J. Striessnig, J. Wratten, A. Davies, and A.C. Dolphin. 2008. Pharmacological disruption of calcium channel trafficking by the α2δ ligand gabapentin. Proc. Natl. Acad. Sci. USA. 105 :3628–3633. 10.1073/pnas.0708930105 18299583 Hines, K.E., T.R. Middendorf, and R.W. Aldrich. 2014. Determination of parameter identifiability in nonlinear biophysical models: A Bayesian approach. J. Gen. Physiol. 143 :401–416. 10.1085/jgp.201311116 24516188 Kohout, S.C., M.H. Ulbrich, S.C. Bell, and E.Y. Isacoff. 2008. Subunit organization and functional transitions in Ci-VSP. Nat. Struct. Mol. Biol. 15 :106–108. 10.1038/nsmb1320 18084307 Kurshan, P.T., A. Oztan, and T.L. Schwarz. 2009. Presynaptic alpha2delta-3 is required for synaptic morphogenesis independent of its Ca2+-channel functions. Nat. Neurosci. 12 :1415–1423. 10.1038/nn.2417 19820706 Lacroix, J.J., and F. Bezanilla. 2011. Control of a final gating charge transition by a hydrophobic residue in the S2 segment of a K+ channel voltage sensor. Proc. Natl. Acad. Sci. USA. 108 :6444–6449. 10.1073/pnas.1103397108 21464282 Larsson, H.P., A.V. Tzingounis, H.P. Koch, and M.P. Kavanaugh. 2004. Fluorometric measurements of conformational changes in glutamate transporters. Proc. Natl. Acad. Sci. USA. 101 :3951–3956. 10.1073/pnas.0306737101 15001707 Li, Y. 2012. MOMCMC: An efficient Monte Carlo method for multi-objective sampling over real parameter space. Comput. Math. Appl. 64 :3542–3556. 10.1016/j.camwa.2012.09.003 Li, C.Y., X.L. Zhang, E.A. Matthews, K.W. Li, A. Kurwa, A. Boroujerdi, J. Gross, M.S. Gold, A.H. Dickenson, G. Feng, and Z.D. Luo. 2006. Calcium channel alpha2delta1 subunit mediates spinal hyperexcitability in pain modulation. Pain. 125 :20–34. 10.1016/j.pain.2006.04.022 16764990 Long, S.B., X. Tao, E.B. Campbell, and R. MacKinnon. 2007. Atomic structure of a voltage-dependent K+ channel in a lipid membrane-like environment. Nature. 450 :376–382. 10.1038/nature06265 18004376 Lörinczi, É., Y. Bhargava, S.F. Marino, A. Taly, K. Kaczmarek-Hájek, A. Barrantes-Freer, S. Dutertre, T. Grutter, J. Rettinger, and A. Nicke. 2012. Involvement of the cysteine-rich head domain in activation and desensitization of the P2X1 receptor. Proc. Natl. Acad. Sci. USA. 109 :11396–11401. 10.1073/pnas.1118759109 22745172 Mannuzzu, L.M., M.M. Moronne, and E.Y. Isacoff. 1996. Direct physical measure of conformational rearrangement underlying potassium channel gating. Science. 271 :213–216. 10.1126/science.271.5246.213 8539623 Meinild, A.K., B.A. Hirayama, E.M. Wright, and D.D. Loo. 2002. Fluorescence studies of ligand-induced conformational changes of the Na(+)/glucose cotransporter. Biochemistry. 41 :1250–1258. 10.1021/bi011661r 11802724 Nakai, J., B.A. Adams, K. Imoto, and K.G. Beam. 1994. Critical roles of the S3 segment and S3-S4 linker of repeat I in activation of L-type calcium channels. Proc. Natl. Acad. Sci. USA. 91 :1014–1018. 10.1073/pnas.91.3.1014 8302825 Neely, A., and P. Hidalgo. 2014. Structure-function of proteins interacting with the α1 pore-forming subunit of high-voltage-activated calcium channels. Front. Physiol. 5 :209. 10.3389/fphys.2014.00209 24917826 Neely, A., R. Olcese, X. Wei, L. Birnbaumer, and E. Stefani. 1994. Ca(2+)-dependent inactivation of a cloned cardiac Ca2+ channel alpha 1 subunit (alpha 1C) expressed in Xenopus oocytes. Biophys. J. 66 :1895–1903. 10.1016/S0006-3495(94)80983-X 8075326 Neely, G.G., A. Hess, M. Costigan, A.C. Keene, S. Goulas, M. Langeslag, R.S. Griffin, I. Belfer, F. Dai, S.B. Smith, 2010. A genome-wide Drosophila screen for heat nociception identifies α2δ3 as an evolutionarily conserved pain gene. Cell. 143 :628–638. 10.1016/j.cell.2010.09.047 21074052 Osteen, J.D., C. Gonzalez, K.J. Sampson, V. Iyer, S. Rebolledo, H.P. Larsson, and R.S. Kass. 2010. KCNE1 alters the voltage sensor movements necessary to open the KCNQ1 channel gate. Proc. Natl. Acad. Sci. USA. 107 :22710–22715. 10.1073/pnas.1016300108 21149716 Palovcak, E., L. Delemotte, M.L. Klein, and V. Carnevale. 2014. Evolutionary imprint of activation: The design principles of VSDs. J. Gen. Physiol. 143 :145–156. 10.1085/jgp.201311103 24470486 Pantazis, A., and R. Olcese. 2013. Cut-open oocyte voltage clamp technique. In Encyclopedia of Biophysics. G.C.K. Roberts, editor. Springer, Berlin, Heidelberg. 406–413. 10.1007/978-3-642-16712-6_371 Pantazis, A., N. Savalli, D. Sigg, A. Neely, and R. Olcese. 2014. Functional heterogeneity of the four voltage sensors of a human L-type calcium channel. Proc. Natl. Acad. Sci. USA. 111 :18381–18386. 10.1073/pnas.1411127112 25489110 Papazian, D.M., X.M. Shao, S.A. Seoh, A.F. Mock, Y. Huang, and D.H. Wainstock. 1995. Electrostatic interactions of S4 voltage sensor in Shaker K+ channel. Neuron. 14 :1293–1301. 10.1016/0896-6273(95)90276-7 7605638 Payandeh, J., T.M. Gamal El-Din, T. Scheuer, N. Zheng, and W.A. Catterall. 2012. Crystal structure of a voltage-gated sodium channel in two potentially inactivated states. Nature. 486 :135–139.22678296 Peterson, B.Z., C.D. DeMaria, J.P. Adelman, and D.T. Yue. 1999. Calmodulin is the Ca2+ sensor for Ca2+-dependent inactivation of L-type calcium channels. Neuron. 22 :549–558. 10.1016/S0896-6273(00)80709-6 10197534 Platano, D., N. Qin, F. Noceti, L. Birnbaumer, E. Stefani, and R. Olcese. 2000. Expression of the alpha(2)delta subunit interferes with prepulse facilitation in cardiac L-type calcium channels. Biophys. J. 78 :2959–2972. 10.1016/S0006-3495(00)76835-4 10827975 Qin, N., R. Olcese, M. Bransby, T. Lin, and L. Birnbaumer. 1999. Ca2+-induced inhibition of the cardiac Ca2+ channel depends on calmodulin. Proc. Natl. Acad. Sci. USA. 96 :2435–2438. 10.1073/pnas.96.5.2435 10051660 Ruscic, K.J., F. Miceli, C.A. Villalba-Galea, H. Dai, Y. Mishina, F. Bezanilla, and S.A. Goldstein. 2013. IKs channels open slowly because KCNE1 accessory subunits slow the movement of S4 voltage sensors in KCNQ1 pore-forming subunits. Proc. Natl. Acad. Sci. USA. 110 :E559–E566. 10.1073/pnas.1222616110 23359697 Savalli, N., A. Kondratiev, L. Toro, and R. Olcese. 2006. Voltage-dependent conformational changes in human Ca(2+)- and voltage-activated K(+) channel, revealed by voltage-clamp fluorometry. Proc. Natl. Acad. Sci. USA. 103 :12619–12624. 10.1073/pnas.0601176103 16895996 Savalli, N., A. Kondratiev, S.B. de Quintana, L. Toro, and R. Olcese. 2007. Modes of operation of the BKCa channel beta2 subunit. J. Gen. Physiol. 130 :117–131. 10.1085/jgp.200709803 17591990 Sigg, D. 2014. Modeling ion channels: Past, present, and future. J. Gen. Physiol. 144 :7–26. 10.1085/jgp.201311130 24935742 Smith, P.L., and G. Yellen. 2002. Fast and slow voltage sensor movements in HERG potassium channels. J. Gen. Physiol. 119 :275–293. 10.1085/jgp.20028534 11865022 Soldatov, N.M. 1992. Molecular diversity of L-type Ca2+ channel transcripts in human fibroblasts. Proc. Natl. Acad. Sci. USA. 89 :4628–4632. 10.1073/pnas.89.10.4628 1316612 Starace, D.M., and F. Bezanilla. 2004. A proton pore in a potassium channel voltage sensor reveals a focused electric field. Nature. 427 :548–553. 10.1038/nature02270 14765197 Stefani, E., and F. Bezanilla. 1998. Cut-open oocyte voltage-clamp technique. Methods Enzymol. 293 :300–318. 10.1016/S0076-6879(98)93020-8 9711615 Swartz, K.J. 2008. Sensing voltage across lipid membranes. Nature. 456 :891–897. 10.1038/nature07620 19092925 Talwar, S., and J.W. Lynch. 2015. Investigating ion channel conformational changes using voltage clamp fluorometry. Neuropharmacology. 98 :3–12. 10.1016/j.neuropharm.2015.03.018 25839896 Tao, X., A. Lee, W. Limapichat, D.A. Dougherty, and R. MacKinnon. 2010. A gating charge transfer center in voltage sensors. Science. 328 :67–73. 10.1126/science.1185954 20360102 Templin, C., J.R. Ghadri, J.S. Rougier, A. Baumer, V. Kaplan, M. Albesa, H. Sticht, A. Rauch, C. Puleo, D. Hu, 2011. Identification of a novel loss-of-function calcium channel gene mutation in short QT syndrome (SQTS6). Eur. Heart J. 32 :1077–1088. 10.1093/eurheartj/ehr076 21383000 Tombola, F., M.M. Pathak, and E.Y. Isacoff. 2006. How does voltage open an ion channel? Annu. Rev. Cell Dev. Biol. 22 :23–52. 10.1146/annurev.cellbio.21.020404.145837 16704338 Tombola, F., M.H. Ulbrich, S.C. Kohout, and E.Y. Isacoff. 2010. The opening of the two pores of the Hv1 voltage-gated proton channel is tuned by cooperativity. Nat. Struct. Mol. Biol. 17 :44–50. 10.1038/nsmb.1738 20023640 Tuluc, P., G. Kern, G.J. Obermair, and B.E. Flucher. 2007. Computer modeling of siRNA knockdown effects indicates an essential role of the Ca2+ channel α2δ-1 subunit in cardiac excitation-contraction coupling. Proc. Natl. Acad. Sci. USA. 104 :11091–11096. 10.1073/pnas.0700577104 17563358 Tuluc, P., N. Molenda, B. Schlick, G.J. Obermair, B.E. Flucher, and K. Jurkat-Rott. 2009. A CaV1.1 Ca2+ channel splice variant with high conductance and voltage-sensitivity alters EC coupling in developing skeletal muscle. Biophys. J. 96 :35–44. 10.1016/j.bpj.2008.09.027 19134469 Tuluc, P., V. Yarov-Yarovoy, B. Benedetti, and B.E. Flucher. 2016. Molecular interactions in the voltage sensor controlling gating properties of CaV calcium channels. Structure. 24 :261–271. 10.1016/j.str.2015.11.011 26749449 Uchitel, O.D., M.N. Di Guilmi, F.J. Urbano, and C. Gonzalez-Inchauspe. 2010. Acute modulation of calcium currents and synaptic transmission by gabapentinoids. Channels (Austin). 4 :490–496. 10.4161/chan.4.6.12864 21150315 Vergult, S., A. Dheedene, A. Meurs, F. Faes, B. Isidor, S. Janssens, A. Gautier, C. Le Caignec, and B. Menten. 2015. Genomic aberrations of the CACNA2D1 gene in three patients with epilepsy and intellectual disability. Eur. J. Hum. Genet. 23 :628–632. 10.1038/ejhg.2014.141 25074461 Walsh, C.P., A. Davies, A.J. Butcher, A.C. Dolphin, and A. Kitmitto. 2009 a. Three-dimensional structure of CaV3.1: Comparison with the cardiac L-type voltage-gated calcium channel monomer architecture. J. Biol. Chem. 284 :22310–22321. 10.1074/jbc.M109.017152 19520861 Walsh, C.P., A. Davies, M. Nieto-Rostro, A.C. Dolphin, and A. Kitmitto. 2009 b. Labelling of the 3D structure of the cardiac L-type voltage-gated calcium channel. Channels (Austin). 3 :387–392. 10.4161/chan.3.6.10225 19875947 Wu, D., K. Delaloye, M.A. Zaydman, A. Nekouzadeh, Y. Rudy, and J. Cui. 2010. State-dependent electrostatic interactions of S4 arginines with E1 in S2 during Kv7.1 activation. J. Gen. Physiol. 135 :595–606. 10.1085/jgp.201010408 20479111 Wu, J., Z. Yan, Z. Li, C. Yan, S. Lu, M. Dong, and N. Yan. 2015. Structure of the voltage-gated calcium channel Cav1.1 complex. Science. 350 :aad2395. 10.1126/science.aad2395 26680202 Wycisk, K.A., C. Zeitz, S. Feil, M. Wittmer, U. Forster, J. Neidhardt, B. Wissinger, E. Zrenner, R. Wilke, S. Kohl, and W. Berger. 2006. Mutation in the auxiliary calcium-channel subunit CACNA2D4 causes autosomal recessive cone dystrophy. Am. J. Hum. Genet. 79 :973–977. 10.1086/508944 17033974 Zamponi, G.W., J. Striessnig, A. Koschak, and A.C. Dolphin. 2015. The physiology, pathology, and pharmacology of voltage-gated calcium channels and their future therapeutic potential. Pharmacol. Rev. 67 :821–870. 10.1124/pr.114.009654 26362469 Zhu, W., Z. Varga, and J.R. Silva. 2016. Molecular motions that shape the cardiac action potential: Insights from voltage clamp fluorometry. Prog. Biophys. Mol. Biol. 120 :3–17. 10.1016/j.pbiomolbio.2015.12.003 26724572

PMC004xxxxxx/PMC4969798.txt

==== Front J Gen Physiol J Gen Physiol jgp jgp The Journal of General Physiology 0022-1295 1540-7748 The Rockefeller University Press 27481712 201611619 10.1085/jgp.201611619 Reviews Review Insights into the structure and function of HV1 from a meta-analysis of mutation studies Annotated compendium of HV1 mutations http://orcid.org/0000-0002-4263-180X DeCoursey Thomas E. 1 Morgan Deri 1 Musset Boris 2 Cherny Vladimir V. 1 1 Department of Molecular Biophysics and Physiology, Rush University, Chicago, IL 60612 2 Institut für Physiologie, PMU Klinikum Nürnberg, 90419 Nürnberg, Germany Correspondence to Thomas E. DeCoursey: tdecours@rush.edu 8 2016 148 2 97118 04 5 2016 30 6 2016 © 2016 DeCoursey et al. 2016 https://creativecommons.org/licenses/by-nc-sa/3.0/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 3.0 Unported license, as described at http://creativecommons.org/licenses/by-nc-sa/3.0/). The voltage-gated proton channel (HV1) is a widely distributed, proton-specific ion channel with unique properties. Since 2006, when genes for HV1 were identified, a vast array of mutations have been generated and characterized. Accessing this potentially useful resource is hindered, however, by the sheer number of mutations and interspecies differences in amino acid numbering. This review organizes all existing information in a logical manner to allow swift identification of studies that have characterized any particular mutation. Although much can be gained from this meta-analysis, important questions about the inner workings of HV1 await future revelation. National Institutes of Health 10.13039/100000002 GM102336 National Science Foundation 10.13039/100000001 MCB-1242985 ==== Body pmcIntroduction Voltage-gated proton channels are found in highly diverse species, from unicellular marine creatures such as dinoflagellates, diatoms, and coccolithophores (Smith et al., 2011; Taylor et al., 2011, 2012) to insects (Chaves et al., 2016), snails (Thomas and Meech, 1982; Byerly et al., 1984; Doroshenko et al., 1986), and human beings, where they are found in a variety of cells and perform many disparate functions (DeCoursey, 2013). Their unique properties (perfect H+ selectivity, ΔpH-dependent gating, extreme temperature dependence, and the ability to shift into a strikingly enhanced gating mode) are paralleled by a unique structure, the reconciliation of which is a goal of this review. Most voltage-gated ion channels are tetramers or quasi-tetramers of monomers comprising a voltage-sensing domain (VSD) S1­–S4 (transmembrane [TM] segments 1–4) and a pore domain S5–S6, four of which combine to produce a single central conduction pathway. In contrast, HV1 consists of S1–S4 alone, a VSD without an explicit pore domain (Ramsey et al., 2006; Sasaki et al., 2006). In mammals and many other species (Koch et al., 2008; Lee et al., 2008; Tombola et al., 2008; Smith and DeCoursey, 2013), HV1 forms a dimer in cell membranes. However, each monomer, or protomer, has its own pore and other necessary parts and can function as a monomer (Koch et al., 2008; Tombola et al., 2008). The properties of monomeric constructs are similar in most respects to those of the dimer, but monomeric constructs open faster (Koch et al., 2008; Tombola et al., 2008; Musset et al., 2010b,c; Fujiwara et al., 2012). Several lines of evidence indicate that the two protomers comprising the HV1 dimer do not function independently, but gate cooperatively in the sense that each must undergo a voltage-dependent conformational change before either can conduct current (Gonzalez et al., 2010; Musset et al., 2010b; Tombola et al., 2010; Smith and DeCoursey, 2013). Fig. 1 illustrates schematically the entire 273–amino acid sequence of hHV1. The signature sequence that has been used successfully to identify HV1 in new species is RxWRxxR in the S4 helix (Smith et al., 2011; Rodriguez et al., 2015; Chaves et al., 2016). This sequence also identifies c15orf27 proteins, of unknown function, but these all lack Asp112 in S1, which is required for proton selectivity (Table 1). Another reported conserved motif in S2, [F,Y,W]xx[E,D]xxx[R,K], identifies some HV1 channels but is not specific to HV1, and instead identifies VSDs in general (Kang and Baker, 2016). This motif is not present in all HV1; for example, it is not present in several unicellular marine species (Taylor et al., 2011), including kHV1, which was identified by using the S4 motif (Smith et al., 2011). Figure 1. The amino acid sequence and schematic topology of the human voltage-gated proton channel, hHV1. Within the TM domain, acidic residues are red, basic residues are blue, aromatic residues are orange, and polar residues are gray. Specific amino acids of note, beginning at the N terminus: deletion of 1–20 (green) produces a “short” isoform common in malignant B cells (Hondares et al., 2014); Thr29 is a PKC phosphorylation site responsible for enhanced gating (Musset et al., 2010a); M91T is the first identified hHV1 mutation (Iovannisci et al., 2010); Asp112 is crucial to H+ selectivity (Musset et al., 2011); His140 and His193 coordinate Zn2+ binding (Ramsey et al., 2006); the three Arg in S4 are thought to open the conductance pathway in response to voltage (Ramsey et al., 2006; Sasaki et al., 2006; Gonzalez et al., 2013); and the C terminus has an extensive coiled-coil region (black) that holds the dimer together (Koch et al., 2008; Lee et al., 2008; Tombola et al., 2008; Fujiwara et al., 2014). The image was drawn with TOPO2 (Johns, 2016). Table 1. Numerical key to HV1 in species verified by heterologous expression Species are indicated by one- or two-letter abbreviations: h, human; m, mouse; Ci, Ciona intestinalis; Np, Nicoletia phytophila; Sp, Strongylocentrotus purpuratus; k, Karlodinium veneficum; Pt, Phaeodactylum tricornutum; Cb, Coccolithus braarudii; Eh, Emiliania huxleyi. HG, hydrophobic gasket residues (gray); Sel, selectivity filter (yellow). Zn2+, one of the two Zn2+-binding His (in some species). Green-shaded residues sense voltage. Okamura et al. (2015) propose a slightly different hydrophobic plug based on the mHV1 crystal structure: F146, M147, L150, and F178. Several alignments of the S4 helix have been produced, which result in shifts in the register of the basic residues. Kv1.2 is human, NCBI Reference Sequence accession no. NP_004965.1; Shaker is UniProt/Swiss-Prot accession no. P08510.3. Many reviews describe in detail the biological functions proposed for HV1 (Eder and DeCoursey, 2001; DeCoursey, 2010, 2012, 2013; Capasso et al., 2011; Demaurex, 2012; Fischer, 2012; Lishko et al., 2012; Taylor et al., 2012; Smith and DeCoursey, 2013; DeCoursey and Hosler, 2014; Seredenina et al., 2015). Here we will summarize a few aspects of functions that are relevant to the analysis that follows. The main function of HV1 in most cells is acid extrusion, although the specific consequences in each cell vary drastically. For example, HV1-mediated acid extrusion triggers capacitation in human sperm (Lishko et al., 2010), enables histamine release by basophils (Musset et al., 2008b), and exacerbates several cancers (Wang et al., 2012, 2013; Hondares et al., 2014). Acid extrusion requires extreme proton selectivity because the concentration of H+ in biological solutions is a million-fold lower than that of other major ions. HV1 are brilliantly designed and extremely efficient acid extrusion devices, changing pHi at least an order of magnitude faster than other H+ exporters (DeCoursey and Cherny, 1994), mainly because of their unique ΔpH-dependent gating mechanism (Cherny et al., 1995), which is discussed in detail below in the section Table entries defined. A second type of function of HV1 in many cells reflects the electrical consequences of the charge movement that occurs during H+ extrusion. For example, in phagocytes and certain other cells, H+ efflux serves to compensate electrically for the electron extrusion that occurs as a direct consequence of the electrogenic activity of nicotinamide adenine dinucleotide phosphate (NADPH; reduced form) oxidase and related NOX (cytochrome subunit of NADPH oxidase) isoforms (Henderson et al., 1987, 1988; DeCoursey et al., 2003). In dinoflagellates, HV1 are thought to mediate the action potential that triggers the flash in bioluminescent species (Smith et al., 2011; Taylor et al., 2012). The respiratory burst of phagocytes, which can be elicited by pathogenic microbes, chemotactic peptides, or the phorbol ester PMA, is the manifestation of NADPH oxidase activation. During the respiratory burst, HV1 properties undergo a drastic transformation (Bánfi et al., 1999; DeCoursey et al., 2000; Murphy and DeCoursey, 2006; DeCoursey, 2016), resulting in a much higher level of activity in what has been called the enhanced gating mode (DeCoursey, 2003b). The bulk of evidence indicates that enhanced gating results from phosphorylation of HV1 (Morgan et al., 2007, Musset et al., 2010a; DeCoursey, 2016); mutations to putative phosphorylation sites are listed in Table 2. Table 2. Changes in HV1 properties in N-terminal (1–100) mutants versus WT channels Mutant Species Expr. system I? τ act τ tail ΔV thr ΔpH slope Selectivity Other Reference hHV1S LK35.2 wc Yes 2.5 0.75 5.4 43.2 H+ More profound enhanced gating Hondares et al., 2014 T9A hHV1S LK35.2 pp Yes 0.28 0.9 −18.2 Enhanced gating lost Hondares et al., 2014 S77A hHV1S LK35.2 pp Yes 1.75 1.1 −3.6 nc Hondares et al., 2014 T9A/S77A hHV1S LK35.2 pp Yes 0.63 0.8 −5.8 Enhanced gating lost Hondares et al., 2014 T29A hHV1 LK35.2 pp Yes 0.23 1.8 −27.6 Enhanced gating lost Musset et al., 2010a T29D hHV1 LK35.2 pp Yes 0.41 1.0 −12.5 Enhanced gating lost Musset et al., 2010a M91T hHV1 COS wc Yes 20 47 First identified naturally occurring hHV1 mutation Iovannisci et al., 2010 S97A hHV1 LK35.2 pp Yes 0.96 1.2 3.9 nc Musset et al., 2010a S97D hHV1 LK35.2 pp Yes 0.34 0.9 −3.0 nc Musset et al., 2010a T29A/S97A hHV1 LK35.2 pp Yes 0.25 1.7 −19.4 Enhanced gating lost Musset et al., 2010a S98A hHV1 HEK wc Yes −7 Ramsey et al., 2010 H99A hHV1 HEK wc Yes 13 Ramsey et al., 2010 R100A hHV1 HEK wc Yes −7 Ramsey et al., 2010 ΔN (1–96 deleted) hHV1 HEK wc Yes 15 40 Ramsey et al., 2010 ΔN/ΔC mHV1 (1–77 deleted, V216stop) HEK wc Yes 0.20 nc nc Loss of dimer formation Koch et al., 2008 ΔN/ΔC mHV1 (1–77 deleted, V216stop) HEK i-o Yes Weaker Zn2+ effects Musset et al., 2010b That numerical entries are shown does not imply that any given change was significant. The entries for hHV1S are in a short isoform and are compared with full-length hHV1. HEK, HEK-293, HEK-293T, tsA, or HM1; COS, COS-7; pp, perforated patch; wc, whole cell; i-o, inside-out patch configuration. Blank entries indicate that the parameter was not examined. nc, measured, but no change. Parameters are given relative to WT in each study. For I?, yes means currents are detectable. Time constants are ratios of mutant/WT. The ΔVthreshold value is the change in absolute position of the gH–V relationship versus WT. The ΔpH slope is the slope in millivolts of the relationship between Vthreshold (or other parameters reflecting the absolute position of the gH–V relationship) and Vrev or EH (which are not identical; see section Table entries defined). When C-terminal truncations (ΔC) are indicated as XNNNstop, this means STOP replaces X at position NNN; hence, position NNN and all subsequent residues are truncated, and the last position remaining is NNN-1. The mouse N-terminal deletions (ΔN) were done by replacing P78M to initiate translation at that position. Table 1 presents the position numbers of several key amino acids in HV1 from the nine species in which the channel has been verified by electrophysiological studies in heterologous expression systems, and the corresponding positions in two closely related molecules (c15orf27 and CiVSP) as well as in two exhaustively studied K+ channels. HV1 contains two highly conserved Asp that other voltage-gated ion channels lack, Asp112 and Asp185, as well as an anomalously located Trp207. However, HV1 lack the equivalent of Glu283 of Shaker and Na+ channels, having Ser143 instead. Fig. 2 shows where several key amino acids are located in the closed crystal structure of a mouse HV1 (mHV1) chimera (Fig. 2 A; Takeshita et al., 2014) and an open state model of hHV1 (Fig. 2 B; Kulleperuma et al., 2013). Numbering for both corresponds to hHV1, although the closed structure is of mHV1. Figure 2. Location of key amino acids. Location of some key amino acids in the crystal structure of the mHV1 chimera (Takeshita et al., 2014), labeled with hHV1 numbers (A), and in an open state model of hHV1, R2D (B; Kulleperuma et al., 2013). The channels are viewed from the side with the extracellular end at the top. An EPR study of hHV1 generally agreed with the structure of the mHV1 chimera, except in the EPR study the S2 helix (with F150 and E153) was one turn of the helix lower and S3 (with D174 and D185) was one turn higher relative to S1 and S4 (DeCoursey, 2015a; Li et al., 2015). E153 is the first amino acid replaced by the spliced-in CiVSP segment, and is actually D in the crystal. The images were produced with the PyMOL Molecular Graphics System (version 1.8; Schrödinger, LLC). We hope that this assembly of information will in itself allow some general conclusions about structure–function relationships. Most existing data appear reasonably consistent, but in some instances, there appear to be species differences. It is unclear whether these are real or simply examples of laboratory to laboratory variation; such observations are inevitably somewhat anecdotal. We also identify examples of qualitatively different outcomes for the same mutation studied in different expression systems. In addition to listing the outcomes of mutations, the motivation for several strategies for generating mutants is discussed. The mutation studies have resulted in rapid progress in understanding how HV1 works, but many important questions remain. One word of caution must be stated: mutations are designed to test the effect of changing one or more amino acids, usually with the assumption that the rest of the protein will assemble and function exactly as the WT does. This assumption can be tested rigorously only by determining the structure of every mutant. Although this procedure became almost routine for the bacterial reaction center (Xu et al., 2004), it is not remotely possible for HV1. Nevertheless, many mutants function with little overt change beyond what might be imagined. On the other hand, mutations that alter charge may have powerful effects on structure and hence function at least locally if not globally. Data organization and exclusions The information in this review is organized according to the amino acid numbering of the human voltage-gated proton channel, hHV1. To make the tables more manageable, we present the mutants in numerical order (with a few exceptions whose logic may or may not become apparent) and include separate tables for the N terminus, for each TM helix (S1–S4), and for the C terminus. The boundaries of the TM helices are defined according to the electron paramagnetic resonance (EPR) study of Li et al. (2015). Double or triple mutants are listed according to the first (i.e., nearest the N terminus) position mutated (again with a few exceptions whose rationale may become apparent). The intention is to make these tables exhaustive as far as is practical, but our bias is toward electrophysiological descriptions of individual point mutants. Thus, we do not list all 109 mutants studied by Cys scanning and assessed for accessibility by PEGylation protection (Sakata et al., 2010; Kurokawa and Okamura, 2014), for example. Nor do we list all 149 positions at which Cys was introduced for EPR measurements (Li et al., 2015). These blanket mutations are interpretable mainly within the context of the entire study. We exclude most mutations examining the link between S4 and the C terminus, which involve a large variety of deletions and insertions (Fujiwara et al., 2012, 2014). Mutations to the coiled-coil region of the C terminus resulting in trimeric and tetrameric channels (Fujiwara et al., 2013a) are not discussed here. Only a fraction of a large series of Trp scanning mutants in both monomeric and dimeric constructs is listed (Okuda et al., 2016). A series of mutants and tandem constructs in which one or both of the His that bind Zn2+ (His140 and His193) were replaced (Musset et al., 2010b) is not included in the tables. A series of Cys cross-linking mutations aimed at identifying the dimer interface (Lee et al., 2008) is also omitted. Finally, we do not include domain-swap mutants, such as those of Alabi et al. (2007), in part because chimerae do not logically fit into the format of the tables. Table entries defined The first column lists the mutations as per the usual convention (single-letter amino acid abbreviations: WT, position counting from the N terminus, and replacement). When a study used a nonhuman species, the hHV1 equivalent is given in italics in the first column, and the actual mutation is listed in the second column. The third column gives the expression system and the voltage-clamp method used. To our knowledge, no studies exist in which different properties were observed in HEK versus COS cells (Musset et al., 2008a). However, mammalian versus amphibian studies sometimes differ. For example, D112S from three different species all expressed well and exhibited anion permeation in mammalian cells (Musset et al., 2011; Smith et al., 2011; Chaves et al., 2016), whereas currents were not observed in Xenopus laevis oocytes (Berger and Isacoff, 2011). Some proteins function better in certain expression systems: CiHV1 works well in Xenopus oocytes, whereas mHV1 does not and prefers mammalian (HEK) cells (Okuda et al., 2016). Among mammalian cells, hHV1 expressed in the B cell–related LK35.2 cell line exhibits an enhanced gating response to stimulation with PMA (Musset et al., 2010a; Hondares et al., 2014), whereas hHV1 expressed in HEK or COS cells did not respond (Musset et al., 2008a). The fourth column (I?) simply reports whether interpretable currents were observed. A positive answer means the mutant protein is produced, reaches the plasma membrane, and functions. A negative result may have various undetermined explanations (protein misfolding, failure to traffic to the plasma membrane, disruption of gating or permeation) but is nevertheless potentially important because amino acids that play crucial roles in function may be difficult to replace without disrupting molecular function. For example, Asp112 and Arg208 form a crucial nexus that in general cannot be meddled with without altering or eliminating function. However, different laboratories may have different criteria for deciding whether small currents are “real,” small enough to be negligible, or nonexistent. This evaluation is complicated by the fact that all common mammalian expression systems (HEK, Chinese hamster ovary [CHO], and COS cells) frequently display (typically small) native voltage-gated proton currents (Cherny et al., 1997; Musset et al., 2011). To address this concern, we often introduce mutations into hHV1 in a Zn2+-insensitive background, meaning the two His primarily responsible for Zn2+ inhibition are mutated (H140A/H193A; Ramsey et al., 2006; Musset et al., 2010b). Then, if we see small currents that might be due either to native currents or to expression of a poorly conducting mutant, we add 10 µM Zn2+, which profoundly inhibits WT hHV1, but will have negligible effects on a Zn2+-insensitive mutant. This approach cannot be applied to HV1 from species lacking these His (Table 1). The gating kinetics columns are self-evident: τact is the activation (channel opening) time constant, and τtail is the tail current or deactivation (channel closing) time constant. These are expressed as the ratio τmutant/τWT so that 1 means no change, a ratio <1 means faster than WT, and a ratio >1 means slower than WT. Because HV1 gating kinetics depends very strongly on temperature, with Q10 6–9 (DeCoursey and Cherny, 1998; Kuno et al., 2009), and is also influenced profoundly by experimental artifacts including proton depletion-induced current decay (“droop”) and pH changes, anything less than a twofold change should be viewed with skepticism. The columns labeled ΔVthreshold (ΔVthr) and ΔpH slope embody one of the crucial and unique properties of this channel, namely its ΔpH-dependent gating. Decreasing pHi or increasing pHo shifts the gH–V relationship negatively by roughly 40 mV/U of change in pH, as originally described by Cherny et al. (1995):Vthreshold= 20−40ΔpH(1) or, generalized:Vthreshold=V0–VslopeΔpH,(2) where Vthreshold is the most negative voltage at which detectable current can be elicited, V0 is Vthreshold at symmetrical pH (roughly 20 mV), Vslope is the steepness (in millivolts/unit pH) of the relationship (nominally 40 mV), and ΔpH = pHo − pHi. The precise value for Vslope depends on whether the abscissae are EH (the Nernst potential for H+ based on the nominal pH of the solutions) or the measured Vrev. Because Vrev often changes by less than a Nernstian amount in real experiments (almost certainly as a result of our inability to perfectly control pH, in combination with pH changes due to the measurement itself), plotting Vthreshold against the measured Vrev usually produces a larger slope. Thus, the shift of Vthreshold (vs. Vrev) for native proton currents measured in rat alveolar epithelial cells was 44 mV/U (Cherny et al., 1995). The mean shift reported in 15 types of cells was 46 mV/U (DeCoursey, 2003b). The slope for hHV1 transfected into HEK or COS cells was 39–43 mV/U (Musset et al., 2008a). There are essentially no reports of mutants in which Vslope departs convincingly from this range. Two nominal deviations from this rule are R211A (53 mV) and ΔN (28 mV; Ramsey et al., 2010), but their P-values versus WT are 0.04 and 0.02, and 2/31 values in this study could easily fall just under the arbitrary P = 0.05 cutoff by chance. Rather than give an absolute value for the position of the gH–V relationship, such as using the parameter V0 in Eq. 2 above (which is not in common parlance), we list the change of Vthreshold (Vthr) or an equivalent parameter from control values in each study. Actual numbers will depend on conventions in each laboratory. When data were reported for asymmetrical pH, the Vthreshold value was “corrected” by shifting it by 40 mV/U change in ΔpH (Cherny et al., 1995). For reasons discussed at length elsewhere (Musset et al., 2008a), we consider it a highly questionable practice to fit whole-cell gH–V data with a Boltzmann function, as is routinely done for other voltage-gated channels. Because the distortion of current amplitudes and kinetics resulting from proton depletion are ubiquitous and profound, we prefer to quantify absolute voltage dependence by Vthreshold (the most negative voltage at which discernable time-dependent H+ currents are detected) or VgH,max/10 (the voltage at which the gH is 10% of its maximal value), both measured during small depolarizations and thus minimizing depletion. The absolute position of the gH–V relationship appears quite mutable with mutation; the extensive study of Ramsey et al. (2010) produced examples spanning >200 mV for various mutants. Given the technical difficulty and intrinsic variability of Vthreshold estimation, it would be dangerous to draw conclusions about shifts of less than ∼20 mV. There is a 30-mV range of values reported for WT hHV1 (see Table 3 in DeCoursey, 2013). Selectivity is given only when it was explicitly evaluated. The WT channel is perfectly selective for protons, so H+ is entered. The entry Cl– means that the channel is permeable to Cl– and likely to other anions as well; Na+ means the channel is permeable to Na+ and likely other cations besides H+. The column Other simply provides concise information that does not fit elsewhere in the tables. Table 2: The N terminus (positions 1–100 in hHV1) N terminus The N terminus of hHV1 comprises 100 amino acids and is intracellular. The effects of truncating the entire N terminus (ΔN) are not dramatic. Deleting both the N and C termini (ΔN/ΔC) simultaneously results in five- to sixfold faster activation, presumably because these truncations result in monomeric constructs. Although coiled-coil interactions in the C terminus are generally considered to be the main interaction that stabilizes the dimer (Koch et al., 2008; Lee et al., 2008; Tombola et al., 2008; Fujiwara et al., 2013b, 2014; Smith and DeCoursey, 2013), deleting both N and C termini (ΔN/ΔC) appeared to produce monomers more reliably (Koch et al., 2008). Nevertheless, even when both N and C termini are deleted, the VSD-only construct spontaneously dimerizes with a Kd of ∼3 µM (Li et al., 2015). The short isoform, hHV1S Some B cells, especially B lymphocytes from chronic lymphocytic leukemia patients or malignant B cell lines (Hondares et al., 2014), express a short isoform of hHV1 (hHV1S) that lacks the first 20 amino acids (Capasso et al., 2010). Fig. 1 shows that the 21st amino acid is Met (ATG), which acts as an alternative start site (Hondares et al., 2014). Compared with the full-length protein, hHV1S opens more slowly, and its enhanced gating response to PMA is more profound. Furthermore, hHV1S interacts less with the B cell receptor, resulting in less internalization. Together, its properties suggest that its expression may contribute to the pathogenesis of B cell malignancies (Hondares et al., 2014). The phosphorylation site responsible for enhanced gating When phagocytes are stimulated to undergo the respiratory burst (i.e., activation of NADPH oxidase, or NOX2), the properties of proton channels change so dramatically (Bánfi et al., 1999; DeCoursey et al., 2000; Musset et al., 2009) that at first, the appearance of a second, distinct type of proton channel (proposed to be a component of the active NADPH oxidase complex) was hypothesized (Henderson et al., 1995; Henderson and Chappell, 1996; Bánfi et al., 1999). A decade of controversy ensued (Henderson et al., 1997; Henderson, 1998; Henderson and Meech, 1999, 2002; DeCoursey et al., 2001b, 2002, 2003; Maturana et al., 2002; Touret and Grinstein, 2002; DeCoursey, 2003a,b, 2016). The idea that the gp91phox component of NADPH oxidase could function as a proton channel was not dispelled completely until well after the HVCN1 gene was identified (Ramsey et al., 2006; Sasaki et al., 2006) when the HVCN1 knockout mouse was developed, which provided the final nail in the coffin (Morgan et al., 2009; El Chemaly et al., 2010). In activated phagocytes, four characteristics of H+ currents change, all in the direction of increasing proton flux: the maximum gH increases two- to fourfold, the gH–V relationship shifts negatively by 30–40 mV, τact becomes two to five times faster, and τtail slows two- to sixfold (DeCoursey et al., 2000, 2001a,b; Cherny et al., 2001; DeCoursey, 2003a; Musset et al., 2009). This is referred to as the enhanced gating mode to emphasize that the properties of the HV1 channel change as a result of phosphorylation, as opposed to a second type of channel appearing. The original mechanism proposed for enhanced gating of HV1 was not phosphorylation, but rather modulation of the channel by arachidonic acid generated by cPLA2α (Henderson and Chappell, 1992). Although arachidonic acid does enhance HV1 gating by a direct pharmacological effect (DeCoursey and Cherny, 1993; Kapus et al., 1994; Suszták et al., 1997; Kawanabe and Okamura, 2016), neither specific cPLA2α inhibition nor genetic knockout of cPLA2α affects the activation of NADPH oxidase or the enhanced gating of HV1 channels during the respiratory burst (Morgan et al., 2007). Two predicted PKC phosphorylation sites, Thr29 and Ser97, were studied in hHV1 expressed in the B cell–related LK35.2 cell line (Musset et al., 2010a). Although both were detectably phosphorylated, mutation of Thr29 but not Ser97 abolished the enhanced gating response, implicating Thr29 as the key PKC phosphorylation site in hHV1 (Musset et al., 2010a). Analogous studies of the short isoform identified the same residue, Thr9, as the main phosphorylation site (Hondares et al., 2014). The first identified human hHV1 mutation The first naturally occurring hHV1 mutation from a human subject, M91T (Table 2), was identified by Iovannisci et al. (2010), who cloned HVCN1 genes from primary human airway tissue cultures. Unfortunately, the mutation was discovered only after the death of the donor, who consequently had no opportunity to mourn the defective nature of his/her proton channels, and we lack the opportunity to evaluate the effects of the mutation on his/her quality of life. The main effect of the mutation on HV1 expressed in COS cells is to decrease the likelihood of channel opening. It requires ∼20 mV more depolarization or ∼0.5 U of greater ΔpH (for airway epithelia, this likely means a higher pHo) to open mutant M91T channels (Iovannisci et al., 2010). Table 3: The S1 helix (positions 101–125) and the S1–S2 linker (126–133) Table 3. Changes in HV1 properties in S1 (101–125) and S1–S2 linker (126–133) mutants versus WT channels Mutant Species Expr. system I? τ act τ tail ΔV thr ΔpH slope Selectivity Other Reference Q102C CiHV1 [H150C] Xenopus i-o Yes MTSi access Mony et al., 2015 V103C CiHV1 [V151C] Xenopus i-o Yes MTSi open> closed Mony et al., 2015 I105C CiHV1 [I153C] Xenopus i-o Yes MTSi open> closed Mony et al., 2015 I106C CiHV1 [I154C] Xenopus i-o Yes MTSi open> closed Mony et al., 2015 C107A hHV1 HEK wc Yes −17 Ramsey et al., 2010 C107S hHV1 90% dimer Li et al., 2015 V109C CiHV1 [V157C] Xenopus i-o Yes MTSi open> closed Mony et al., 2015 V109A hHV1 Xenopus i-o Yes −11 Hong et al., 2014 D112A hHV1 HEK wc Yes 59 38 Ramsey et al., 2010 D112Aa hHV1 COS/HEK wc Yes 2.2 3.0 41b 43 Cl– Musset et al., 2011 D112A hHV1 Vesicle flux Slows H+ flux Letts, 2014 D112A hHV1 Xenopus i-o No Hong et al., 2014 D112A CiHV1 [D160A] Xenopus TEVC No Chamberlin et al., 2015 D112A kHV1 [D51A] COS/HEK wc Yes Cl– Smith et al., 2011 D112A NpHv1 [D66A] HEK wc Yes Cl– Chaves et al., 2016 D112C CiHV1 [D160C] Xenopus TEVC No Chamberlin et al., 2015 D112C NpHv1 [D66C] HEK wc No Chaves et al., 2016 D112C/R211C CiHV1 [D160C/ R261C] Xenopus TEVC Yes nc Na+ Chamberlin et al., 2015 D112N hHV1 HEK wc Yes 31 42 Ramsey et al., 2010 D112Na hHV1 COS/HEK wc Yes 2.4 3.0 23b 35 Cl– Musset et al., 2011 D112N/D185A hHV1 HEK wc Yes 103 Ramsey et al., 2010 D112A hHV1 Xenopus i-o No Hong et al., 2014 D112E hHV1 Xenopus i-o Yes −13 Hong et al., 2014 D112E hHV1 COS/HEK wc Yes 0.18 7.4/.085 −11b 34 H+ Biexponential tails Musset et al., 2011 D112E hHV1 Xenopus i-o Yes −15 Berger and Isacoff, 2011 D112E kHV1 [D51E] COS/HEK wc Yes H+ Smith et al., 2011 D112E NpHv1 [D66E] HEK wc Yes H+ Chaves et al., 2016 D112E/I127C CiHV1 [D160A/I175C] Xenopus i-o Yes nc 4.0 Mony et al., 2015 D112H hHV1 COS/HEK wc Yes 2.0 0.85 13b 38 Cl– Musset et al., 2011 D112H kHV1 [D51H] COS/HEK wc Yes Cl– Smith et al., 2011 D112H NpHv1 [D66H] HEK wc Yes Cl– Chaves et al., 2016 D112Ka hHV1 COS/HEK wc Yes 0.8 0.22 46b 40 Cl– Musset et al., 2011 D112S hHV1 COS/HEK wc Yes 2.0 4.1 25b 38 Cl– Musset et al., 2011 D112S hHV1 Xenopus i-o No Berger and Isacoff, 2011 D112S hHV1 Vesicle flux Slows H+ flux Letts, 2014 D112S kHV1 [D51S] COS/HEK wc Yes Cl– Smith et al., 2011 D112S NpHv1 [D66S] HEK wc Yes Cl– Chaves et al., 2016 D112S/R211S hHV1 Xenopus i-o Yes 24 44 Gu+ At pH 8//8 Berger and Isacoff, 2011 D112Fa hHV1 COS/HEK wc Yes 1.6 0.03 44b 38 Cl– Musset et al., 2011 D112R/R211D hHV1 Xenopus i-o Yes H+ Berger and Isacoff, 2011 D112V hHV1 COS/HEK wc No Musset et al., 2011 D112L hHV1 Vesicle flux Slows H+ flux Letts, 2014 D112Ia hHV1 COS/HEK wc No DeCoursey, 2015b D112Q hHV1 Xenopus i-o No Hong et al., 2014 D112A/L108Da hHV1 COS/HEK wc No Morgan et al., 2013 D112V/V109Da hHV1 COS/HEK wc No Morgan et al., 2013 D112A/V109Da hHV1 COS/HEK wc Yes Cl– Morgan et al., 2013 D112A/V110Da hHV1 COS/HEK wc No Morgan et al., 2013 D112A/L111Da hHV1 COS/HEK wc No Morgan et al., 2013 D112A/A113Da hHV1 COS/HEK wc No Morgan et al., 2013 D112A/L114Da hHV1 COS/HEK wc No Morgan et al., 2013 D112A/L115Da hHV1 COS/HEK wc No Morgan et al., 2013 D112A/V116Da hHV1 COS/HEK wc Yes H+ Morgan et al., 2013 D112V/V116Da hHV1 COS/HEK wc Yes H+ Morgan et al., 2013 D112V/V116Ea hHV1 COS/HEK wc Yes H+ Morgan et al., 2013 D112V/V116Sa hHV1 COS/HEK wc Yes Cl– Morgan et al., 2013 D112V/V116Na hHV1 COS/HEK wc Yes Cl– Morgan et al., 2013 D112A/L117Da hHV1 COS/HEK wc No Morgan et al., 2013 D112A/A118Da hHV1 COS/HEK wc No Morgan et al., 2013 D112N/I127C CiHV1 [D160N/I175C] Xenopus i-o No Mony et al., 2015 D112N/R211S/I127C CiHV1 [D160N/R261S/I175C] Xenopus i-o Yes 15 Mony et al., 2015 D112N/R211S/G199C CiHV1 [D160N/R261S/G249C] Xenopus i-o Yes 44 Mony et al., 2015 D112N/G199C CiHV1 [D160N/G249C] Xenopus i-o No Mony et al., 2015 A113Da hHV1 COS/HEK wc Yes H+ Morgan et al., 2013 E119A hHV1 HEK wc Yes 20 47 Ramsey et al., 2010 E119L hHV1 Vesicle flux nc H+ flux Letts, 2014 E119S mHV1 [E115S] HEK wc Yes nc Zn2+ inhibition Takeshita et al., 2014 E119S/D123S mHV1 [E115S/D119S] HEK wc Yes Weaker Zn2+ inhibition Takeshita et al., 2014 E119A CiHV1 [E167A] Xenopus TEVC Yes 4 Chamberlin et al., 2014 E119C/R205C CiHV1 [E167C/R255C] Xenopus TEVC Yes −2 Chamberlin et al., 2014 E119C/R208 CiHV1 [E167C/R258C] Xenopus TEVC Yes −52 Chamberlin et al., 2014 D123A hHV1 HEK wc Yes 20 48 Ramsey et al., 2010 D123S mHV1 [D119S] HEK wc Yes nc Zn2+ inhibition Takeshita et al., 2014 D123C CiHV1 [D171C] Xenopus i-o Yes MTSo open> closed Mony et al., 2015 D123A CiHV1 [D171A] Xenopus TEVC Yes 72 Chamberlin et al., 2014 D123A/R205N CiHV1 [D171A/R255N] Xenopus TEVC Yes 11 Chamberlin et al., 2014 K125A hHV1 HEK wc Yes 19 47 Ramsey et al., 2010 K125C CiHV1 [K173C] Xenopus i-o Yes MTSo access Mony et al., 2015 I127C CiHV1 [I175C] Xenopus i-o Yes 0 Mony et al., 2015 D130A hHV1 HEK wc Yes 13 Ramsey et al., 2010 K131A hHV1 HEK wc Yes 33 Ramsey et al., 2010 That numerical entries are shown does not imply that any given change was significant. Italicized mutant entries from nonhuman species show the hHV1 equivalent. HEK, HEK-293, HEK-293T, tsA, or HM1; COS, COS-7; Xenopus, Xenopus laevis oocyte; wc, whole cell; i-o, inside-out patch configuration; TEVC, two-electrode voltage clamp. Blank entries indicate that the parameter was not examined. nc, measured, but no change. Parameters are given relative to WT in each study. For I?, yes means currents are detectable. Time constants are ratios of mutant/WT. The ΔVthreshold value is the change in absolute position of the gH–V relationship versus WT. The ΔpH slope is the slope in millivolts of the relationship between Vthreshold (or other parameters reflecting the absolute position of the gH–V relationship) and Vrev or EH (which are not identical; see section Table entries defined). For column Other, MTS access from inside or outside (MTSi or MTSo, respectively) is listed as open>closed if the open channel was more accessible. a In an H140A/H193A (Zn2+ insensitive) background. b Previously unpublished, analyzed from data for Musset et al. (2011). The selectivity filter, Asp112 The most intensively studied position in hHV1 is Asp112, which was implicated in proton flux (Letts, 2014) and was identified as a crucial part of the selectivity filter (Musset et al., 2011). An indication of the importance of this position is that most mutants malfunction or fail to function altogether, although with some unexplained apparent variability between species or expression systems. Mutation of Asp112 to a neutral residue in most cases results in anion conduction. Specifically, replacing the large anion methanesulfonate– in the external solution with the smaller Cl– shifts Vrev negatively, demonstrating permeability to Cl– (Musset et al., 2011). Lowering the ionic strength by 90% shifts Vrev positively, confirming anion over cation permeation (Musset et al., 2011). Quite similar phenomenology supports an identical role for the analogous Asp in the middle of the S1 helix in two evolutionarily distant species, Karlodinium veneficum (Smith et al., 2011) and Nicoletia phytophila (Chaves et al., 2016), which respectively are only 15% and 33% identical to hHV1. The conductance of Asp112 mutants appears to vary inversely with the hydrophobicity of the substituent at position 112, with two of the most hydrophobic amino acids tested, Val and Ile, eliminating current flow altogether (Musset et al., 2011; DeCoursey, 2015b). The conservative Asp→Glu mutant retains proton selectivity. Clearly, Asp112 is crucial to the proton selectivity of hHV1. However, other Asp are present in the presumed conduction pathway, such as Asp185 (Fig. 2), but Table 5 shows that its mutation does not impair H+ flux (Letts, 2014) or H+ selectivity (Musset et al., 2011). In an attempt to determine what other requirements exist for selectivity, Asp was moved along the S1 helix to each position from 108 to 118 (Morgan et al., 2013). At most positions where Asp faced away from the pore, no current was observed. Asp produced proton selectivity at just one other position, 116. Molecular dynamics (MD) simulations suggest that the D112V/V116D construct is proton selective only when Asp116 interacts with one or more S4 Arg residues (Morgan et al., 2013). This result shows that not every near-neighbor interaction of Asp112 in its native position is necessary, but there are clearly strong constraints. The requirements for proton selectivity in hHV1 deduced from these and many other mutations include the following: (a) a carboxyl group (Asp or Glu) is required; (b) it must face the pore; (c) it must be located at a narrow point in the channel; and (d) it must be able to interact with a basic group (Arg or Lys). These conditions apparently exist only in the outer vestibule of hHV1. Furthermore, several attempts to reposition Asp into S2 or S3 failed to produce a proton-selective conductance, suggesting that, for reasons that are not at all clear, the carboxyl group must be on S1. A quantum model of the selectivity filter of hHV1 illustrates how interacting Asp and Arg side chains can selectively conduct protons while excluding other ions (Dudev et al., 2015). Intriguingly, proton-selective conduction is preserved when Asp112 is replaced by Glu112 (Musset et al., 2011) or when Arg208 is replaced by Lys208 (Dudev et al., 2015). Clearly, this critical interaction has leeway with respect to chain length. The F1Fo ATP synthase (H+ translocating ATPase) remarkably parallels HV1 in that the proton pathway in its c subunit has an essential Asp61–Arg210 pair and Asp61 can be moved to a different location or replaced by Glu with only partial loss of function (Miller et al., 1990). It is noteworthy that in several other molecules with critical proton transport pathways, analogous substitutions impair function: Asp→Glu (Chen et al., 2000; Ruivo et al., 2012; Luoto et al., 2013), Glu→Asp (Thorndycroft et al., 2007; Cornish et al., 2011), and Lys→Arg (Balashov et al., 2013). Li et al. (2015) found that hHV1 is more mobile and dynamic than VSDs of other voltage-gated ion channels. The series of mutations to Asp112 nicely illustrates the difficulty of interpreting mutations. For example, D112A, D112S, and D112N (Table 3) all open and close more slowly than WT. So is the function of Asp112 to speed gating? No, because D112K produced faster kinetics, and D112F changed activation and deactivation in opposite directions. So does Asp112 regulate gating kinetics? Not really, because practically every mutation for which kinetics were reported alters kinetics, by up to 100-fold. Considering that gating reflects large conformational changes and perhaps other subtler changes that involve a large fraction of the amino acids in the protein, and HV1 is a compact molecule, it is not surprising that most mutations affect gating kinetics. Their interpretation requires semantic judiciousness. Not every position whose mutation affects gating can reasonably be said to regulate gating kinetics. Nevertheless, the effects of mutations are real and suggest involvement in the process, whose mechanism may, however, be difficult to disentangle. Countercharges in S1 A fundamental principle in the conception of how voltage gating works is that the periodically spaced cationic residues in S4 (Arg and sometimes Lys) that sense voltage interact electrostatically with anionic amino acids elsewhere in the channel protein to stabilize both closed and open states (Papazian et al., 1995; Tiwari-Woodruff et al., 1997; Lecar et al., 2003). In the General conclusions section below, we discuss alternative interpretations. Within the charge/countercharge conceptual framework, externally accessible acidic residues stabilize the open state, and internally accessible acidic groups stabilize the closed state, presumably by interacting with the cationic groups in S4. If this is the case, a neutralizing mutation to such an externally accessible amino acid should shift the gH–V relationship positively because the mutant will lose open state stabilization. Conversely, an internal acidic residue would normally stabilize the closed state, and its mutation should promote channel opening, thus shifting the gH–V relationship negatively. By these criteria, three acidic amino acids in S1 in the outer vestibule, Asp112, Glu119, and Asp123, and possibly Lys125 may be considered weak stabilizers of the open state because their neutralization by mutation produces modest positive shifts. Both Asp112 and Glu119 interact with S4 Arg residues in MD simulations of open state homology models of hHV1 (Wood et al., 2012; Kulleperuma et al., 2013) and CiHV1 (Chamberlin et al., 2014). Table 4: The S2 helix (positions 134–156) and the S2–S3 linker (157–165) Table 4. Changes in HV1 properties in S2 (134–156) and S2–S3 linker (157–165) mutants versus WT channels Mutant Species Expr. system I? τ act τ tail ΔV thr ΔpH slope Selectivity Other Reference Y134A hHV1 HEK wc Yes 3 Ramsey et al., 2010 H140A hHV1 HEK wc Yes H+ 9 × Kda Zn2+ Ramsey et al., 2006 H140A hHV1 Vesicle flux nc H+ flux Letts, 2014 H193A hHV1 HEK wc Yes H+ 39 × Kda Zn2+ Ramsey et al., 2006 H140A/ H193A hHV1 HEK wc Yes −12 46 H+ 2,000 × Kda Zn2+ Ramsey et al., 2006 H140A/ H193A hHV1 COS/HEK wc Yes H+ Musset et al., 2011 Y141A hHV1 HEK wc Yes −27 Ramsey et al., 2010 S143A hHV1 HEK wc Yes 11 41 Ramsey et al., 2010 S143A hHV1 Vesicle flux nc H+ flux Letts, 2014 D112V/S143D hHV1 COS/HEK wc Yes Cl– Morgan et al., 2013 D112V/I146D hHV1 COS/HEK wc No Morgan et al., 2013 D112V/L147D hHV1 COS/HEK wc No Morgan et al., 2013 F150A hHV1 Xenopus i-o Yes −24 Hong et al., 2013 F150C hHV1 Xenopus i-o Yes −22 Hong et al., 2013 F150W hHV1 Xenopus i-o Yes −55 Hong et al., 2013 E153A hHV1 HEK wc Yes −55 42 Ramsey et al., 2010 E153A CiHV1 [E201A] Xenopus TEVC No Chamberlin et al., 2014 E153G CiHV1 [E201G] Xenopus TEVC No Chamberlin et al., 2014 E153N hHV1 HEK wc Yes −1.17 45 Ramsey et al., 2010 E153A hHV1 Vesicle flux nc H+ flux Letts, 2014 E153D hHV1 HEK wc Yes −23 37 Ramsey et al., 2010 E153D/D174E hHV1 HEK wc Yes −102 40 Ramsey et al., 2010 E153C hHV1 Xenopus i-o Yes −55 Tombola et al., 2010 E153C CiHV1 [E201C] Xenopus TEVC Yes −101 Chamberlin et al., 2014 E153C/R205C CiHV1 [E201C/R255C] Xenopus TEVC Yes −43 Chamberlin et al., 2014 E153C/R208 CiHV1 [E201C/R258C] Xenopus TEVC Yes 37 Chamberlin et al., 2014 E153C/R211C CiHV1 [E201CR261C] Xenopus TEVC Yes −60 Chamberlin et al., 2014 K157A hHV1 HEK wc Yes 1 39 Ramsey et al., 2010 K157A hHV1 Vesicle flux nc H+ flux Letts, 2014 R162A hHV1 HEK wc Yes 13 Ramsey et al., 2010 That numerical entries are shown does not imply that any given change was significant. Italicized mutant entries from nonhuman species show the hHV1 equivalent. HEK, HEK-293, HEK-293T, tsA, or HM1; COS, COS-7; Xenopus, Xenopus laevis oocyte; wc, whole cell; i-o, inside-out patch configuration; TEVC, two-electrode voltage clamp. Blank entries indicate that the parameter was not examined. nc, measured, but no change. Parameters are given relative to WT in each study. For I?, yes means currents are detectable. The ΔVthreshold value is the change in absolute position of the gH–V relationship versus WT. The ΔpH slope is the slope in millivolts of the relationship between Vthreshold (or other parameters reflecting the absolute position of the gH–V relationship) and Vrev or EH (which are not identical; see section Table entries defined). a Nominal Kd values are nearly meaningless for Zn2+ inhibition of HV1 because its main effects are slowing activation and shifting the gH–V relationship positively (Cherny and DeCoursey, 1999). As demonstrated in the Appendix of DeCoursey et al. (2001a), the apparent Kd derived from the ratio IH(Zn2+)/IH can vary more than three orders of magnitude depending on the test potential selected. If all measurements are done the same way, relative Kd values have meaning. The Zn2+-binding site The most potent inhibitor of voltage-gated proton currents is Zn2+ (Mahaut-Smith, 1989; DeCoursey, 2003b). Unlike traditional channel blockers that occlude the pore, Zn2+ shifts the gH–V relationship positively and slows activation (Cherny and DeCoursey, 1999). These effects were strongly inhibited at low pHo, indicating competition between Zn2+ and H+ for a binding site. To model the competition between H+ and Zn2+ quantitatively required assuming that Zn2+ prevents channel opening by binding to an externally accessible site on the closed channel comprising at least two titratable groups with a pKa of 6.2–6.6 (near that of His; Cherny and DeCoursey, 1999). Seven years later, the identification of the hHV1 gene confirmed this deduction because two His residues, His140 in S2 and His193 in the S3–S4 linker, were found to comprise the main sites at which Zn2+ binds to inhibit proton currents (Ramsey et al., 2006). The single mutants H140A and H193A each have diminished sensitivity to Zn2+, and the double mutant is nearly impervious. The Zn2+ sensitivity of a series of mutants in which one or both of these His were mutated to Ala, including various tandem dimers, is described elsewhere (Musset et al., 2010b) and is not included in the tables. Remarkably, the crystal structure of the closed mHV1 channel contained a Zn2+ atom tetrahedrally coordinated by the corresponding two His in the mouse (His136 and His189 in mHV1), with weaker binding to Glu115 and Asp119 (Glu119 and Asp123 in hHV1). Countercharges in S2 S2 contains an important countercharge, Glu153, which, as seen in Table 1, is highly conserved among VSD-containing molecules (Smith et al., 2011). In neutral mutants, Vthreshold is shifted consistently negatively, in some cases by >100 mV, suggesting that this internal acidic residue stabilizes the closed state. Charge transfer center or hydrophobic gasket The S2 helix contains Phe150, another highly conserved residue among VSD-containing molecules (Tao et al., 2010; Smith et al., 2011) whose K+ channel correlate was described as the outer limit of the charge transfer center (Tao et al., 2010). As the positively charged Arg residues in S4 move outwards during a depolarization that opens the channel, they move past Phe150, which serves as a delimiter of internal and external accessibility. Bezanilla and colleagues include Phe150 along with two other hydrophobic residues, Val109 and Val178, in a hydrophobic gasket (or “plug”) that functions similarly (labeled HG in Table 1; Lacroix et al., 2014; Li et al., 2014, 2015; DeCoursey, 2015a). The global purpose of having a narrow isthmus of protein between the two aqueous vestibules is to focus the electric field (Yang et al., 1996, 1997; Starace and Bezanilla, 2001, 2004). This means that each gating charge (e.g., Arg) needs to move only a small distance to effectively cross the entire membrane electrical field. Mutations to Phe150 in hHV1, like those of the corresponding Phe in K+ channels, shift the gH–V relationship (Tao et al., 2010; Hong et al., 2013). Table 5: The S3 helix (positions 166–188) and the S3–S4 linker (189–196) Table 5. Changes in HV1 properties in S3 (166–188) and S3–S4 linker (189–196) mutants versus WT channels Mutant Species Expr. system I? τ act τ tail ΔV thr ΔpH slope Selectivity Other Reference E164A/E171A hHV1 HEK wc Yes −7 Ramsey et al., 2010 H167N/H168V/K169N hHV1 HEK wc Yes −13 44 Ramsey et al., 2010 E171A/D174A hHV1 HEK wc Yes −116 39 Ramsey et al., 2010 E171A hHV1 Vesicle flux Δ H+ fluxa Letts, 2014 D174A hHV1 HEK wc Yes −111 38 Ramsey et al., 2010 D174A hHV1 Vesicle flux Δ H+ fluxa Letts, 2014 D174N hHV1 HEK wc Yes −142 36 Ramsey et al., 2010 D174H hHV1 HEK wc Yes −136 37 Ramsey et al., 2010 D174E hHV1 HEK wc Yes −52 46 Ramsey et al., 2010 D174A CiHV1 [D222A] Xenopus TEVC Yes −111 Chamberlin et al., 2014 D174C/R205C CiHV1 [D222C/R255C] Xenopus TEVC Yes −95 Chamberlin et al., 2014 D174C/R208C CiHV1 [D222C/R258C] Xenopus TEVC Yes 48 Chamberlin et al., 2014 V178A hHV1 Xenopus i-o Yes −27 Hong et al., 2014 D112V/V178D hHV1 COS/HEK wc No Morgan et al., 2013 S181A hHV1 HEK wc Yes 18 46 Ramsey et al., 2010 S181A hHV1 Xenopus i-o Yes 0 Hong et al., 2014 D112V/S181D hHV1 COS/HEK wc No Morgan et al., 2013 F182A hHV1 Xenopus i-o Yes −9 Hong et al., 2014 D185A hHV1 HEK wc Yes 58 47 Ramsey et al., 2010 D185M hHV1 COS/HEK wc Yes H+ Musset et al., 2011 D185V hHV1 COS/HEK wc Yes 20b 43b H+ Musset et al., 2011 D185A hHV1 COS/HEK wc Yes 42b 40b H+ Musset et al., 2011 D185A hHV1 Vesicle flux nc H+ flux Letts, 2014 D185N hHV1 COS/HEK wc Yes 36b 47b H+ Musset et al., 2011 D185C CiHV1 [D233C] Xenopus TEVC Yes 76 Chamberlin et al., 2014 E185C/R208C CiHV1 [D233C/R258C] Xenopus TEVC Yes 4 Chamberlin et al., 2014 E192A/E196A hHV1 HEK wc Yes 13 Ramsey et al., 2010 H193A hHV1 HEK wc Yes H+ 39 × Kd Zn2+ Ramsey et al., 2006 H193A hHV1 Vesicle flux nc H+ flux Letts, 2014 H140A/H193A hHV1 HEK wc Yes −12 46 H+ 2,000 × Kd Zn2+ Ramsey et al., 2006 That numerical entries are shown does not imply that any given change was significant. Italicized mutant entries from nonhuman species show the hHV1 equivalent. In the Expression system column: HEK, HEK-293, HEK-293T, tsA, or HM1; COS, COS-7; Xenopus, Xenopus laevis oocyte; wc, whole cell; i-o, inside-out patch configuration; TEVC, two-electrode voltage clamp. Blank entries indicate that the parameter was not examined. nc, measured, but no change. Parameters are given relative to WT in each study. For I?, yes means currents are detectable. Time constants are ratios of mutant/WT. The ΔVthreshold value is the change in absolute position of the gH–V relationship versus WT. The ΔpH slope is the slope in millivolts of the relationship between Vthreshold (or other parameters reflecting the absolute position of the gH–V relationship) versus Vrev or EH (which are not identical; see section Table entries defined). a Normal initial H+ flux followed by recovery ascribed to leak induced in vesicles. b Previously unpublished, analyzed from data for Musset et al. (2011). Countercharges in S3 S3 contains two important countercharges. Asp174 is internally accessible and stabilizes the closed state, and neutral mutants shift the gH–V relationship strongly negatively. In the closed structure of mHV1, the Asp174 equivalent appears to interact with Arg211 in an internal pocket (Takeshita et al., 2014; Cherny et al., 2015). Conversely, Asp185 (which Table 1 shows is unique to the HV1 family) is externally accessible and stabilizes the open state, and neutral mutants shift the gH–V relationship positively. The milder effect of the Asp185 mutation mirrors its moderate interaction with Arg205 observed in MD simulations of an open state model of hHV1 (Kulleperuma et al., 2013) and with all three Arg in a model of CiHV1 (Chamberlin et al., 2014). Table 6: The S4 helix (positions 197–218) Table 6. Changes in HV1 properties in S4 (197–218) mutants versus WT channels Mutant Species Expr. system I? τ act τ tail ΔV thr ΔpH slope Selectivity Other Reference E196C CiHV1 [A246C] Xenopus TEVC Yes MTSo open>closed Gonzalez et al., 2010 L198C CiHV1 [I248C] Xenopus TEVC Yes MTSo open>closed Gonzalez et al., 2010 G199C CiHV1 [G249C] Xenopus i-o Yes 2 Mony et al., 2015 L200W CiHV1 [L250W] Xenopus TEVC Yes 0.27 0.67 Okuda et al., 2016 L200W/ΔC CiHV1 [L250W/ΔC] Xenopus TEVC Yes 0.20 0.019 Okuda et al., 2016 I202C CiHV1 [V252C] Xenopus TEVC Yes MTSo open>closed Gonzalez et al., 2010 R205A hHV1 HEK wc Yes 0.0053 0.086 −1 48 0.63 × WT gating chargea Ramsey et al., 2006, 2010 R205A hHV1 Vesicle flux Δ H+ fluxb Letts, 2014 R205Hc/T222stop hHV1 COS wc Yes H+ Accessible to external Zn2+ Kulleperuma et al., 2013 R205Q Mouse [R201Q] HEK wc Yes Faster −50 1.36 × WT gating chargea Sasaki et al., 2006 R205N CiHV1 [R255N] Xenopus i-o Yes 0.33 × WT gating charged Gonzalez et al., 2013 R205N CiHV1 [R255N] Xenopus TEVC Yes −36 0.78 × WT gating chargea Chamberlin et al., 2014 R205C CiHV1 [R255C] Xenopus TEVC Yes −37 1.3 × WT gating chargea Chamberlin et al., 2014 R205A/R208A hHV1 HEK wc Yes 128 51 Ramsey et al., 2010 R205A/R211A hHV1 HEK wc Yes 96 45 Ramsey et al., 2010 L206C CiHV1 [L256C] Xenopus TEVC Yes No MTSo/i access Gonzalez et al., 2010 W207A,c W207S,c or W207Fc hHV1 HEK/COS wc Yes 0.01 0.034 −17.9 40 H+ Loss of selectivity at pHo > 8 Cherny et al., 2015 W207I Mouse [W203I] HEK wc Yes 0.019 0.059 Tandem dimer Okuda et al., 2016 W207A, W207S, or W207F kHV1 [W176A, W176S, W176F] HEK/COS wc Yes 0.025 40 H+ Cherny et al., 2015 W207A, W207S, or W207F EhHV1 [W278A, W278S, W278F] HEK/COS wc Yes 0.2 −28.2 50 H+ Cherny et al., 2015 W207I CiHV1 [W257I] Xenopus TEVC Yes 0.29 0.030 Okuda et al., 2016 W207I/A210A CiHV1 [W257I/F260A] Xenopus TEVC Yes 0.23 0.026 Okuda et al., 2016 R208A hHV1 HEK wc Yes 0.0965 0.075 7 45 Ramsey et al., 2006, 2010 R208A hHV1 Vesicle flux Δ H+ fluxb Letts, 2014 R208A hHV1 Xenopus i-o No Hong et al., 2014 R208K hHV1 Xenopus i-o Yes −40 1.7 × WT gating chargea Hong et al., 2014 R208K hHV1 HEK/COS wc Yes H+ Dudev et al., 2015 R208Hc/T222stop hHV1 COS wc Yes H+ Accessible to external & maybe internal Zn2+ Kulleperuma et al., 2013 R208Q hHV1 Xenopus i-o No Hong et al., 2014 R208Q Mouse [R204Q] HEK wc No Sasaki et al., 2006 R208N hHV1 Xenopus i-o No Hong et al., 2014 R208N CiHV1 [R258N] Xenopus i-o Yes 0.50 × WT gating charged Gonzalez et al., 2013 R208C CiHV1 [R258C] Xenopus TEVC Yes −10 0.87 × WT gating chargea Chamberlin et al., 2014 V209C CiHV1 [V259C] Xenopus TEVC Yes No MTSo/i access Gonzalez et al., 2010 R211A hHV1 HEK wc Yes 2.24 0.092 70 53 Ramsey et al., 2006, 2010 R211A hHV1 Vesicle flux Δ H+ fluxb Letts, 2014 R211S hHV1 Xenopus i-o Yes 35 0.72 × WT gating chargea Hong et al., 2014 R211S hHV1 Xenopus i-o Yes 87 49 Gu+ at pH 8//8 Berger and Isacoff, 2011 R211S/I127C CiHV1 [R261S/I175C] Xenopus i-o Yes 15 Mony et al., 2015 R211Hc/T222stop hHV1 COS wc Yes H+ Accessible to internal Zn2+ when open Kulleperuma et al., 2013 R211H/D112V/V116Dc hHV1 COS/HEK wc Yes Accessible to internal Zn2+ when open Morgan et al., 2013 R211Q mHV1 [R207Q] HEK wc Yes nc Sasaki et al., 2006 R211N CiHV1 [R261N] Xenopus i-o Yes 0.38 × WT gating charged Gonzalez et al., 2013 R211C CiHV1 [R261C] Xenopus TEVC Yes 54 H+ 0.78 × WT gating chargea Chamberlin et al., 2014 R211C NpHv1 [R163C] HEK wc Yes H+ Chaves et al., 2016 I212C CiHV1 [I262C] Xenopus TEVC Yes MTSi closed>open Gonzalez et al., 2010 N214K hHV1 HEK wc Yes −3 43 Inward rectification Ramsey et al., 2010 N214R hHV1 HEK wc Yes 10 40 Inward rectification Ramsey et al., 2010 N214R hHV1 Xenopus i-o No Tombola et al., 2008 N214A hHV1 HEK wc Yes −3 42 Ramsey et al., 2010 N214A hHV1 Vesicle flux nc H+ flux Letts, 2014 N214R mHV1 [N210R] HEK wc Yes Slow Very slow −V H+ Sakata et al., 2010 N214D hHV1 COS/HEK wc Yes H+ Musset et al., 2011 N214C CiHV1 [N264C] Xenopus TEVC Yes MTSi closed>open Gonzalez et al., 2010 G215A hHV1 COS/HEK wc Yes Fast H+ Musset et al., 2011 I217stop mHV1 [I213stop] HEK wc Yes Sakata et al., 2010 G215stop mHV1 [G211stop] HEK wc Yes Sakata et al., 2010 I213stop mHV1 [I209stop] HEK wc Yes Sakata et al., 2010 A210stop mHV1 [A206stop] HEK wc Yes +V H+ τact has weak V dependence Sakata et al., 2010 L204stop mHV1 [L200stop] HEK wc No Sakata et al., 2010 That numerical entries are shown does not imply that any given change was significant. Italicized mutant entries from nonhuman species show the hHV1 equivalent. HEK, HEK-293, HEK-293T, tsA, or HM1; COS, COS-7; Xenopus, Xenopus laevis oocyte; wc, whole cell; i-o, inside-out patch configuration; TEVC, two-electrode voltage clamp. Blank entries indicate that the parameter was not examined. nc, measured, but no change. Parameters are given relative to WT in each study. For I?, yes means currents are detectable. Time constants are ratios of mutant/WT. The ΔVthreshold value is the change in absolute position of the gH–V relationship versus WT. The ΔpH slope is the slope in millivolts of the relationship between Vthreshold (or other parameters reflecting the absolute position of the gH–V relationship) and Vrev or EH (which are not identical; see section Table entries defined). When C-terminal truncations are indicated as XNNNstop, this means STOP replaces X at position NNN; hence, position NNN and all subsequent residues are truncated, and the last position remaining is NNN-1. For column Other, MTS access from inside or outside (MTSi or MTSo, respectively) is listed as open>closed if the open channel was more accessible. a Slope factor of gH–V relationship. b Normal initial H+ flux followed by recovery ascribed to leak induced in vesicles. c In an H140A/H193A (Zn2+ insensitive) background. d By limiting slope method. Cys scanning reveals aqueous accessibility A now standard approach to demonstrate aqueous accessibility of amino acids in a protein is to convert the target to Cys and then challenge the mutant with MTS reagents that react with Cys sulfhydryl groups. Whatever functional effect this reaction produces can be examined as a function of the time of exposure to determine accessibility of the Cys. Gonzalez et al. (2010) identified E196C, L198C, and I202C (all external to R1) that were accessible externally preferentially in the open state, suggesting that S4 moves outward and/or rotates. However, a smaller probe for accessibility, n-ethylmaleimide (NEM) in a PEGylation protection assay, revealed that all S4 residues external to position 203 (including the three mentioned above) are accessible, presumably in the closed state (Kurokawa and Okamura, 2014). It should be noted that the former study examined kinetics of MTS effects under voltage clamp, so the gating state was well defined. In the latter study, voltage clamp was not involved, and although at 0 mV most WT channels are closed, many mutations shift the gH–V relationship, and thus mutant channels could be open at 0 mV. Accessibility assays are limited by the size of the probe but will also be influenced by charge. For example, aqueous accessibility determined by Cys scanning with NEM as a probe revealed greater accessibility than when using the larger AMS (4-acetamido-4′-maleimidylstilbene-2,2′-disulfonic acid) as a probe (Kurokawa and Okamura, 2014). Furthermore, Ag+ as a probe revealed much greater accessibility than the larger NEM (Fillingame et al., 2002). Fillingame et al. (2002) point out that because Ag+ has an ionic radius like H3O+, it is ideal for probing proton pathways. Zn2+ has a smaller ionic radius than Ag+ (Robinson and Stokes, 1959) but is divalent. As a probe of HV1, it reveals greater accessibility than the bulkier MTS reagents (Kulleperuma et al., 2013; Morgan et al., 2013). In bulk solution, the proton diffuses almost exclusively as protonated buffer (DeCoursey, 1991; DeCoursey and Cherny, 1994, 1996). One expects that the proton permeates the aqueous vestibules of HV1 as H3O+ (DeCoursey, 2003b) and the selectivity filter as H+ (Dudev et al., 2015). Internal accessibility assessed rigorously by Cys scanning and MTS reaction kinetics under voltage clamp indicated that I212C and N214C were both more accessible at negative voltages, indicating greater accessibility in closed channels (Gonzalez et al., 2010). The residues with state-dependent accessibility in S4 thus span two positions internally and four externally. These results strongly support the idea that S4 moves outwardly during opening, but the extent of movement could be one turn of the helix, consistent with the “one-click” model (Li et al., 2015). Also consistent with a small excursion of S4 during opening are studies using Zn2+ to probe for accessibility of Arg→His mutants. In hHV1, R205H was externally accessible and R208H was accessible externally and possibly also internally (currents were tiny), whereas R211H was accessible only internally even in the open state (Kulleperuma et al., 2013; Morgan et al., 2013). Accessibility of S1 was explored by Cys scanning, and five residues were found to be more accessible at positive (open) voltages (Mony et al., 2015). One was external, and the rest were internal. Although the internal residues span seven positions, the fact that both the external and internal residues were more exposed in open channels suggests a widening of the vestibules rather than a large inward translational movement. Gating charge Numerous mutations have been performed in the S4 helix with the goal of determining the extent to which channel opening involves outward movement of positively charged groups in S4 during depolarization, as is thought to occur in most other voltage-gated ion channels. Each of the three Arg in S4 has been mutated, and the effect on gating charge was evaluated in various ways. Unfortunately, the methods for estimating gating charge are challenging, and several problems unique to HV1 make the task even more difficult. A direct approach is to measure the integral of the gating current and divide by the number of channels. However, it is nearly impossible to measure gating current in HV1 because the permeant ion cannot be removed, simple blockers have not been identified, and gating is extremely slow. The most potent inhibitor, Zn2+, does not occlude the pore, as would be required to reveal gating charge, but rather shifts the gH–V relationship positively and slows activation (Cherny and DeCoursey, 1999). Block by guanidine derivatives is also state dependent (Hong et al., 2013). Another procedure that is technically straightforward but of limited information value is to determine the slope factor of a Boltzmann function fit to the gH–V relationship. The slope factor does include the effective gating charge, but in a highly model-dependent manner; nevertheless, the steepness of the voltage dependence should diminish if gating charges are removed by mutation. Such estimates are indicated in Table 6 by footnote b. There are numerous pitfalls in this measurement, a major one being proton depletion, which produces artificial saturation of current (DeCoursey, 1991; DeCoursey and Cherny, 1994; Musset et al., 2008a), and at best, the slope provides a model-dependent estimate that almost always underestimates the true gating charge (Bezanilla and Villalba-Galea, 2013). Early studies reported slope factors corresponding to a gating charge of 1.4 e0 (Sasaki et al., 2006) for WT mHV1 or 0.9 e0 for WT hHV1 (Ramsey et al., 2006). A more meaningful approach for HV1 has been the limiting slope method, devised by Wolf Almers (Almers, 1978). This method works for a wide range of gating models, but not all (Sigg and Bezanilla, 1997), and provides gating charge estimates of ∼6 e0 for native rat proton currents (DeCoursey and Cherny, 1996, 1997), for CiHV1 (Gonzalez et al., 2010), and for hHV1 (Musset et al., 2008a) and 4 e0 for mHV1 (Fujiwara et al., 2012). The difficulty is mainly that measurements need to be extended to large negative voltages to achieve sufficiently low gH values to determine the limiting slope. When critical amino acids such as the Arg in S4 are mutated, the resulting currents are often quite small, which leads to underestimates of the gating charge. Another vexing source of error is that mutation, for example, of Arg205 (R255N in CiHV1) appears to reduce the extent of S4 movement during gating (Gonzalez et al., 2013), which in retrospect is not a very surprising result of removing one of the charges in the protein thought to move in response to voltage changes! Three studies reported lower gating charge (Ramsey et al., 2006; Gonzalez et al., 2013; Chamberlain et al., 2014) and two reported higher gating charge (Sasaki et al, 2006; Chamberlain et al, 2014) when Arg205 was neutralized (Table 6). Two studies reported lower gating charge when Arg208 was neutralized (Gonzalez et al., 2013; Chamberlain et al., 2014). The conservative Arg→Lys mutation increased gating charge by 70% (Hong et al., 2014). Finally, both studies of neutralized Arg211 reported lower gating charge (Gonzalez et al., 2013; Chamberlain et al., 2014). It might be noted that all studies reporting higher gating charge were based on the slope of Boltzmann functions. The tables do not list a number of studies of monomeric constructs, which consistently exhibit only half the gating charge of native dimers, a manifestation of cooperative gating (Gonzalez et al., 2010, 2013; Fujiwara et al., 2012; Okuda et al., 2016). Why is there a Trp in the middle of S4? Tryptophan prefers the interfacial environment near membrane lipid head groups (Landolt-Marticorena et al., 1993; Killian and von Heijne, 2000), but in HV1, a perfectly conserved Trp residue (Trp207 in hHV1) is located right in the middle of the S4 TM helix. Two studies have been conducted to determine why it is there, exploring mutations in multiple species (Table 1). The most prominent effect of Trp mutation was drastic acceleration of channel gating. In hHV1, activation and deactivation were 100 times and 30 times faster, respectively (Cherny et al., 2015), whereas in CiHV1, deactivation was more profoundly accelerated (Okuda et al., 2016). Effects of Trp mutation differ dramatically among species. Table 6 shows that the acceleration of channel opening was 100-fold in hHV1, 5-fold in EhHV1, 40-fold in kHV1 (Cherny et al., 2015), 3.4-fold in CiHV1, and 53-fold in mHV1 (Okuda et al., 2016). It is noteworthy that the properties of W207A, W207S, and W207F (three substituents with quite different properties) were all modified identically in hHV1 and in kHV1, which strongly implicates a unique property of Trp at this location (Cherny et al., 2015). Two mechanisms have been proposed to explain how Trp slows gating. Based on the proximity of Trp207 and Arg211 in the closed crystal structure (Takeshita et al., 2014), cation–π interaction between Trp207 and Arg211 was postulated to stabilize closed hHV1 channels, with this interaction broken during channel opening (Cherny et al., 2015). In CiHV1, π stacking of Trp from each protomer at the dimer interface was proposed to slow deactivation (Okuda et al., 2016). This proposal was supported by Trp slowing deactivation preferentially in dimeric versus monomeric constructs (Okuda et al., 2016). Mutant cycle analysis supported the idea that π stacking of Trp at the dimer interface contributed to slow deactivation, but the slowing of activation was independent of channel dimerization. Consistent with this latter conclusion, kHV1 lacks predicted coiled coil in its C terminus and thus is likely a monomer (Smith and DeCoursey, 2013), and τact was two orders of magnitude faster in Trp mutants of kHV1. Trp207 mutants not only opened and closed faster, but the Q10 of their gating kinetics dropped from the astronomical 6–9 of WT HV1 (DeCoursey and Cherny, 1998; Ramsey et al., 2006; Kuno et al., 2009) into the realm of ordinary voltage-gated ion channels, 3.5–4.0 (Cherny et al., 2015). Trp207 is a key component in several of the unique properties of HV1 (Cherny et al., 2015). Does Arg211 contribute to proton selectivity? One study concluded that Arg211 is part of the selectivity filter because 15 R211x mutants were permeable to guanidinium, Gu+, at symmetrical pH 8, whereas outward current in WT hHV1 expressed in Xenopus oocytes was blocked (Berger and Isacoff, 2011). However, this result could not be reproduced in hHV1 expressed in HEK cells, suggesting that the expression system alters this property. Large outward currents were seen in both WT and R211A channels in HEK cells at high pH (DeCoursey, 2013). Neither WT nor R211A was detectably permeant to smaller cations (unpublished data), so the Gu+ result appears anomalous, perhaps related to the ability of this chaotropic ion to disrupt hydrogen bonds, interact with hydrophobic regions of proteins, bind to sites normally occupied by water, and denature proteins (Makhatadze and Privalov, 1992; Courtenay et al., 2001; England and Haran, 2011). Although molar Gu+ is typically required for wholesale denaturation, 50 mM Gu+ sufficed to perturb the permeation pathway of voltage-gated K+ channels (Kalia and Swartz, 2011). It is likely that ions are highly concentrated in the pores of channels just as they are in the active sites of enzymes (Jimenez-Morales et al., 2012) because of the high charge density in the protein. We imagine that Gu+ tunnels through the pore in a manner that no physiological ion can reproduce, perhaps by breaking the hydrogen bonds between Asp112 and Arg208 that prevent ions other than H3O+ from entering the selectivity filter (Dudev et al., 2015). It is difficult to imagine a role for Arg211 in selectivity because the C terminus can be truncated along with the inner part of S4 (between Arg208 and Arg211) without loss of selectivity (Sakata et al., 2010). One might argue that when Arg211 is removed, Arg208 may take over its function. However, R211H in hHV1 (Kulleperuma et al., 2013) and R211C in NpHV1 (Chaves et al., 2016) and CiHV1 (Chamberlin et al., 2015) are all proton selective. What about Asn214? In S4, where many VSD-containing proteins have a fourth Arg or Lys (Table 1), HV1 has Asn214. Noting that the N214R mutant did not conduct, an early suggestion was that, assuming that S4 in HV1 moves outward as it does in other channels, Asn214 might occupy a narrow constriction where it would allow protons to pass (Tombola et al., 2008). Two other groups reported that N214R did conduct (Ramsey et al., 2010; Sakata et al., 2010), but both used mammalian expression systems, as opposed to Xenopus. This may be another example in which the expression system alters the outcome. However, the currents in N214R mutants were small, so this may be a case of different laboratories having different definitions of what comprises a detectable current. Given evidence that Arg211 remains internally accessible in open hHV1 (Kulleperuma et al., 2013; Morgan et al., 2013; Li et al., 2015), Asn214 most likely remains well inside the internal vestibule. Table 7: The C terminus (positions 219–273) Table 7. Changes in HV1 properties in C-terminal (219–273) mutants versus WT channels Mutant Species Expr. system I? τ act τ tail ΔV thr ΔpH slope Selectivity Other Reference S219P hHV1 COS/HEK wc Yes H+ Musset et al., 2011 C249S hHV1 Loss of dimer formation Li et al., 2015 ΔC (T222stop) hHV1 HEK wc Yes 1 28 Ramsey et al., 2010 ΔC (T222stop) hHV1 Yes 0.15 Weaker Zn2+ effects Musset et al., 2010b ΔC mHV1 (V216stop) HEK wc Yes 0.34 nc nc Loss of dimer formation Koch et al., 2008 ΔC mHV1 (V216stop) HEK wc Yes 0.15 0.16 Okuda et al., 2016 ΔC CiHV1 (D275stop) Xenopus TEVC Yes 0.14 0.075 Okuda et al., 2016 V220G/ K221G/T222G mHV1 V216G/K217G/T218G HEK wc Loss of cooperative gating Fujiwara et al., 2012 That numerical entries are shown does not imply that any given change was significant. HEK, HEK-293, HEK-293T, tsA, or HM1; COS, COS-7; Xenopus, Xenopus laevis oocyte; wc, whole cell; i-o, inside-out patch configuration; o-o, outside-out patch configuration; TEVC, two-electrode voltage clamp. Blank entries indicate that the parameter was not examined. nc, measured, but no change. Parameters are given relative to WT in each study. For I?, yes means currents are detectable. Time constants are ratios of mutant/WT. The ΔVthreshold value is the change in absolute position of the gH–V relationship versus WT. The ΔpH slope is the slope in millivolts of the relationship between Vthreshold (or other parameters reflecting the absolute position of the gH–V relationship) and Vrev or EH (which are not identical; see section Table entries defined). When C-terminal truncations are indicated as XNNNstop, this means STOP replaces X at position NNN; hence, position NNN and all subsequent residues are truncated, and the last position remaining is NNN-1. The mouse N-terminal deletions were done by replacing P78M to initiate translation at that position. Note that a large number of mutations to the linker between S4 and the C terminus have been studied (Fujiwara et al., 2012, 2014), but their results are beyond the scope of this table and thus are not included. The coiled-coil region holds the dimer together The C terminus of hHV1 contains extensive typical coiled-coil sequences. When it was learned that mammalian and some other HV1 assemble into dimers in cell membranes, the coiled-coil regions were implicated in dimerization (Koch et al., 2008; Lee et al., 2008; Tombola et al., 2008; Li et al., 2010). A cross-linking study with 15 strategically located Cys mutants confirmed that the C terminus was a major point of attachment of the dimer (Lee et al., 2008). By modifying the C terminus, Okamura’s group was able to generate functioning trimers and tetramers (Fujiwara et al., 2013a). A crystal structure of the C terminus revealed that the lone Cys (Cys249 in hHV1) in the C terminus forms a disulfide bond, increasing the stability of the dimer (Fujiwara et al., 2013b). This result was confirmed by mutation of the two native Cys in hHV1; C107S was still 90% dimer, whereas C249S was <5% dimer (Li et al., 2015). C terminus truncation was determined to produce mainly monomeric constructs (Koch et al., 2008; Tombola et al., 2008). The monomers behave differently electrophysiologically because the dimer exhibits cooperative gating (Gonzalez et al., 2010; Tombola et al., 2010). The dimer opens with sigmoid kinetics, like a classical Hodgkin-Huxley n2 mechanism (Hodgkin and Huxley, 1952). The monomer opens approximately five times faster and with exponential kinetics (Koch et al., 2008; Musset et al., 2010b,c; Tombola et al., 2010; Fujiwara et al., 2012). Because both monomers must move in response to voltage before either can open, the gating charge is twice as large in the dimer as in the monomer (Gonzalez et al., 2010, 2013; Fujiwara et al., 2012; Okuda et al., 2016). Does the C terminus modulate gating? The Okamura group has studied the C terminus extensively using creative approaches. They conclude that the C terminus and the S4 helix form a single rigid monolithic rod, which moves during cooperative gating (Fujiwara et al., 2012, 2014). Two types of evidence support this conclusion. First, when a rigid (AAA) or flexible (GGG) triplet was inserted between S4 and the C terminus (at 220–222, replacing VKT; Table 7), the rigid linker behaved like a WT dimer (sigmoid kinetics, slow activation, and full gating charge), whereas the floppy linker produced monomer-like behavior (exponential kinetics, fast activation, and half the gating charge). This showed that simply being a dimer was insufficient to produce cooperative gating; a continuous α helix from the C terminus through S4 was required (Fujiwara et al., 2012). In another intriguing study, 1–10 amino acids were inserted or deleted from the connection between the S4 TM segment and the C terminus. Gating kinetics exhibited a periodic dependence on the linker length, presenting slower WT-like kinetics when the whole S4-C domain was in register (Fujiwara et al., 2014). In this context, even though the VSD-only construct (lacking both N and C termini) spontaneously associates into dimers in liposomes (Li et al., 2015), it likely still functions as two independent monomers. There is also evidence that interactions at the extracellular end of the S1 segment contribute to the dimer interface (Lee et al., 2008; Qiu et al., 2013; Hong et al., 2015). To enable this to occur, the outer ends of S4 were proposed to relax or unwind (Hong et al., 2015). General conclusions and remaining questions A large number of mutations result in functional proton channels. This might be inferred from the fact that there are sequence differences at most positions in HV1 among different species (Smith et al., 2011). Perhaps not surprisingly, mutations that produce nonfunctional protein tend to occur at highly conserved positions and often involve changes in charge. Both Arg208 mutation (Table 6) and Asp112 mutation (Table 3) are severely detrimental to HV1 function, which we ascribe to both their central location and essential roles in proton permeation (Kulleperuma et al., 2013; Morgan et al., 2013; Dudev et al., 2015). Several examples exist in which HV1 channels apparently behave differently in mammalian and amphibian expression systems. Whether these discrepancies reflect differences in as yet unknown posttranslational modification can only be speculated. Charge, countercharge, and contra-countercharge Electrostatics is clearly of central importance in ion channel function. Voltage sensing almost certainly involves charged groups within the membrane electrical field. The idea that interaction of charge pairs (oppositely charged) helps stabilize open or closed states is well entrenched among enthusiasts of HV1 as well as other voltage-gated ion channels (Papazian et al., 1995; Tiwari-Woodruff et al., 1997). It seems reasonable to conclude that mutations that neutralize a charge will preclude this element from performing this function. The extensive study by Ramsey et al. (2010) together with other table entries is consistent with the interpretation that Asp112, Glu119, Asp123, and Asp185 in the outer vestibule stabilize the open state; internally accessible acidic groups (Glu153 and Asp174) more strongly stabilize the closed state. Interacting external and internal charge clusters have been observed consistently in MD simulations of HV1 (Ramsey et al., 2010; Kulleperuma et al., 2013; Morgan et al., 2013; Chamberlin et al., 2014). However, large shifts of the gH–V relationship can also be observed upon replacing one uncharged residue with another one (e.g., F150W in Table 4). In fact, almost every mutation that has been examined changes Vthreshold one way or the other. Pless et al. (2011) showed convincingly that two highly conserved acids in the Shaker K+ channel VSD (equivalent to Glu153 and Asp174 in hHV1; Table 1) could be neutralized with little effect on the gK–V relationship. Four external and two internal acidic residues were identified here as counterchanges based on the observation that in neutral mutants the gH–V relationship shifted in the predicted direction; thus, by this definition, the mechanism appears to involve charge. Alternatively, these charged groups might function primarily to create an aqueous vestibule and consequently a highly focused electrical field (Pless et al., 2011). We cannot resolve the mechanism without strategically designed experiments. If the function of the acidic residues is to create aqueous vestibules, one might predict that polar substituents that replace charges should produce less drastic effects than nonpolar ones. Mining the tables, we find that existing data are neither extensive nor self-consistent enough to provide a clear answer. The reported gH–V relationship shifts (polar vs. nonpolar) are as follows: Asp112 (31, 23, 13, and 25 vs. 59, 41, and 44); Asp185 (36 vs. 58, 20, 42, and 76); Glu153 (−117 vs. −55, −55, and −101), and Asp174 (−142 and −136 vs. −111 and −111). Another point is that Lys157 is predicted by MD to be involved in an internal salt bridge in closed HV1 (Chamberlin et al., 2014), yet its neutralization has little effect on the gH–V relationship (Table 4). This question clearly requires future study. Outstanding remaining problems It seems surprising that the N terminus contains several sites that influence gating, despite being topologically remote from S4 (at least in the image in Fig. 1). It also seems paradoxical that truncation of the entire N terminus has less overt effect than point mutations within the N terminus. However, the N terminus is disordered, and its tertiary structure is unknown. How it interacts with the rest of the molecule is completely unknown. For example, how does phosphorylation of Thr29 alter gating so profoundly? Why does the innocuous-appearing M91T mutation impede hHV1 activation? Voltage gating is a difficult structure–function problem because it is dynamic, not to mention three-dimensional. Despite Cys scanning and other studies designed to examine state-dependent accessibility and molecular movement, it remains unclear which parts of the HV1 molecule move during channel opening (S4, S1?), in which direction, and how far. For example, accessibility may change as a result of movement “up” or “down,” or into or out of a hydrophobic region, but it can also change simply by expansion or contraction of an aqueous vestibule, or by the helix rotating toward or away from an aqueous space. Although much has been learned about selectivity and a model has been proposed (Dudev et al., 2015), questions still remain. Is the model correct? Do other parts of the channel contribute to selectivity? Where is the rate-limiting point of the permeation pathway? Without a doubt, the most important question remains completely unsolved; namely, the mechanism of ΔpH-dependent gating. This unique ΔpH dependence is crucial to all of the known biological functions of HV1. ACKNOWLEDGMENTS We appreciate helpful corrections and comments provided by I. Scott Ramsey and Robert H. Fillingame. This work was supported by National Institutes of Health grant GM102336 and National Science Foundation grant MCB-1242985. The authors declare no competing financial interests. Author contributions: T.E. DeCoursey wrote the paper. D. Morgan, B. Musset, and V.V. Cherny analyzed existing data for Tables 3 and 5. All authors read and approved the manuscript. Lesley C. Anson served as editor. Abbreviations used: EPR electron paramagnetic resonance MD molecular dynamics NADPH nicotinamide adenine dinucleotide phosphate NEM n-ethylmaleimide TM transmembrane VSD voltage-sensing domain ==== Refs Alabi, A.A., M.I. Bahamonde, H.J. Jung, J.I. Kim, and K.J. Swartz. 2007. Portability of paddle motif function and pharmacology in voltage sensors. Nature. 450 :370–375. 10.1038/nature06266 18004375 Almers, W. 1978. Gating currents and charge movements in excitable membranes. Rev. Physiol. Biochem. Pharmacol. 82 :96–190.356157 Balashov, S.P., L.E. Petrovskaya, E.S. Imasheva, E.P. Lukashev, A.K. Dioumaev, J.M. Wang, S.V. Sychev, D.A. Dolgikh, A.B. Rubin, M.P. Kirpichnikov, and J.K. Lanyi. 2013. Breaking the carboxyl rule: lysine 96 facilitates reprotonation of the Schiff base in the photocycle of a retinal protein from Exiguobacterium sibiricum. J. Biol. Chem. 288 :21254–21265. 10.1074/jbc.M113.465138 23696649 Bánfi, B., J. Schrenzel, O. Nüsse, D.P. Lew, E. Ligeti, K.H. Krause, and N. Demaurex. 1999. A novel H+ conductance in eosinophils: Unique characteristics and absence in chronic granulomatous disease. J. Exp. Med. 190 :183–194. 10.1084/jem.190.2.183 10432282 Berger, T.K., and E.Y. Isacoff. 2011. The pore of the voltage-gated proton channel. Neuron. 72 :991–1000. 10.1016/j.neuron.2011.11.014 22196334 Bezanilla, F., and C.A. Villalba-Galea. 2013. The gating charge should not be estimated by fitting a two-state model to a Q-V curve. J. Gen. Physiol. 142 :575–578. 10.1085/jgp.201311056 24218396 Byerly, L., R. Meech, and W. Moody Jr. 1984. Rapidly activating hydrogen ion currents in perfused neurones of the snail, Lymnaea stagnalis. J. Physiol. 351 :199–216. 10.1113/jphysiol.1984.sp015241 6086903 Capasso, M., M.K. Bhamrah, T. Henley, R.S. Boyd, C. Langlais, K. Cain, D. Dinsdale, K. Pulford, M. Khan, B. Musset, 2010. HVCN1 modulates BCR signal strength via regulation of BCR-dependent generation of reactive oxygen species. Nat. Immunol. 11 :265–272. 10.1038/ni.1843 20139987 Capasso, M., T.E. DeCoursey, and M.J.S. Dyer. 2011. pH regulation and beyond: unanticipated functions for the voltage-gated proton channel, HVCN1. Trends Cell Biol. 21 :20–28. 10.1016/j.tcb.2010.09.006 20961760 Chamberlin, A., F. Qiu, S. Rebolledo, Y. Wang, S.Y. Noskov, and H.P. Larsson. 2014. Hydrophobic plug functions as a gate in voltage-gated proton channels. Proc. Natl. Acad. Sci. USA. 111 :E273–E282. 10.1073/pnas.1318018111 24379371 Chamberlin, A., F. Qiu, Y. Wang, S.Y. Noskov, and H.P. Larsson. 2015. Mapping the gating and permeation pathways in the voltage-gated proton channel Hv1. J. Mol. Biol. 427 :131–145. 10.1016/j.jmb.2014.11.018 25481746 Chaves, G., C. Derst, A. Franzen, Y. Mashimo, R. Machida, and B. Musset. 2016. Identification of an HV1 voltage-gated proton channel in insects. FEBS J. 283 :1453–1464. 10.1111/febs.13680 26866814 Chen, K., J. Hirst, R. Camba, C.A. Bonagura, C.D. Stout, B.K. Burgess, and F.A. Armstrong. 2000. Atomically defined mechanism for proton transfer to a buried redox centre in a protein. Nature. 405 :814–817. 10.1038/35015610 10866206 Cherny, V.V., and T.E. DeCoursey. 1999. pH-dependent inhibition of voltage-gated H+ currents in rat alveolar epithelial cells by Zn2+ and other divalent cations. J. Gen. Physiol. 114 :819–838. 10.1085/jgp.114.6.819 10578017 Cherny, V.V., V.S. Markin, and T.E. DeCoursey. 1995. The voltage-activated hydrogen ion conductance in rat alveolar epithelial cells is determined by the pH gradient. J. Gen. Physiol. 105 :861–896. 10.1085/jgp.105.6.861 7561747 Cherny, V.V., L.M. Henderson, and T.E. DeCoursey. 1997. Proton and chloride currents in Chinese hamster ovary cells. Membr. Cell Biol. 11 :337–347.9460053 Cherny, V.V., L.M. Henderson, W. Xu, L.L. Thomas, and T.E. DeCoursey. 2001. Activation of NADPH oxidase-related proton and electron currents in human eosinophils by arachidonic acid. J. Physiol. 535 :783–794. 10.1111/j.1469-7793.2001.00783.x 11559775 Cherny, V.V., D. Morgan, B. Musset, G. Chaves, S.M.E. Smith, and T.E. DeCoursey. 2015. Tryptophan 207 is crucial to the unique properties of the human voltage-gated proton channel, hHV1. J. Gen. Physiol. 146 :343–356. 10.1085/jgp.201511456 26458876 Cornish, A.J., K. Gärtner, H. Yang, J.W. Peters, and E.L. Hegg. 2011. Mechanism of proton transfer in [FeFe]-hydrogenase from Clostridium pasteurianum. J. Biol. Chem. 286 :38341–38347. 10.1074/jbc.M111.254664 21900241 Courtenay, E.S., M.W. Capp, and M.T. Record Jr. 2001. Thermodynamics of interactions of urea and guanidinium salts with protein surface: relationship between solute effects on protein processes and changes in water-accessible surface area. Protein Sci. 10 :2485–2497. 10.1110/ps.ps.20801 11714916 DeCoursey, T.E. 1991. Hydrogen ion currents in rat alveolar epithelial cells. Biophys. J. 60 :1243–1253. 10.1016/S0006-3495(91)82158-0 1722118 DeCoursey, T.E. 2003 a. Interactions between NADPH oxidase and voltage-gated proton channels: why electron transport depends on proton transport. FEBS Lett. 555 :57–61. 10.1016/S0014-5793(03)01103-7 14630319 DeCoursey, T.E. 2003 b. Voltage-gated proton channels and other proton transfer pathways. Physiol. Rev. 83 :475–579. 10.1152/physrev.00028.2002 12663866 DeCoursey, T.E. 2010. Voltage-gated proton channels find their dream job managing the respiratory burst in phagocytes. Physiology (Bethesda). 25 :27–40. 10.1152/physiol.00039.2009 20134026 DeCoursey, T.E. 2012. Voltage-gated proton channels. Compr. Physiol. 2 :1355–1385.23798303 DeCoursey, T.E. 2013. Voltage-gated proton channels: molecular biology, physiology, and pathophysiology of the HV family. Physiol. Rev. 93 :599–652. 10.1152/physrev.00011.2012 23589829 DeCoursey, T.E. 2015 a. Structural revelations of the human proton channel. Proc. Natl. Acad. Sci. USA. 112 :13430–13431. 10.1073/pnas.1518486112 26466610 DeCoursey, T.E. 2015 b. The voltage-gated proton channel: a riddle, wrapped in a mystery, inside an enigma. Biochemistry. 54 :3250–3268. 10.1021/acs.biochem.5b00353 25964989 DeCoursey, T.E. 2016. The intimate and controversial relationship between voltage gated proton channels and the phagocyte NADPH oxidase. Immunol. Rev. In press. DeCoursey, T.E., and V.V. Cherny. 1993. Potential, pH, and arachidonate gate hydrogen ion currents in human neutrophils. Biophys. J. 65 :1590–1598. 10.1016/S0006-3495(93)81198-6 7506066 DeCoursey, T.E., and V.V. Cherny. 1994. Voltage-activated hydrogen ion currents. J. Membr. Biol. 141 :203–223. 10.1007/BF00235130 7528804 DeCoursey, T.E., and V.V. Cherny. 1996. Effects of buffer concentration on voltage-gated H+ currents: does diffusion limit the conductance? Biophys. J. 71 :182–193. 10.1016/S0006-3495(96)79215-9 8804602 DeCoursey, T.E., and V.V. Cherny. 1997. Deuterium isotope effects on permeation and gating of proton channels in rat alveolar epithelium. J. Gen. Physiol. 109 :415–434. 10.1085/jgp.109.4.415 9101402 DeCoursey, T.E., and V.V. Cherny. 1998. Temperature dependence of voltage-gated H+ currents in human neutrophils, rat alveolar epithelial cells, and mammalian phagocytes. J. Gen. Physiol. 112 :503–522. 10.1085/jgp.112.4.503 9758867 DeCoursey, T.E., and J. Hosler. 2014. Philosophy of voltage-gated proton channels. J. R. Soc. Interface. 11 :20130799. 10.1098/rsif.2013.0799 24352668 DeCoursey, T.E., V.V. Cherny, W. Zhou, and L.L. Thomas. 2000. Simultaneous activation of NADPH oxidase-related proton and electron currents in human neutrophils. Proc. Natl. Acad. Sci. USA. 97 :6885–6889. 10.1073/pnas.100047297 10823889 DeCoursey, T.E., V.V. Cherny, A.G. DeCoursey, W. Xu, and L.L. Thomas. 2001 a. Interactions between NADPH oxidase-related proton and electron currents in human eosinophils. J. Physiol. 535 :767–781. 10.1111/j.1469-7793.2001.00767.x 11559774 DeCoursey, T.E., V.V. Cherny, D. Morgan, B.Z. Katz, and M.C. Dinauer. 2001 b. The gp91phox component of NADPH oxidase is not the voltage-gated proton channel in phagocytes, but it helps. J. Biol. Chem. 276 :36063–36066. 10.1074/jbc.C100352200 11477065 DeCoursey, T.E., D. Morgan, and V.V. Cherny. 2002. The gp91phox component of NADPH oxidase is not a voltage-gated proton channel. J. Gen. Physiol. 120 :773–779. 10.1085/jgp.20028704 12451047 DeCoursey, T.E., D. Morgan, and V.V. Cherny. 2003. The voltage dependence of NADPH oxidase reveals why phagocytes need proton channels. Nature. 422 :531–534. 10.1038/nature01523 12673252 Demaurex, N. 2012. Functions of proton channels in phagocytes. Wiley Interdiscip. Rev. Membr. Transp. Signal. 1 :3–15. 10.1002/wmts.2 Doroshenko, P.A., P.G. Kostyuk, and A.E. Martynyuk. 1986. Transmembrane outward hydrogen current in intracellularly perfused neurones of the snail Helix pomatia. Gen. Physiol. Biophys. 5 :337–350.3021566 Dudev, T., B. Musset, D. Morgan, V.V. Cherny, S.M.E. Smith, K. Mazmanian, T.E. DeCoursey, and C. Lim. 2015. Selectivity mechanism of the voltage-gated proton channel, HV1. Sci. Rep. 5 :10320. 10.1038/srep10320 25955978 Eder, C., and T.E. DeCoursey. 2001. Voltage-gated proton channels in microglia. Prog. Neurobiol. 64 :277–305. 10.1016/S0301-0082(00)00062-9 11240310 El Chemaly, A., Y. Okochi, M. Sasaki, S. Arnaudeau, Y. Okamura, and N. Demaurex. 2010. VSOP/Hv1 proton channels sustain calcium entry, neutrophil migration, and superoxide production by limiting cell depolarization and acidification. J. Exp. Med. 207 :129–139. 10.1084/jem.20091837 20026664 England, J.L., and G. Haran. 2011. Role of solvation effects in protein denaturation: from thermodynamics to single molecules and back. Annu. Rev. Phys. Chem. 62 :257–277. 10.1146/annurev-physchem-032210-103531 21219136 Fillingame, R.H., C.M. Angevine, and O.Y. Dmitriev. 2002. Coupling proton movements to c-ring rotation in F1Fo ATP synthase: aqueous access channels and helix rotations at the a-c interface. Biochim. Biophys. Acta. 1555 :29–36. 10.1016/S0005-2728(02)00250-5 12206887 Fischer, H. 2012. Function of proton channels in lung epithelia. Wiley Interdiscip. Rev. Membr. Transp. Signal. 1 :247–258. 10.1002/wmts.17 22662311 Fujiwara, Y., T. Kurokawa, K. Takeshita, M. Kobayashi, Y. Okochi, A. Nakagawa, and Y. Okamura. 2012. The cytoplasmic coiled-coil mediates cooperative gating temperature sensitivity in the voltage-gated H+ channel Hv1. Nat. Commun. 3 :816. 10.1038/ncomms1823 22569364 Fujiwara, Y., T. Kurokawa, K. Takeshita, A. Nakagawa, H.P. Larsson, and Y. Okamura. 2013 a. Gating of the designed trimeric/tetrameric voltage-gated H+ channel. J. Physiol. 591 :627–640. 10.1113/jphysiol.2012.243006 23165764 Fujiwara, Y., K. Takeshita, A. Nakagawa, and Y. Okamura. 2013 b. Structural characteristics of the redox-sensing coiled coil in the voltage-gated H+ channel. J. Biol. Chem. 288 :17968–17975. 10.1074/jbc.M113.459024 23667254 Fujiwara, Y., T. Kurokawa, and Y. Okamura. 2014. Long α helices projecting from the membrane as the dimer interface in the voltage-gated H+ channel. J. Gen. Physiol. 143 :377–386. 10.1085/jgp.201311082 24567511 Gonzalez, C., H.P. Koch, B.M. Drum, and H.P. Larsson. 2010. Strong cooperativity between subunits in voltage-gated proton channels. Nat. Struct. Mol. Biol. 17 :51–56. 10.1038/nsmb.1739 20023639 Gonzalez, C., S. Rebolledo, M.E. Perez, and H.P. Larsson. 2013. Molecular mechanism of voltage sensing in voltage-gated proton channels. J. Gen. Physiol. 141 :275–285. 10.1085/jgp.201210857 23401575 Henderson, L.M. 1998. Role of histidines identified by mutagenesis in the NADPH oxidase-associated H+ channel. J. Biol. Chem. 273 :33216–33223. 10.1074/jbc.273.50.33216 9837891 Henderson, L.M., and J.B. Chappell. 1992. The NADPH-oxidase-associated H+ channel is opened by arachidonate. Biochem. J. 283 :171–175. 10.1042/bj2830171 1373602 Henderson, L.M., and J.B. Chappell. 1996. NADPH oxidase of neutrophils. Biochim. Biophys. Acta. 1273 :87–107. 10.1016/0005-2728(95)00140-9 8611594 Henderson, L.M., and R.W. Meech. 1999. Evidence that the product of the human X-linked CGD gene, gp91-phox, is a voltage-gated H+ pathway. J. Gen. Physiol. 114 :771–786. 10.1085/jgp.114.6.771 10578014 Henderson, L.M., and R.W. Meech. 2002. Proton conduction through gp91phox. J. Gen. Physiol. 120 :759–765. 10.1085/jgp.20028708 12451045 Henderson, L.M., J.B. Chappell, and O.T.G. Jones. 1987. The superoxide-generating NADPH oxidase of human neutrophils is electrogenic and associated with an H+ channel. Biochem. J. 246 :325–329. 10.1042/bj2460325 2825632 Henderson, L.M., J.B. Chappell, and O.T.G. Jones. 1988. Superoxide generation by the electrogenic NADPH oxidase of human neutrophils is limited by the movement of a compensating charge. Biochem. J. 255 :285–290.2848506 Henderson, L.M., G. Banting, and J.B. Chappell. 1995. The arachidonate-activable, NADPH oxidase-associated H+ channel. Evidence that gp91-phox functions as an essential part of the channel. J. Biol. Chem. 270 :5909–5916. 10.1074/jbc.270.11.5909 7890722 Henderson, L.M., S. Thomas, G. Banting, and J.B. Chappell. 1997. The arachidonate-activatable, NADPH oxidase-associated H+ channel is contained within the multi-membrane-spanning N-terminal region of gp91-phox. Biochem. J. 325 :701–705. 10.1042/bj3250701 9271091 Hodgkin, A.L., and A.F. Huxley. 1952. A quantitative description of membrane current and its application to conduction and excitation in nerve. J. Physiol. 117 :500–544. 10.1113/jphysiol.1952.sp004764 12991237 Hondares, E., M.A. Brown, B. Musset, D. Morgan, V.V. Cherny, C. Taubert, M.K. Bhamrah, D. Coe, F. Marelli-Berg, J.G. Gribben, 2014. Enhanced activation of an amino-terminally truncated isoform of the voltage-gated proton channel HVCN1 enriched in malignant B cells. Proc. Natl. Acad. Sci. USA. 111 :18078–18083. 10.1073/pnas.1411390111 25425665 Hong, L., M.M. Pathak, I.H. Kim, D. Ta, and F. Tombola. 2013. Voltage-sensing domain of voltage-gated proton channel Hv1 shares mechanism of block with pore domains. Neuron. 77 :274–287. 10.1016/j.neuron.2012.11.013 23352164 Hong, L., I.H. Kim, and F. Tombola. 2014. Molecular determinants of Hv1 proton channel inhibition by guanidine derivatives. Proc. Natl. Acad. Sci. USA. 111 :9971–9976. 10.1073/pnas.1324012111 24912149 Hong, L., V. Singh, H. Wulff, and F. Tombola. 2015. Interrogation of the intersubunit interface of the open Hv1 proton channel with a probe of allosteric coupling. Sci. Rep. 5 :14077. 10.1038/srep14077 26365828 Iovannisci, D., B. Illek, and H. Fischer. 2010. Function of the HVCN1 proton channel in airway epithelia and a naturally occurring mutation, M91T. J. Gen. Physiol. 136 :35–46. 10.1085/jgp.200910379 20548053 Jimenez-Morales, D., J. Liang, and B. Eisenberg. 2012. Ionizable side chains at catalytic active sites of enzymes. Eur. Biophys. J. 41 :449–460. 10.1007/s00249-012-0798-4 22484856 Johns, S.J. 2016. TOPO2. Transmembrane protein display software. http://www.sacs.ucsf.edu/TOPO2 (accessed April 21, 2016). Kalia, J., and K.J. Swartz. 2011. Elucidating the molecular basis of action of a classic drug: guanidine compounds as inhibitors of voltage-gated potassium channels. Mol. Pharmacol. 80 :1085–1095. 10.1124/mol.111.074989 21926190 Kang, B.E., and B.J. Baker. 2016. Pado, a fluorescent protein with proton channel activity can optically monitor membrane potential, intracellular pH, and map gap junctions. Sci. Rep. 6 :23865. 10.1038/srep23865 27040905 Kapus, A., R. Romanek, and S. Grinstein. 1994. Arachidonic acid stimulates the plasma membrane H+ conductance of macrophages. J. Biol. Chem. 269 :4736–4745.7508928 Kawanabe, A., and Y. Okamura. 2016. Effects of unsaturated fatty acids on the kinetics of voltage-gated proton channels heterologously expressed in cultured cells. J. Physiol. 594 :595–610. 10.1113/JP271274 26563684 Killian, J.A., and G. von Heijne. 2000. How proteins adapt to a membrane-water interface. Trends Biochem. Sci. 25 :429–434. 10.1016/S0968-0004(00)01626-1 10973056 Koch, H.P., T. Kurokawa, Y. Okochi, M. Sasaki, Y. Okamura, and H.P. Larsson. 2008. Multimeric nature of voltage-gated proton channels. Proc. Natl. Acad. Sci. USA. 105 :9111–9116. 10.1073/pnas.0801553105 18583477 Kulleperuma, K., S.M.E. Smith, D. Morgan, B. Musset, J. Holyoake, N. Chakrabarti, V.V. Cherny, T.E. DeCoursey, and R. Pomès. 2013. Construction and validation of a homology model of the human voltage-gated proton channel hHV1. J. Gen. Physiol. 141 :445–465. 10.1085/jgp.201210856 23530137 Kuno, M., H. Ando, H. Morihata, H. Sakai, H. Mori, M. Sawada, and S. Oiki. 2009. Temperature dependence of proton permeation through a voltage-gated proton channel. J. Gen. Physiol. 134 :191–205. 10.1085/jgp.200910213 19720960 Kurokawa, T., and Y. Okamura. 2014. Mapping of sites facing aqueous environment of voltage-gated proton channel at resting state: a study with PEGylation protection. Biochim. Biophys. Acta. 1838 :382–387. 10.1016/j.bbamem.2013.10.001 24140009 Lacroix, J.J., H.C. Hyde, F.V. Campos, and F. Bezanilla. 2014. Moving gating charges through the gating pore in a Kv channel voltage sensor. Proc. Natl. Acad. Sci. USA. 111 :E1950–E1959. 10.1073/pnas.1406161111 24782544 Landolt-Marticorena, C., K.A. Williams, C.M. Deber, and R.A. Reithmeier. 1993. Non-random distribution of amino acids in the transmembrane segments of human type I single span membrane proteins. J. Mol. Biol. 229 :602–608. 10.1006/jmbi.1993.1066 8433362 Lecar, H., H.P. Larsson, and M. Grabe. 2003. Electrostatic model of S4 motion in voltage-gated ion channels. Biophys. J. 85 :2854–2864. 10.1016/S0006-3495(03)74708-0 14581190 Lee, S.Y., J.A. Letts, and R. Mackinnon. 2008. Dimeric subunit stoichiometry of the human voltage-dependent proton channel Hv1. Proc. Natl. Acad. Sci. USA. 105 :7692–7695. 10.1073/pnas.0803277105 18509058 Letts, J.A. 2014. Functional and structural studies of the human voltage-gated proton channel. PhD thesis. The Rockefeller University, New York. 209 pp. Li, Q., S. Wanderling, M. Paduch, D. Medovoy, A. Singharoy, R. McGreevy, C.A. Villalba-Galea, R.E. Hulse, B. Roux, K. Schulten, 2014. Structural mechanism of voltage-dependent gating in an isolated voltage-sensing domain. Nat. Struct. Mol. Biol. 21 :244–252. 10.1038/nsmb.2768 24487958 Li, Q., R. Shen, J.S. Treger, S.S. Wanderling, W. Milewski, K. Siwowska, F. Bezanilla, and E. Perozo. 2015. Resting state of the human proton channel dimer in a lipid bilayer. Proc. Natl. Acad. Sci. USA. 112 :E5926–E5935. 10.1073/pnas.1515043112 26443860 Li, S.J., Q. Zhao, Q. Zhou, H. Unno, Y. Zhai, and F. Sun. 2010. The role and structure of the carboxyl-terminal domain of the human voltage-gated proton channel Hv1. J. Biol. Chem. 285 :12047–12054. 10.1074/jbc.M109.040360 20147290 Lishko, P.V., I.L. Botchkina, A. Fedorenko, and Y. Kirichok. 2010. Acid extrusion from human spermatozoa is mediated by flagellar voltage-gated proton channel. Cell. 140 :327–337. 10.1016/j.cell.2009.12.053 20144758 Lishko, P.V., Y. Kirichok, D. Ren, B. Navarro, J.J. Chung, and D.E. Clapham. 2012. The control of male fertility by spermatozoan ion channels. Annu. Rev. Physiol. 74 :453–475. 10.1146/annurev-physiol-020911-153258 22017176 Luoto, H.H., E. Nordbo, A.A. Baykov, R. Lahti, and A.M. Malinen. 2013. Membrane Na+-pyrophosphatases can transport protons at low sodium concentrations. J. Biol. Chem. 288 :35489–35499. 10.1074/jbc.M113.510909 24158447 Mahaut-Smith, M.P. 1989. The effect of zinc on calcium and hydrogen ion currents in intact snail neurones. J. Exp. Biol. 145 :455–464.22912993 Makhatadze, G.I., and P.L. Privalov. 1992. Protein interactions with urea and guanidinium chloride: A calorimetric study. J. Mol. Biol. 226 :491–505. 10.1016/0022-2836(92)90963-K 1322462 Maturana, A., K.H. Krause, and N. Demaurex. 2002. NOX family NADPH oxidases: Do they have built-in proton channels? J. Gen. Physiol. 120 :781–786. 10.1085/jgp.20028713 12451048 Miller, M.J., M. Oldenburg, and R.H. Fillingame. 1990. The essential carboxyl group in subunit c of the F1F0 ATP synthase can be moved and H+-translocating function retained. Proc. Natl. Acad. Sci. USA. 87 :4900–4904. 10.1073/pnas.87.13.4900 2142302 Mony, L., T.K. Berger, and E.Y. Isacoff. 2015. A specialized molecular motion opens the Hv1 voltage-gated proton channel. Nat. Struct. Mol. Biol. 22 :283–290. 10.1038/nsmb.2978 25730777 Morgan, D., V.V. Cherny, A. Finnegan, J. Bollinger, M.H. Gelb, and T.E. DeCoursey. 2007. Sustained activation of proton channels and NADPH oxidase in human eosinophils and murine granulocytes requires PKC but not cPLA2 α activity. J. Physiol. 579 :327–344. 10.1113/jphysiol.2006.124248 17185330 Morgan, D., M. Capasso, B. Musset, V.V. Cherny, E. Ríos, M.J.S. Dyer, and T.E. DeCoursey. 2009. Voltage-gated proton channels maintain pH in human neutrophils during phagocytosis. Proc. Natl. Acad. Sci. USA. 106 :18022–18027. 10.1073/pnas.0905565106 19805063 Morgan, D., B. Musset, K. Kulleperuma, S.M.E. Smith, S. Rajan, V.V. Cherny, R. Pomès, and T.E. DeCoursey. 2013. Peregrination of the selectivity filter delineates the pore of the human voltage-gated proton channel hHV1. J. Gen. Physiol. 142 :625–640. 10.1085/jgp.201311045 24218398 Murphy, R., and T.E. DeCoursey. 2006. Charge compensation during the phagocyte respiratory burst. Biochim. Biophys. Acta. 1757 :996–1011. 10.1016/j.bbabio.2006.01.005 16483534 Musset, B., V.V. Cherny, D. Morgan, Y. Okamura, I.S. Ramsey, D.E. Clapham, and T.E. DeCoursey. 2008 a. Detailed comparison of expressed and native voltage-gated proton channel currents. J. Physiol. 586 :2477–2486. 10.1113/jphysiol.2007.149427 18356202 Musset, B., D. Morgan, V.V. Cherny, D.W. MacGlashan Jr., L.L. Thomas, E. Ríos, and T.E. DeCoursey. 2008 b. A pH-stabilizing role of voltage-gated proton channels in IgE-mediated activation of human basophils. Proc. Natl. Acad. Sci. USA. 105 :11020–11025. 10.1073/pnas.0800886105 18664579 Musset, B., V.V. Cherny, D. Morgan, and T.E. DeCoursey. 2009. The intimate and mysterious relationship between proton channels and NADPH oxidase. FEBS Lett. 583 :7–12. 10.1016/j.febslet.2008.12.005 19084015 Musset, B., M. Capasso, V.V. Cherny, D. Morgan, M. Bhamrah, M.J.S. Dyer, and T.E. DeCoursey. 2010 a. Identification of Thr29 as a critical phosphorylation site that activates the human proton channel Hvcn1 in leukocytes. J. Biol. Chem. 285 :5117–5121. 10.1074/jbc.C109.082727 20037153 Musset, B., S.M.E. Smith, S. Rajan, V.V. Cherny, S. Sujai, D. Morgan, and T.E. DeCoursey. 2010 b. Zinc inhibition of monomeric and dimeric proton channels suggests cooperative gating. J. Physiol. 588 :1435–1449. 10.1113/jphysiol.2010.188318 20231140 Musset, B., S.M.E. Smith, S. Rajan, V.V. Cherny, D. Morgan, and T.E. DeCoursey. 2010 c. Oligomerization of the voltage-gated proton channel. Channels (Austin). 4 :260–265. 10.4161/chan.4.4.12789 20676047 Musset, B., S.M.E. Smith, S. Rajan, D. Morgan, V.V. Cherny, and T.E. DeCoursey. 2011. Aspartate 112 is the selectivity filter of the human voltage-gated proton channel. Nature. 480 :273–277. 10.1038/nature10557 22020278 Okamura, Y., Y. Fujiwara, and S. Sakata. 2015. Gating mechanisms of voltage-gated proton channels. Annu. Rev. Biochem. 84 :685–709. 10.1146/annurev-biochem-060614-034307 26034892 Okuda, H., Y. Yonezawa, Y. Takano, Y. Okamura, and Y. Fujiwara. 2016. Direct interaction between the voltage sensors produces cooperative sustained deactivation in voltage-gated H+ channel dimers. J. Biol. Chem. 291 :5935–5947. 10.1074/jbc.M115.666834 26755722 Papazian, D.M., X.M. Shao, S.A. Seoh, A.F. Mock, Y. Huang, and D.H. Wainstock. 1995. Electrostatic interactions of S4 voltage sensor in Shaker K+ channel. Neuron. 14 :1293–1301. 10.1016/0896-6273(95)90276-7 7605638 Pless, S.A., J.D. Galpin, A.P. Niciforovic, and C.A. Ahern. 2011. Contributions of counter-charge in a potassium channel voltage-sensor domain. Nat. Chem. Biol. 7 :617–623. 10.1038/nchembio.622 21785425 Qiu, F., S. Rebolledo, C. Gonzalez, and H.P. Larsson. 2013. Subunit interactions during cooperative opening of voltage-gated proton channels. Neuron. 77 :288–298. 10.1016/j.neuron.2012.12.021 23352165 Ramsey, I.S., M.M. Moran, J.A. Chong, and D.E. Clapham. 2006. A voltage-gated proton-selective channel lacking the pore domain. Nature. 440 :1213–1216. 10.1038/nature04700 16554753 Ramsey, I.S., Y. Mokrab, I. Carvacho, Z.A. Sands, M.S.P. Sansom, and D.E. Clapham. 2010. An aqueous H+ permeation pathway in the voltage-gated proton channel Hv1. Nat. Struct. Mol. Biol. 17 :869–875. 10.1038/nsmb.1826 20543828 Robinson, R.A., and R.H. Stokes. 1959. Electrolyte Solutions, the Measurement and Interpretation of Conductance, Chemical Potential, and Diffusion in Solutions of Simple Electrolytes. Second edition. Butterworths Scientific Publications, London. 559 pp. Rodriguez, J.D., S. Haq, K.F. Nowak, D. Morgan, V.V. Cherny, S. Bernstein, M.S. Sapp, J.R. Curcuru, C. Antchouey, S.J. Nowak, 2015. Characterization and subcellular localization of HV1 in Lingulodinium polyedrum confirms its role in bioluminescence. Biophys. J. 108 :425a. 10.1016/j.bpj.2014.11.2325 Ruivo, R., G.C. Bellenchi, X. Chen, G. Zifarelli, C. Sagné, C. Debacker, M. Pusch, S. Supplisson, and B. Gasnier. 2012. Mechanism of proton/substrate coupling in the heptahelical lysosomal transporter cystinosin. Proc. Natl. Acad. Sci. USA. 109 :E210–E217. 10.1073/pnas.1115581109 22232659 Sakata, S., T. Kurokawa, M.H. Nørholm, M. Takagi, Y. Okochi, G. von Heijne, and Y. Okamura. 2010. Functionality of the voltage-gated proton channel truncated in S4. Proc. Natl. Acad. Sci. USA. 107 :2313–2318. 10.1073/pnas.0911868107 20018719 Sasaki, M., M. Takagi, and Y. Okamura. 2006. A voltage sensor-domain protein is a voltage-gated proton channel. Science. 312 :589–592. 10.1126/science.1122352 16556803 Seredenina, T., N. Demaurex, and K.H. Krause. 2015. Voltage-gated proton channels as novel drug targets: From NADPH oxidase regulation to sperm biology. Antioxid. Redox Signal. 23 :490–513. 10.1089/ars.2013.5806 24483328 Sigg, D., and F. Bezanilla. 1997. Total charge movement per channel. The relation between gating charge displacement and the voltage sensitivity of activation. J. Gen. Physiol. 109 :27–39. 10.1085/jgp.109.1.27 8997663 Smith, S.M.E., and T.E. DeCoursey. 2013. Consequences of dimerization of the voltage-gated proton channel. Prog. Mol. Biol. Transl. Sci. 117 :335–360. 10.1016/B978-0-12-386931-9.00012-X 23663974 Smith, S.M.E., D. Morgan, B. Musset, V.V. Cherny, A.R. Place, J.W. Hastings, and T.E. DeCoursey. 2011. Voltage-gated proton channel in a dinoflagellate. Proc. Natl. Acad. Sci. USA. 108 :18162–18167. 10.1073/pnas.1115405108 22006335 Starace, D.M., and F. Bezanilla. 2001. Histidine scanning mutagenesis of basic residues of the S4 segment of the Shaker K+ channel. J. Gen. Physiol. 117 :469–490. 10.1085/jgp.117.5.469 11331357 Starace, D.M., and F. Bezanilla. 2004. A proton pore in a potassium channel voltage sensor reveals a focused electric field. Nature. 427 :548–553. 10.1038/nature02270 14765197 Suszták, K., A. Mócsai, E. Ligeti, and A. Kapus. 1997. Electrogenic H+ pathway contributes to stimulus-induced changes of internal pH and membrane potential in intact neutrophils: role of cytoplasmic phospholipase A2. Biochem. J. 325 :501–510. 10.1042/bj3250501 9230134 Takeshita, K., S. Sakata, E. Yamashita, Y. Fujiwara, A. Kawanabe, T. Kurokawa, Y. Okochi, M. Matsuda, H. Narita, Y. Okamura, and A. Nakagawa. 2014. X-ray crystal structure of voltage-gated proton channel. Nat. Struct. Mol. Biol. 21 :352–357. 10.1038/nsmb.2783 24584463 Tao, X., A. Lee, W. Limapichat, D.A. Dougherty, and R. MacKinnon. 2010. A gating charge transfer center in voltage sensors. Science. 328 :67–73. 10.1126/science.1185954 20360102 Taylor, A.R., A. Chrachri, G. Wheeler, H. Goddard, and C. Brownlee. 2011. A voltage-gated H+ channel underlying pH homeostasis in calcifying coccolithophores. PLoS Biol. 9 :e1001085. 10.1371/journal.pbio.1001085 21713028 Taylor, A.R., C. Brownlee, and G.L. Wheeler. 2012. Proton channels in algae: reasons to be excited. Trends Plant Sci. 17 :675–684. 10.1016/j.tplants.2012.06.009 22819465 Thomas, R.C., and R.W. Meech. 1982. Hydrogen ion currents and intracellular pH in depolarized voltage-clamped snail neurones. Nature. 299 :826–828. 10.1038/299826a0 7133121 Thorndycroft, F.H., G. Butland, D.J. Richardson, and N.J. Watmough. 2007. A new assay for nitric oxide reductase reveals two conserved glutamate residues form the entrance to a proton-conducting channel in the bacterial enzyme. Biochem. J. 401 :111–119. 10.1042/BJ20060856 16961460 Tiwari-Woodruff, S.K., C.T. Schulteis, A.F. Mock, and D.M. Papazian. 1997. Electrostatic interactions between transmembrane segments mediate folding of Shaker K+ channel subunits. Biophys. J. 72 :1489–1500. 10.1016/S0006-3495(97)78797-6 9083655 Tombola, F., M.H. Ulbrich, and E.Y. Isacoff. 2008. The voltage-gated proton channel Hv1 has two pores, each controlled by one voltage sensor. Neuron. 58 :546–556. 10.1016/j.neuron.2008.03.026 18498736 Tombola, F., M.H. Ulbrich, S.C. Kohout, and E.Y. Isacoff. 2010. The opening of the two pores of the Hv1 voltage-gated proton channel is tuned by cooperativity. Nat. Struct. Mol. Biol. 17 :44–50. 10.1038/nsmb.1738 20023640 Touret, N., and S. Grinstein. 2002. Voltage-gated proton “channels”: a spectator’s viewpoint. J. Gen. Physiol. 120 :767–771. 10.1085/jgp.20028706 12451046 Wang, Y., S.J. Li, X. Wu, Y. Che, and Q. Li. 2012. Clinicopathological and biological significance of human voltage-gated proton channel Hv1 protein overexpression in breast cancer. J. Biol. Chem. 287 :13877–13888. 10.1074/jbc.M112.345280 22367212 Wang, Y., X. Wu, Q. Li, S. Zhang, and S.J. Li. 2013. Human voltage-gated proton channel hv1: a new potential biomarker for diagnosis and prognosis of colorectal cancer. PLoS One. 8 :e70550. 10.1371/journal.pone.0070550 23940591 Wood, M.L., E.V. Schow, J.A. Freites, S.H. White, F. Tombola, and D.J. Tobias. 2012. Water wires in atomistic models of the Hv1 proton channel. Biochim. Biophys. Acta. 1818 :286–293. 10.1016/j.bbamem.2011.07.045 21843503 Xu, Q., H.L. Axelrod, E.C. Abresch, M.L. Paddock, M.Y. Okamura, and G. Feher. 2004. X-Ray structure determination of three mutants of the bacterial photosynthetic reaction centers from Rb. sphaeroides; altered proton transfer pathways. Structure. 12 :703–715. 10.1016/j.str.2004.03.001 15062092 Yang, N., A.L. George Jr., and R. Horn. 1996. Molecular basis of charge movement in voltage-gated sodium channels. Neuron. 16 :113–122. 10.1016/S0896-6273(00)80028-8 8562074 Yang, N., A.L. George Jr., and R. Horn. 1997. Probing the outer vestibule of a sodium channel voltage sensor. Biophys. J. 73 :2260–2268. 10.1016/S0006-3495(97)78258-4 9370423

PMC005xxxxxx/PMC5037345.txt

==== Front J Gen Physiol J Gen Physiol jgp jgp The Journal of General Physiology 0022-1295 1540-7748 The Rockefeller University Press 27619418 201611625 10.1085/jgp.201611625 Reviews Viewpoint 508 509 501 Pore size matters for potassium channel conductance Pore size determines K+ channel conductance http://orcid.org/0000-0003-3482-5126 Naranjo David 1 Moldenhauer Hans 1 Pincuntureo Matías 14 Díaz-Franulic Ignacio 123 1 Centro Interdisciplinario de Neurociencia de Valparaíso, Universidad de Valparaíso, Playa Ancha, Valparaíso 2360103, Chile 2 Center for Bioinformatics and Integrative Biology, Universidad Andrés Bello, Santiago 8370146, Chile 3 Fraunhofer Chile Research, Las Condes 7550296, Chile 4 Programa de Doctorado en Ciencias, mención Biofísica y Biología Computacional, Universidad de Valparaíso, Valparaíso 2360103, Chile Correspondence to David Naranjo: david.naranjo@uv.cl; or Ignacio Díaz-Franulic: ignacio.diaz@cinv.cl 10 2016 148 4 277291 19 5 2016 10 8 2016 © 2016 Naranjo et al. 2016 https://creativecommons.org/licenses/by-nc-sa/3.0/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 3.0 Unported license, as described at http://creativecommons.org/licenses/by-nc-sa/3.0/). Ion channels are membrane proteins that mediate efficient ion transport across the hydrophobic core of cell membranes, an unlikely process in their absence. K+ channels discriminate K+ over cations with similar radii with extraordinary selectivity and display a wide diversity of ion transport rates, covering differences of two orders of magnitude in unitary conductance. The pore domains of large- and small-conductance K+ channels share a general architectural design comprising a conserved narrow selectivity filter, which forms intimate interactions with permeant ions, flanked by two wider vestibules toward the internal and external openings. In large-conductance K+ channels, the inner vestibule is wide, whereas in small-conductance channels it is narrow. Here we raise the idea that the physical dimensions of the hydrophobic internal vestibule limit ion transport in K+ channels, accounting for their diversity in unitary conductance. Fondo Nacional de Desarrollo Científico y Tecnológico 10.13039/501100002850 #1120819 Fondecyt 10.13039/501100002850 #3160321 Ministerio de Economía, Fomento y Turismo 10.13039/501100005886 ==== Body pmcIntroduction That the charged nature of K+ ions impairs their free movement across the plasma membrane derives from elementary physics. The calculation of the Born self-energy for K+ within the low dielectric constant of the membrane shows the nonspontaneity of this process (Parsegian, 1969). Nevertheless, in K+ channels, nature found a low-energy mechanism to move K+ ions across the plasma membrane by developing proteins able to mimic its water coordination (Zhou et al., 2001). Thanks to these membrane proteins, K+ is the most permeable ion in resting cells, and because K+ is also the most abundant intracellular ion, the resting membrane potential in most living cells is close to the Nernst potential for K+ (Hodgkin and Huxley, 1952). K+ channels are probably an ancient protein family and are present in every living being (Armstrong, 2015). These membrane proteins belong to one of the biggest gene families, with ∼90 representatives in the mammalian genome (Yu et al., 2005). Their physiological role is widespread: they guard the resting membrane potential, stabilize osmotic imbalance, set the excitability threshold in excitable membranes, and shape the neuronal action potential (Hille, 2001; Armstrong, 2015). K+ channels are endowed with an unsurpassed architectural mechanism that allows K+ ions to permeate selectively across the cell membrane. However, they show a wide variability in unitary conductance (or ion transport rate), which spans approximately two orders of magnitude when measured under similar experimental conditions. In this viewpoint, we propose that the structural determinants for selectivity and conductance are segregated to two structures within the pore of K+ channels: the selectivity filter and the internal vestibule, respectively. We raise the idea that the structure of the selectivity filter seems to be so dedicated to selective and efficient K+ transport that it is unlikely to be the structural determinant of conductance diversity. On the contrary, the physical dimensions of the hydrophobic inner vestibule seem to be the factors that limit K+ transport, accounting for the difference in unitary conductance among K+ channels. The structure of the K+ permeation pathway K+ channels allow selective passage of K+ ions, thermodynamically lured to flow against their own electrochemical gradient, to the exclusion of all other physiological cations. K+ channels select for K+ over Na+ by almost 1,000-fold, a surprising task considering a difference of <0.5 Å between the ionic radii of these two cations (Hille, 1973). Nevertheless, because the difference in their hydration energies is ∼16 kcal/mol, removal of the hydration water should be ∼1010 harder for Na+ (Robinson and Stokes, 2002). Thus, the simplest explanation for the high K+ selectivity should involve, at least in part, the need for partial dehydration of the ion, excluding Na+ because replacing its hydration waters is energetically costlier (Bezanilla and Armstrong, 1972). It has also been argued that the binding sites within the selectivity pore must be precisely shaped around a partially dehydrated K+ so that it fits snugly (Mullins, 1959). The selectivity sequence of K+ channels for alkali metal cations (K+ ≈ Rb+ > Cs+ > Na+ > Li+) indicates that K+ channel permeation is biased against larger hydration energy and larger size, as expected for a relatively “low-field-strength” site (Eisenman, 1962). Thus, Bezanilla and Armstrong (1972) postulated “Na+ ions do not enter the narrower part of the pore because they are too small to fit well in the coordination cages provided by the pore as replacements for the water molecules surrounding an ion.” Because the observed selectivity sequence is virtually identical among K+ channels, it anticipates a highly conserved ion selectivity structure (Latorre and Miller, 1983; Heginbotham and MacKinnon, 1993). The crystallographic structure of the KcsA bacterial K+ channel resolved at 3.2 Å by Doyle et al. (1998) revealed for the first time how the pore of a K+ channel looks (Fig. 1, A and B). The protein has a tetrameric organization around the pore placed in its axis of symmetry. Although the structure corresponded to that of a closed channel, it showed several, previously anticipated, functional features: (a) The pore hosts several K+ ions in single-file order as Hodgkin and Keynes predicted 60 years ago (Hodgkin and Keynes, 1955). (b) The pore has a narrow selectivity filter located toward the external entrance and is flanked internally by a wider internal vestibule as anticipated by Armstrong and Bezanilla (Armstrong, 1971; Bezanilla and Armstrong, 1972; Miller, 1982; Latorre and Miller, 1983). (c) K+ ions are partially hydrated in the narrow section of the pore, as Mullins, Bezanilla, and Armstrong proposed half a century ago (Mullins, 1959; Armstrong, 1971; Bezanilla and Armstrong, 1972; Hille, 1973). Although the structural analysis could not resolve interatomic bond orientation, it was hypothesized that carbonyl oxygens from the signature sequence in the peptide backbone, TVGYG, shape the anticipated low-field-strength K+ binding sites by forming surrogate hydration cages in the filter (Eisenman, 1962; Heginbotham et al., 1994). The presence of these expected features in a single crystallographic structure gave this study immediate acceptance. Later on, crystallization of KcsA channels with improved resolution (2.2 Å) provided a detailed picture of K+ ions and their carbonyl cages in the selectivity filter (Fig. 1, C and D). At the internal and external entrances, K+ ions are fully or partially hydrated, whereas those located inside the filter fit perfectly into the four carbonyl-lined binding sites of the selectivity filter (Zhou et al., 2001). Constrained by the K+/water 1:1 stoichiometry flux ratio, determined from streaming potentials by Alcayaga et al. (1989), it was proposed that the four ion-binding sites at the selectivity filter are energetically equivalent for K+ in alternate occupancy of sites 1–3 and 2–4, with intervening waters at the vacant sites (Bernèche and Roux, 2000; Morais-Cabral et al., 2001). This arrangement makes near to zero the energy cost to put two K+ ions inside the selectivity filter. In contrast, a solitary K+ would not be able to permeate measurably because it would be too tightly bound (Neyton and Miller, 1988; Liu and Lockless, 2013). In contrast, double occupancy in sites separated by ∼7 Å (either sites 1 and 3 or 2 and 4) in the selectivity filter affords enough electrostatic repulsion to allow efficient ion translocation along the pore (Åqvist and Luzhkov, 2000; Morais-Cabral et al., 2001). Figure 1. Structural features of the KcsA channel and K+ coordination structure in the pore. (A and B) Membrane-omitted side and top views of the KcsA K+ channel (PDB ID 1K4C). Each monomer is a two–transmembrane segment peptide position around the pore at the axis of fourfold symmetry forming the K+ selective pore (green spheres). (C) High-resolution electronic density map showing the two diagonal subunits and the orientation of the carbonyl oxygen atoms to coordinate K+ ions. The numbers correspond to the four binding sites determined by the sequence TVGYG. (D) Antiprism and cubic cages forming the selectivity filter binding sites, the distances d1–d5 and heights h1–h4 correspond to the inter-oxygen separations described in Table 1 for several K+ channel structures. A and B were inspired by Doyle et al. (1998), C was modified from Zhou et al. (2001) with permission from Macmillan Publishers Ltd., and D was inspired by Chen et al. (2014). The geometry of cation coordination in the selectivity filter The K+ ions along the selectivity filter are coordinated in a square prism fashion (Fig. 1, C and D), with eight carbonyls groups each contributing a binding site in the selectivity filter, four on top and four below the cation. For sites 1, 2, and 3, the top four carbonyls are rotated ∼45°, forming a squared antiprism with vertices separated by 3–4 Å, whereas a cube encases site 4. Notably, the K+ located internally to the selectivity filter appears to coordinate eight water molecules also in antiprism fashion, and the most external K+ is coordinated on top by four waters as if these two cations were caught “in flagrante” getting stripped from their waters before entering the selectivity filter (Miller, 2001; Zhou et al., 2001). Neutron scattering, spectroscopy, statistical mechanical, and molecular dynamic simulation converge on a mean center to center distance between the K+ ion and the oxygen atoms of the hydration shell of ∼2.6–2.8 Å (Enderby, 1995; Glezakou et al., 2006; Mancinelli et al., 2007; Bankura et al., 2013). Such a distance matches the center to center separation between K+ and the carbonyl’s oxygen atoms in the selectivity filter. The mean K+ coordination geometry in solution is unknown; however, using the above K+-O separation of 2.6–2.8 Å, it is possible to calculate a vertex to vertex distance of 3.0–3.2 Å in a squared hydration cage. These distances fit well for most of the selectivity filter cages in Table 1. Moreover, a cube, or a squared antiprism, formed by a cage composed of eight water molecules separated by 3.0–3.2 Å would fill a volume of 195–220 Å3, which is the volume of a 3.6–3.8-Å-radius sphere, consistent with the estimated hydrodynamic radius of K+ (Díaz-Franulic et al., 2015; Moldenhauer et al., 2016). Thus, from the geometrical point of view, the oxygen cages can be regarded as a surrogate water cages. Moreover, this argument favors a homotetrameric structure in K+ channels as a requisite for highly selective K+ binding sites, as was proposed by Zhou et al. (2001). Table 1. O–O distances at the edges of the water surrogating cages in the selectivity filter, Å Selectivity filter edge Structure (resolution) KcsA 1K4C (2.0 Å) Kv1.2/2.1 2R9R (2.4 Å) MthK 4HYO (1.65 Å) KvAP 1ORQ (3.2 Å) d1 3.6 3.5 3.5 4.1 d2 3.3 3.3 3.2 3.9 d3 3.3 3.3 3.3 3.9 d4 3.2 3.3 3.1 3.1 d5 3.7 3.9 3.6 3.4 h1 3.1 3.1 3.1 2.9 h2 3.0 3.0 3.0 2.7 h3 3.1 3.1 3.2 3.3 h4 3.0 2.9 2.9 2.9 Center to center inter-oxygen distances at the selectivity filter edges shown in Fig. 1 D. Distance calculations are directly from the Biological Assembly PDB file coordinates. Distances di result from the construction of the Biological Assembly. We chose the above structures because they showed the best resolution available at the time. The rigidity of the K+ selective filter Electrophysiological, structural, and calorimetry studies support the equilibrium-binding hypothesis for ionic selectivity in K+ channels: selectivity is attained by making the energy wells along the reaction coordinate deeper for K+ than for Na+ (Eisenman, 1962; Neyton and Miller, 1988; Zhou et al., 2001; Piasta et al., 2011; Liu et al., 2012). In light of the conserved structure of the selectivity filter among different channels (Fig. 1, C and D; and Table 1), we are tempted to consider this structure as static and immutable. Nevertheless, functional evidence and intuitive thinking indicate that this temptation is dangerous. Proteins are flexible, and the carbonyl atoms of the selectivity filter fluctuate around 0.4 Å root mean square (Allen et al., 2004), nearly the difference among Na+ and K+ ionic radii, suggesting that the snug-fit mechanism requires some amendments (Allen et al., 2004; Noskov et al., 2004). Functionally speaking, selectivity filter flexibility is evidenced in, for instance, C-type inactivation of Kv channels. C-type inactivation is a Na+-permeable pore conformation in K+ channel usually triggered by the removal of external K+ or by long depolarizations (see, for example, Starkus et al. [1997]). This phenomenon has also been seen in the KcsA channel, being associated with the loss of the second and third selectivity filter K+ binding sites (Cuello et al., 2010). Selectivity filter stability has been proposed to be important for K+ discrimination in KcsA, as mutations to surrounding residues (E71; Fig. 1 C) increase Na+ permeability (Cheng et al., 2011). Thus, a complex network of interactions must support the functional integrity of the selectivity filter. In fact, despite their identical signature sequence and ion distribution along the KcsA and MthK channels selectivity filter, the pore of KcsA, but not MthK, collapses upon K+ removal (Morais-Cabral et al., 2001; Ye et al., 2010). Thus, the selectivity filter emerges as a dynamic structure, able to adopt a collection of stable conformations, of which the fully conducting ones seem to be also the most often crystallized. The bacterial nonselective NaK channel, which allows permeation of both K+ and Na+, has contributed significantly to our understanding of ionic selectivity in K+ channels. The NaK channel selectivity filter possesses the sequence TVGDG instead of the canonical TVGYG, failing to show S1 and S2 ion-binding sites (Alam and Jiang, 2009). Variants having the number of binding sites restored to four are K+ selective, indicating that the ability to make transitions between 1,3 and 2,4 configurations is essential for K+-selective permeation, revealing a strong marriage between selectivity and conductance (Derebe et al., 2011; Liu and Lockless, 2013; Sauer et al., 2013). Excellent reviews covering the mechanisms of ion selectivity in rich detail are available (Noskov and Roux, 2006; Nimigean and Allen, 2011; Lockless, 2015). We excluded the inward rectifier family from consideration in this review. Although these channels share with BK and Kvs the selectivity filter and the internal vestibule structures, unlike the latter, their tetramerization domain contributes significantly to ion conduction, adding several K+-binding sites in series with the pore. Their conduction mechanism looks more complex, and their structural determinants for unitary conductance may be different from Kv and BK channels (Nishida and MacKinnon, 2002; Lu, 2004). Pore architecture of K+ channels: Implications for ion permeation Selectivity filters among different K+ channels are so similar that we hypothesized that, regardless of their specific unitary conductance, they should be equally competent to allow permeation of K+ at high rates. However, different K+ channels display large differences in their unitary conductance measured under similar experimental conditions. Permeation should occur at very high rates because the energy cost of putting two K+ ions in the selectivity filter is to zero (Åqvist and Luzhkov, 2000; Morais-Cabral et al., 2001; Zhou et al., 2001; Bernèche and Roux, 2003; Allen et al., 2004; Jensen et al., 2013). Accordingly, in standard recording solutions, the selectivity filter accounts for, at most, 3 GΩ (333 pS) of the total pore resistance in both small- and large-conductance K+ channels (Díaz-Franulic et al., 2015). Direct estimation of electrical resistance, or electrostatic field calculation along the pore, showed that 3 GΩ accounts for 50–90% of the total resistance of BK or MthK channels (Morais-Cabral et al., 2001; Contreras et al., 2010; Díaz-Franulic et al., 2015). In contrast, in Shaker Kv channels, such resistance accounts for only ∼10% of the total (Díaz-Franulic et al., 2015). Thus, in small-conductance Kv channels, ion transport must be limited in structures that are separate from the selectivity filter (Moscoso et al., 2012). Table 2 compares the amino acidic sequences of several K+-selective channels between the pore helix and the C terminus of S6, where the structural determinants for single-channel conductance reside. It is apparent that, although K+ channels share identical selectivity filter sequences, their unitary conductance ranges from 5 to 270 pS and, accordingly, they fit into two categories: small and large conductance (top and bottom groups in Table 2). Sequence diversity in both groups leads to the idea that the determinants of unitary conductance are located at or near the internal cavity. Table 2. Single-channel conductance in selected K+ channels Accession no. Gene Protein Pore helix-S6 sequence pS Reference Pore helix Selectivity filter Pore-S6 Small conductance NP_728123.1 SHAKER Shaker AFWWAVVTMT TVGYGD MTPVGVWGKIVGSLCAIAGVLTIALPV P VIVSNFNYFYHR 487 20 Carvacho et al., 2008 NP_000208.2 KCNA1 Kv1.1 AFWWAVVSMT TVGYGD MYPVTIGGKIVGSLCAIAGVLTIALPV P VIVSNFNYFYHR 417 10 Gutman et al., 2005 NP_004965.1 KCNA2 Kv1.2 AFWWAVVSMT TVGYGD MVPTTIGGKIVGSLCAIAGVLTIALPV P VIVSNFNYFYHR 419 14 Carvacho et al., 2008 AAC31761.1 KCNA3 Kv1.3 AFWWAVVTMT TVGYGD MHPVTIGGKIVGSLCAIAGVLTIALPV P VIVSNFNYFYHR 737 13 Gutman et al., 2005 NP_002224.1 KCNA4 Kv1.4 AFWWAVVTMT TVGYGD MKPITVGGKIVGSLCAIAGVLTIALPV P VIVSNFNYFYHR 569 5 Gutman et al., 2005 AAA61276.1 KCNA5 Kv1.5 AFWWAVVTMT TVGYGD MRPITVGGKIVGSLCAIAGVLTIALPV P VIVSNFNYFYHR 523 10 Carvacho et al., 2008 NP_002226.1 KCNA6 Kv1.6 AFWWAVVTMT TVGYGD MYPMTVGGKIVGSLCAIAGVLTIALPV P VIVSNFNYFYHR 467 9 Gutman et al., 2005 AAX11186.1 KCNA7 Kv1.7 SFWWAVVTMT TVGYGD MAPVTVGGKIVGSLCAIAGVLTISLPV P VIVSNFSYFYHR 403 21 Carvacho et al., 2008 NP_005540.1 KCA10 Kv1.8 GFWWAVVTMT TVGYGD MCPTTPGGKIVGTLCAIAGVLTIALPV P VIVSNFNYFYHR 466 12 Gutman et al., 2005 NP_004761.2 KCNB2 Kv2.2 SFWWATITMT TVGYGD IYPKTLLGKIVGGLCCIAGVLVIALPI P IIVNNFSEFYKE 426 15 Carvacho et al., 2008 Large conductance P0A334.1 KCSA KcsA ALWWSVETAT TVGYGD LYPVTLWGRLVAVVVMVAGITSFGLVT A ALATWFVGREQE 120 100 Nimigean et al., 2003 Q9YDF8.1 KVAP KvAP ALWWAVVTAT TVGYGD VVPATPIGKVIGIAVMLTGISALTLLI G TVSNMFQKILVG 253 170 Ruta et al., 2003 O27564.1 MTHK MthK SLYWTFVTIA TVGYGD YSPSTPLGMYFTVTLIVLGIGTFAVAV E RLLEFLINREQM 104 200 Shi et al., 2011 NP_001240307.1 KCMA1 MSlo1 CVYLLMVTMS TVGYGD VYAKTTLGRLFMVFFILGGLAMFASYV P EIIELIGNRKKY 332 270 Brelidze et al., 2003 Multiple alignment of the primary structure of K+ channel pores. The signature sequence TVGYGD on the selectivity filter, used as reference for the alignments, is separated for clarity. Shaker’s Gly466 and aligning residues are underlined. Shaker’s 475 and aligning residues in Kv channels are individualized for clarity. Aligning with Shaker 475 are KcsA’s Ala108, KvAP’s Gly241, MthK’s Glu92 and MSlo1’s Pro320. Aspartates and glutamates are in bold (table modified from Moscoso et al. [2012] with permission from Elsevier). Architecture of BK and MthK channels BK channels (also known as Slo1 or Maxi-K) display the largest unitary conductance among K+ channels, ranging from ∼250 pS under standard recording conditions (100–150 mM K+) to 600 pS under saturating K+ concentrations (Eisenman et al., 1986; Brelidze et al., 2003). Sequence analysis of BK channels reveals two conserved glutamate rings (Glu rings) at the internal entrance, comprising Glu321 and Glu324 in each subunit of MSlo (Fig. 2 A). In 100 mM K+, these eight negatively charged residues double unitary conductance (Brelidze et al., 2003; Zhang et al., 2006). However, such increments vanish at saturating K+ concentration, indicating that the charged rings contribute to channel conductance by attracting cations and are not an essential part of the efficient K+ transport mechanism (Fig. 2 B). Therefore, maximum conductance, measured at saturating K+ concentrations, is required to separate permeation from binding (Díaz-Franulic et al., 2015; Sack and Tilley, 2015). In contrast, Phe380, located at the inner cavity of HSlo (F315 in MSlo), was shown to be critical for ion permeation when replacement with isoleucine or tyrosine decreased unitary conductance by ∼70% or ∼50%, respectively (Carrasquel-Ursulaez et al., 2015). However, the impact of these mutations on the maximum conductance is unknown. Figure 2. Occupancy and dimensions of the internal cavity and the maximum conductance of Kv and BK channel. (A) Alignment of the Kv1.2/2.1 chimera (gray backbone) with MthK (blue) pore domain structures. Shown are the diagonal subunits, with front and back subunits omitted for clarity. K+/water in the selectivity filter are green spheres. The side chains shown on CPK color correspond to the glutamate ring equivalent positions (right for MthK and left for Kv1.2/2.1). The residues in yellow are the internal cavity residues able to tolerate aspartate substitution. Kv2.1/1.2 chimera: PDB ID 2R9R; MthK: PDB ID 4HYO. (B) Role of charged residues in the inner cavity on maximum conductance of Shaker and BK. Maximal conductance is defined as the unitary conductance at saturating K+ concentration. The internal entrance of BK channels seems wider than in Kv channels. The association rate of internally applied quaternary ammonium to the pore is higher than for Kv channels (Li and Aldrich, 2004; Wilkens and Aldrich, 2006). Also, overlapping dual cysteine modification in BK channels revealed an inner vestibule wider than that of Kv channels (Zhou et al., 2011). Thus, a wide, and possibly short, vestibule could afford a low-resistance pathway for ion permeation in BK channels. This suggests that the dimension of the internal pore entrance limits ionic conduction, such that the presence of large side chain amino acids at the inner entrance reduce BK’s unitary conductance, whereas smaller side chain substitutions have little effect (Geng et al., 2011). In the absence of crystallographic data of BK in the open conformation, the bacterial calcium activated MthK K+ channel has been validated as a bona fide structural model for the pore domain of BK channels (Geng et al., 2011; Shi et al., 2011; Posson et al., 2013; Moldenhauer et al., 2016). MthK is a large-conductance two–transmembrane segment (TM) channel coupled to a calcium gating ring similar to that of BK (Jiang et al., 2002; Yuan et al., 2011). Consistent with BK’s guessed internal pore dimensions, the MthK structure displays a wide ∼15-Å internal entrance (Fig. 2 A), lined with hydrophobic residues (Table 2). A direct functional assessment of BK’s pore architecture came from Magleby’s laboratory with measurements of the “radius of capture” (Brelidze and Magleby, 2005). If the pore is assumed to be a hemispheric sink into which approaching ions vanish, it is possible to use diffusional collisions theory to infer the dimension of the pore entrance in a condition where the rate-limiting step for ion transport is the diffusion of K+ ions into the entrance of the pore (Läuger, 1976; Andersen, 1983). Assuming K+ is a rigid sphere approaching the mouth of the channel, the radius of capture corresponds to the difference between the radius of the ion and the radius of the pore entrance (Ferry, 1936; Läuger, 1976) asrP=rC+ri,(1) where rP, rC, and ri are the pore, the capture, and the ion radii, respectively. If the ion is a point charge (ri = 0), the radius of capture is the same as the radius of the pore. Experimentally, increasing the viscosity of the solution through the addition of high concentrations of sugar, the amplitude of K+ current becomes asymptotically voltage independent, revealing the limiting K+ diffusional access to the pore (Läuger, 1976; Brelidze and Magleby, 2005; Díaz-Franulic et al., 2015). In BK channels, the internal rC is 2.2 Å, a surprising, and debated, number suggesting that the pore is barely wide enough to fit a hydrated K+ ion (Brelidze and Magleby, 2005). Furthermore, we must consider the hydrodynamic dimensions of K+ to calculate the pore radius, but the size of hydrated K+ is not well defined because hydrating water molecules bind with dissimilar energies and lifetimes, forming a fuzzy arrangement. Then, one has to decide how many hydration shells should be added to the radius of the ion. The simplest assumption is to add just one water layer, but this seems arbitrary (Brelidze and Magleby, 2005). Let’s consider that rP, the pore radius (a functional estimate arising from the radius of capture), is equivalent to the “effective pore radius” (rE; a structural estimate defined as the radius of the largest sphere that is able to enter the pore cavity; Ferry, 1936). rE is very easy to estimate from the wealth of K+ channel structures available, and for MthK it is 5.7–5.9 Å (Moldenhauer et al., 2016). Thus, by replacing rE for rP in Eq. 1, we obtain ri = 3.5–3.7 Å for K+. These figures are consistent with K+ ions carrying only one hydration layer (Enderby, 1995; Glezakou et al., 2006; Mancinelli et al., 2007; Bankura et al., 2013). Bearing in mind that the radius of capture is measured at room temperature, in liquid phase, and with operative friction forces, the harmonious outcome with structural studies, made at very low temperature, in solid phase, and in equilibrium, is satisfying. Architecture of Shaker and Kv channels We have seen that side chain volume and charge of residues at the internal entrance of large-conductance channels restrict K+ conductance (Brelidze et al., 2003; Nimigean et al., 2003; Geng et al., 2011). Thus, K+ channels having narrower pores should display smaller unitary conductance. Among the small-conductance K+ channels the most extensively studied is the Drosophila melanogaster Shaker K+ channel. As expected, Shaker ionic selectivity is lost upon mutation of some of the selectivity filter residues (Heginbotham et al., 1994). Shaker’s unitary conductance ranges from ∼20 pS in ∼100 mM K+ to 60 pS in saturating 3,000 mM K+ (Heginbotham and MacKinnon, 1993; Díaz-Franulic et al., 2015). Intra- and intersubunit metal coordination at the internal entrance suggests that this region is narrower than that in BK or MthK and/or that the opening conformational change implies a small physical displacement (Webster et al., 2004; del Camino et al., 2005). Consistent with a narrow pore, the crystal structure of Kv1.2, a mammalian homologue of Shaker, shows an internal entrance of ∼10 Å, which, as occurs in MthK, is lined with hydrophobic side chains (Long et al., 2005a, 2007). Thus, the narrow Kv internal cavity is just large enough to let ions go across, and “…may help to explain why Shaker has an approximately tenfold lower conductance than its bacterial relatives” (Webster et al., 2004). As with BK, measurements of diffusion limited outward currents allowed us to estimate that Shaker’s cytosolic radius of capture was ∼0.8 Å, ∼1.4 Å narrower than BK’s (Díaz-Franulic et al., 2015). As seen with MthK/BK in section Architecture of BK and MthK channels, this estimate of rC is consistent with the effective opening of the open Kv1.2/2.1 chimera if we assume that the hydrodynamic radius for K+ is ∼3.5–4.4 Å (Díaz-Franulic et al., 2015; Moldenhauer et al., 2016). As hypothesized, such a narrow cavity contributes decisively to the total electrical resistance of the pore. Estimations of the sectional resistance along the pore of Shaker showed that, in contrast to BK, the inner cavity entails an electrical resistance ∼55–fold larger (or 1/55 lower conductance) than that of BK channels (Díaz-Franulic et al., 2015). Thus, these results reinforce the idea that the size of the internal vestibule severely limits single-channel conductance of small-conductance K+ channels. The internal pore dimensions and K+ channels cytosolic structure For both Kv and MthK structures, the effective entrance is consistent with the estimates of the radius of capture of Shaker and BK channels, respectively, when the K+ hydrodynamic radius is taken as ∼4 Å (Díaz-Franulic et al., 2015; Moldenhauer et al., 2016). However, the canonical Kv channel activation bundle crossing gate, located at the lower end of S6, appears to be missing in BK channels. Instead, these K channels may gate the ion pathway near the selectivity filter (Wilkens and Aldrich [2006], Garg et al. [2013], and Posson et al. [2013]; but see Hite et al. [2015]). Thus, the estimates of radius of capture could represent the pore dimensions at different depths in the permeation pathway. Because the estimates of the hydrodynamic size of K+ produced by both comparisons (MthK with BK and Kv1.2 with Shaker) are so similar, the possible discrepancy in deepness should not account for more than 1 Å in radius. In addition, the resulting size of K+ agrees well with estimations for the hydrated K+ in aqueous solutions obtained using unrelated techniques (Enderby, 1995; Glezakou et al., 2006; Mancinelli et al., 2007; Bankura et al., 2013). There is one more problem with the estimation of the internal pore dimension based on the radius of capture for BK and Kv channels. These measurements disregard the contribution of important intracellular domains in both proteins, leading to an oversimplified structural image (Long et al., 2005a; Hite et al., 2015). Both BK and Kv channels have tetramerization domains near the internal entrance (Kobertz and Miller, 1999; Krishnamoorthy et al., 2005; Hite et al., 2015). In Kv channels, the four tetramerization domains (T1) form a structure known as “the hanging gondola” because it hangs from the pore domain through four linkers, leaving four side-facing openings for ion transport (Fig. 3; Kobertz and Miller, 1999; Long et al., 2007). Meanwhile, the structure of Slo2.2, a BK channel, shows the calcium-dependent gating ring forming a funnel structure that could guide ions into the pore cavity (Hite et al., 2015). Both structures project negative electrostatic potential into the permeation pathway, raising local K+ concentration, but contribute, at most, to 30% to the unitary current (Kobertz and Miller, 1999; Budelli et al., 2013; Hite et al., 2015). Figure 3. K+ channel structural topology. Surface representation of Kv (Kv1.2/2.1 chimera; left) and BK (Slo2.2; right) structures. Green and yellow colors represent the voltage-sensing domain (VSD) and the pore domain (PD), respectively. The green arrows show the putative conduction paths for ions that in Kv channels K+ access/exit through the lateral windows of the “hanging” T1 domain (the gondola in cyan), whereas in BK the ions cross the entire gating ring formed by the RCK domains (in cyan). The pink spheres are K+ ions, and the horizontal discontinued lines indicate the approximate inner and outer boundaries of the membrane. External side is up. The Kv figure is a 6-Å slab prepared with VMD, with −6 < x < 0 (Humphrey et al., 1996). BK front and rear subunits are removed for clarity (inspired by Hite et al. [2015]). Why does a narrow pore cause larger resistance? How is it possible that the internal cavity of Kv channels, with a radius only 1.4 Å smaller than that of BK, causes a 55-fold increase in resistance? The seminal paper by Parsegian (1969) predicted that the energy required to put an ion inside an aqueous pore embedded in the low dielectric membrane could decrease by up to 28 kcal/mol for every Å increase in sectional radius. Thus, a mere 1-Å narrower pore would represent a much larger energy barrier for the ion to overcome. Energy calculations show that K+ transport in the nonpolar nanotube membrane model is highly sensitive to the dimensions of the pore. In narrow pores (<5-Å opening diameter) ion transport cannot occur, but wider pores (∼10-Å opening diameter) are highly permeable, with ionic mobilities comparable with those seen in bulk solution (Peter and Hummer, 2005). In contrast, narrow pores would strip off solvation waters, paying a large energetic penalty; in contrast, wider pores would keep the first hydration shell intact. The free energy calculated for K+ ions inside the cavity of K+ channels is indeed a few kcal lower in the wider parts of the pore (Chung et al., 2002; Jogini and Roux, 2005; Treptow and Tarek, 2006). Thus, higher resistances may result from the reduced probability of finding an ion inside narrower pores, and, conversely, we expect that decreasing the energy required to put an ion inside the cavity will result in an increased conductance (see section Turning Shaker into a large-conductance K+ channel below). We have to keep in mind that these considerations involve using equilibrium energies to describe the nonequilibrium phenomenon that is ion transport. Nevertheless, because diffusion and water turnover are orders of magnitude faster than permeation, an equilibrium approximation should be at least partially satisfactory (Hille, 2001; Grossfield, 2005; Mancinelli et al., 2007). Because ion conduction occurs away from equilibrium, we should also consider the action of frictional forces on unitary conductance. We are required to acknowledge that ions must interact physically with their surroundings (Eisenberg, 2013). Also, molecular dynamics simulations have shown drastic reductions of ion and water diffusion coefficients inside wide pores (Wilson et al., 2011; Zhu and Hummer, 2012b). As expected for friction, such reduction seems to be caused by the pore shape and wall tortuosities. Because of the lack of room inside the inner vestibule of Kv channels, fully or partially hydrated ions would be bumping into the walls of the pore. In contrast, in BK channels, with a 1.4-Å wider cavity, there is enough room for a loosely attached extra layer of water molecules interfacing with the walls of the inner pore. In principle, these interfacial waters should bear low mobility (because of their low entropy); nevertheless, recent molecular dynamic simulations show highly mobile and chaotic interfacial waters in narrow nanotubes organic-aqueous contact (Garate et al., 2014). Such loosely attached water molecules may lubricate the passage of the K+ cage, working as ball bearing sliders. Turning Shaker into a large-conductance K+ channel In agreement with the equilibrium energy hypothesis, lowering the K+ energy inside the cavity is expected to increase unitary conductance, by increasing pore occupancy. Indeed, introducing negatively charged residues at Shaker’s internal entrance (position Pro475) increases unitary conductance by eightfold (Sukhareva et al., 2003; Moscoso et al., 2012). Such an increment was correlated with the appearance of a cytosolically accessible K+ site that could enhance Mg2+ blockade by preventing its exit toward the cytosol. Molecular dynamics simulations showed a largely increased K+ occupancy and Mg2+ binding at the inner cavity. Two distal K+-binding sites flanked Mg2+ exit (Moscoso et al., 2012). Because of the deep impact of mutations of Pro475 on channel gating (Hackos et al., 2002), it was quite possible that P475D had enlarged pore size, accounting for such an increased conductance. However, the radius of capture was found to be identical to that of Shaker-WT (Díaz-Franulic et al., 2015). This is in satisfying agreement with the Parsegian hypothesis because it shows that it is possible to raise conductance by several-fold just by increasing pore occupancy, i.e., by lowering the work needed to put an ion in the cavity. Several other charges added to the internal vestibule (471D, 476D, and 479D) also increased unitary conductance; nevertheless, no additional change could increase maximum conductance beyond 200 pS, which is one third of BK’s conductance (Díaz-Franulic et al., 2015). Part of the other two thirds of the difference in unitary conductance between Kv and BK channels might have different causes: fluctuations of the selectivity filter or friction. In the former case, differences in the selectivity sequence should be expected (Starkus et al., 1997; Allen et al., 2004; Cheng et al., 2011). Electromechanical coupling and permeation determinants map to the same hotspot Voltage-activated K+ channels are impressive examples of evolutionary functional tuning. On one hand, the addition of several sensing modules to the seemingly permeation-optimized voltage- and calcium-activated BK pore domain gives them the ability to integrate complex stimuli into transmembrane K+ fluxes. On the other hand, voltage sensitivity of Kv channels seems to be near the maximal possible, limited only by the capacity of the voltage-sensing domain to host more charged residues without disrupting the trans-protein electric field (Ahern and Horn, 2004; Tombola et al., 2007; González-Pérez et al., 2010). The voltage-sensing domain is energetically linked to the cytosolic end of S6 in the pore domain, mostly through the S4-S5 linker. Open probability increases 20-fold the for every 6-mV depolarization (Aggarwal and MacKinnon, 1996; Seoh et al., 1996; Islas and Sigworth, 1999; Lu et al., 2002; Long et al., 2005b; Batulan et al., 2010; González-Pérez et al., 2010). In the Shaker Kv channel, residues located toward the internal end of S6 are important for both permeation and electromechanical coupling (Ding and Horn, 2002; Hackos et al., 2002; Sukhareva et al., 2003; Batulan et al., 2010). Because the electromechanical coupling is tight, mutual interference between permeant cations and channel gating is expected. In fact, we’ve known for a long time that Rb+ and K+ slow channel closure in the squid axon and in oocytes, resembling a “foot in the door” mechanism, suggesting the existence of a cation-selective site near the inner end of the pore (Swenson and Armstrong, 1981; Matteson and Swenson, 1986). Such a site was supported by the observation that K+ ions applied internally lock in quaternary ammonia or Mg2+ (Thompson and Begenisich, 2001; Moscoso et al., 2012). Thus, tampering with pore occupancy to increase conductance also affects voltage sensitivity (Sukhareva et al., 2003; Moscoso et al., 2012). Is it possible that the reduced unitary conductance of Kv channels is a consequence of the tight functional coupling between the voltage sensor and the pore domain? Faure et al. (2012) proposed an ∼4-Å radial displacement during the opening transition based on luminescence resonance energy transfer (LRET) measurements of the bacterial Kv channel, KvAP. This displacement is about the same as the effective radial difference between closed and open pore structures of KcsA and Kv1.2/2.1, respectively (Moldenhauer et al., 2016). Molecular dynamics simulations of Kv1.2-2.1 under an external electric field show channel openings as small conformational changes at the activation gate. Ensuing pore hydration drives the channel toward its conductive state (Jensen et al., 2012). This process, coined “hydrophobic gating,” has been reported for several other ion channels. Small, not physically occluded, pores reside in a de-wetted state because water molecules refuse to interact with their hydrophobic linings (Beckstein et al., 2003; Anishkin and Sukharev, 2004; Jensen et al., 2010, 2012; Zhu and Hummer, 2012a; Neale et al., 2015). The critical radius for pore hydration is ∼4–5 Å, very close to the radius of the internal entrance in the Kv1.2/2.1 structure (Beckstein et al., 2003; Webster et al., 2004; Long et al., 2005b; Wang et al., 2008; Díaz-Franulic et al., 2015). Thus, according to these considerations, pore opening in Kv channels might be barely wide enough to allow permeation. The closed–open transitions seem to be ruled by economical conformational changes in other ion channels too. In fact, several ion channels (K+-CNG and TRP) switch between open and closed states with physically small conformational changes in the inner cavity and/or at the selectivity filter (Flynn and Zagotta, 2001; Bruening-Wright et al., 2002; Proks et al., 2003; Salazar et al., 2009; Rapedius et al., 2012; Cao et al., 2013; Garg et al., 2013; Liao et al., 2013; Posson et al., 2013). Although small movements during channel gating are not mandatory by any biophysical principle, they may possibly result from an evolutionary selective pressure to reduce the energy required to control the pore open probability. Assuming 25–33 cal/mol per Å2 of hydrocarbon exposed to water (Reynolds et al., 1974), a cylindrical greasy cavity of 15-Å depth and 5-Å radius requires 11–15 kcal/mol to open. This figure may be considered a lower limit for the work the electromechanic gear has to perform to open the pore of a Kv channel because it prefers to be closed (Yifrach and MacKinnon, 2002; Jensen et al., 2010, 2012). On the opposite sidewalk, a large-conductance channel such as BK, endowed with a wider pore, would require larger, energetically expensive gate movements (Beckstein and Sansom, 2004). To open the gate of a BK-like channel, a 7-Å radius pore would require 16–21 kcal/mol, that is 5–6 kcal/mol more than those required to open a Kv channel (Reynolds et al., 1974). If, as recent crystal structures suggest, the Slo2.2 BK channel primary gate is located at the internal bundle crossing, as in Kv channels (Hite et al., 2015), the extra energy required to open BK’s pore can be translated into a shift along the voltage axis. The voltage shift, ΔV, is ΔV = ΔΔG/zF, where ΔΔG is the difference between the energies required to open both pores, z is the effective valence of the voltage dependence, and F the Faraday constant. Interestingly, if we take BK’s voltage dependency as z = 1.5–2.0, the additional energy required to gate the wider pore would correspond to a positive 100–170-mV shift in the conductance-voltage relationship with respect to Kv channels. Aware that this calculation oversimplifies gating energetics, and assuming similar voltage-responsive gears, it is remarkable that such a voltage shift agrees well with the actual difference in half-activation voltage between BK and Kv channels (Díaz et al., 1998; Horrigan et al., 1999). In addition, other larger-conductance ion channels such as TRPV1 also exhibit similarly shifted conductance-voltage relationships (Matta and Ahern, 2007). Several larger-conductance and multimodal ion channels have evolved additional, physiologically relevant gates, distinct from those at the cytoplasmic end of the pore, which gate ion access with 1–2-Å movements (Flynn and Zagotta, 2001; Bruening-Wright et al., 2002; Proks et al., 2003; Salazar et al., 2009; Rapedius et al., 2012; Garg et al., 2013). This rationale of small movement gates is per se speculative and simplistic because there aren’t obvious biophysical or physiological constraints to limit pore gating to small conformational adjustments. Worse, x-ray structures have to be taken with certain skepticism because they are obtained under nonphysiological conditions. Nevertheless, we should bear in mind that, on the one hand, energy economy is not uncommon in the operation of K+ channels (for example, in cation coordination in the selectivity filter), and, on the other hand, smaller energy requirements to open the pore would permit a sensitive tuning of the channel open probability. The physiological role of unitary conductance is mostly unknown The large functional diversity and regulatory mechanisms of K+ channels surely underlie physiological fine-tuning within each cell type. These channels are present in almost every tissue, and most cells express different K+ channels. Possibly, because of this functional redundancy, K+ channel pathologies are rarely fatal, although they often seriously challenge an individual’s happiness. Finding the physiological significance for the spectrum of unitary conductances in K+ channels is also challenging because there is not a clear connection with cell type. In this regard, congenital channelopathy analysis could inform us about the physiological relevance of unitary conductance if we focus on missense mutations occurring at the internal end of S6, the functional hotspot for conductance and electromechanical coupling (Ding and Horn, 2002; Lu et al., 2002; Sukhareva et al., 2003; Long et al., 2005b; Posson et al., 2013). A myriad of mutations in S6, with pathological consequences, may potentially be affecting unitary conductance (Table 3), but in most cases, research has focused largely on the macroscopic phenotype, such as current density and kinetics, failing to explore microscopic electrophysiological behavior. For example, episodic ataxia type 1 (EA1; characterized by spells of incoordination and imbalance) is caused by mutations in the KCNA1 gene (Kv1.1). Likewise, cerebellar ataxia (CA), characterized by coordination imbalance, and epileptic encephalopathy early infantile 32 (EIEE32), characterized by refractory seizures and neurodevelopmental impairment, are both caused by mutation of the KCNA2 gene (Kv1.2; Imbrici et al., 2006; Xie et al., 2010; Syrbe et al., 2015). Mutations could diminish ionic currents by decreasing protein maturation, trafficking, or activity level, or interaction with accessory subunits, and/or by dominant-negative effects. Nevertheless, we will miss the whole pathological picture as long as we ignore their impact on unitary conductance. Table 3. S6 channelopathies of voltage-dependent K+ channels Channel Gene Unitary cond. Tissue expression S6 limits Mutation Disease Reference pS Kv1.1 KCNA1 8.7–20 Central nervous system, kidney, and heart 387–415 V404I, V408A Episodic ataxia EA1 Browne et al., 1994; Scheffer et al., 1998 Kv1.2 KCNA2 14–18 Neocortex, hippocampus, main olfactory bulb, and cerebellum 389–417 P405L Early infantile epileptic encephalopathy, 32 EIEE32 Syrbe et al., 2015 Kv2.1 KCNB1 14 Hippocampal neurons and cortical neurons 392–420 G401R Early infantile epileptic encephalopathy, 26 EIEE26 Saitsu et al., 2015 Kv3.3 KCNC3 32–38 Cerebellum, basal ganglia, and spinal cord 518–539 V535M Spinocerebellar ataxia 13 SCA13 Duarri et al., 2015 Kv4.3 KCND3 4 Substantia nigra pars compact, retrosplenial cortex, superior colliculus, the raphe nuclei and amygdala, olfactory bulb, and dentate gyrus 382–402 S390N; V392I Spinocerebellar ataxia 19; Brugada syndrome SCA19; BRGDA9 Duarri et al., 2012; Giudicessi et al., 2012 Kv7.1 KCNQ1 0.7–4 Heart, uterus, stomach, small and large intestine, kidney and pancreas; smooth muscle 328–348 F339Y, A341E/G/V, L342F, P343L/R/S, A344E/V, G345E/R Long QT síndrome type 1 LQT1 Jongbloed et al., 1999; Tester et al., 2005; Kapplinger et al., 2009 Kv7.2 KCNQ2 6.5 Hippocampal and cortical neurons 292–312 A306T Benign familial neonatal seizures BFNS1 Singh et al., 2003 Kv7.4 KCNQ4 2.1 Brain, cochlea, heart, and skeletal muscle; neuron derived from embryonic stem cells 297–317 G321S Deafness autosomal dominant 2A DFNA2A Coucke et al., 1999 Kv10.1 EAG KCNH1 8.5 Brain, kidney, lung, and pancreas; in brain: in cortex, hippocampus, caudate, putamen, amígdala, and substantia nigra 478–498 L489F, I494V Temple-Baraitser syndrome and epilepsy TMBTS Simons et al., 2015 Kv11.1 ERG KCNH2 10–13 Brian: reticular thalamic nucleus, cerebral cortex, cerebellum, and hippocampus; heart 639–659 F640L/V, S641F, V644F/L, M645I/L, G648S, F656C, G657R Long QT syndrome 2 LGT2 Napolitano et al., 2005; Tester et al., 2005; Kapplinger et al., 2009 KCa3.1 KCNN4 30–80 Nonexcitable tissues 265–285 V282E/M Dehydrated hereditary stomatocytosis DHS2 Glogowska et al., 2015 TASK 3 KCNK9 16–32 Cerebellum and external plexiform layer of the olfactory bulb; hippocampus 219–239 G236R Birk-Barel mental retardation dysmorphism BIBAS Barel et al., 2008 Non-exhaustive listing of mutations potentially affecting unitary conductance in voltage-gated K+ channel. Mutational data as well as topological composition of S6 transmembrane segments were obtained from UniProt (http://www.uniprot.org/uniprot/). We must not think that the unitary conductance is a fixed character of the channel. There are beautiful examples of how accessory subunits change conductance. For example, the unitary conductances of Kv4.2 and Kv7.1 (KCNQ1) increase several fold when they are coexpressed with accessory subunits: dipeptidyl-peptidase–like protein-6 (DPP6) and KCNE1 (MinK), respectively (Melman et al., 2004; Kaulin et al., 2009). The mechanism behind this DPP6 gain of function is similar to that of the Glu ring in BK (Kaulin et al., 2009), whereas in MinK it seems to involve intimate interaction with the S6 domain of Kv7 (Melman et al., 2004). Are these changes in unitary conductance relevant or just collateral consequences of the protein–protein interaction controlling gating kinetics and inactivation? By dissociating kinetics from conductance phenotypes, these cases present an opportunity to understand how relevant the changes in the unitary conductance are to physiology. Concluding remarks K+ channels are finely tuned to allow the selective passage of K+ across the membrane at high rates. The channel selectivity filter is in charge of this task, dropping to near zero the energy for ion transfer from the bulk solution (Morais-Cabral et al., 2001). K+ ions pass across the selectivity filter so efficiently that, even in the largest conductance channels, the physical dimensions of the internal vestibule limit channel conductance (Geng et al., 2011). Thus, we propose that the main difference between large- and small-conductance channels arises from the size of the entrance to the internal pore; large-conductance channels have wider vestibules than do smaller conductance ones. The effect of vestibule size on unitary conductance is clearly nonsteric because it is not proportional to the sectional area available for permeation. Inside narrow aqueous pores, embedded in low dielectric lipid membranes, ions require larger energies to become stabilized, limiting K+ current (Parsegian, 1969). These equilibrium energy considerations would reduce the conductance gap from approximately two orders of magnitude to one third of the maximal transport rate, corresponding to ∼1 kT in activation energy terms. The rest of the difference remains to be accounted for (Díaz-Franulic et al., 2015). Although Kv channels open just wide enough to let hydrated ions enter the permeation pathway, larger-conductance channels would require larger energies to open because of the hydrophobic nature of their inner walls. Thus, larger-conductance channels may host other activation gates as functional and structural data suggest (Zhou et al., 2011; Hite et al., 2015). Because structural determinants for unitary conductance and for electromechanical coupling colocalize toward the cytosolic end of S6, a mutual interference between pore occupancy and gating is expected. Therefore, S6 mutant Kv channelopathies require unitary conductance studies to fully understand their pathophysiology. ACKNOWLEDGMENTS We thank John Ewer (Centro Interdisciplinario de Neurociencia de Valparaíso [CINV]) for critical reading of the manuscript. This work is supported by Fondo Nacional de Desarrollo Científico y Tecnológico (Fondecyt) grant #1120819 and by Iniciativa Cientifica Milenio grant PO9-022. I. Díaz-Franulic and H. Moldenhauer were funded by Fraunhofer Chile Research and Fondecyt postdoctoral grant #3160321, respectively. The CINV is a Millennium Institute supported by the Millennium Scientific Initiative of the Ministerio de Economía, Fomento y Turismo. The authors declare no competing financial interests. Lesley C. Anson served as editor. ==== Refs Aggarwal, S.K., and R. MacKinnon. 1996. Contribution of the S4 segment to gating charge in the Shaker K+ channel. Neuron. 16 :1169–1177. 10.1016/S0896-6273(00)80143-9 8663993 Ahern, C.A., and R. Horn. 2004. Specificity of charge-carrying residues in the voltage sensor of potassium channels. J. Gen. Physiol. 123 :205–216. 10.1085/jgp.200308993 14769847 Alam, A., and Y. Jiang. 2009. High-resolution structure of the open NaK channel. Nat. Struct. Mol. Biol. 16 :30–34. 10.1038/nsmb.1531 19098917 Alcayaga, C., X. Cecchi, O. Alvarez, and R. Latorre. 1989. Streaming potential measurements in Ca2+-activated K+ channels from skeletal and smooth muscle. Coupling of ion and water fluxes. Biophys. J. 55 :367–371. 10.1016/S0006-3495(89)82814-0 2713449 Allen, T.W., O.S. Andersen, and B. Roux. 2004. On the importance of atomic fluctuations, protein flexibility, and solvent in ion permeation. J. Gen. Physiol. 124 :679–690. 10.1085/jgp.200409111 15572347 Andersen, O.S. 1983. Ion movement through gramicidin A channels. Studies on the diffusion-controlled association step. Biophys. J. 41 :147–165. 10.1016/S0006-3495(83)84416-6 6188502 Anishkin, A., and S. Sukharev. 2004. Water dynamics and dewetting transitions in the small mechanosensitive channel MscS. Biophys. J. 86 :2883–2895. 10.1016/S0006-3495(04)74340-4 15111405 Åqvist, J., and V. Luzhkov. 2000. Ion permeation mechanism of the potassium channel. Nature. 404 :881–884. 10.1038/35009114 10786795 Armstrong, C.M. 1971. Interaction of tetraethylammonium ion derivatives with the potassium channels of giant axons. J. Gen. Physiol. 58 :413–437. 10.1085/jgp.58.4.413 5112659 Armstrong, C.M. 2015. Packaging life: the origin of ion-selective channels. Biophys. J. 109 :173–177. 10.1016/j.bpj.2015.06.012 26200853 Bankura, A., V. Carnevale, and M.L. Klein. 2013. Hydration structure of salt solutions from ab initio molecular dynamics. J. Chem. Phys. 138 :014501. 10.1063/1.4772761 23298049 Barel, O., S.A. Shalev, R. Ofir, A. Cohen, J. Zlotogora, Z. Shorer, G. Mazor, G. Finer, S. Khateeb, N. Zilberberg, and O.S. Birk. 2008. Maternally inherited Birk Barel mental retardation dysmorphism syndrome caused by a mutation in the genomically imprinted potassium channel KCNK9. Am. J. Hum. Genet. 83 :193–199. 10.1016/j.ajhg.2008.07.010 18678320 Batulan, Z., G.A. Haddad, and R. Blunck. 2010. An intersubunit interaction between S4-S5 linker and S6 is responsible for the slow off-gating component in Shaker K+ channels. J. Biol. Chem. 285 :14005–14019. 10.1074/jbc.M109.097717 20202932 Beckstein, O., and M.S. Sansom. 2004. The influence of geometry, surface character, and flexibility on the permeation of ions and water through biological pores. Phys. Biol. 1 :42–52. 10.1088/1478-3967/1/1/005 16204821 Beckstein, O., P.C. Biggin, P. Bond, J.N. Bright, C. Domene, A. Grottesi, J. Holyoake, and M.S. Sansom. 2003. Ion channel gating: insights via molecular simulations. FEBS Lett. 555 :85–90. 10.1016/S0014-5793(03)01151-7 14630324 Bernèche, S., and B. Roux. 2000. Molecular dynamics of the KcsA K+ channel in a bilayer membrane. Biophys. J. 78 :2900–2917. 10.1016/S0006-3495(00)76831-7 10827971 Bernèche, S., and B. Roux. 2003. A microscopic view of ion conduction through the K+ channel. Proc. Natl. Acad. Sci. USA. 100 :8644–8648. 10.1073/pnas.1431750100 12837936 Bezanilla, F., and C.M. Armstrong. 1972. Negative conductance caused by entry of sodium and cesium ions into the potassium channels of squid axons. J. Gen. Physiol. 60 :588–608. 10.1085/jgp.60.5.588 4644327 Brelidze, T.I., and K.L. Magleby. 2005. Probing the geometry of the inner vestibule of BK channels with sugars. J. Gen. Physiol. 126 :105–121. 10.1085/jgp.200509286 16043773 Brelidze, T.I., X. Niu, and K.L. Magleby. 2003. A ring of eight conserved negatively charged amino acids doubles the conductance of BK channels and prevents inward rectification. Proc. Natl. Acad. Sci. USA. 100 :9017–9022. 10.1073/pnas.1532257100 12843404 Browne, D.L., S.T. Gancher, J.G. Nutt, E.R. Brunt, E.A. Smith, P. Kramer, and M. Litt. 1994. Episodic ataxia/myokymia syndrome is associated with point mutations in the human potassium channel gene, KCNA1. Nat. Genet. 8 :136–140. 10.1038/ng1094-136 7842011 Bruening-Wright, A., M.A. Schumacher, J.P. Adelman, and J. Maylie. 2002. Localization of the activation gate for small conductance Ca2+-activated K+ channels. J. Neurosci. 22 :6499–6506.12151529 Budelli, G., Y. Geng, A. Butler, K.L. Magleby, and L. Salkoff. 2013. Properties of Slo1 K+ channels with and without the gating ring. Proc. Natl. Acad. Sci. USA. 110 :16657–16662. 10.1073/pnas.1313433110 24067659 Cao, E., M. Liao, Y. Cheng, and D. Julius. 2013. TRPV1 structures in distinct conformations reveal activation mechanisms. Nature. 504 :113–118. 10.1038/nature12823 24305161 Carrasquel-Ursulaez, W., G.F. Contreras, R.V. Sepúlveda, D. Aguayo, F. González-Nilo, C. González, and R. Latorre. 2015. Hydrophobic interaction between contiguous residues in the S6 transmembrane segment acts as a stimuli integration node in the BK channel. J. Gen. Physiol. 145 :61–74. 10.1085/jgp.201411194 25548136 Carvacho, I., W. Gonzalez, Y.P. Torres, S. Brauchi, O. Alvarez, F.D. Gonzalez-Nilo, and R. Latorre. 2008. Intrinsic electrostatic potential in the BK channel pore: role in determining single channel conductance and block. J. Gen. Physiol. 131 :147–161. 10.1085/jgp.200709862 18227273 Chen, H., F.C. Chatelain, and F. Lesage. 2014. Altered and dynamic ion selectivity of K+ channels in cell development and excitability. Trends Pharmacol. Sci. 35 :461–469. 10.1016/j.tips.2014.06.002 25023607 Cheng, W.W., J.G. McCoy, A.N. Thompson, C.G. Nichols, and C.M. Nimigean. 2011. Mechanism for selectivity-inactivation coupling in KcsA potassium channels. Proc. Natl. Acad. Sci. USA. 108 :5272–5277. 10.1073/pnas.1014186108 21402935 Chung, S.H., T.W. Allen, and S. Kuyucak. 2002. Conducting-state properties of the KcsA potassium channel from molecular and Brownian dynamics simulations. Biophys. J. 82 :628–645. 10.1016/S0006-3495(02)75427-1 11806907 Contreras, J.E., J. Chen, A.Y. Lau, V. Jogini, B. Roux, and M. Holmgren. 2010. Voltage profile along the permeation pathway of an open channel. Biophys. J. 99 :2863–2869. 10.1016/j.bpj.2010.08.053 21044583 Coucke, P.J., P. Van Hauwe, P.M. Kelley, H. Kunst, I. Schatteman, D. Van Velzen, J. Meyers, R.J. Ensink, M. Verstreken, F. Declau, 1999. Mutations in the KCNQ4 gene are responsible for autosomal dominant deafness in four DFNA2 families. Hum. Mol. Genet. 8 :1321–1328. 10.1093/hmg/8.7.1321 10369879 Cuello, L.G., V. Jogini, D.M. Cortes, and E. Perozo. 2010. Structural mechanism of C-type inactivation in K+ channels. Nature. 466 :203–208. 10.1038/nature09153 20613835 del Camino, D., M. Kanevsky, and G. Yellen. 2005. Status of the intracellular gate in the activated-not-open state of Shaker K+ channels. J. Gen. Physiol. 126 :419–428. 10.1085/jgp.200509385 16260836 Derebe, M.G., D.B. Sauer, W. Zeng, A. Alam, N. Shi, and Y. Jiang. 2011. Tuning the ion selectivity of tetrameric cation channels by changing the number of ion binding sites. Proc. Natl. Acad. Sci. USA. 108 :598–602. 10.1073/pnas.1013636108 21187421 Díaz, L., P. Meera, J. Amigo, E. Stefani, O. Alvarez, L. Toro, and R. Latorre. 1998. Role of the S4 segment in a voltage-dependent calcium-sensitive potassium (hSlo) channel. J. Biol. Chem. 273 :32430–32436. 10.1074/jbc.273.49.32430 9829973 Díaz-Franulic, I., R.V. Sepúlveda, N. Navarro-Quezada, F. González-Nilo, and D. Naranjo. 2015. Pore dimensions and the role of occupancy in unitary conductance of Shaker K channels. J. Gen. Physiol. 146 :133–146. 10.1085/jgp.201411353 26216859 Ding, S., and R. Horn. 2002. Tail end of the S6 segment: role in permeation in Shaker potassium channels. J. Gen. Physiol. 120 :87–97. 10.1085/jgp.20028611 12084778 Doyle, D.A., J. Morais Cabral, R.A. Pfuetzner, A. Kuo, J.M. Gulbis, S.L. Cohen, B.T. Chait, and R. MacKinnon. 1998. The structure of the potassium channel: molecular basis of K+ conduction and selectivity. Science. 280 :69–77. 10.1126/science.280.5360.69 9525859 Duarri, A., J. Jezierska, M. Fokkens, M. Meijer, H.J. Schelhaas, W.F. den Dunnen, F. van Dijk, C. Verschuuren-Bemelmans, G. Hageman, P. van de Vlies, 2012. Mutations in potassium channel kcnd3 cause spinocerebellar ataxia type 19. Ann. Neurol. 72 :870–880. 10.1002/ana.23700 23280838 Duarri, A., E.A. Nibbeling, M.R. Fokkens, M. Meijer, M. Boerrigter, C.C. Verschuuren-Bemelmans, B.P. Kremer, B.P. van de Warrenburg, D. Dooijes, E. Boddeke, 2015. Functional analysis helps to define KCNC3 mutational spectrum in Dutch ataxia cases. PLoS One. 10 :e0116599. 10.1371/journal.pone.0116599 25756792 Eisenberg, B. 2013. Interacting ions in biophysics: real is not ideal. Biophys. J. 104 :1849–1866. 10.1016/j.bpj.2013.03.049 23663828 Eisenman, G. 1962. Cation selective glass electrodes and their mode of operation. Biophys. J. 2 :259–323. 10.1016/S0006-3495(62)86959-8 13889686 Eisenman, G., R. Latorre, and C. Miller. 1986. Multi-ion conduction and selectivity in the high-conductance Ca++-activated K+ channel from skeletal muscle. Biophys. J. 50 :1025–1034. 10.1016/S0006-3495(86)83546-9 2432947 Enderby, J.E. 1995. Ion solvation via neutron scattering. Chem. Soc. Rev. 24 :159–168. 10.1039/cs9952400159 Faure, É., G. Starek, H. McGuire, S. Bernèche, and R. Blunck. 2012. A limited 4 Å radial displacement of the S4-S5 linker is sufficient for internal gate closing in Kv channels. J. Biol. Chem. 287 :40091–40098. 10.1074/jbc.M112.415497 23019337 Ferry, J.D. 1936. Statistical evaluation of sieve constants in ultrafiltration. J. Gen. Physiol. 20 :95–104. 10.1085/jgp.20.1.95 19872986 Flynn, G.E., and W.N. Zagotta. 2001. Conformational changes in S6 coupled to the opening of cyclic nucleotide-gated channels. Neuron. 30 :689–698. 10.1016/S0896-6273(01)00324-5 11430803 Garate, J.A., T. Perez-Acle, and C. Oostenbrink. 2014. On the thermodynamics of carbon nanotube single-file water loading: free energy, energy and entropy calculations. Phys. Chem. Chem. Phys. 16 :5119–5128. 10.1039/c3cp54554g 24477412 Garg, P., A. Gardner, V. Garg, and M.C. Sanguinetti. 2013. Structural basis of ion permeation gating in Slo2.1 K+ channels. J. Gen. Physiol. 142 :523–542. 10.1085/jgp.201311064 24166878 Geng, Y., X. Niu, and K.L. Magleby. 2011. Low resistance, large dimension entrance to the inner cavity of BK channels determined by changing side-chain volume. J. Gen. Physiol. 137 :533–548. 10.1085/jgp.201110616 21576375 Giudicessi, J.R., D. Ye, C.J. Kritzberger, V.V. Nesterenko, D.J. Tester, C. Antzelevitch, and M.J. Ackerman. 2012. Novel mutations in the KCND3-encoded Kv4.3 K+ channel associated with autopsy-negative sudden unexplained death. Hum. Mutat. 33 :989–997. 10.1002/humu.22058 22457051 Glezakou, V.-A., Y. Chen, J.L. Fulton, G.K. Schenter, and L.X. Dang. 2006. Electronic structure, statistical mechanical simulations, and EXAFS spectroscopy of aqueous potassium. Theor. Chem. Acc. 115 :86–99. 10.1007/s00214-005-0054-4 Glogowska, E., K. Lezon-Geyda, Y. Maksimova, V.P. Schulz, and P.G. Gallagher. 2015. Mutations in the Gardos channel (KCNN4) are associated with hereditary xerocytosis. Blood. 126 :1281–1284. 10.1182/blood-2015-07-657957 26198474 González-Pérez, V., K. Stack, K. Boric, and D. Naranjo. 2010. Reduced voltage sensitivity in a K+-channel voltage sensor by electric field remodeling. Proc. Natl. Acad. Sci. USA. 107 :5178–5183. 10.1073/pnas.1000963107 20194763 Grossfield, A. 2005. Dependence of ion hydration on the sign of the ion’s charge. J. Chem. Phys. 122 :024506. 10.1063/1.1829036 15638597 Gutman, G.A., K.G. Chandy, S. Grissmer, M. Lazdunski, D. McKinnon, L.A. Pardo, G.A. Robertson, B. Rudy, M.C. Sanguinetti, W. Stühmer, and X. Wang. 2005. International Union of Pharmacology. LIII. Nomenclature and molecular relationships of voltage-gated potassium channels. Pharmacol. Rev. 57 :473–508. 10.1124/pr.57.4.10 16382104 Hackos, D.H., T.H. Chang, and K.J. Swartz. 2002. Scanning the intracellular S6 activation gate in the shaker K+ channel. J. Gen. Physiol. 119 :521–531. 10.1085/jgp.20028569 12034760 Heginbotham, L., and R. MacKinnon. 1993. Conduction properties of the cloned Shaker K+ channel. Biophys. J. 65 :2089–2096. 10.1016/S0006-3495(93)81244-X 8298038 Heginbotham, L., Z. Lu, T. Abramson, and R. MacKinnon. 1994. Mutations in the K+ channel signature sequence. Biophys. J. 66 :1061–1067. 10.1016/S0006-3495(94)80887-2 8038378 Hille, B. 1973. Potassium channels in myelinated nerve. Selective permeability to small cations. J. Gen. Physiol. 61 :669–686. 10.1085/jgp.61.6.669 4541077 Hille, B. 2001. Ion channels of excitable membranes. Third edition. Sinauer, Sunderland, MA. 814 pp. Hite, R.K., P. Yuan, Z. Li, Y. Hsuing, T. Walz, and R. MacKinnon. 2015. Cryo-electron microscopy structure of the Slo2.2 Na+-activated K+ channel. Nature. 527 :198–203. 10.1038/nature14958 26436452 Hodgkin, A.L., and A.F. Huxley. 1952. A quantitative description of membrane current and its application to conduction and excitation in nerve. J. Physiol. 117 :500–544. 10.1113/jphysiol.1952.sp004764 12991237 Hodgkin, A.L., and R.D. Keynes. 1955. The potassium permeability of a giant nerve fibre. J. Physiol. 128 :61–88. 10.1113/jphysiol.1955.sp005291 Horrigan, F.T., J. Cui, and R.W. Aldrich. 1999. Allosteric voltage gating of potassium channels I. Mslo ionic currents in the absence of Ca2+. J. Gen. Physiol. 114 :277–304. 10.1085/jgp.114.2.277 10436003 Humphrey, W., A. Dalke, and K. Schulten. 1996. VMD: Visual molecular dynamics. J. Mol. Graph. 14 :33–38.8744570 Imbrici, P., M.C. D’Adamo, D.M. Kullmann, and M. Pessia. 2006. Episodic ataxia type 1 mutations in the KCNA1 gene impair the fast inactivation properties of the human potassium channels Kv1.4-1.1/Kvβ1.1 and Kv1.4-1.1/Kvβ1.2. Eur. J. Neurosci. 24 :3073–3083. 10.1111/j.1460-9568.2006.05186.x 17156368 Islas, L.D., and F.J. Sigworth. 1999. Voltage sensitivity and gating charge in Shaker and Shab family potassium channels. J. Gen. Physiol. 114 :723–742. 10.1085/jgp.114.5.723 10539976 Jensen, M.O., D.W. Borhani, K. Lindorff-Larsen, P. Maragakis, V. Jogini, M.P. Eastwood, R.O. Dror, and D.E. Shaw. 2010. Principles of conduction and hydrophobic gating in K+ channels. Proc. Natl. Acad. Sci. USA. 107 :5833–5838. 10.1073/pnas.0911691107 20231479 Jensen, M.O., V. Jogini, D.W. Borhani, A.E. Leffler, R.O. Dror, and D.E. Shaw. 2012. Mechanism of voltage gating in potassium channels. Science. 336 :229–233. 10.1126/science.1216533 22499946 Jensen, M.O., V. Jogini, M.P. Eastwood, and D.E. Shaw. 2013. Atomic-level simulation of current-voltage relationships in single-file ion channels. J. Gen. Physiol. 141 :619–632. 10.1085/jgp.201210820 23589581 Jiang, Y., A. Lee, J. Chen, M. Cadene, B.T. Chait, and R. MacKinnon. 2002. Crystal structure and mechanism of a calcium-gated potassium channel. Nature. 417 :515–522. 10.1038/417515a 12037559 Jogini, V., and B. Roux. 2005. Electrostatics of the intracellular vestibule of K+ channels. J. Mol. Biol. 354 :272–288. 10.1016/j.jmb.2005.09.031 16242718 Jongbloed, R.J., A.A. Wilde, J.L. Geelen, P. Doevendans, C. Schaap, I. Van Langen, J.P. van Tintelen, J.M. Cobben, G.C. Beaufort-Krol, J.P. Geraedts, and H.J. Smeets. 1999. Novel KCNQ1 and HERG missense mutations in Dutch long-QT families. Hum. Mutat. 13 :301–310. 10.1002/(SICI)1098-1004(1999)13:4<301::AID-HUMU7>3.0.CO;2-V 10220144 Kapplinger, J.D., D.J. Tester, B.A. Salisbury, J.L. Carr, C. Harris-Kerr, G.D. Pollevick, A.A. Wilde, and M.J. Ackerman. 2009. Spectrum and prevalence of mutations from the first 2,500 consecutive unrelated patients referred for the FAMILION long QT syndrome genetic test. Heart Rhythm. 6 :1297–1303. 10.1016/j.hrthm.2009.05.021 19716085 Kaulin, Y.A., J.A. De Santiago-Castillo, C.A. Rocha, M.S. Nadal, B. Rudy, and M. Covarrubias. 2009. The dipeptidyl-peptidase-like protein DPP6 determines the unitary conductance of neuronal Kv4.2 channels. J. Neurosci. 29 :3242–3251. 10.1523/JNEUROSCI.4767-08.2009 19279261 Kobertz, W.R., and C. Miller. 1999. K+ channels lacking the ‘tetramerization’ domain: implications for pore structure. Nat. Struct. Biol. 6 :1122–1125. 10.1038/70061 10581553 Krishnamoorthy, G., J. Shi, D. Sept, and J. Cui. 2005. The NH2 terminus of RCK1 domain regulates Ca2+-dependent BKCa channel gating. J. Gen. Physiol. 126 :227–241. 10.1085/jgp.200509321 16103277 Latorre, R., and C. Miller. 1983. Conduction and selectivity in potassium channels. J. Membr. Biol. 71 :11–30. 10.1007/BF01870671 6300405 Läuger, P. 1976. Diffusion-limited ion flow through pores. Biochim. Biophys. Acta. 455 :493–509. 10.1016/0005-2736(76)90320-5 999924 Li, W., and R.W. Aldrich. 2004. Unique inner pore properties of BK channels revealed by quaternary ammonium block. J. Gen. Physiol. 124 :43–57. 10.1085/jgp.200409067 15197222 Liao, M., E. Cao, D. Julius, and Y. Cheng. 2013. Structure of the TRPV1 ion channel determined by electron cryo-microscopy. Nature. 504 :107–112. 10.1038/nature12822 24305160 Liu, S., and S.W. Lockless. 2013. Equilibrium selectivity alone does not create K+-selective ion conduction in K+ channels. Nat. Commun. 4 :2746.24217508 Liu, S., X. Bian, and S.W. Lockless. 2012. Preferential binding of K+ ions in the selectivity filter at equilibrium explains high selectivity of K+ channels. J. Gen. Physiol. 140 :671–679. 10.1085/jgp.201210855 23148260 Lockless, S.W. 2015. Determinants of cation transport selectivity: Equilibrium binding and transport kinetics. J. Gen. Physiol. 146 :3–13. 10.1085/jgp.201511371 26078056 Long, S.B., E.B. Campbell, and R. Mackinnon. 2005 a. Crystal structure of a mammalian voltage-dependent Shaker family K+ channel. Science. 309 :897–903. 10.1126/science.1116269 16002581 Long, S.B., E.B. Campbell, and R. Mackinnon. 2005 b. Voltage sensor of Kv1.2: structural basis of electromechanical coupling. Science. 309 :903–908. 10.1126/science.1116270 16002579 Long, S.B., X. Tao, E.B. Campbell, and R. MacKinnon. 2007. Atomic structure of a voltage-dependent K+ channel in a lipid membrane-like environment. Nature. 450 :376–382. 10.1038/nature06265 18004376 Lu, Z. 2004. Mechanism of rectification in inward-rectifier K+ channels. Annu. Rev. Physiol. 66 :103–129. 10.1146/annurev.physiol.66.032102.150822 14977398 Lu, Z., A.M. Klem, and Y. Ramu. 2002. Coupling between voltage sensors and activation gate in voltage-gated K+ channels. J. Gen. Physiol. 120 :663–676. 10.1085/jgp.20028696 12407078 Mancinelli, R., A. Botti, F. Bruni, M.A. Ricci, and A.K. Soper. 2007. Hydration of sodium, potassium, and chloride ions in solution and the concept of structure maker/breaker. J. Phys. Chem. B. 111 :13570–13577. 10.1021/jp075913v 17988114 Matta, J.A., and G.P. Ahern. 2007. Voltage is a partial activator of rat thermosensitive TRP channels. J. Physiol. 585 :469–482. 10.1113/jphysiol.2007.144287 17932142 Matteson, D.R., and R.P. Swenson Jr. 1986. External monovalent cations that impede the closing of K channels. J. Gen. Physiol. 87 :795–816. 10.1085/jgp.87.5.795 2425039 Melman, Y.F., S.Y. Um, A. Krumerman, A. Kagan, and T.V. McDonald. 2004. KCNE1 binds to the KCNQ1 pore to regulate potassium channel activity. Neuron. 42 :927–937. 10.1016/j.neuron.2004.06.001 15207237 Miller, C. 1982. Bis-quaternary ammonium blockers as structural probes of the sarcoplasmic reticulum K+ channel. J. Gen. Physiol. 79 :869–891. 10.1085/jgp.79.5.869 6284862 Miller, C. 2001. See potassium run. Nature. 414 :23–24. 10.1038/35102126 11689922 Moldenhauer, H., I. Díaz-Franulic, F. González-Nilo, and D. Naranjo. 2016. Effective pore size and radius of capture for K+ ions in K-channels. Sci. Rep. 6 :19893. 10.1038/srep19893 26831782 Morais-Cabral, J.H., Y. Zhou, and R. MacKinnon. 2001. Energetic optimization of ion conduction rate by the K+ selectivity filter. Nature. 414 :37–42. 10.1038/35102000 11689935 Moscoso, C., A. Vergara-Jaque, V. Márquez-Miranda, R.V. Sepúlveda, I. Valencia, I. Díaz-Franulic, F. González-Nilo, and D. Naranjo. 2012. K+ conduction and Mg2+ blockade in a shaker Kv-channel single point mutant with an unusually high conductance. Biophys. J. 103 :1198–1207. 10.1016/j.bpj.2012.08.015 22995492 Mullins, L.J. 1959. An analysis of conductance changes in squid axon. J. Gen. Physiol. 42 :1013–1035. 10.1085/jgp.42.5.1013 13654748 Napolitano, C., S.G. Priori, P.J. Schwartz, R. Bloise, E. Ronchetti, J. Nastoli, G. Bottelli, M. Cerrone, and S. Leonardi. 2005. Genetic testing in the long QT syndrome: development and validation of an efficient approach to genotyping in clinical practice. JAMA. 294 :2975–2980. 10.1001/jama.294.23.2975 16414944 Neale, C., N. Chakrabarti, P. Pomorski, E.F. Pai, and R. Pomès. 2015. Hydrophobic gating of ion permeation in magnesium channel CorA. PLOS Comput. Biol. 11 :e1004303. 10.1371/journal.pcbi.1004303 26181442 Neyton, J., and C. Miller. 1988. Discrete Ba2+ block as a probe of ion occupancy and pore structure in the high-conductance Ca2+-activated K+ channel. J. Gen. Physiol. 92 :569–586. 10.1085/jgp.92.5.569 3235974 Nimigean, C.M., and T.W. Allen. 2011. Origins of ion selectivity in potassium channels from the perspective of channel block. J. Gen. Physiol. 137 :405–413. 10.1085/jgp.201010551 21518829 Nimigean, C.M., J.S. Chappie, and C. Miller. 2003. Electrostatic tuning of ion conductance in potassium channels. Biochemistry. 42 :9263–9268. 10.1021/bi0348720 12899612 Nishida, M., and R. MacKinnon. 2002. Structural basis of inward rectification: cytoplasmic pore of the G protein-gated inward rectifier GIRK1 at 1.8 Å resolution. Cell. 111 :957–965. 10.1016/S0092-8674(02)01227-8 12507423 Noskov, S.Y., and B. Roux. 2006. Ion selectivity in potassium channels. Biophys. Chem. 124 :279–291. 10.1016/j.bpc.2006.05.033 16843584 Noskov, S.Y., S. Bernèche, and B. Roux. 2004. Control of ion selectivity in potassium channels by electrostatic and dynamic properties of carbonyl ligands. Nature. 431 :830–834. 10.1038/nature02943 15483608 Parsegian, A. 1969. Energy of an ion crossing a low dielectric membrane: solutions to four relevant electrostatic problems. Nature. 221 :844–846. 10.1038/221844a0 5765058 Peter, C., and G. Hummer. 2005. Ion transport through membrane-spanning nanopores studied by molecular dynamics simulations and continuum electrostatics calculations. Biophys. J. 89 :2222–2234. 10.1529/biophysj.105.065946 16006629 Piasta, K.N., D.L. Theobald, and C. Miller. 2011. Potassium-selective block of barium permeation through single KcsA channels. J. Gen. Physiol. 138 :421–436. 10.1085/jgp.201110684 21911483 Posson, D.J., J.G. McCoy, and C.M. Nimigean. 2013. The voltage-dependent gate in MthK potassium channels is located at the selectivity filter. Nat. Struct. Mol. Biol. 20 :159–166. 10.1038/nsmb.2473 23262489 Proks, P., J.F. Antcliff, and F.M. Ashcroft. 2003. The ligand-sensitive gate of a potassium channel lies close to the selectivity filter. EMBO Rep. 4 :70–75. 10.1038/sj.embor.embor708 12524524 Rapedius, M., M.R. Schmidt, C. Sharma, P.J. Stansfeld, M.S. Sansom, T. Baukrowitz, and S.J. Tucker. 2012. State-independent intracellular access of quaternary ammonium blockers to the pore of TREK-1. Channels (Austin). 6 :473–478. 10.4161/chan.22153 22991046 Reynolds, J.A., D.B. Gilbert, and C. Tanford. 1974. Empirical correlation between hydrophobic free energy and aqueous cavity surface area. Proc. Natl. Acad. Sci. USA. 71 :2925–2927. 10.1073/pnas.71.8.2925 16578715 Robinson, R.A., and R.H. Stokes. 2002. Electrolyte solutions. Second revised edition. Dover Publications, Mineola, NY. 590 pp. Ruta, V., Y. Jiang, A. Lee, J. Chen, and R. MacKinnon. 2003. Functional analysis of an archaebacterial voltage-dependent K+ channel. Nature. 422 :180–185. 10.1038/nature01473 12629550 Sack, J.T., and D.C. Tilley. 2015. What keeps Kv channels small? The molecular physiology of modesty. J. Gen. Physiol. 146 :123–127. 10.1085/jgp.201511469 26216857 Saitsu, H., T. Akita, J. Tohyama, H. Goldberg-Stern, Y. Kobayashi, R. Cohen, M. Kato, C. Ohba, S. Miyatake, Y. Tsurusaki, 2015. De novo KCNB1 mutations in infantile epilepsy inhibit repetitive neuronal firing. Sci. Rep. 5 :15199. 10.1038/srep15199 26477325 Salazar, H., A. Jara-Oseguera, E. Hernández-García, I. Llorente, I.I. Arias-Olguín, M. Soriano-García, L.D. Islas, and T. Rosenbaum. 2009. Structural determinants of gating in the TRPV1 channel. Nat. Struct. Mol. Biol. 16 :704–710. 10.1038/nsmb.1633 19561608 Sauer, D.B., W. Zeng, J. Canty, Y. Lam, and Y. Jiang. 2013. Sodium and potassium competition in potassium-selective and non-selective channels. Nat. Commun. 4 :2721. 10.1038/ncomms3721 24217363 Scheffer, H., E.R. Brunt, G.J. Mol, P. van der Vlies, R.P. Stulp, E. Verlind, G. Mantel, Y.N. Averyanov, R.M. Hofstra, and C.H. Buys. 1998. Three novel KCNA1 mutations in episodic ataxia type I families. Hum. Genet. 102 :464–466. 10.1007/s004390050722 9600245 Seoh, S.A., D. Sigg, D.M. Papazian, and F. Bezanilla. 1996. Voltage-sensing residues in the S2 and S4 segments of the Shaker K+ channel. Neuron. 16 :1159–1167. 10.1016/S0896-6273(00)80142-7 8663992 Shi, N., W. Zeng, S. Ye, Y. Li, and Y. Jiang. 2011. Crucial points within the pore as determinants of K+ channel conductance and gating. J. Mol. Biol. 411 :27–35. 10.1016/j.jmb.2011.04.058 21554888 Simons, C., L.D. Rash, J. Crawford, L. Ma, B. Cristofori-Armstrong, D. Miller, K. Ru, G.J. Baillie, Y. Alanay, A. Jacquinet, 2015. Mutations in the voltage-gated potassium channel gene KCNH1 cause Temple-Baraitser syndrome and epilepsy. Nat. Genet. 47 :73–77. 10.1038/ng.3153 25420144 Singh, N.A., P. Westenskow, C. Charlier, C. Pappas, J. Leslie, J. Dillon, V.E. Anderson, M.C. Sanguinetti, and M.F. Leppert. BFNC Physician Consortium. 2003. KCNQ2 and KCNQ3 potassium channel genes in benign familial neonatal convulsions: expansion of the functional and mutation spectrum. Brain. 126 :2726–2737. 10.1093/brain/awg286 14534157 Starkus, J.G., L. Kuschel, M.D. Rayner, and S.H. Heinemann. 1997. Ion conduction through C-type inactivated Shaker channels. J. Gen. Physiol. 110 :539–550. 10.1085/jgp.110.5.539 9348326 Sukhareva, M., D.H. Hackos, and K.J. Swartz. 2003. Constitutive activation of the Shaker Kv channel. J. Gen. Physiol. 122 :541–556. 10.1085/jgp.200308905 14557403 Swenson, R.P. Jr., and C.M. Armstrong. 1981. K+ channels close more slowly in the presence of external K+ and Rb+. Nature. 291 :427–429. 10.1038/291427a0 6264306 Syrbe, S., U.B. Hedrich, E. Riesch, T. Djémié, S. Müller, R.S. Møller, B. Maher, L. Hernandez-Hernandez, M. Synofzik, H.S. Caglayan, EuroEPINOMICS RES. 2015. De novo loss- or gain-of-function mutations in KCNA2 cause epileptic encephalopathy. Nat. Genet. 47 :393–399. 10.1038/ng.3239 25751627 Tester, D.J., M.L. Will, C.M. Haglund, and M.J. Ackerman. 2005. Compendium of cardiac channel mutations in 541 consecutive unrelated patients referred for long QT syndrome genetic testing. Heart Rhythm. 2 :507–517. 10.1016/j.hrthm.2005.01.020 15840476 Thompson, J., and T. Begenisich. 2001. Affinity and location of an internal K+ ion binding site in Shaker K channels. J. Gen. Physiol. 117 :373–384. 10.1085/jgp.117.5.373 11331347 Tombola, F., M.M. Pathak, P. Gorostiza, and E.Y. Isacoff. 2007. The twisted ion-permeation pathway of a resting voltage-sensing domain. Nature. 445 :546–549. 10.1038/nature05396 17187057 Treptow, W., and M. Tarek. 2006. Molecular restraints in the permeation pathway of ion channels. Biophys. J. 91 :L26–L28. 10.1529/biophysj.106.087437 16751240 Wang, W., S.S. Black, M.D. Edwards, S. Miller, E.L. Morrison, W. Bartlett, C. Dong, J.H. Naismith, and I.R. Booth. 2008. The structure of an open form of an E. coli mechanosensitive channel at 3.45 Å resolution. Science. 321 :1179–1183. 10.1126/science.1159262 18755969 Webster, S.M., D. Del Camino, J.P. Dekker, and G. Yellen. 2004. Intracellular gate opening in Shaker K+ channels defined by high-affinity metal bridges. Nature. 428 :864–868. 10.1038/nature02468 15103379 Wilkens, C.M., and R.W. Aldrich. 2006. State-independent block of BK channels by an intracellular quaternary ammonium. J. Gen. Physiol. 128 :347–364. 10.1085/jgp.200609579 16940557 Wilson, M.A., C. Wei, P. Bjelkmar, B.A. Wallace, and A. Pohorille. 2011. Molecular dynamics simulation of the antiamoebin ion channel: linking structure and conductance. Biophys. J. 100 :2394–2402. 10.1016/j.bpj.2011.03.054 21575573 Xie, G., J. Harrison, S.J. Clapcote, Y. Huang, J.Y. Zhang, L.Y. Wang, and J.C. Roder. 2010. A new Kv1.2 channelopathy underlying cerebellar ataxia. J. Biol. Chem. 285 :32160–32173. 10.1074/jbc.M110.153676 20696761 Ye, S., Y. Li, and Y. Jiang. 2010. Novel insights into K+ selectivity from high-resolution structures of an open K+ channel pore. Nat. Struct. Mol. Biol. 17 :1019–1023. 10.1038/nsmb.1865 20676101 Yifrach, O., and R. MacKinnon. 2002. Energetics of pore opening in a voltage-gated K+ channel. Cell. 111 :231–239. 10.1016/S0092-8674(02)01013-9 12408867 Yu, F.H., V. Yarov-Yarovoy, G.A. Gutman, and W.A. Catterall. 2005. Overview of molecular relationships in the voltage-gated ion channel superfamily. Pharmacol. Rev. 57 :387–395. 10.1124/pr.57.4.13 16382097 Yuan, P., M.D. Leonetti, Y. Hsiung, and R. MacKinnon. 2011. Open structure of the Ca2+ gating ring in the high-conductance Ca2+-activated K+ channel. Nature. 481 :94–97. 10.1038/nature10670 22139424 Zhang, Y., X. Niu, T.I. Brelidze, and K.L. Magleby. 2006. Ring of negative charge in BK channels facilitates block by intracellular Mg2+ and polyamines through electrostatics. J. Gen. Physiol. 128 :185–202. 10.1085/jgp.200609493 16847096 Zhou, Y., J.H. Morais-Cabral, A. Kaufman, and R. MacKinnon. 2001. Chemistry of ion coordination and hydration revealed by a K+ channel–Fab complex at 2.0 Å resolution. Nature. 414 :43–48. 10.1038/35102009 11689936 Zhou, Y., X.M. Xia, and C.J. Lingle. 2011. Cysteine scanning and modification reveal major differences between BK channels and Kv channels in the inner pore region. Proc. Natl. Acad. Sci. USA. 108 :12161–12166. 10.1073/pnas.1104150108 21730134 Zhu, F., and G. Hummer. 2012 a. Drying transition in the hydrophobic gate of the GLIC channel blocks ion conduction. Biophys. J. 103 :219–227. 10.1016/j.bpj.2012.06.003 22853899 Zhu, F., and G. Hummer. 2012 b. Theory and simulation of ion conduction in the pentameric GLIC channel. J. Chem. Theory Comput. 8 :3759–3768. 10.1021/ct2009279 23413364

PMC005xxxxxx/PMC5223608.txt

==== Front J Cell Biol J Cell Biol jcb jcb The Journal of Cell Biology 0021-9525 1540-8140 The Rockefeller University Press 28003333 201608071 10.1083/jcb.201608071 Research Articles Article 28 34 35 Regulation of clathrin-mediated endocytosis by hierarchical allosteric activation of AP2 AP2 activation in clathrin-mediated endocytosis Kadlecova Zuzana 1 Spielman Stephanie J. 2 Loerke Dinah 3 Mohanakrishnan Aparna 1 http://orcid.org/0000-0002-3980-4239 Reed Dana Kim 1 http://orcid.org/0000-0002-1690-7024 Schmid Sandra L. 1 1 Department of Cell Biology, University of Texas Southwestern Medical Center, Dallas, TX 75390 2 Institute for Genomics and Evolutionary Medicine, Temple University, Philadelphia, PA 19122 3 Department of Physics and Astronomy, University of Denver, Denver, CO 80208 Correspondence to Sandra L. Schmid: sandra.schmid@utsouthwestern.edu 2 1 2017 216 1 167179 21 8 2016 31 10 2016 30 11 2016 © 2017 Kadlecova et al. 2017 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License(Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). The adaptor AP2 is required for initiation of clathrin-mediated endocytosis. Kadlecova et al. delineate the functional hierarchy of AP2 interactions with phosphatidylinositol lipids and cargo and their relationship to distinct steps in clathrin-coated pit nucleation and maturation in living cells. The critical initiation phase of clathrin-mediated endocytosis (CME) determines where and when endocytosis occurs. Heterotetrameric adaptor protein 2 (AP2) complexes, which initiate clathrin-coated pit (CCP) assembly, are activated by conformational changes in response to phosphatidylinositol-4,5-bisphosphate (PIP2) and cargo binding at multiple sites. However, the functional hierarchy of interactions and how these conformational changes relate to distinct steps in CCP formation in living cells remains unknown. We used quantitative live-cell analyses to measure discrete early stages of CME and show how sequential, allosterically regulated conformational changes activate AP2 to drive both nucleation and subsequent stabilization of nascent CCPs. Our data establish that cargoes containing Yxxφ motif, but not dileucine motif, play a critical role in the earliest stages of AP2 activation and CCP nucleation. Interestingly, these cargo and PIP2 interactions are not conserved in yeast. Thus, we speculate that AP2 has evolved as a key regulatory node to coordinate CCP formation and cargo sorting and ensure high spatial and temporal regulation of CME. Seventh Framework Programme https://doi.org/10.13039/501100004963 Swiss National Science Foundation https://doi.org/10.13039/501100001711 National Institutes of Health https://doi.org/10.13039/100000002 MH61345 GM73165 ==== Body pmcIntroduction Clathrin-mediated endocytosis (CME) is the major pathway by which receptors and their ligands are concentrated and taken up into cells (Conner and Schmid, 2003; McMahon and Boucrot, 2011). CME is fundamental to cell nutrition, neurotransmission, and cellular signaling. CME begins with an initiation step in which adaptors nucleate clathrin assembly, forming nascent clathrin-coated pits (CCPs; Owen et al., 2004; Cocucci et al., 2012; Traub and Bonifacino, 2013). CCPs recruit cargo, grow, and gain curvature through continued adaptor-dependent polymerization of clathrin (Godlee and Kaksonen, 2013; Kirchhausen et al., 2014). CCPs then undergo a maturation process involving multiple endocytic accessory proteins that results in formation of deeply invaginated CCPs (Schmid and McMahon, 2007; Merrifield and Kaksonen, 2014). Finally, the GTPase dynamin assembles into collar-like structures at the necks of CCPs, where it catalyzes membrane fission and vesicle release (Schmid and Frolov, 2011; Ferguson and De Camilli, 2012; Morlot and Roux, 2013). Adaptor protein 2 (AP2), the major clathrin adaptor protein, is a heterotetramer (α, β2, µ2, and σ2 subunits) that forms a large globular core structure with two appendage domains connected via long flexible linkers (Collins et al., 2002; Jackson et al., 2010; Kirchhausen et al., 2014). The α and β2 subunits contribute the appendage domains, and interactions of the β2 appendage domain and linker with clathrin are required for clathrin assembly (Shih et al., 1995; Traub et al., 1999; Kelly et al., 2014). The appendage domain of the α subunit binds to and recruits endocytic accessory proteins during the maturation process (Owen et al., 1999; Praefcke et al., 2004). The core is composed of the N-terminal domains of α and β2 subunits, as well as the µ2 and σ2 subunits that, respectively, bind to either Yxxφ-based (where φ indicates a hydrophobic residue) or dileucine (diLeu)-based (Ohno et al., 1996; Owen and Evans, 1998; Kelly et al., 2008; Mattera et al., 2011) internalization motifs on transmembrane cargo proteins. AP2 also harbors three spatially distinct phosphatidylinositol-4,5-bisphosphate (PIP2) binding sites, one on each of the α, β2, and µ2 subunits (Gaidarov and Keen, 1999; Collins et al., 2002; Höning et al., 2005). A comparison of the crystal structures of the AP2 core, solved in the presence or absence of a bound cargo peptide, shows that AP2 undergoes a large conformational change from a “closed,” cargo-inaccessible state to an “open” (i.e., active) conformation (Jackson et al., 2010). In the closed state, the clathrin binding site in the linker is buried within the core; hence AP2 is also unable to bind clathrin (Kelly et al., 2014). In vitro biochemical studies have suggested that the transition from the closed to open state requires PIP2 binding, is further stabilized by binding cargo peptides (Höning et al., 2005; Jackson et al., 2010; Kelly et al., 2014), and may be favored by phosphorylation of the µ2 subunit by adaptor-associated kinase 1 (AAK1; Ricotta et al., 2002). Which of these multiple interactions is required in vivo, their functional hierarchy, and how the different conformational states relate to the dynamic sequence of early events in CME has not been explored. In this work, we used sensitive live-cell total internal reflection fluorescence (TIRF) microscopy (Merrifield et al., 2002) in combination with biochemical measurements to dissect the role of low-affinity interactions with PIP2 or cargo as regulators of AP2 activation. We asked which of these interactions controls successful CCP nucleation and what is the functional and temporal relationship between the three distinct PIP2 and two cargo binding sites for CCP initiation and maturation. Finally, we investigated whether Yxxφ and diLeu cargo play identical roles in CCP initiation. To address these outstanding questions in a cell-based system, we generated stable cell lines in which wild-type (WT) AP2 subunits are replaced with mutant subunits expressed at endogenous levels. These cell lines also stably overexpress CLCa-EGFP, which incorporates into clathrin triskelions without affecting the concentration of clathrin heavy chains or perturbing CME (Gaidarov et al., 1999; Ehrlich et al., 2004; Taylor et al., 2011; Aguet et al., 2013). This approach allows simultaneous, unbiased, live-cell visualization of thousands of CCPs at a time. The comprehensive nature of this analysis allows measurement of the rates of CCP nucleation, initiation, growth, and maturation (Mettlen and Danuser, 2014) and provides robust detection and tracking of even dim, nascent CCPs (Aguet et al., 2013). Most importantly, it allows measurements of the rate and extent of clathrin assembly at nascent CCPs (Loerke et al., 2011), which is a proxy for AP2 activity at the plasma membrane (Kelly et al., 2014). Our results establish that the allosteric regulation of AP2 plays a critical role in vivo not only for directing nucleation, as was predicted, but also in subsequent stages to regulate the stepwise growth, stabilization, and maturation of nascent CCPs. This regulation involves the hierarchical interaction of multiple subunits with PIP2, as well as µ2 interactions with Yxxφ-containing cargo. We also found that AP2 interactions with cargoes containing Yxxφ motif, but not diLeu motif, critically regulate early stages of AP2 activation and CCP nucleation. Results To define the functional consequences and temporal hierarchy of AP2 activation by low-affinity interactions with PIP2 and cargo, we first established a library of htertRPE cell lines stably expressing siRNA-resistant AP2 mutants. These included hypomorphic mutations in α and β2 subunits defective in PIP2 binding, mutations in the µ2 subunit defective in PIP2 or Yxxφ-based cargo binding, and mutations in the σ2 subunit defective in binding diLeu-based cargo (Fig. 1, A and B). All mutations were designed based on known AP2 structures. Importantly, all mutants have been previously analyzed for correct folding, and their biochemical properties have been confirmed and characterized in vitro (Fig. 1 B; Owen and Evans, 1998; Collins et al., 2002; Höning et al., 2005; Kelly et al., 2008). The stable cell lines were selected by FACS to ensure that the levels of expression of mutant subunits were comparable to those of the endogenous subunits. To ensure full substitution, we also treated cells with siRNA to knock down endogenous subunits. Finally, we verified full incorporation of mutant subunits within AP2 complexes by immunoprecipitation of AP2α and immunoblotting for the mutated subunit (Fig. S1). Expression of mutant subunit and its localization in CCPs was also monitored by immunofluorescence (Fig. S1 F). Figure 1. AP2 mutants and TIRFM-based assays used in this study. (A) Table listing the in vitro biochemical phenotypes of mutant AP2 subunits, their designations, and the residues mutated. (B) Schematic representation showing the positions of mutated residues with respect to the closed (left) and open (right) conformations of AP2. (C) Schematic representation of the CCP parameters measured by TIRF imaging and quantitative image analysis in this work. Tyr, tyrosine. These mutations in individual subunits reduce, but do not completely eliminate, intact AP2 activity, which is required for CCP initiation (Motley et al., 2006; Aguet et al., 2013). Hence the temporal and functional hierarchy of these low-affinity interactions can be inferred through the quantitative analyses of CCP dynamics using live-cell TIRF microscopy (Fig. 1 C). We previously developed a highly sensitive object-based detection method to accurately detect all clathrin-labeled structures (CLSs) and quantitative methods to distinguish subthreshold CLSs (sCLSs) from bona fide CCPs (Aguet et al., 2013). sCLSs, which are short-lived and dim, include AP2-independent stochastic clathrin assemblies that occur within the TIRF field (i.e., <100 nm from the cell surface), as well as AP2-dependent early nucleation events that fail to grow (Fig. 1 C; Aguet et al., 2013). In contrast, bona fide CCPs are quantitatively defined by the continued accumulation of clathrin and growth past a threshold intensity (Fig. 1 C; see Materials and methods). In addition, a fraction (30–50%) of bona fide CCPs fail to mature and are aborted (Ehrlich et al., 2004; Loerke et al., 2009; Aguet et al., 2013). Abortive CCPs fail to gain curvature or recruit a burst of dynamin-2 before disappearing from the TIRF field. We quantified the effect of the AP2 mutations on sequential stages of CCV formation by measuring multiple independent properties of CCP dynamics, including the initiation density of sCLSs and bona fide CCPs, the rate and extent of clathrin assembly, and the efficiency of CCP maturation, as assessed by the fraction of short-lived (<20 s) abortive and longer-lived (40–60 s) productive CCPs (Fig. 1 C). AP2α–PIP2 interactions are required at multiple early stages of CME Previous studies have shown that bulk depletion of plasma membrane PIP2 abolishes CCP assembly (Boucrot et al., 2006; Zoncu et al., 2007). However, because many components of the endocytic and cytoskeletal machinery bind PIP2 and these experiments involved its prolonged depletion, the specific roles of AP2–PIP2 interactions in CCP nucleation and their potential roles after initiation remain unknown. Moreover, the functional and temporal hierarchy of the three spatially distinct PIP2 binding sites (Fig. 1 B) and their roles in AP2 activation in cells have not been defined. Therefore, we analyzed the effects of mutating each site independently. To validate our approach, verify the sensitivity of our live-cell imaging tools, and set a point of reference for other AP2 mutations, we first analyzed the role of the surface-accessible PIP2 binding site on the α subunit (Fig. 2 A). The consequence of abrogation of α–PIP2 interaction on clathrin assembly is relatively well understood in vitro (Kelly et al., 2014), and previous studies reported an ∼50% decrease in CME of transferrin (Tfn) in cells expressing this mutant (Motley et al., 2006), but the effects on CCP dynamics have not been studied. Figure 2. Binding of PIP2 to the α subunit is essential for allosteric activation of AP2 to trigger clathrin polymerization and CCP initiation. (A) Schematic representation of the possible early roles of α–PIP2 interactions. The question mark and double arrowheads point to potential roles in enhancing the rates and or extents of clathrin polymerization, CCP nucleation, cargo recruitment, and CCP maturation. (B) Initiation density of all detected subthreshold CLSs and bona fide CCPs for the indicated wt or mutant cells (≥15 cells per condition, #CCPs αwt 32,320, #CCPs αPIP2− 20,510). Box plots show medians, 25th and 75th percentiles, and outermost data points. ***, P ≤ 0.001, unpaired t test. (C) Mean clathrin fluorescence intensity traces of lifetime cohorts of CCPs from αwt (gray) and αPIP2− (blue) cells. Intensities are shown as mean ± SE calculated from ≥15 cells per condition. (D) The slope of intensity trace (averaged in the time interval 3–8 s of the elapsed lifetime) in αwt and αPIP2− cells. A.U., arbitrary units; Fluo., fluorescence. In cells expressing αPIP2−, the rates of initiation of bona fide CCPs were reduced by ∼60% (Fig. 2 B), providing a mechanism for the reported defect in CME. Importantly, we also observed an ∼50% decrease in the rate of appearance of sCLSs, which represent failed nucleation attempts. Thus, the α–PIP2 binding site plays a critical role at the very earliest stages of CCP nucleation to ensure activation of AP2 only at the plasma membrane (PM). To more directly assess AP2 activity, we measured the rates and extents of clathrin recruitment to nascent CCPs by measuring changes in CLC-EGFP intensity for bona fide CCPs binned within distinct lifetime cohorts. We observed a 30% decrease in the extent of clathrin assembly at CCPs in αPIP2− mutant cells for all lifetime cohorts (Fig. 2 C). Because we confine our analyses to diffraction-limited CCPs (Aguet et al., 2013), these data indicate that even productive CCPs formed in the presence of this AP2 mutant were smaller. Importantly, the rate of clathrin polymerization, measured during the initial phase of CCP growth, (Fig. 2 D) also decreased by ∼45% in the αPIP2− mutant cells compared with WT cells (Fig. 2 D and Fig. S2). These results establish that α–PIP2 interactions are required to regulate clathrin polymerization at nascent CCPs. The initial rate of clathrin recruitment is critical for stabilizing nascent CCPs, as shorter-lived CCPs recruited clathrin at slower initial rates than longer-lived CCPs (Figs. 2 D and S2). This was true in both WT and mutant cells (Fig. 2 C). Whether AP2–PIP2 interactions are required for subsequent stages of CCP maturation remains an open question. Indeed, it has been suggested based on theoretical considerations (Schmid and McMahon, 2007) and in vitro studies (Dannhauser and Ungewickell, 2012) that clathrin assembly is sufficient to drive subsequent stages of CME. To test this, we next analyzed the role of α–PIP2 interactions on the lifetime distribution of bona fide CCPs, a measure of CCP maturation. αPIP2− cells exhibited a significant increase in the fraction of shorter-lived (≤20 s) CCPs compared with WT cells (Fig. 3 A); whereas longer-lived (40–60 s) populations were unaffected. These data suggest a defect in CCP maturation and an increase in abortive events. Consistent with this interpretation, we also observed a decrease in the fraction of CCPs that recruit dynamin-2 (from 55% in WT cells to ∼37% in αPIP2− cells; Fig. 3 C; Taylor et al., 2012; Aguet et al., 2013; Grassart et al., 2014). These results establish that α–PIP2 interactions are required beyond initiation to stabilize nascent CCPs. Figure 3. Sustained binding of PIP2 to AP2 α subunit is required for CCP maturation. (A) Fraction of CCPs found in short-lived versus longer-lived lifetime cohorts in αWT (gray) and αPIP2− (blue) cells. Data shown are mean ± SD (n > 100,000 CCPs from three independent experiments); n.s., not significant; ***, P < 0.001. Lifetime distributions of all bona fide CCPs (black lines), dynamin-2 (DYN2)-positive CCPs (green lines), and DYN2-negative CCPs (blue lines) in αWT (B) and αPIP2− (C) cells. (D) The ratio of epifluorescence (EPI):TIRF intensity levels for individual CCPs is indicative of curvature acquisition. EPI:TIRF ratio for individual CCPs plotted as a function of CCP lifetime in αWT (E) and αPIP2− (F) cells. Heatmap indicates frequency. EM images of “unroofed” htertRPE cells reconstituted with either αWT (G) or αPIP2− (H). Bottom panels show higher-magnification view of the indicated area. Bars: (top) 500 nm; (bottom) 200 nm. To further characterize the defect in CCP maturation, we analyzed CCP dynamics using near-simultaneous TIRF and epifluorescence microscopy. In TIRF, the evanescent illumination field decays exponentially as it penetrates the adherent cell surface; therefore, the ratio between TIRF and epifluorescence intensities provides a measure of CCP curvature acquisition as a function of lifetime (Saffarian and Kirchhausen, 2008; Aguet et al., 2013; Fig. 3 D). We detected a twofold increase in the number of flat structures (TIRF/epifluorescence ≤1) with short lifetimes (≤20 s) in αPIP2− cells (Fig. 3 F) compared with WT cells (Fig. 3 E). These small, flat CCPs could be directly observed by negative-stain electron microscopy in αPIP2− cells (Fig. 3, G and H). Additionally, abrogation of α–PIP2 binding led to a decreased concentration of both cargo and AP2 at CCPs (Fig. S3). Together, these effects lead to an increase in the number of abortive events. Altogether, our data explain the observed defect in CME in αPIP2− cells (Motley et al., 2006). We show that CCP nucleation in cells is dependent on activation of AP2 through interactions between plasma membrane PIP2 and the surface exposed binding site on the α-subunit. Moreover, these interactions play a critical role not just in nucleating clathrin assembly at nascent CCPs, but in all subsequent steps of CME, including CCP stabilization, growth, maturation, and cargo recruitment. AP2 β2–PIP2 interactions are equally critical for early stages of CME The β2-subunit of AP2 encodes a second PIP2 binding site, formed by six surface-exposed lysine residues (Collins et al., 2002; Jackson et al., 2010), which has not been studied in vivo. We find that β2PIP2− cells (Fig. 1 and Fig. S1, B and F). phenocopied αPIP2− cells in that they exhibited decreased rates of initiation of both sCLSs and bona fide CCPs (Fig. 4 A), reduced extents of clathrin polymerization at bona fide CCPs (Fig. 4 B), and a shift in lifetime distribution toward short-lived versus productive CCPs (Fig. 4 C). Consistent with these findings, CME of transferrin receptors was inhibited by ∼50% in β2PIP2− cells, the same extent seen in αPIP2− cells (Fig. 4 D). Thus, we conclude that although the α– and β2–PIP2 binding sites can operate independently, full allosteric activation of AP2, which is needed in cells to trigger rapid clathrin assembly and efficient CCP maturation, requires binding of PIP2 to both α and β2 subunits. Figure 4. Binding of PIP2 to AP2 β subunit is equally essential for efficient CCP initiation and clathrin polymerization. (A) Initiation density of subthreshold CLSs and bona fide CCPs for the indicated wt or mutant cells (18 cells per condition). Box plots show medians, 25th and 75th percentiles, and outermost data points. ***, P ≤ 0.001, unpaired t test. (B) Mean clathrin fluorescence intensity traces in lifetime cohorts of CCPs from β2wt (gray) and β2PIP2- (blue) reconstituted cells. Intensities are shown as mean ± SE calculated from 18 cells per condition. A.U., arbitrary units; Fluo., fluorescence. (C) Fraction of CCPs found in short-lived versus longer-lived cohorts for βWT (gray) and βPIP2− (blue) cells. ***, P < 0.001. (D) Transferrin receptor internalization was measured in βWT, αPIP2−, and βPIP2− cells using a monoclonal anti-TfnR antibody as ligand. Percentage of TfnR uptake was calculated relative to the initial total of surface-bound antibody. Data represent mean ± SD, n = 4. ***, P ≤ 0.005, unpaired t test. µ2–PIP2 binding provides downstream stabilization of AP2 at the PM The µ2 subunit is pivotal for AP2 function. It harbors a third PIP2 binding site (Collins et al., 2002; Jackson et al., 2010; Fig. 5 A) and is essential for Yxxφ-based cargo recognition (Ohno et al., 1995; Owen and Evans, 1998). It is also a substrate for phosphorylation by AAK1 (Olusanya et al., 2001; Conner and Schmid, 2002; Ricotta et al., 2002). The PIP2 binding site on µ2 is located at the C terminus and is in the closed conformation (Figs. 1 B and 5 A; Jackson et al., 2010). There are contradictory findings as to whether µ2–PIP2 interactions are required for CME. One study reported a strong dominant negative effect of overexpression of a µ2 mutant defective in PIP2 binding (Rohde et al., 2002), whereas a second study showed that the same µ2 mutant could fully rescue CME in µ2-deficient cells (Motley et al., 2006). We reproduced the latter finding by showing that Tfn uptake was unaffected in htertRPE cells reconstituted with the identical µ2PIP2− mutant (Fig. S4 A). Figure 5. Binding of PIP2 to the µ2 subunit is required downstream of initial AP2 activation for later stages of CCP maturation. (A) Schematic representation of the possible early roles of α–PIP2 interactions. The question mark and double arrowheads point to potential roles in enhancing the rates and or extents of clathrin polymerization, CCP nucleation, cargo recruitment, and CCP maturation. (B) Initiation density of all subthreshold CLSs and bona fide CCPs for the indicated wt or mutant cells (≥22 cells per condition, #CCPs µ2wt 27,943, #CCPs µ2PIP2− 36,523). Box plots show medians, 25th and 75th percentiles, and outermost data points. ***, P ≤ 0.0005, unpaired t test. n.s., not significant. (C) Mean clathrin fluorescence intensity traces in lifetime cohorts of CCPs from µ2wt (gray) and µ2PIP2− (blue) reconstituted cells. Intensities are shown as mean ± SE calculated from 17 cells per condition. (D) Slope of intensity trace (averaged in the time interval 3–8 s of the elapsed lifetime) in µ2wt (gray) and µ2PIP2− (blue) cells. (E) Fraction of CCPs found in short-lived versus longer-lived lifetime cohorts in µ2WT (gray) and µ2PIP2− (blue) cells. ***, P < 0.001. (F) Lifetime distributions of all bona fide CCPs (black lines), dynamin-2 (DYN2)-positive CCPs (green lines), and Dyn2-negative CCPs (blue lines) in µ2PIP2− cells. We next applied our more sensitive live-cell imaging assays to more directly test whether the PIP2 binding site on µ2 plays a role in AP2 activation and early stages of CME. Unexpectedly, and in diametric contrast to results obtained when surface-localized PIP2 binding sites were disrupted, we observed an increased rate of initiation of bona fide CCPs (Fig. 5 B) and an increased rate and extent of clathrin recruitment to CCPs in the µ2PIP2− cells (Fig. 5, C and D). No significant changes in the rates of appearance of stochastic assemblies or early nucleation events (i.e., sCLSs) were observed, indicating that µ2–PIP2 binding is not required for these earliest steps. Importantly, the increased rate of CCP initiation and growth of CCPs in µ2PIP2− cells was not associated with a higher likelihood of vesicle formation. Thus, as observed with the αPIP2− and β2PIP2− mutants, there was a significant increase in the fraction of short-lived, presumably abortive CCPs in the µ2PIP2− cells compared with µ2WT cells (Fig. 5 E). Correspondingly, we observed an increase in short-lived, flat CCPs (Fig. S4, B and C) and a decrease in the fraction of dynamin-2–positive CCPs (from 55% to 42%; Fig. 5 F). Unlike in α/β2PIP2- mutants, the recruitment of cargo into CCPs and the quantity of AP2 in µ2PIP2− CCPs was not affected (Fig. S3). Finally, given that the PIP2 binding site on µ2 corresponds to two large patches (Jackson et al., 2010) we also examined the effect of mutating additional positively charged residues implicated in PIP2 binding (K167, R169, R170, K365, and K367) and obtained results similar to those described for µ2PIP2− (unpublished data). From these studies, we conclude that µ2–PIP2 binding functions downstream of α– and β2–PIP2 interactions and is required for sustained activation of AP2 and efficient CCP maturation. Increased AAK1 activity compensates for µ2–PIP2 binding deficiency The increased rate of CCP initiation observed in µ2PIP2− cells may reflect activation of a compensatory mechanism in response to the decreased efficiency of CCP maturation to restore CME to normal levels. Indeed, we have previously identified compensatory mechanisms that restored CME in cells expressing an α-appendage domain deletion mutant of AP2 (ΔαAD cells; Aguet et al., 2013). Tfn uptake was unaffected in the ΔαAD cells despite profound defects in CCP maturation (Aguet et al., 2013; Reis et al., 2015). We therefore tested whether alternate, compensatory mechanisms might also have restored efficient clathrin recruitment and CME in µ2PIP2− cells. The phosphorylation of µ2 on T156 by AAK1 has been proposed to stabilize the open conformation of AP2 (Ricotta et al., 2002; Jackson et al., 2010). Moreover, both in vitro (Conner et al., 2003) and in vivo (Jackson et al., 2003) evidence suggests that AAK1 activity is stimulated by assembled clathrin. Thus, we wondered whether the increase in the rate and extent of clathrin assembly in µ2PIP2− cells might reflect the activation of an AAK1-dependent feed-forward loop (Fig. 6 A). To test whether increased AAK1 activity might be compensating for defects in CCP maturation in µ2PIP2− cells, we measured levels of phosphorylation of T156 on µ2. µ2PIP2− cells exhibited a 30% increase in T156 phosphorylation compared with µ2WT cells (Fig. S4 D; quantified in Fig. 6 B). In contrast, µ2 T156 phosphorylation levels were reduced in αPIP2− cells. To determine whether enhanced AAK1 activity was required to compensate for the PIP2 binding defect in µ2, we compared Tfn receptor (TfnR) internalization in control, µ2PIP2−, and αPIP2− cells with and without treatment with a specific chemical inhibitor of AAK1 (Compound 2; Bamborough et al., 2008). Control experiments verified that Compound 2 effectively reduced µ2 phosphorylation (Fig. S4 E). As we predicted, TfnR uptake in µ2PIP2− cells was much more sensitive to AAK1 inhibition than that in WT cells (∼50% inhibition in µ2PIP2− cells compared with ∼20% in WT cells; Fig. 6 C). In contrast, the residual levels of TfnR internalization in αPIP2− cells were unaffected by Compound 2, indicating that AAK1 is not active in these cells. Because AAK1 is recruited to CCPs through interactions with both AP2 and clathrin, this finding likely reflects the impaired rates of CCP initiation and growth in these mutant cells. We were unable to study the consequence of total ablation of µ2T156 phosphorylation because we found that the µ2T156A mutant was poorly expressed and inefficiently incorporated into AP2 complexes (Fig. S1 E). Figure 6. Increased phosphorylation of µ2 at T156 compensates to maintain efficient TfnR internalization in µ2PIP2− cells. (A) Schematic representation of the potential roles for AAK1 activation and phosphorylation of T156 on µ2. The question mark and arrows point to possible compensatory mechanisms and a positive feed-forward loop that can restore efficient CME in µ2PIP2− cells. (B) Quantification (mean ± SD, n = 3; two-tailed Student’s t tests were used to assess statistical significance: *, P < 0.05) of total and phosphorylated µ2pT156 subunit in µ2WT, µ2PIP2-, and αPIP2− cells (see also Fig. S5 B). (C) TfnR uptake measured at 10 min in control, µ2PIP2−, and αPIP2− cells with or without treatment with the 10-µM AAK1 inhibitor (inhib), Compound 2. Data shown are mean ± SD, n = 3; normalized to total surface bound; *, P < 0.05; ***, P < 0.0005. n.s., not significant. These results suggest that µ2–PIP2 binding promotes CCP maturation but that defects in this activity can be compensated for by activation of AAK1 and phosphorylation of µ2. Our results provide in vivo evidence that AAK1 phosphorylation can indeed drive a conformational change similar to that triggered by PIP2 and/or cargo binding to release clathrin binding sites and activate AP2 for clathrin assembly. Our data are also consistent with previous studies (Conner et al., 2003; Jackson et al., 2003) showing that clathrin assembly, which is enhanced in µ2PIP2− cells but decreased in αPIP2− cells, stimulates AAK1 activity. Together, these findings provide strong evidence for the role of AAK1-mediated phosphorylation of µ2 in providing positive feedback to enhance the clathrin assembly activity of AP2 (Fig. 6 A). We conclude that µ2–PIP2 binding operates in cells downstream of α– and β2–PIP2 interactions to stabilize active AP2 complexes already on the PM, thus enhancing the efficiency of growth and stabilization of bona fide CCPs. Importantly, our data show that sustained interactions of PIP2 with binding sites on all three subunits are required to maintain AP2 activation to ensure efficient CCP maturation and CME. Essential role for cargo binding in CCP nucleation Although a consensus is emerging that cargo recruitment can stabilize growing CCPs (Ehrlich et al., 2004; Loerke et al., 2009; Traub, 2009), whether AP2–cargo interactions are required for CCP nucleation remains a matter of debate (Godlee and Kaksonen, 2013). One group reported that AP2 and clathrin assembled several seconds before detection of cargo (Cocucci et al., 2012), and another group reported that TfnR are recruited concomitantly with AP2 and clathrin in nascent CCPs (Liu et al., 2010). Other studies approached this question by manipulating the levels, activities, or clustering of single cargo receptors (Loerke et al., 2009; Liu et al., 2010; Mettlen et al., 2010). We decided to take a different approach by globally eliminating the binding of one of the two major classes of internalization motifs, the Yxxφ motif and the diLeu motif, to AP2 (Fig. 7 A). Thus, we replaced endogenous subunits with either a µ2cargo− mutant that is unable to bind cognate cargoes carrying the Yxxφ motif or a σ2cargo− mutant that is unable to bind cargoes carrying the diLeu motif (Fig. 1 A). We first confirmed these phenotypes biochemically. As expected, internalization of a model Yxxφ cargo (CD8-YAAL; Fig. S5) was strongly impaired in the µ2cargo− mutant cell line (>60% inhibition), whereas internalization of a model diLeu cargo (CD8-EAAALL; Fig. S5) was strongly impaired (>50% inhibition) in cells expressing σ2cargo− (Fig. 7 B). Interestingly, whereas the uptake of the orthogonal, Yxxφ-based cargo was not significantly affected in the σ2cargo− cells, diLeu cargo uptake was reduced in the µ2cargo− cell line, albeit to a lesser extent (∼30% inhibition) than either the cognate Yxxφ-containing (Fig. 7 B) or diLeu-containing cargo in the σ2cargo− cells. These relative cargo-sorting activities were confirmed by directly measuring the concentration of different cargo molecules in CCPs (Fig. 7 C). σ2cargo− cells were specifically defective in recruitment of diLeu cargo to CCPs, whereas µ2cargo− cells showed a general defect in all cargo classes, including FXNPXY-containing cargo and the EGF receptor. Figure 7. Activation of AP2 by binding to YXXφ sorting signals is necessary for CCP nucleation. (A) Schematic representation of the potential role of σ2 and µ2 interactions with their cognate (diLeu and YXXφ, respectively) or orthogonal cargo on AP2 activation and CCP nucleation. (B) Internalization of CD8 chimeras containing either the YXXφ (cognate for µ2, orthogonal for σ2) or diLeu (cognate for σ2, orthogonal for µ2) sorting signals was followed using CD8 mAb (data shown are mean ± SD, n = 3; normalized to total surface bound; **, P < 0.05; ***, P < 0.005). (C) Heatmap representing the changes in Pearson correlation coefficient (PC) between EGFP-CLCa, respective CD8 chimeras, and other CME cargo showing efficiency of cargo loading into CCPs in µ2cargo− and σ2cargo− cells. (D) Initiation density of all detected CLSs and bona fide CCPs for the indicated wt or mutant cells (≥35 cells per condition, #CCPs µ2wt 45,054, #CCPs µ2cargo− 38,120). Box plots show medians, 25th and 75th percentiles, and outermost data points. ***, P < 0.0005; ****, P < 0.0001, t test. n.s., not significant. (E) Mean clathrin fluorescence (fluo.) intensity traces in lifetime cohorts of CCPs from µ2WT (gray) and µ2cargo− (blue) reconstituted cells. Intensities are shown as mean ± SE calculated from 20 cells per condition. A.U., arbitrary units. (F) Slope of intensity trace (averaged in the time interval 3–8 s of the elapsed lifetime) in µ2wt (gray) and µ2cargo− (blue) cells. To explore the basis for the more general defect in CME caused by the µ2cargo− mutant, we next looked at CCP dynamics. Cells expressing the µ2cargo− mutant showed a significant decrease in initiation rates of both sCLSs and CCPs, whereas there was no effect in σ2cargo− cells (Fig. 7 D). Similarly to αPIP2− cells, we detected a decrease in the rate and extent of clathrin recruitment in all lifetime cohorts of bona fide CCPs (Fig. 7, E and F). Thus the µ2cargo− mutant phenocopies the αPIP2− and βPIP2− mutants with regard to its effects on CCP initiation (Fig. 7 F). From this, we conclude that AP2–cargo interactions play an essential, early role in the activation of AP2 and productive nucleation of CCPs. Our studies also reveal an apparent functional hierarchy in cargo binding, because interactions with Yxxφ-containing cargo showed a stronger and more global effect on CME than those of diLeu-containing cargo. This hierarchy could reflect the reported approximately threefold difference in binding affinity of AP2 to Yxxφ- versus diLeu-containing cargo motifs (Höning et al., 2005; Jackson et al., 2010) or differences in the ability of these two motifs to stabilize the open conformation of AP2 during early, critical stages of CCP assembly. If the former, then we would predict that overexpressing diLeu motif–containing cargo should, by mass action, rescue the µ2cargo− defect in CCP initiation. However, adenoviral-driven overexpression of the diLeu-containing CD8 chimera failed to rescue the µ2cargo− defect (Fig. 8 A), indicating that there is indeed a functional hierarchy of cargo binding sites in the core of the AP2 complex, and that the binding of Yxxφ-based cargo to µ2 plays a pivotal role in AP2 activation and CCP nucleation. Figure 8. Selective role of YXXφ-bearing cargo for AP2 activation and model for sequential allosteric regulation of AP2 by PIP2 and cargo interactions. (A) Overexpression of CD8 chimera containing a diLeu sorting motif does not rescue initiation of CCPs in µ2cargo− cells (≥16 cells per condition; CCPs µ2wt 18,235, #CCPs µ2cargo− 16,573, #CCPs µ2cargo−[CD8-EAAALL] 15,836). n.s., not significant. (B) Bioinformatics search for human transmembrane proteins containing YXXφ- or diLeu-based sorting motif revealed overrepresentation of YXXφ motif–bearing cargo. (C) Model for regulation of CCP nucleation and maturation through allosteric activation of AP2 by sequential and hierarchical interactions with its ligands, PIP2, and YXXφ-bearing cargo. Initiation (step 1) involves interactions between surface-exposed PIP2-binding sites on both the α and β2 subunits, as well as interactions between YXXφ-bearing cargo and the µ2 subunit to allow AP2 to recruit clathrin. Rapidly assembling clathrin (step 2) can activate AAK1 kinase, which phosphorylates µ2, creating a positive feed-forward loop that drives efficient CCP maturation. Full activation of AP2, which is required for CCP growth and stabilization, occurs when the µ2 subunit also engages PIP2. Early CCP intermediates formed by AP2 mutants defective in α–, β2–, or µ2–PIP2 binding are impaired in curvature generation and the recruitment of dynamin-2 (Dyn2) and exhibit a greater tendency to abort. To further probe this potential functional hierarchy, we performed a bioinformatics analysis of the PM transmembrane-proteome based on identified internalization-coding motifs, Yxxφ and diLeu. To account for the higher complexity of the diLeu motif, we included all previously reported variations (see Materials and Methods). Our analysis revealed that the internalization motifs are not equally distributed (Fig. 8 B): of 3,705 plasma membrane receptors, 1,250 (33%) uniquely contain a Yxxφ motif, whereas only 275 (7%) uniquely contain a diLeu motif. These results are consistent with our finding that Yxxφ motif–containing cargo function synergistically with PIP2 in vivo to efficiently nucleate CCPs and initiation cargo sorting and CME. Discussion AP2, which is absolutely required for CCP initiation in higher eukaryotes, is the most abundant endocytic adaptor, with an exceptionally high degree of conservation of all subunits from yeast to human (Schledzewski et al., 1999). Initiation of productive CCPs requires nucleation, rapid growth, and stabilization. Quantitative analysis in living cells allowed us to measure these discrete early stages of CME and revealed how sequential allosterically regulated conformational changes in AP2 adaptors are required to drive the vectorial nature of these events. We show that interactions with both PIP2 and cargo are required for full activation of AP2 to couple cargo detection and sorting with efficient CCP formation and maturation. Our data establish the following sequence: (1) AP2 activation via α– and β2–PIP2 binding and Yxxφ cargo recruitment drives clathrin polymerization at the PM and CCP nucleation; (2) stabilization of active AP2 complexes on the PM occurs via µ2–PIP2 engagement or µ2-T156 phosphorylation, which itself is regulated via assembled clathrin through an AAK1-dependent feedback loop to enhance clathrin polymerization; and (3) sustained interactions of PIP2 with binding sites on α, β2, and µ2 are required for CCP stabilization and efficient maturation (Fig. 8 C). Thus, in contrast to previous suggestions that clathrin interactions substitute for early AP2 interactions in stabilizing nascent CCPs and driving CCP maturation (Schmid and McMahon, 2007), our results establish a sustained requirement for active AP2 complexes. Recent studies have proposed that the muniscin family of endocytic accessory proteins (FCho and SGIP) can activate AP2 in cells (Hollopeter et al., 2014; Umasankar et al., 2014), and while this study was in preparation, a new article (Ma et al., 2016) suggested that AP2 recruitment to sites of CCP formation is assisted by preformed FCHO-Eps15 complexes. In that model, the capture of AP2 reflects the formation of FCho–Eps15–AP2 nanoclusters, and AP2–PIP2 interactions are required only subsequently. Although the formation of nanoclusters may lead to an increase in AP2 residence time on the PM, our data show that in the presence of endogenous levels of FCho and Eps15, efficient AP2 recruitment to the PM and CME initiation is entirely dependent on the AP2–PIP2 and AP2–cargo interactions. Consistent with this, redirection of the FCho µ homology domain to the Golgi is sufficient to recruit Eps15, but not AP2, complexes (Ma et al., 2016), presumably because of the lack of PIP2 on Golgi membranes. Altogether, our findings reveal that a temporal hierarchy of interactions govern AP2 activation to regulate early stages of CCP formation and stabilization. They also provide a molecular explanation for the contributions of both PIP2 (Antonescu et al., 2011) and cargo (Ehrlich et al., 2004; Loerke et al., 2009) to CCP maturation. Like AP2, most of the components of the CME machinery interact via low-affinity binding to divergent motifs that are broadly expressed on multiple components (Owen et al., 1999; Praefcke et al., 2004; Schmid and McMahon, 2007). Therefore, understanding the functional hierarchy of AP2 interactions with molecular precision has broad relevance, as it represents a general paradigm for cargo transport and sorting events (Pandey, 2009; Perkins et al., 2010; Van Roey et al., 2012). Previous studies have shown that acute and bulk depletion of PIP2 from the PM results in loss of all CCPs (Boucrot et al., 2006; Zoncu et al., 2007). Our dissection of the specific roles of the three AP2–PIP2 binding sites reveals a previously unappreciated complexity and multiple roles for PIP2 throughout the lifetime of CCPs. Thus, we found that nucleation of all detectable clathrin assemblies was reduced by ∼50% when either of the surface-exposed PIP2 binding sites on α or β2 was mutated. Because previous studies showed that the rate of initiation of bona fide CCPs was nearly ablated upon siRNA knockdown of AP2 (Aguet et al., 2013), we interpret the difference to reflect the residual (∼50%) activity of a single surface PIP2 binding site on AP2. These data refine structural models (Kelly et al., 2014) and show that both α– and β2–PIP2 interactions are equally important for full activation of AP2 to expel the β2 clathrin binding site from the AP2 core and restrict clathrin assembly to the plasma membrane (Fig. 8 C). Although in vitro studies have established a role for AAK1 in phosphorylating µ2 and stabilizing the open conformation of AP2, in vivo evidence of a role for AAK1 in CME is lacking (Pelkmans et al., 2005). Our finding that AAK1 activation can compensate for defects in µ2–PIP2 interactions suggests that AAK1 activity can be fine-tuned in response to defects in early stages of CME, in part through a positive feedback loop dependent on clathrin assembly (Fig. 8 C, step 2). The role of cargo in CCP nucleation has been debated. Our data show that µ2 interactions with Yxxφ motif–containing cargo efficiently modulate AP2 activation at the PM and are required for CCP nucleation. In contrast, interactions between σ2 and diLeu-based cargo have little effect. This hierarchy in AP2–cargo binding in vivo is determined by (a) Yxxφ-containing cargo abundance and (b) their allosteric effect on AP2 activity. The differential allosteric effect of the two sorting signals on AP2 observed in our experiments could be explained by the different structural requirements for exposing the Yxxφ and diLeu motif binding sites on AP2. Only a minimal conformation change from the closed state is absolutely required to expose the diLeu binding site, and in this partially unlocked conformation, the canonical Yxxφ binding site remains inaccessible (Canagarajah et al., 2013). Full activation is necessary to accommodate Yxxφ motif, and this might be accompanied by efficient release of auto-inhibitory binding between the β2 and µ2 subunits to fully expose the clathrin binding site (Kelly et al., 2008; Jackson et al., 2010). Hence, unlike diLeu motif binding to AP2, the global disruption of interaction of Yxxφ with AP2 has profound consequences on CCP nucleation. Based on our observations and existing structural evidence, we can conclude that a broad spectrum of attainable AP2 conformations can drive cargo sorting. It still needs to be determined whether and how this differential regulation affects sorting of the 858 receptors identified in our screen containing both types of the motifs. Interestingly, both diLeu and Yxxφ motifs are subject to regulation by phosphorylation. For example, phosphorylation of Ser residues in the vicinity of diLeu motifs is known to enhance their interaction with AP2 (Pitcher et al., 1999). In contrast, tyrosine phosphorylation within Yxxφ motifs negatively regulates their interaction with AP2 (Marchese et al., 2008; Traub and Bonifacino, 2013). Most likely, various posttranslational modifications in cargo, as well as adaptors, act together with internalization motifs to regulate cargo sorting into CCPs. Therefore future experiments will unravel the temporal and functional hierarchy of kinase and phosphatase networks in signaling cascades that orchestrate cargo sensing during CCP formation. AP2 is indispensable for normal embryonic development and CME in vertebrates, Drosophila melanogaster, and Caenorhabditis elegans (González-Gaitán and Jäckle, 1997; Mitsunari et al., 2005; Gu et al., 2008); however, it is not required for viability or CME in yeast (Weinberg and Drubin, 2012). In yeast, the initiation phase of CME is remarkably flexible, such that many early-arriving adaptors, including homologues for Eps15 and FCHo, share the initiation function in a potentially redundant manner (Brach et al., 2014). Unlike in mammalian cells, monoubiquitylation is the main internalization signal, and all of the yeast initiation components contain ubiquitin-binding domains (Weinberg and Drubin, 2012). Strikingly, neither PIP2 (Sun and Drubin, 2012) nor Yxxφ interactions are required for nucleation of endocytic sites in yeast. There are fewer endocytic Yxxφ-containing cargoes (Munn, 2001; Weinberg and Drubin, 2012), and the critical residue in the Yxxφ-binding site on µ2 is not conserved in yeast. Thus, AP2 complexes appear to have evolved their role in mammals as allosteric regulators of CCP nucleation, through PIP2 and cargo interactions. We propose that this functional hierarchy, based on allosteric regulation of AP2 activity, provides a mechanism to link cargo capture and sorting with CCP initiation and to provide greater spatial and temporal control of CME in mammalian cells. Materials and methods Generation of constructs and viruses AP2 µ and α constructs The AP2 µ2 and α sequences were generated using standard site-directed mutagenesis within siRNA-resistant cDNA encoding the full-length subunits, provided by M.S. Robinson (Cambridge Institute for Medical Research, Cambridge, England, UK; Motley et al., 2006). The α subunit contained a brain-specific splice insert, whereas the µ2 subunit contained a myc tag within flexible linkers (Motley et al., 2006). Resulting cDNAs were inserted into retroviral bicistronic IRES vector PMIB6 (Aguet et al., 2013) using conventional restriction enzyme cloning techniques. AP2 β construct cDNA encoding human isoform 1 of the full-length AP2β subunit (937 aa; Uniprot identifier P63010-1) was obtained from C. Antonescu (Ryerson University, Toronto, ON, Canada). An siRNA-resistant form was created by silent mutation of the siRNA target sequence 5′-TGGCAGAACTGAAAGAATA-3′ to 5′-TGGCAGAGTTAAAAGAATA-3′ (underline denotes changes). For the recognition of ectopically expressed AP2β, a Flag tag sequence (DYKDDDDK) was inserted by PCR into the cDNA at residue 602 within the flexible linker region. The β2PIP2− plasmid was generated by inserting a mutant fragment containing residues 5E/12E/26E/27E/29E/36E acquired as “gblock” (Integrated DNA Technologies) using conventional cloning techniques with restriction enzymes. AP2 σ construct cDNA encoding AP2σ with a C-terminal Flag-tag was acquired as a gBlock DNA fragment (Integrated DNA Technologies), designed as a Megaprimer (Miyazaki, 2011) that contained flanking 24-bp-long regions of homology to PMIB6 for its insertion by PCR. Four silent mutations conferring siRNA resistance were designed in the siRNA target sequence 5′-CTTCGTGGAGGTCTTAAACGA-3′ to 5′-TTTTGTAGAAGTCTTAAACGA-3′. The σ2 sequence was altered by point mutation V88D to abrogate the diLeu-motif binding using standard site-directed mutagenesis. CD8 chimeras cDNAs coding for CD8-YAAL or CD8-EAAALL chimeras were provided by M.S. Robinson (Kozik et al., 2010). cDNAs were subcloned from the original PiresNeo2 vector into an adenoviral pADTET T3T7 vector by seamless cloning and in vivo recombination (Lu, 2005). Preparation of viruses Recombinant adenoviruses for tet-regulated CD8 chimeras were generated as previously described (Hardy et al., 1997; Damke et al., 2001). Retroviruses encoding AP2 adaptins were generated as previously described (Aguet et al., 2013). Cell culture htertRPE-1 cells were obtained from ATCC and used because they have a normal karyotype, are suitable for genome editing, are nontransformed, and have a diffraction-limited and dynamic population of CCPs when seeded on gelatin. htertRPE-1 cells stably expressing CLCa-EGFP and reconstituted with WT or mutant AP2 subunits were derived as previously described (Aguet et al., 2013). In brief, CLCa-EGFP–expressing cells were infected with retrovirus and FACS-sorted 3 d after infection into cohorts based on BFP fluorescence. Stable cell lines expressing near-endogenous levels of α, β2, µ2, and σ2 adaptins as determined by Western blotting using the anti-α (#AC1-M11; Thermo Fisher Scientific), anti-µ2 polyclonal antibody R11-29 (gift of J. Bonifacino, National Institutes of Health, Bethesda, MD; Aguilar et al., 1997), anti-AP2σ mAb ab128950 (Abcam), anti-AP2β ab75158 (Abcam), mAb anti–Flag tag M1 (Sigma-Aldrich), and anti–Myc tag antibody clone 9E10 (EMD Millipore) were chosen for further experiments. All cell lines were grown under 5% CO2 at 37°C in DMEM high-glucose medium (Thermo Fisher Scientific) supplemented with 10% (vol/vol) FCS (HyClone). siRNA transfection Cells grown in 6-cm dishes were treated with siRNA sequences to silence the endogenous AP2 subunit. 170 pmol of siRNA was mixed with RNAiMAX reagent in 0.5 ml of OptiMEM (Thermo Fisher Scientific). The mixture was incubated at RT for 20 min, added to cells, and incubated with cells for 4 h. Transfection was performed 12, 36, and 60 h after plating, and experiments were performed on the fifth day. The control “AllStars negative” siRNA nontargeting sequence was purchased from QIAGEN. Transferrin receptor internalization TfnR internalization was assessed using a modified protocol described by (Reis et al., 2015). Mouse anti-TfnR mAb (HTR-D65, generated in-house from hybridomas; Schmid and Smythe, 1991) at a concentration of 4 µg/ml was added to the cells at 37°C at time 0 of internalization. After 5 min at 37°C, during the linear phase of uptake, cells were transferred to 4°C to stop internalization. To assess TfnR internalization upon AAK1 inhibition, cells were preincubated for 3 h at 37°C with 10 µM Compound 2 (a previously published AAK1 inhibitor [Bamborough et al., 2008]) or the equivalent final concentration of DMSO (0.1% vol/vol) as a control. For uptake experiments, the anti-TfnR mAb solution was made in medium containing the same concentration of DMSO or inhibitor. CD8 chimera internalization and immunofluorescence Cells were coinfected in suspension with adenoviruses encoding the CD8 chimeras and adenoviruses encoding a tet repressible transcription activator. In brief, cells were detached by trypsinization, washed once with DMEM, and resuspended in 4 ml of DMEM containing both types of adenovirus and tetracycline at a concentration of 15 ng/ml. Cells were seeded either on gelatin-coated coverslips (for immunofluorescence) or into 96-well plates at a density of 3 × 104 cells per well (for internalization assays). Localization or internalization of CD8 chimera was followed by anti-CD8 mAb UCHT-4 (C7423; Sigma-Aldrich) 12–16 h after infection. TIRF microscopy and quantification TIRF microscopy was performed as previously described (Loerke et al., 2009). Cells were imaged on gelatin-coated coverslips 5–12 h after seeding. In brief, cells expressing EGFP-CLCa and AP2 subunits were imaged using a 100×, 1.49-NA Apo TIRF objective (Nikon) mounted on a Ti-Eclipse inverted microscope equipped with the Perfect Focus System (Nikon). Time-lapse image sequences from different cells were acquired at a frame rate of 1 frame/s and exposure time of 150 ms using a CoolSNAP HQ2 monochrome CCD camera with 6.45 × 6.45 µm2 pixels (Photometrics). Similarly, nearly simultaneous two-channel (e.g., 488-nm epifluorescence/TIRF or 488/561-nm TIRF) movies were acquired at 0.5 frame/s with exposure times of 200 ms (EGFP-CLCa; for epifluorescence excitation), and 200–300 ms (overexpressed Dnm2-mRuby2; for TIRF excitation). Quantitative analysis to distinguish bona fide CCPs from all detected CLSs, and to measure CCP initiation rates and lifetime distributions, was performed exactly as previously described (Aguet et al., 2013). The quantitative analysis of CCPs and sCLSs was previously described and developed (Aguet et al., 2013). In brief, our analysis focuses on diffraction-limited objects; therefore, the detection of all CLSs is based on the assumption that the fluorescent signals measured can be described by a Gaussian point spread function. Signals were selected as valid CLS detections if the amplitude was higher than a 95th percentile confidence threshold in the local background noise distribution. CLS trajectories were calculated from the detections obtained in individual frames using the u-track software package (Jaqaman et al., 2008). Bona fide CCPs that undergo stabilization and maturation are distinguished from transient sCLCs based on quantitative analysis of the progression of their CLCa-EGFP fluorescence intensity during early stages of growth, as previously described (Aguet et al., 2013). CCP initiation rates were calculated as the number of bona fide CCPs per surface area and unit time. Initiation rates of sCLCs were determined by subtracting the rate of bona fide CCP initiation from the rate of initiation of all detected CLSs. Calculation of clathrin recruitment rates Clathrin recruitment rates were determined using the approach outlined in Loerke et al. (2011): all available CCP trajectories of a given total lifetime τ were averaged and smoothed with a box filter to produce a single lifetime bin, i.e., the representative intensity time course I(t,τ), with t being the elapsed time. The clathrin recruitment rate for each total lifetime τ was the slope of the intensity time course I(t,τ) averaged in the time interval t = 3–8 s after the CCP’s first visible (i.e., detected) time point. Raw recruitment rates were calculated in units of intensity counts per second. For all conditions, the measured recruitment rates increased with total CCP lifetime τ and typically stabilized at approximately τ = 45 s. Immunofluorescence microscopy Cells seeded on gelatin-coated coverslips were fixed and permeabilized according to previously published protocols (Mettlen et al., 2010). AP2 was detected using mouse anti-AP2 mAb (AP6, generated in-house from hybridomas; Chin et al., 1989). Transmembrane proteins were detected using mouse anti–EGF receptor mAb AB11 (199.12; NeoMarkers), mAb anti-TfnR HTR-D65, and mouse anti-CD8 mAb (UCHT-4, C7423; Sigma-Aldrich). Fixed cell images were acquired by TIRF microscopy. Quantification of colocalization Quantitative colocalization analysis was used to compare the stoichiometry of AP2 with respect to CLCa-EGFP in CCPs of mutant and control cells. The Pearson correlation coefficient was calculated using image analysis software Imaris 7.4 and ImarisColoc according to established protocols (Pompey et al., 2013). In brief, images were preprocessed to exclude background fluorescence in the green channel so as to analyze only CCPs. Next, intensity thresholds were automatically established for both channels using a point spread function value of 0.3 µm. The Pearson correlation coefficient for CCPs was calculated for five images containing up to three cells, and values for 20 images were pooled for statistical analysis. Computational screen for short endocytic motifs We performed a computational screen to determine the relative abundance of diLeu- and Yxxφ-based motifs across cytosolic domains of human transmembrane proteins that annotate as being localized to the plasma membrane. The amino acid composition of the short motifs included in the screen were as follows. For Yxxφ-based motifs, we allowed φ = L, I, M, F, or V. For diLeu motifs, we used the following search X[D/E]XXXL[L/I]; for phosphorylation-dependent diLeu motifs, [S/T*]XXXXL[L/I]; and for noncanonical diLeu motifs, X[R/H/Q] XXXL[L/I] to ensure capture of their greater heterogeneity (Traub and Bonifacino, 2013). We queried the UnitProt database for all proteins matching the following search term: annotation:(type:topo_dom cytoplasmic) annotation:(type:transmem) AND reviewed:yes AND organism“:Homo sapiens (Human) [9606].” This query returned 3,730 sequences, from which duplicate records were removed, yielding a total of 3,705 proteins. We then searched the cytosolic domains of each sequence, using UniProt’s domain range annotations, for the presence of either diLeu motifs or Yxxφ-based motifs. We identified a total of 2,383 proteins with at least one motif, and we compared the presence of tyrosine and diLeu motifs across these proteins. Overall, tyrosine motifs were approximately five times more prevalent. All computational analyses were conducted using custom Python and R scripts, which are included in the supplemental material. Online supplemental material Fig. S1 shows Western blots of AP2 complex immunoprecipitations and immunofluorescence images that validate the mutants studied. Fig. S2 shows fluorescence analyses of clathrin assembly during CCP maturation in αWT and αPIP2− cells. Fig. S3 shows Pearson correlation coefficient data for localization of cargo and AP2 complexes with CCPs in αWT and αPIP2− cells. Fig. S4 shows internalization efficiency of TfnRs and epifluorescence/TIRF data for CCPs in µ2WT and µ2PIP2− cells, as well as Western blots showing levels of phosphorylation of µ2 at T156 under various conditions. Fig. S5 shows immunofluorescence and Western blots validating expression levels of model Y–based and diLeu-based cargo receptors in htertRPE cells. Supplementary Material Supplemental Materials (PDF) Python and R Scripts (zipped files) Acknowledgments We thank Philippe Roudot and Gaudenz Danuser for help with quantification of our data. We are grateful to Wesley Burford for adenovirus production, Margaret S. Robinson for CD8 and AP2 constructs, David J. Owen for helpful discussions and insights, and W. Mike Henne for comments that greatly improved the manuscript. We thank Peter Michaely, Assaf Zaritsky, and all members of the Schmid laboratory for helpful discussions. This research was supported by the People Program (Marie Curie Actions) of the European Union’s Seventh Framework Programme grant agreement no. PIOF-GA-2012-330268, the Swiss National Science Foundation Fellowship for Prospective Researchers (to Z. Kadlecova), and National Institutes of Health grants MH61345 and GM73165 to S.L. Schmid. The authors declare no competing financial interests. Abbreviations used: AAK1 adaptor-associated kinase 1 CCP clathrin-coated pit CLS clathrin-labeled structure CME clathrin-mediated endocytosis PIP2 phosphatidylinositol-4,5-bisphosphate PM plasma membrane sCLS subthreshold CLS Tfn transferrin TfnR Tfn receptor TIRF total internal reflection fluorescence WT wild-type ==== Refs Aguet, F., C.N. Antonescu, M. Mettlen, S.L. Schmid, and G. Danuser. 2013. Advances in analysis of low signal-to-noise images link dynamin and AP2 to the functions of an endocytic checkpoint. Dev. Cell. 26 :279–291. 10.1016/j.devcel.2013.06.019 23891661 Aguilar, R.C., H. Ohno, K.W. Roche, and J.S. Bonifacino. 1997. Functional domain mapping of the clathrin-associated adaptor medium chains μ1 and μ2. J. Biol. Chem. 272 :27160–27166. 10.1074/jbc.272.43.27160 9341158 Antonescu, C.N., F. Aguet, G. Danuser, and S.L. Schmid. 2011. Phosphatidylinositol-(4,5)-bisphosphate regulates clathrin-coated pit initiation, stabilization, and size. Mol. Biol. Cell. 22 :2588–2600. 10.1091/mbc.E11-04-0362 21613550 Bamborough, P., D. Drewry, G. Harper, G.K. Smith, and K. Schneider. 2008. Assessment of chemical coverage of kinome space and its implications for kinase drug discovery. J. Med. Chem. 51 :7898–7914. 10.1021/jm8011036 19035792 Boucrot, E., S. Saffarian, R. Massol, T. Kirchhausen, and M. Ehrlich. 2006. Role of lipids and actin in the formation of clathrin-coated pits. Exp. Cell Res. 312 :4036–4048. 10.1016/j.yexcr.2006.09.025 17097636 Brach, T., C. Godlee, I. Moeller-Hansen, D. Boeke, and M. Kaksonen. 2014. The initiation of clathrin-mediated endocytosis is mechanistically highly flexible. Curr. Biol. 24 :548–554. 10.1016/j.cub.2014.01.048 24530066 Canagarajah, B.J., X. Ren, J.S. Bonifacino, and J.H. Hurley. 2013. The clathrin adaptor complexes as a paradigm for membrane-associated allostery. Protein Sci. 22 :517–529. 10.1002/pro.2235 23424177 Chin, D.J., R.M. Straubinger, S. Acton, I. Näthke, and F.M. Brodsky. 1989. 100-kDa polypeptides in peripheral clathrin-coated vesicles are required for receptor-mediated endocytosis. Proc. Natl. Acad. Sci. USA. 86 :9289–9293. 10.1073/pnas.86.23.9289 2574457 Cocucci, E., F. Aguet, S. Boulant, and T. Kirchhausen. 2012. The first five seconds in the life of a clathrin-coated pit. Cell. 150 :495–507. 10.1016/j.cell.2012.05.047 22863004 Collins, B.M., A.J. McCoy, H.M. Kent, P.R. Evans, and D.J. Owen. 2002. Molecular architecture and functional model of the endocytic AP2 complex. Cell. 109 :523–535. 10.1016/S0092-8674(02)00735-3 12086608 Conner, S.D., and S.L. Schmid. 2002. Identification of an adaptor-associated kinase, AAK1, as a regulator of clathrin-mediated endocytosis. J. Cell Biol. 156 :921–929. 10.1083/jcb.200108123 11877461 Conner, S.D., and S.L. Schmid. 2003. Regulated portals of entry into the cell. Nature. 422 :37–44. 10.1038/nature01451 12621426 Conner, S.D., T. Schröter, and S.L. Schmid. 2003. AAK1-mediated micro2 phosphorylation is stimulated by assembled clathrin. Traffic. 4 :885–890. 10.1046/j.1398-9219.2003.0142.x 14617351 Damke, H., D.D. Binns, H. Ueda, S.L. Schmid, and T. Baba. 2001. Dynamin GTPase domain mutants block endocytic vesicle formation at morphologically distinct stages. Mol. Biol. Cell. 12 :2578–2589. 10.1091/mbc.12.9.2578 11553700 Dannhauser, P.N., and E.J. Ungewickell. 2012. Reconstitution of clathrin-coated bud and vesicle formation with minimal components. Nat. Cell Biol. 14 :634–639. 10.1038/ncb2478 22522172 Ehrlich, M., W. Boll, A. Van Oijen, R. Hariharan, K. Chandran, M.L. Nibert, and T. Kirchhausen. 2004. Endocytosis by random initiation and stabilization of clathrin-coated pits. Cell. 118 :591–605. 10.1016/j.cell.2004.08.017 15339664 Ferguson, S.M., and P. De Camilli. 2012. Dynamin, a membrane-remodelling GTPase. Nat. Rev. Mol. Cell Biol. 13 :75–88. 10.1038/nrm3266 22233676 Gaidarov, I., and J.H. Keen. 1999. Phosphoinositide-AP-2 interactions required for targeting to plasma membrane clathrin-coated pits. J. Cell Biol. 146 :755–764. 10.1083/jcb.146.4.755 10459011 Gaidarov, I., F. Santini, R.A. Warren, and J.H. Keen. 1999. Spatial control of coated-pit dynamics in living cells. Nat. Cell Biol. 1 :1–7.10559856 Godlee, C., and M. Kaksonen. 2013. From uncertain beginnings: Initiation mechanisms of clathrin-mediated endocytosis. J. Cell Biol. 203 :717–725. 10.1083/jcb.201307100 24322426 González-Gaitán, M., and H. Jäckle. 1997. Role of Drosophila alpha-adaptin in presynaptic vesicle recycling. Cell. 88 :767–776. 10.1016/S0092-8674(00)81923-6 9118220 Grassart, A., A.T. Cheng, S.H. Hong, F. Zhang, N. Zenzer, Y. Feng, D.M. Briner, G.D. Davis, D. Malkov, and D.G. Drubin. 2014. Actin and dynamin2 dynamics and interplay during clathrin-mediated endocytosis. J. Cell Biol. 205 :721–735. 10.1083/jcb.201403041 24891602 Gu, M., K. Schuske, S. Watanabe, Q. Liu, P. Baum, G. Garriga, and E.M. Jorgensen. 2008. Mu2 adaptin facilitates but is not essential for synaptic vesicle recycling in Caenorhabditis elegans. J. Cell Biol. 183 :881–892. 10.1083/jcb.200806088 19047463 Hardy, S., M. Kitamura, T. Harris-Stansil, Y. Dai, and M.L. Phipps. 1997. Construction of adenovirus vectors through Cre-lox recombination. J. Virol. 71 :1842–1849.9032314 Hollopeter, G., J.J. Lange, Y. Zhang, T.N. Vu, M. Gu, M. Ailion, E.J. Lambie, B.D. Slaughter, J.R. Unruh, L. Florens, and E.M. Jorgensen. 2014. The membrane-associated proteins FCHo and SGIP are allosteric activators of the AP2 clathrin adaptor complex. eLife. 3 :. 10.7554/eLife.03648 Höning, S., D. Ricotta, M. Krauss, K. Späte, B. Spolaore, A. Motley, M. Robinson, C. Robinson, V. Haucke, and D.J. Owen. 2005. Phosphatidylinositol-(4,5)-bisphosphate regulates sorting signal recognition by the clathrin-associated adaptor complex AP2. Mol. Cell. 18 :519–531. 10.1016/j.molcel.2005.04.019 15916959 Jackson, A.P., A. Flett, C. Smythe, L. Hufton, F.R. Wettey, and E. Smythe. 2003. Clathrin promotes incorporation of cargo into coated pits by activation of the AP2 adaptor micro2 kinase. J. Cell Biol. 163 :231–236. 10.1083/jcb.200304079 14581451 Jackson, L.P., B.T. Kelly, A.J. McCoy, T. Gaffry, L.C. James, B.M. Collins, S. Höning, P.R. Evans, and D.J. Owen. 2010. A large-scale conformational change couples membrane recruitment to cargo binding in the AP2 clathrin adaptor complex. Cell. 141 :1220–1229. 10.1016/j.cell.2010.05.006 20603002 Jaqaman, K., D. Loerke, M. Mettlen, H. Kuwata, S. Grinstein, S.L. Schmid, and G. Danuser. 2008. Robust single-particle tracking in live-cell time-lapse sequences. Nat. Methods. 5 :695–702. 10.1038/nmeth.1237 18641657 Kelly, B.T., A.J. McCoy, K. Späte, S.E. Miller, P.R. Evans, S. Höning, and D.J. Owen. 2008. A structural explanation for the binding of endocytic dileucine motifs by the AP2 complex. Nature. 456 :976–979. 10.1038/nature07422 19140243 Kelly, B.T., S.C. Graham, N. Liska, P.N. Dannhauser, S. Höning, E.J. Ungewickell, and D.J. Owen. 2014. Clathrin adaptors. AP2 controls clathrin polymerization with a membrane-activated switch. Science. 345 :459–463. 10.1126/science.1254836 25061211 Kirchhausen, T., D. Owen, and S.C. Harrison. 2014. Molecular structure, function, and dynamics of clathrin-mediated membrane traffic. Cold Spring Harb. Perspect. Biol. 6 :a016725. 10.1101/cshperspect.a016725 24789820 Kozik, P., R.W. Francis, M.N. Seaman, and M.S. Robinson. 2010. A screen for endocytic motifs. Traffic. 11 :843–855. 10.1111/j.1600-0854.2010.01056.x 20214754 Liu, A.P., F. Aguet, G. Danuser, and S.L. Schmid. 2010. Local clustering of transferrin receptors promotes clathrin-coated pit initiation. J. Cell Biol. 191 :1381–1393. 10.1083/jcb.201008117 21187331 Loerke, D., M. Mettlen, D. Yarar, K. Jaqaman, H. Jaqaman, G. Danuser, and S.L. Schmid. 2009. Cargo and dynamin regulate clathrin-coated pit maturation. PLoS Biol. 7 :e57. 10.1371/journal.pbio.1000057 19296720 Loerke, D., M. Mettlen, S.L. Schmid, and G. Danuser. 2011. Measuring the hierarchy of molecular events during clathrin-mediated endocytosis. Traffic. 12 :815–825. 10.1111/j.1600-0854.2011.01197.x 21447041 Lu, Q. 2005. Seamless cloning and gene fusion. Trends Biotechnol. 23 :199–207. 10.1016/j.tibtech.2005.02.008 15780712 Ma, L., P.K. Umasankar, A.G. Wrobel, A. Lymar, A.J. McCoy, S.S. Holkar, A. Jha, T. Pradhan-Sundd, S.C. Watkins, D.J. Owen, and L.M. Traub. 2016. Transient Fcho1/2⋅Eps15/R⋅AP-2 nanoclusters prime the AP-2 clathrin adaptor for cargo binding. Dev. Cell. 37 :428–443. 10.1016/j.devcel.2016.05.003 27237791 Marchese, A., M.M. Paing, B.R. Temple, and J. Trejo. 2008. G protein-coupled receptor sorting to endosomes and lysosomes. Annu. Rev. Pharmacol. Toxicol. 48 :601–629. 10.1146/annurev.pharmtox.48.113006.094646 17995450 Mattera, R., M. Boehm, R. Chaudhuri, Y. Prabhu, and J.S. Bonifacino. 2011. Conservation and diversification of dileucine signal recognition by adaptor protein (AP) complex variants. J. Biol. Chem. 286 :2022–2030. 10.1074/jbc.M110.197178 21097499 McMahon, H.T., and E. Boucrot. 2011. Molecular mechanism and physiological functions of clathrin-mediated endocytosis. Nat. Rev. Mol. Cell Biol. 12 :517–533. 10.1038/nrm3151 21779028 Merrifield, C.J., and M. Kaksonen. 2014. Endocytic accessory factors and regulation of clathrin-mediated endocytosis. Cold Spring Harb. Perspect. Biol. 6 :a016733. 10.1101/cshperspect.a016733 25280766 Merrifield, C.J., M.E. Feldman, L. Wan, and W. Almers. 2002. Imaging actin and dynamin recruitment during invagination of single clathrin-coated pits. Nat. Cell Biol. 4 :691–698. 10.1038/ncb837 12198492 Mettlen, M., and G. Danuser. 2014. Imaging and modeling the dynamics of clathrin-mediated endocytosis. Cold Spring Harb. Perspect. Biol. 6 :a017038. 10.1101/cshperspect.a017038 25167858 Mettlen, M., D. Loerke, D. Yarar, G. Danuser, and S.L. Schmid. 2010. Cargo- and adaptor-specific mechanisms regulate clathrin-mediated endocytosis. J. Cell Biol. 188 :919–933. 10.1083/jcb.200908078 20231386 Mitsunari, T., F. Nakatsu, N. Shioda, P.E. Love, A. Grinberg, J.S. Bonifacino, and H. Ohno. 2005. Clathrin adaptor AP-2 is essential for early embryonal development. Mol. Cell. Biol. 25 :9318–9323. 10.1128/MCB.25.21.9318-9323.2005 16227583 Miyazaki, K. 2011. MEGAWHOP cloning: A method of creating random mutagenesis libraries via megaprimer PCR of whole plasmids. Methods Enzymol. 498 :399–406. 10.1016/B978-0-12-385120-8.00017-6 21601687 Morlot, S., and A. Roux. 2013. Mechanics of dynamin-mediated membrane fission. Annu. Rev. Biophys. 42 :629–649. 10.1146/annurev-biophys-050511-102247 23541160 Motley, A.M., N. Berg, M.J. Taylor, D.A. Sahlender, J. Hirst, D.J. Owen, and M.S. Robinson. 2006. Functional analysis of AP-2 α and μ2 subunits. Mol. Biol. Cell. 17 :5298–5308. 10.1091/mbc.E06-05-0452 17035630 Munn, A.L. 2001. Molecular requirements for the internalisation step of endocytosis: Insights from yeast. Biochim. Biophys. Acta. 1535 :236–257. 10.1016/S0925-4439(01)00028-X 11278164 Ohno, H., J. Stewart, M.C. Fournier, H. Bosshart, I. Rhee, S. Miyatake, T. Saito, A. Gallusser, T. Kirchhausen, and J.S. Bonifacino. 1995. Interaction of tyrosine-based sorting signals with clathrin-associated proteins. Science. 269 :1872–1875. 10.1126/science.7569928 7569928 Ohno, H., M.C. Fournier, G. Poy, and J.S. Bonifacino. 1996. Structural determinants of interaction of tyrosine-based sorting signals with the adaptor medium chains. J. Biol. Chem. 271 :29009–29015. 10.1074/jbc.271.46.29009 8910552 Olusanya, O., P.D. Andrews, J.R. Swedlow, and E. Smythe. 2001. Phosphorylation of threonine 156 of the mu2 subunit of the AP2 complex is essential for endocytosis in vitro and in vivo. Curr. Biol. 11 :896–900. 10.1016/S0960-9822(01)00240-8 11516654 Owen, D.J., and P.R. Evans. 1998. A structural explanation for the recognition of tyrosine-based endocytotic signals. Science. 282 :1327–1332. 10.1126/science.282.5392.1327 9812899 Owen, D.J., Y. Vallis, M.E. Noble, J.B. Hunter, T.R. Dafforn, P.R. Evans, and H.T. McMahon. 1999. A structural explanation for the binding of multiple ligands by the alpha-adaptin appendage domain. Cell. 97 :805–815. 10.1016/S0092-8674(00)80791-6 10380931 Owen, D.J., B.M. Collins, and P.R. Evans. 2004. Adaptors for clathrin coats: Structure and function. Annu. Rev. Cell Dev. Biol. 20 :153–191. 10.1146/annurev.cellbio.20.010403.104543 15473838 Pandey, K.N. 2009. Functional roles of short sequence motifs in the endocytosis of membrane receptors. Front. Biosci. (Landmark Ed.). 14 :5339–5360. 10.2741/3599 19482617 Pelkmans, L., E. Fava, H. Grabner, M. Hannus, B. Habermann, E. Krausz, and M. Zerial. 2005. Genome-wide analysis of human kinases in clathrin- and caveolae/raft-mediated endocytosis. Nature. 436 :78–86. 10.1038/nature03571 15889048 Perkins, J.R., I. Diboun, B.H. Dessailly, J.G. Lees, and C. Orengo. 2010. Transient protein-protein interactions: structural, functional, and network properties. Structure. 18 :1233–1243. 10.1016/j.str.2010.08.007 20947012 Pitcher, C., S. Höning, A. Fingerhut, K. Bowers, and M. Marsh. 1999. Cluster of differentiation antigen 4 (CD4) endocytosis and adaptor complex binding require activation of the CD4 endocytosis signal by serine phosphorylation. Mol. Biol. Cell. 10 :677–691. 10.1091/mbc.10.3.677 10069811 Pompey, S.N., P. Michaely, and K. Luby-Phelps. 2013. Quantitative fluorescence co-localization to study protein-receptor complexes. Methods Mol. Biol. 1008 :439–453. 10.1007/978-1-62703-398-5_16 23729262 Praefcke, G.J., M.G. Ford, E.M. Schmid, L.E. Olesen, J.L. Gallop, S.Y. Peak-Chew, Y. Vallis, M.M. Babu, I.G. Mills, and H.T. McMahon. 2004. Evolving nature of the AP2 alpha-appendage hub during clathrin-coated vesicle endocytosis. EMBO J. 23 :4371–4383. 10.1038/sj.emboj.7600445 15496985 Reis, C.R., P.H. Chen, S. Srinivasan, F. Aguet, M. Mettlen, and S.L. Schmid. 2015. Crosstalk between Akt/GSK3β signaling and dynamin-1 regulates clathrin-mediated endocytosis. EMBO J. 34 :2132–2146. 10.15252/embj.201591518 26139537 Ricotta, D., S.D. Conner, S.L. Schmid, K. von Figura, and S. Honing. 2002. Phosphorylation of the AP2 mu subunit by AAK1 mediates high affinity binding to membrane protein sorting signals. J. Cell Biol. 156 :791–795. 10.1083/jcb.200111068 11877457 Rohde, G., D. Wenzel, and V. Haucke. 2002. A phosphatidylinositol (4,5)-bisphosphate binding site within mu2-adaptin regulates clathrin-mediated endocytosis. J. Cell Biol. 158 :209–214. 10.1083/jcb.200203103 12119359 Saffarian, S., and T. Kirchhausen. 2008. Differential evanescence nanometry: Live-cell fluorescence measurements with 10-nm axial resolution on the plasma membrane. Biophys. J. 94 :2333–2342. 10.1529/biophysj.107.117234 17993495 Schledzewski, K., H. Brinkmann, and R.R. Mendel. 1999. Phylogenetic analysis of components of the eukaryotic vesicle transport system reveals a common origin of adaptor protein complexes 1, 2, and 3 and the F subcomplex of the coatomer COPI. J. Mol. Evol. 48 :770–778. 10.1007/PL00006521 10229581 Schmid, E.M., and H.T. McMahon. 2007. Integrating molecular and network biology to decode endocytosis. Nature. 448 :883–888. 10.1038/nature06031 17713526 Schmid, S.L., and V.A. Frolov. 2011. Dynamin: functional design of a membrane fission catalyst. Annu. Rev. Cell Dev. Biol. 27 :79–105. 10.1146/annurev-cellbio-100109-104016 21599493 Schmid, S.L., and E. Smythe. 1991. Stage-specific assays for coated pit formation and coated vesicle budding in vitro. J. Cell Biol. 114 :869–880. 10.1083/jcb.114.5.869 1908470 Shih, W., A. Gallusser, and T. Kirchhausen. 1995. A clathrin-binding site in the hinge of the β2 chain of mammalian AP-2 complexes. J. Biol. Chem. 270 :31083–31090. 10.1074/jbc.270.52.31083 8537368 Sun, Y., and D.G. Drubin. 2012. The functions of anionic phospholipids during clathrin-mediated endocytosis site initiation and vesicle formation. J. Cell Sci. 125 :6157–6165. 10.1242/jcs.115741 23097040 Taylor, M.J., D. Perrais, and C.J. Merrifield. 2011. A high precision survey of the molecular dynamics of mammalian clathrin-mediated endocytosis. PLoS Biol. 9 :e1000604. 10.1371/journal.pbio.1000604 21445324 Taylor, M.J., M. Lampe, and C.J. Merrifield. 2012. A feedback loop between dynamin and actin recruitment during clathrin-mediated endocytosis. PLoS Biol. 10 :e1001302. 10.1371/journal.pbio.1001302 22505844 Traub, L.M. 2009. Tickets to ride: Selecting cargo for clathrin-regulated internalization. Nat. Rev. Mol. Cell Biol. 10 :583–596. 10.1038/nrm2751 19696796 Traub, L.M., and J.S. Bonifacino. 2013. Cargo recognition in clathrin-mediated endocytosis. Cold Spring Harb. Perspect. Biol. 5 :a016790. 10.1101/cshperspect.a016790 24186068 Traub, L.M., M.A. Downs, J.L. Westrich, and D.H. Fremont. 1999. Crystal structure of the alpha appendage of AP-2 reveals a recruitment platform for clathrin-coat assembly. Proc. Natl. Acad. Sci. USA. 96 :8907–8912. 10.1073/pnas.96.16.8907 10430869 Umasankar, P.K., L. Ma, J.R. Thieman, A. Jha, B. Doray, S.C. Watkins, and L.M. Traub. 2014. A clathrin coat assembly role for the muniscin protein central linker revealed by TALEN-mediated gene editing. eLife. 3 :e04137. 10.7554/eLife.04137 25303365 Van Roey, K., T.J. Gibson, and N.E. Davey. 2012. Motif switches: Decision-making in cell regulation. Curr. Opin. Struct. Biol. 22 :378–385. 10.1016/j.sbi.2012.03.004 22480932 Weinberg, J., and D.G. Drubin. 2012. Clathrin-mediated endocytosis in budding yeast. Trends Cell Biol. 22 :1–13. 10.1016/j.tcb.2011.09.001 22018597 Zoncu, R., R.M. Perera, R. Sebastian, F. Nakatsu, H. Chen, T. Balla, G. Ayala, D. Toomre, and P.V. De Camilli. 2007. Loss of endocytic clathrin-coated pits upon acute depletion of phosphatidylinositol 4,5-bisphosphate. Proc. Natl. Acad. Sci. USA. 104 :3793–3798. 10.1073/pnas.0611733104 17360432

PMC005xxxxxx/PMC5299621.txt

==== Front J Gen Physiol J Gen Physiol jgp jgp The Journal of General Physiology 0022-1295 1540-7748 The Rockefeller University Press 28108595 201611675 10.1085/jgp.201611675 Research Articles Research Article 500 501 514 Identification of the ADPR binding pocket in the NUDT9 homology domain of TRPM2 ADPR binding on TRPM2 Yu Peilin 12* Xue Xiwen 3* Zhang Jianmin 1 Hu Xupang 1 Wu Yan 1 Jiang Lin-Hua 456 Jin Hongwei 3 Luo Jianhong 1 Zhang Liangren 3 Liu Zhenming 3 Yang Wei 1 1 Department of Neurobiology, Key Laboratory of Medical Neurobiology of the Ministry of Health of China, Zhejiang University School of Medicine, Hangzhou, Zhejiang 310058, China 2 Department of Toxicology, School of Public Health, Zhejiang University, Hangzhou, Zhejiang 310058, China 3 State Key Laboratory of Natural and Biomimetic Drugs, School of Pharmaceutical Sciences, Peking University, Beijing 100191, China 4 School of Biomedical Sciences, Faculty of Biological Sciences, University of Leeds, Leeds LS2 9JT, England, UK 5 Department of Physiology and Neurobiology, Xinxiang Medical University, Henan 453003, China 6 Sino-UK Brain Function Laboratory, Xinxiang Medical University, Henan 453003, China Correspondence to Wei Yang: yangwei@zju.edu.cn; or Zhenming Liu: zmliu@bjmu.edu.cn * P. Yu and X. Xue contributed equally to this paper. 2 2017 149 2 219235 29 7 2016 12 10 2016 30 11 2016 © 2017 Yu et al. 2017 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). ADP ribose (ADPR) is an endogenous ligand for the transient receptor potential melastatin 2 (TRPM2) channel. Yu et al. identify 11 residues in the NUDT9 homology domain of TRPM2 that form a binding site for ADPR involving van der Waals, polar solvation, and electronic interactions. Activation of the transient receptor potential melastatin 2 (TRPM2) channel occurs during the response to oxidative stress under physiological conditions as well as in pathological processes such as ischemia and diabetes. Accumulating evidence indicates that adenosine diphosphate ribose (ADPR) is the most important endogenous ligand of TRPM2. However, although it is known that ADPR binds to the NUDT9 homology (NUDT9-H) domain in the intracellular C-terminal region, the molecular mechanism underlying ADPR binding and activation of TRPM2 remains unknown. In this study, we generate a structural model of the NUDT9-H domain and identify the binding pocket for ADPR using induced docking and molecular dynamics simulation. We find a subset of 11 residues—H1346, T1347, T1349, L1379, G1389, S1391, E1409, D1431, R1433, L1484, and H1488—that are most likely to directly interact with ADPR. Results from mutagenesis and electrophysiology approaches support the predicted binding mechanism, indicating that ADPR binds tightly to the NUDT9-H domain, and suggest that the most significant interactions are the van der Waals forces with S1391 and L1484, polar solvation interaction with E1409, and electronic interactions (including π–π interactions) with H1346, T1347, Y1349, D1431, and H1488. These findings not only clarify the roles of a range of newly identified residues involved in ADPR binding in the TRPM2 channel, but also reveal the binding pocket for ADPR in the NUDT9-H domain, which should facilitate structure-based drug design for the TRPM2 channel. National Basic Research Program of China https://doi.org/10.13039/501100012166 2013CB910204 2014CB910300 Natural Science Foundation of China https://doi.org/10.13039/501100001809 21402171 ==== Body pmcIntroduction The transient receptor potential melastatin 2 (TRPM2) channel is a nonselective Ca2+-permeable cation channel with a broad sensitivity to many factors such as Ca2+, phosphatidylinositol bisphosphate, and temperature (Starkus et al., 2007; Tóth and Csanády, 2012; Song et al., 2016; Tan and McNaughton, 2016). Extensive evidence has demonstrated that TRPM2 functions as a sensor for oxidative stress (Jiang et al., 2010). Several recent studies using transgenic mice have revealed an important role for the TRPM2 channel in processes including insulin secretion (Uchida et al., 2011; Manna et al., 2015), generation of proinflammation cytokines (Yamamoto et al., 2008; Di et al., 2012), delayed neuronal death (Alim et al., 2013; Ye et al., 2014), ischemic kidney injury (Gao et al., 2014), protection of cardiac ischemia-reperfusion damage (Miller et al., 2014), and sepsis (Qian et al., 2014). Therefore, it is critical to understand the mechanism of TRPM2 channel activation during oxidative stress. The homotetrameric TRPM2 channel belongs to the TRP channel family. Each TRPM2 subunit contains a large intracellular N terminus, the typical six transmembrane domains, and a large intracellular C terminus (Fig. 1 A). Like TRPM6 and TRPM7, TRPM2 was considered to be a channel enzyme. The NUDT9 homology (NUDT9-H) domain in TRPM2 C terminus is homologous to the NUDT9 adenosine diphosphate ribose (ADPR) hydrolase (∼50% similarity). TRPM2 activation by ADPR was originally proposed to be mediated by an enzymatic process in which NUDT9-H binds ADPR and converts it to AMP and ribose-5-phophate (R5P; Perraud et al., 2001, 2003). However, this view was refuted by recent work (Tóth et al., 2014; Iordanov et al., 2016), which demonstrated that TRPM2 does not possess ADPR hydrolase activity. Nevertheless, binding of ADPR to the NUDT9-H domain is essential for the channel opening (Perraud et al., 2001; Sano et al., 2001; Csanády and Törocsik, 2009; Tóth and Csanády, 2012). An early mutagenesis study had proposed that several residues in the NUDT9-H domain (e.g., the Nudix motif) may be engaged in ADPR binding and/or activation of the TRPM2 channel (Perraud et al., 2005), whereas the precise and whole perspective of ADPR binding remains poorly defined. Figure 1. Homology modeling of the NUDT9-H domain from the human TRPM2 channel protein. (A) Schematic diagram showing the membrane topology of the TRPM2 channel proteins. Important domains are labeled in the figure. (B) Sequence alignments of the human NUDT9 and NUDT9-H domain. Red shading with purple border denotes identical residues, and white shading with purple border denotes similarities. (C) Ribbon presentation of the homology models of the NUDT9-H domain based on the human NUDT9 structure. The α-helices, β-sheets, and loops are labeled in red, cyan, and gray, respectively. Computer-aided modeling techniques, in particular molecular dynamic (MD) simulations and homology modeling, are powerful tools that facilitate the discovery of molecular and structural insights into both ligand–receptor interactions and ligand-induced receptor activation events. These techniques have been used in elegant combinations with site-directed mutagenesis in many previous studies, including identifying the key residues for cAMP and cGMP binding and selectivity in the cyclic nucleotide-binding domain of the HCN2 channel (Zhou and Siegelbaum, 2007), showing GABA binding sites and conformational changes during the activation of the ionotropic GABA receptors (Ashby et al., 2012). Similar approaches enabled our previous elucidation of the structural mechanisms of capsaicin binding and gating of the TRPV1 channel (Yang et al., 2015). Our present study aimed to define molecular interactions between ADPR and the ligand-binding pocket of TRPM2. Toward this goal, we generated a structural model of the NUDT9-H domain of the TRPM2 channel based on the structure of the human NUDT9 protein (Shen et al., 2003) and used induced docking and MD simulations to identify residues in the NUDT9-H domain that coordinate ADPR via van der Waals (VDW) forces, electrostatic interactions, and hydrophobic interactions. We provided systematic functional evidence from site-directed mutagenesis and patch clamp recording experiments that corroborate their roles in coordinating ADPR. The present study not only identified new residues in the NUDT9-H domain of the TRPM2 channel that are responsible for ADPR binding but also yielded a complete picture of ADPR in the NUDT9-H domain, which holds the promise to facilitate structure-based drug design efforts targeting the TRPM2 channel. Materials and methods cDNA and cell culture The cDNA encoding the human TRPM2 (hTRPM2) channel was provided by A. M. Scharenberg (University of Washington, Seattle, WA; Perraud et al., 2001). Human embryonic kidney (HEK) 293 cells were used to transiently express wild-type (WT) and mutant channels. Cell culture, transfection, and induced TRPM2 expression were each described previously (Yang et al., 2010, 2011; Yu et al., 2014). Mutations were introduced by site-directed mutagenesis and confirmed by sequencing. All chemicals and reagents used were purchased from Sigma-Aldrich, except as otherwise indicated. Electrophysiology Whole-cell current recordings were performed using an Axopatch 200B amplifier (Molecular Devices) at RT, as described previously (Yang et al., 2010, 2011; Yu et al., 2014). For recording ADPR-induced TRPM2 currents, the extracellular solution contained 147 mM NaCl, 2 mM KCl, 1 mM MgCl2, 2 mM CaCl2, 10 mM HEPES, and 13 mM glucose, pH 7.4. The intracellular solution contained 147 mM NaCl, 0.05 mM EGTA, 1 mM MgCl2, 10 mM HEPES, and ADPR at indicated concentrations, pH 7.3. For recording calcium-induced TRPM2 currents, the extracellular solution was the same as that used in ADPR activation; the intracellular solution contained 75 mM NaCl, 1 mM MgCl2, 10 mM HEPES, and 50 mM CaCl2, pH 7.4. For recording H2O2-induced TRPM2 currents, the extracellular and intracellular solutions were the same as that used in ADPR activation, except that 30% H2O2 was added to the extracellular solution with a final concentration of 30 mM. The cell membrane potential was held at 0 mV, and a voltage ramp of 500-ms duration from −100 mV to 100 mV was applied every 5 s. Glass pipettes with a resistance of 3–5 MΩ were used. Data were acquired at 10 kHz and filtered offline during data analysis. Change of the extracellular solution was performed using an RSC-160 system (Bio-Logic Science Instruments). Homology modeling The amino acid sequence of the NUDT9-H domain of the human TRPM2 protein (accession no. O94759; TRPM2_HUMAN) was derived from the UniProt database entry, and a similarity search in Protein Data Bank (PDB) database was performed by BLAST. The crystal structure of human NUDT9 (NUDT9; PDB accession no. 1Q33, chain A) was chosen as the template in building the model for the NUDT9-H domain. Using Discovery Studio 2.5 software (Accelrys Software Inc.), we produced a sequence alignment between the NUDT9 protein (residues 59–350; PDB accession no. 1Q33) and the NUDT9-H domain (UniProt accession no. O94759, residues 1236–1503). A total of 10 models were generated using the Build Homology Models module in Discovery Studio. The final model was chosen by considering probability density functions and discrete optimized protein energy values and further validated using PROCHECK and Profile-3D (Laskowski et al., 1993). Interatomic clashes were removed by 1,500 steps of steepest descent minimization, followed by 1,000 steps of conjugate gradient minimization in a solvated, neutralized simulation box with positional restraints applied to the Cα atoms. Docking The induced-fit docking (IFD) protocol in Glide was used (Friesner et al., 2004). In IFD, the flexibility of the receptor was taken into account by combining a series of Glide and Prime processes. In the preliminary Glide docking step, the VDW radii scaling factor was set at 0.5 by default, and a maximum of 80 poses per ligand was retained. In the subsequent Prime induced-fit section, the residues within a 5-Å vicinity of the ligand were refined, whereas the others were fixed. The receptor–ligand complexes were energy minimized to an induced-fit conformation. Finally, the best receptor–ligand complex was identified using a composite scoring function with the Glide XP scoring mode adopted for docking calculations. The R1433, G1389, and H1488 residues were set as the edges of the docking pocket in an 18 × 18 × 18–Å cube, and induced docking was subsequently performed for ADPR with the NUDT9-H domain of human TRPM2. MD simulations The structure of ADPR complexed with the NUDT9-H domain produced by the IFD procedure was explored with MD simulations. MD calculations were performed using the AMBER 11 software suite (University of California, San Francisco, San Francisco, CA). The general AMBER force field (gaff) and the ff99SB force field were used for the protein and ligand, respectively. To obtain parameters for the small molecules, ab initio quantum chemical methods were employed using Gaussian09 program (Gaussian, Inc.). The geometries were fully optimized, and the electrostatic potentials around them were determined at HF/6-31G* level of theory. Subsequently, the atomic partial charges of the small molecules were obtained by the RESP fitting technique in AMBER 11 (University of California, San Francisco). The charges of protein ionizable groups were found to be in the standard protonation state at neutral pH. The protein–ligand complex was solvated with TIP3P water molecules in a periodically repeating truncated octahedral box. The net charge of the system was brought to neutrality and physiological ionic strength (0.15 M) by addition of dissociated NaCl, keeping an 8-Å distance away from any solute atom. The particle mesh Ewald algorithm was used to handle the long-range electrostatics in the molecular minimization and MD simulations. Before the MD simulations, each system was relaxed using a two-stage minimization strategy: the small molecules and water molecules were first subjected to 1,000 cycles of minimization with the protein backbone constrained (50 kcal · mol−1 · Å−2), and the whole system was minimized by 2,000 cycles of steepest descent minimization without constrain. After minimization, each system was gradually heated in the canonical ensemble from 0 to 300 K over a period of 50 ps, followed by a 20-ns isobaric-isothermal MD simulation with a target temperature of 300 K and a target pressure of 1 atm. The SHAKE procedure was used to constrain all bonds involving hydrogen atoms. The time step was set to 2 fs. The PMEMD program in AMBER 11 was used for the molecular mechanics (MM) optimization and MD simulations. A 15-ns data production run was performed in the isobaric–isothermal ensemble with positional restraints imposed on the last three C-terminal Cα atoms of each subunit (1,000 kJ · mol−1 · nm−2), to mimic the presence of the transmembrane domain. Unless otherwise stated, statistics describing ligand motions and intermolecular interactions refer to the mean ± SEM value, as averaged over all 20 independent binding sites. The coordinates were saved every 10 ps during the MD sampling process. MM/generalized Born surface area (GBSA) binding free energy calculations For all of the calculations below, a total of 50 snapshots from 13–15 ns were evenly extracted from the single MD trajectory at a time interval of 40 ps. The absolute binding free energy (ΔGbinding) was then predicted by applying the MM/GBSA approaches according to the following equation (Massova and Kollman, 2000):ΔGbinding=G¯complex−(G¯protein+G¯ligand)=ΔEMM+ΔGsolv−TΔS, where Gcomplex, Gprotein, and Gligand represent the free energies of the complex, protein, and ligand, respectively. ΔEMM is the gas-phase interaction energy calculated using the sander program in AMBER 11, including the internal, electrostatic, and VDW energies, and the internal energy was cancelled based on the single MD trajectory. The solvation free energy ΔEsolv consists of both the polar and nonpolar parts, which were denoted as ΔGGB and ΔGSA, respectively. MM/GBSA binding energy decomposition analysis To illustrate the interactions between each residue and ADPR, we performed MM/GBSA decomposition analysis supported by the mm_pbsa module in AMBER 11. The binding interaction of each residue–small-molecule pair consists of four components: the VDW contribution (ΔGvdw), the electrostatic contribution (ΔGele), the polar contribution of desolvation (ΔGGB), and the nonpolar contribution of desolvation (ΔGSA). ΔGvdw and ΔGele were calculated using the sander program in AMBER 11. A total of 50 snapshots extracted from 13 to 15 ns MD trajectory was used for the calculations of all energy components. Biotinylation assay and Western blot Biotinylation assays and Western blot were conducted as previously described (Lu et al., 2015; Zhang et al., 2015). In brief, after transfection for 24–36 h, HEK293 cells were rinsed with ice-cold PBS. The cells were subsequently incubated with a fresh preparation of 0.5 mg/ml Sulfo-NHS-SS-Biotin (Thermo Fisher Scientific) dissolved in PBS with Tween-20 for 30 min. Subsequently, unreacted biotin was quenched with PBS containing 100 mM glycine. The cells were lysed with RIPA buffer (10 mM Tris, 150 mM NaCl, 1 mM EDTA, 0.1% SDS, 1% Triton X-100, and 1% sodium deoxycholate, pH 7.4) and subjected to centrifugation at 12,000 g for 15 min. The resulting supernatant was incubated with 40 µl of a 50% slurry of NeutrAvidin beads (Thermo Fisher Scientific) for 2 h at 4°C with continuous rotation. After several washes with RIPA buffer, the biotinylated proteins were eluted from the NeutrAvidin beads with 60 µl of 2× SDS sample buffer. The primary antibody used was rabbit anti–TRPM2 (Ab11168; Abcam), and the secondary antibody was goat anti–rabbit IgG-HRP (1:10,000; 31420; Thermo Fisher Scientific). Data analysis Prism 5 software (GraphPad Software) was used for all statistical analyses. Electrophysiological recordings from at least five cells were averaged and are presented in the text and figures, where appropriate, as means ± SEM. To reduce the variation of TRPM2 currents from different batches in our data analysis, currents of the mutants were normalized with the mean maximal currents of the WT TRPM2 recorded on the same day, which were referred to as the relative currents. The half-maximal effective concentration (EC50) values were derived from fitting the concentration-response relationships to the Hill equation (Weiss, 1997). Online supplemental material Fig. S1 uses Ramachandran plot analysis to validate the reliability of our homology modeling. Fig. S2 indicates that the predicted residues are highly conserved in the TRPM2 proteins of various species. Fig. S3 showed the merged figure of the model similar to the one previously proposed and our prediction. Results Homology modeling of the NUDT9-H domain and molecular docking of ADPR The NUDT9-H domain of the TRPM2 channel (S1237-Y1503) is a close homologue of the human NUDT9 protein for which the apo structure was determined at 1.8 Å resolution (PDB accession no. 1Q33; Shen et al., 2003). Amino acid sequence alignment analysis showed that the NUDT9 and NUDT9-H domains have 52% similarity and 35% identity (Fig. 1 B), which is above the generally accepted threshold of 30% of sequence identity for modeling reliability (Ashby et al., 2012). The location of α-helixes, β-sheets, and loops in both proteins are well conserved (Fig. 1 C). The reliability of our homology modeling was validated by Ramachandran plot analysis (Fig. S1). Further refinement of the structural model by energy minimization resulted in 99.6% of the residues falling within the favored/allowed region of the Ramachandran plot. The tertiary structural model of the NUDT9-H domain is very similar to that of NUDT9 (Fig. 2 A). Figure 2. Docking of ADPR to the NUDT9-H domain of TRPM2. (A) The tertiary structure of human NUDT9 (gray) and the model of NUDT9-H domain. (B) Chemical structures of ADPR. The terminal ribose is circled with orange; the pyrophosphate is boxed with green; and the adenosine is circled with blue. (C) Ribbon presentation of the binding pocket in the NUDT9-H domain (gray) docked with ADPR. The key residues mediating binding of ADPR are highlighted. (D) MD simulation of the NUDT9-H domain in complex with ADPR: the root mean square deviation (RMSD) of ADPR (red) and ADPR-binding protein (black), respectively. To investigate the configuration of the bound ADPR and its interaction with the ligand-binding domain, ADPR (for which the chemical structure is shown in Fig. 2 B) was docked into the structural model. The results predicted that ADPR takes a saddle-like conformation inside the NUDT9-H domain. A subset of 11 residues were identified to interact with ADPR. Specifically, the terminal ribose ring in ADPR makes direct contacts with L1379, G1389, and E1409; the pyrophosphate group interacts with H1346, T1347, S1391, and R1433; and the adenosine base interacts T1349, D1431, L1484, and H1488 (Fig. 2, B and C). Five of these residues—T1349, L1379, S1391, R1433, and H1488—have been previously proposed to participate in ADPR binding, which was based on the observed interactions between the corresponding residues in the NUDT9 protein and R5P in the crystal complex (Shen et al., 2003). Validating the stability of modeled ADPR/NUDT9-H complex with MD simulations MD simulations were performed for the ADPR-bound state of the NUDT9-H domain. We performed 15-ns simulations of the ADPR/NUDT9-H complex. The root mean square deviations of protein backbone atoms for the complex reached equilibrium after 9 ns at ∼5 Å (Fig. 2 D). The bound ligands remained in their original conformation during the 15-ns MD simulations. No large alterations in the secondary structures were detected during the simulations, further demonstrating the stability of this ligand-binding pocket. Furthermore, our simulations provide insight into the local, short-range interactions that occur between ADPR and particular residues in the NUDT9-H domain. Based on the frames from the 15-ns MD simulation trajectory, the component energies of ADPR binding to the NUDT9-H domain were calculated, including the ELE (electrostatic energy), VDW (VDW energy), INT (bond, angle, and dihedral energies), mean electrostatic (Coulombic) and VDW potential energies of interaction GBSUR (hydrophobic contribution to solvent free energy for generalized Born calculations), as well as the GB (reaction field energy calculated by GB; Table 1). Moreover, the interactions between ADPR and each of the 11 residues identified in the molecular docking were assigned into VDW force, electrostatic interaction, polar solvation, and nonpolar solvation groupings, and their contributions to the binding energy for ADPR were calculated (Table 2). This analysis provided an index of the relative contributions of different residues to the strength of ligand binding. MM/GBSA binding free energy calculations supported that the 11 residues make strong interactions with ADPR, which were subject to further mutagenesis study. Table 1. Binding energy of ADPR with NUDT9-H Energy Mean SD kcal/mol kcal/mol ELE (electrostatic energy) −44.42 7.33 VDW (VDW energy) −45.09 4.10 INT (bond, angle, dihedral energies) 0.00 0.00 GAS (ELE + VDW + INT) −89.51 7.13 GBSUR (hydrophobic contributor to solvent free energy for GB calculations) −6.04 0.37 GB (reaction field energy calculated by GB) 71.03 4.94 GBSOL (GBSUR + GB) 65.00 4.81 GBELE (GB + ELE) 26.61 4.67 GBTOT (GBSOL + GAS) −24.52 3.51 Table 2. Binding energy values of residues interacting with ADPR in the NUDT9-H domain Residues VDW Electrostatic Polar solvation Nonpolar solvation Total kcal/mol kcal/mol kcal/mol kcal/mol kcal/mol HIS1346 −1.40 −1.60 2.10 −0.23 −1.13 THR1347 −0.55 −6.53 0.23 −0.06 −6.91 TYR1349 −1.85 −2.99 2.94 −0.11 −2.01 LEU1379 −1.04 0.11 −0.49 −0.17 −1.60 GLY1389 −0.99 −3.22 1.01 −0.20 −3.40 SER1391 −0.14 0.27 −0.03 −7.32E-05 0.10 GLU1409 −0.09 1.08 −0.73 −0.01 0.26 ASP1431 −0.67 −3.14 3.98 −0.13 0.04 ARG1433 −1.17 0.30 0.72 −0.30 −0.45 LEU1484 −0.04 −0.10 −0.05 0.00 −0.18 HIS1488 −2.07 −1.16 0.93 −0.20 −2.51 Alanine scanning validated our predicted binding clefts for ADPR To obtain experimental evidence to verify the predicted binding pockets of ADPR, we initially performed alanine scanning of the 11 critical residues identified by the molecular docking and MD simulations. Whole-cell recordings were performed to measure ADPR-induced currents in HEK293 cells expressing hTRPM2 WT or mutant channels. A representative current recording of the WT channel activated by ADPR is shown in the left panel of Fig. 3 A, with the voltage-dependent current measured with the ramp protocol shown in the right panel. The I-V relationship exhibits linearity and strong sensitivity to inhibition by acidic pH, as reported previously (Yang et al., 2010). Previous studies using whole-cell recording have determined the EC50 value of ADPR from the concentration-response relationship to be 10 to 90 µM (Perraud et al., 2001; Starkus et al., 2007). Our measurements yielded an estimate of the EC50 value of 40 µM; the consistency with previous studies further validated our system. Two concentrations, at approximately EC10 (3 µM) and EC90 (100 µM) of ADPR, were used to activate the mutant forms of TRPM2. None of the mutants, with the exception of S1391A and L1379A, were completely activated by ADPR (Fig. 3 C), supporting the idea that these residues are critical for ADPR binding. To examine whether the observed loss of function in mutants was because of defective membrane trafficking, a biotinylation assay was used to determine surface expression of the mutant proteins. All of the nonfunctional mutants were delivered to the cell surface, as with the WT hTRPM2 channel (Fig. 3 D). Figure 3. Alanine substitution scanning of 11 identified residues in ADPR-binding pockets. (A) 100 µM ADPR-induced TRPM2 currents traces at −80 mV (left panel) and the I-V relationship curves (right panel, at time points indicated by a1 pointed to before pH 5.0 treatment and b1 pointed to after pH 5.0 inhibiting TRPM2 channel). W.C., whole cell. (B) The concentration–current response relationship for ADPR and the WT TRPM2 channel. Data are expressed as mean ± SEM from seven independent repetitions. (C) Summary of the currents induced by EC10 (gray) and EC90 (black) concentrations of ADPR. Data are expressed as mean ± SEM from at least five independent repetitions. (D) Biotinylation assay for surface expression of the TRPM2 and its mutants. The eight mutants that exhibited no channel function were expressed on the cell surface in HEK293 cells. The top panel shows surface expression, and the bottom panel shows total expression. Arrows indicate the specific band size of WT and the indicated TRPM2 mutants. Given that most of these 11 residues have not been investigated previously, we next addressed their roles in ADPR-induced activation of the hTRPM2 channel by characterizing the effects of point mutations on channel function. To avoid current variation resulting from variable expression level of WT or mutants in different batches of cells, we controlled the experimental condition as follows: (a) strictly 50 ng EGFP and 1 µg WT or mutant hTRPM2 plasmids were mixed and used for transfecting HEK293 cells; (b) electrophysiology experiments were performed 24–36 h after transfection; and (c) cells with similar green fluorescence intensity and cell size were chosen for patch-clamp recording. In addition, considering that some of the mutants failed to yield a saturated response even at millimolar concentrations of ADPR, all currents from the mutant channels were normalized to the maximal currents of WT hTRPM2 recorded on the same day and referred to as the relative currents. Residues interacting with the terminal ribose of ADPR ADPR is composed of three moieties: a terminal ribose, a pyrophosphate, and an adenosine (Fig. 2 B). Our results so far predicted that each of these moieties interacts with a different subset of residues in the NUDT9-H domain. Therefore, each of the residues was individually mutated to selected amino acids depending on the nature of the predicted interactions they form with ADPR. The mutational effects on ADPR-induced TRPM2 channel activation were evaluated. L1379, G1389, and E1409 were predicted to interact with the terminal ribose of ADPR (Fig. 4 A). Sequence alignments indicate that these residues are highly conserved in the TRPM2 proteins of various species (Fig. S2). E1409 was predicted in our MD simulation to have a strong polar solvation interaction between its hydroxyl oxygen atom and the terminal ribose of ADPR (Table 2). In support of this prediction, the E1409A mutation caused a dramatic increase in the EC50 value for ADPR (Fig. 4 B and Table 3). Furthermore, we produced the E1409Q mutant that would disrupt electrostatic interaction but maintain polar hydrogen bonding with the ligand. In contrast to E1409A, the E1409Q mutation caused a strong increase in sensitivity to ADPR (Fig. 4 B and Table 3), supporting the importance of the polar solvation interaction between this residue and the terminal ribose of ADPR. Figure 4. Characterization of interactions between the terminal ribose of ADPR and the NUDT9-H domain. (A) Ribbon presentation of the terminal ribose binding pockets in the NUDT9-H domain (gray). The interactions are denoted with yellow dashed lines. (B and C) Characterization of the ADPR concentration–current response relationship for the WT (dashed line) and indicated mutants (continuous line) of the TRPM2 channels. Data are expressed as mean ± SEM from five independent repetitions. (D) The representative traces induced by 50 mM calcium in HEK293 blank cell, TRPM2, and G1389A mutant–transfected cells. (E) Summary of the currents in D. (F) The representative trace induced by 10 mM H2O2 in HEK293 blank cell, TRPM2, and G1389A mutant–transfected cells. (G) Summary of the currents in F). Data in E and G are expressed as mean ± SEM from at least six independent repetitions. Table 3. The EC50 values of ADPR for the mutants in the NUDT9-H domain of TRPM2 Mutants EC50 ± SEM Hill slope µM WT 40.5 ± 4.5 2.6 H1346A NF NA H1346W 112.5 ± 6.4 2.0 T1347A NF NA T1347F 92.9 ± 11.2 1.2 T1347Y 115.1 ± 13.4 1.3 Y1349A NF NA Y1349I >10 mM NA L1379A ND NA L1379S 10.8 ± 3.5 1.0 S1391A 3.5 ± 0.2 2.7 E1409A 506.3 ± 108.7 1.5 E1409Q 5.4 ± 1.6 1.6 D1431A NF NA D1431N >100 mM NA R1433A NF NA R1433K NF NA R1433G >10 mM NA L1484V ND NA H1488A NF NA H1488F >10 mM NA NA, not applicable; ND, EC50 value is not significantly different from the WT, based on the measurements of currents induced by EC10 and EC90 concentrations of ADPR; NF, EC50 value is not defined because the mutation fails to yield a saturated response even at millimolar concentrations of ADPR. L1379 was predicted to interact with the terminal ribose via VDW forces (Table 2). The L1379A mutation, which shortened the side chain of leucine, had no significant effect on the sensitivity to ADPR (Table 3), suggesting that VDW force might not be critical for L1379. However, introducing a polar group by a L1379S mutation not only increased the sensitivity to ADPR, but also resulted in a much shallower slope of the concentration-response curve (Fig. 4 C and Table 3). These strong effects, although not necessarily supporting the existence of specific VDW interactions, suggest the prediction that L1379 directly participates in mediating ligand–host interaction. Our simulation suggested the occurrence of an electrostatic interaction between the carboxyl group of G1389 and the hydroxyl group of the terminal ribose ring of ADPR. To examine this prediction, G1389 was mutated to alanine to add a methyl group to glycine, which has the smallest side chain among all amino acids. Our result showed that the G1389A mutant failed to be activated by ADPR (Fig. 3 C). This result, although consistent with our prediction that G1389 is critical for ADPR binding, prevented us from constructing the concentration-response relationship as we did for mutants of E1409 and L1379. To further exclude the possibility that G1389A mutation affected the channel gating function, we tested with whole-cell patch recording whether this mutant could be activated by calcium or H2O2, which were previously reported to independently activate TRPM2. Consistent with our biotinylation data, G1389A mutant was found to be readily activated by calcium, confirming that this mutant retained channel function (Fig. 4, D and E). Interestingly, our data further showed that H2O2 failed to induce a current from G1389A mutant (Fig. 3, F and G), which suggests that H2O2 may regulate TRPM2 channel through the NUDT9-H domain. In summary, the collective results indicate that the terminal ribose of ADPR likely makes direct contacts with L1379, G1389, and E1409 in the NUDT9-H domain of TRPM2. Differences in the properties of these individual residues lead to their diverse contributions to ligand binding. Residues interacting with the pyrophosphate of ADPR A previous study showed that the pyrophosphate group of ADPR is critical for ADPR binding to the NUDT9-H domain (Moreau et al., 2013). However, channel residues in the ligand-binding domain that interact with the pyrophosphate have not been identified. Our simulations identified H1346, T1347, S1391, and R1433 as the candidate residues that interact with the pyrophosphate group (Fig. 5 A). Of note, these residues are also widely conserved (Fig. S2). Among these four residues, MD simulations suggested the imidazole ring of H1346 makes electrostatic interactions with the pyrophosphate group (Table 2). Functional tests showed that structural perturbations by either H1346A or H1346F mutation left the mutant channels completely insensitive to ADPR (Fig. 5 B), consistent with the idea that the electrostatic interactions mediated by H1346 might be critical for ADPR binding. In further support of this view, the H1346W mutation reduced the sensitivity to ADPR (Fig. 5 B), highlighting the importance of electrostatic interaction with ADPR binding as the replacement with a benzene ring by the mutation is expected to obviate the electronic property of histidine. Figure 5. Characterization of interactions between the pyrophosphate of ADPR and the NUDT9-H domain. (A) Ribbon presentation of the pyrophosphate binding pockets in the NUDT9-H domain (gray). The interactions are denoted with yellow dashed lines. (B–E) Characterization of the ADPR concentration–current response relationship for the WT (dashed line) and the indicated mutant (continuous line) of the TRPM2 channels. The insets in B–E represent summary of the currents from additional mutants of the indicated residues when activated by 100 µM (black) or 3 µM (gray) ADPR. Data in B–E are expressed as mean ± SEM from at least five independent repetitions. Our simulations predicted that T1347, located in close vicinity to H1346, is also important in the electrostatic interaction with ADPR (Table 2). To validate this prediction, we generated T1347A, T1347I, T1347F, and T1347Y mutants. T1347I mutant channels completely lost sensitivity to ADPR, whereas T1347A mutant channels yielded only a minor response to ADPR (Fig. 5 C), consistent with the expectation that eliminating the potential electrostatic interaction would substantially affect ligand–channel interaction. Moreover, the T1347F and T1347Y mutants slightly reduced sensitivity to ADPR (Fig. 5 C), which could be interpreted that the addition of a benzene ring at this position introduced steric hindrance that may interfere with the electrostatic interaction between the hydroxyl group of T1347 and ADPR. It is noted that the slopes of the ADPR concentration-response curves for both the T1347F and T1347Y mutants were substantially altered, indicating a likely change of the binding pattern (Fig. 5 C). These results together supported the notion of a critical role of the side chain of threonine in ADPR binding to the NUDT9-H domain. VDW force interactions were predicted to occur between S1391 and the pyrophosphate group of ADPR (Table 2). Interestingly, the S1391A mutant that increased the distance between the residue and ADPR had a dramatically decreased EC50 for ADPR (Fig. 5 D and Table 3). Furthermore, neither the S1391F mutation, which replaced the polar hydroxyl group with a nonpolar benzene ring, nor the S1391Y mutation, which increased the length of the side chain in serine, had noticeably altered the channel sensitivity to ADPR (Fig. 5 D), suggesting that the hydroxyl group of this residue was not critical for ADPR binding. These observations are consistent with the prediction that VDW force might have a major contribution to ADPR binding at this site. Additionally, the VDW force and a minor nonpolar solvation interaction between R1433 and the pyrophosphate of ADPR were predicted in our simulations (Table 2). This prediction appeared rather strange in that the charged residue arginine unexpectedly provides no contribution to electrostatic or polar solvation interactions. To further investigate this prediction, we produced a series of mutations to change the length, polarity, and charge characteristics of the side chain at this position. Among R1433A, R1433K, R1433L, R1433Q, and R1433G mutants, only the R1433G mutant exhibited a low affinity for ADPR, whereas all of the others had no response to ADPR (Fig. 5 E and Table 3). Although these results did not validate the accuracy of the prediction in our simulations, they are consistent with the importance of this residue for ADPR binding to the NUDT9-H domain. Residues interacting with the adenosine region of ADPR Our simulations predicted that Y1349, D1431, L1484, and H1488 contribute in the interaction with the adenosine group of ADPR (Fig. 6 A). For Y1349, it interacts with the adenine of ADPR through a π–π interaction, which is a form of electrostatic interaction. In addition, an interaction between its hydroxyl group and the terminal ribose was predicted (Fig. 6 A and Table 2). Our mutagenesis experiments showed that the Y1349F mutation did not significantly alter channel activation by ADPR (Fig. 6 B), indicating that the hydroxyl group of Y1349 contributed little to ADPR binding. However, neither Y1349A nor Y1349S mutant could be activated by ADPR (Fig. 6 B). Even the Y1349I mutant, which has the longest side chain, exhibited a very low response to ADPR. These observations suggest that the benzene ring of Y1349 is critical for ADPR binding. Our data are consistent with the idea that a π–π interaction mediated by the benzene ring of Y1349 is required for ADPR binding (Fig. 6 B). Figure 6. Characterization of interactions between the adenosine of ADPR and the NUDT9-H domain. (A) Ribbon presentation of the adenosine binding pockets in the NUDT9-H domain (gray). The interactions are denoted with yellow dashed lines. (B–D) Characterization of the ADPR concentration–current response relationship for the WT (dashed line) and the indicated mutant (continuous line) of the TRPM2 channels. The inset in B represents a summary of the currents from additional mutants of the indicated residue when activated by 100 µM (black) or 3 µM (gray) ADPR. Data in B–D are expressed as mean ± SEM from five independent repetitions. Our simulations also predicted a strong electrostatic interaction between D1431 and ADPR (Table 2). Results from mutagenesis experiments supported this prediction, as the D1431A mutation exhibited almost no response to ADPR, whereas the D1431N mutant, which neutralized the negative charge of aspartic acid, had a much higher EC50 value for ADPR than that of the WT channel (Fig. 6 C and Table 3). These results support the notion that the likely electrostatic interaction from D1431 is necessary for ADPR binding to the NUDT9-H domain. In addition, our MD simulations suggested a weak VDW force between the isobutyl side chain of L1484 and the adenosine (Table 2). The lack of channel function in the L1484A mutant, which has a decreased distance between the side chain of this residue and ADPR, supports the notion of an important role in ADPR binding for the VDW force interaction (Fig. 3 C). In further support of this notion, when L1484 was mutated to valine that has a similar isopropyl side chain, the binding affinity of ADPR did not change (Table 3). H1488 was predicted to interact with the adenosine of ADPR by both VDW force and electrostatic interactions (Table 2). Our mutagenesis experiments showed that the H1488F mutant that lost the positive charge on the imidazole ring of histidine had a much-reduced sensitivity to ADPR (Fig. 6 D), consistent with that disruption of electrostatic interaction did affect the binding affinity between H1488 and ADPR. A working model of ADPR binding inside the NUDT9-H domain Collectively, our mutagenesis and functional assay results not only demonstrated that most of the residues predicted in our MD simulations are indeed critical for ADPR binding, but also lent support to the idea of an interaction force between ADPR and H1346, T1347, Y1349, S1391, E1409, D1431, L1484, and H1488. A working model for ADPR binding to the NUDT9-H domain of TRPM2 that incorporates our observations is presented in Fig. 7. The identified key residues (shown in Fig. 7 C in green) define a clear outline of the binding pocket for ADPR. In this model, the molecular surface of the binding pocket contains two large cavities: one accommodates the adenine by the space parallel to the purine ring of ADPR, which is mainly formed by L1484 and H1488; the other one is formed by L1379, G1389, and E1409 to enwrap the terminal ribose of ADPR. Between these two moieties, the other critical residues of the NUDT9-H domain formed a shallower trench, with the pyrophosphate and the ribose of the adenosine of ADPR lying in it. Figure 7. Binding pocket of ADPR in the NUDT9-H domain presented from different orientations. A and B show the side (A) and top view (B) of the ADPR-binding model in the NUDT9-H domain. The positively and negatively charged residues are colored in blue and red, respectively. (C) The whole view of the binding pocket of ADPR in the NUDT9-H domain. Critical residues for binding are labeled in green. Discussion In this study, we investigated binding of ADPR to the NUDT9-H domain and proposed a detailed structural model for atomic interactions that mediate ligand–host interactions. This model is not only supported by our systematic functional tests but also by results from previous studies. Indeed, recent studies have identified several residues clustered in regions of the NUDT9-H domain that are critical for ADPR-induced activation of the TRPM2 channel (Perraud et al., 2005; Du et al., 2009; Tóth et al., 2014). Although these individual sites alone could not reveal the precise and complete binding mode of ADPR within the NUDT9-H domain, when mapped on our model, it becomes clear how they may contribute to ligand binding. Based on the crystal structure of human NUDT9 and the Escherichia coli ADPRase–ADPR complex, Shen et al. (2003) initially proposed a hypothetical model of ADPR binding to the NUDT9-H domain. A subsequent study generated a homology model of the NUDT9-H domain and docked ADPR into the anticipated ADPR-binding pocket (Perraud et al., 2005). The docked ADPR took a different conformation in that the horseshoe-shaped molecule points its ends in the opposite direction from that of our model. In the present study, we generated a structure model of the NUDT9-H domain guided by the crystal structure of NUDT9 and defined the ligand-binding pocket formed by the NUDT9-H domain using structural modeling, ligand docking simulations, and MD simulations. Our MD simulation results predicted that 11 residues in the binding pocket interact strongly with ADPR through VDW forces, electrostatic interactions, polar solvation interactions, and nonpolar solvation interactions. By combining MD simulations with site-directed mutagenesis studies, we demonstrated that all of the identified residues in the NUDT9-H domain interact with ADPR, supporting our predicted binding pocket model for TRPM2. The homology model of the NUDT9-H domain in our study is very similar to the model reported in a previous study (Perraud et al., 2005), which is not surprising given the high sequence similarity between NUDT9-H and human NUDT9. However, our predicted ADPR binding pocket and the binding orientation of ADPR are clearly different from those in the earlier model. To test whether ADPR can bind in the previously proposed orientation, we also docked ADPR into the NUDT9-H domain in a similar orientation (Fig. 8 A). We found that this orientation yielded a much reduced docking IFD score (Table 4). To more easily compare this model with our predicted one, we merged them together (Fig. S3) and found ADPR in this orientation would interact with NUDT9-H at several additional and potentially important residues, including D1287, S1338, S1340, N1345, S1382, and I1440. Among these residues, D1287, S1340, and N1345 were previously proposed to interact with ADPR but not examined further using functional assays (Shen et al., 2003). Therefore, we mutated these residues to alanine individually and examined by patch-clamp recording of ADPR-induced currents in cells expressing the mutant channels. Both the D1287A and S1340A mutant channels had normal activity induced by ADPR (Fig. 8, B–F), supporting the conclusion that these residues have no major role in ADPR binding. Both the model proposed by Perraud et al. (2005) and our model are based on the crystallographic information of human NUDT9, which contained a ligand molecule, R5P (Shen et al., 2003). In the previous model, ADPR assumes a horseshoe-like conformation that bends over the superficial surface of the ADPR-binding pocket (Perraud et al., 2005). However, this does not match the position of R5P in the crystal structure of NUDT9. To illustrate the difference in these binding models, we highlight the critical residues (W110, N168, M216, R273, M280, and D305) identified in the Perraud model in red in our structure model (Fig. 9, A and B). In our model, ADPR locates in a deeper cleft, a position consistent with that of R5P in the cleft of the NUDT9 structure (Fig. 9, C and D). Of note, key residues that have been previously reported to interact with R5P in the crystal structure of the NUDT9 protein are also identified in our model for ligand binding, including Y1349 (D172), L1379 (R204), E1409 (E234), R1433 (R273), and H1488 (H324) (Fig. 9, E and F). Our mutagenesis and functional results demonstrate that these residues are indeed critical sites for ADPR binding to the NUDT9-H domain. These results lend further support for the accuracy of our ADPR binding model. Figure 8. Evaluation of a previous model for ADPR binding to the NUDT9-H domain. (A) The previously proposed orientation of ADPR shown in the NUDT9-H domain using our present model. In this orientation, ADPR is in close proximity to several residues for which mutations did not affect binding. Current traces of WT TRPM2 (B) and mutants at three positions (C–E) predicted to interact with ADPR in the orientation shown in A. (F) Summary of the currents from the mutants at the three positions when activated by ADPR 100 µM. Data are expressed as mean ± SEM from at least five independent repetitions. Table 4. Comparing of the IFD scoring of the model similar to the one previously proposed and our prediction Models Docking score Glide gscore Glide emodel Prime energy IFD score kcal/mol kcal/mol kcal/mol kcal/mol kcal/mol Our prediction −8.4 −8.4 −98.13 −3,710.1 −3,823.0 The model similar to the one previously proposed −5.9 −5.9 −90.7 −3,644.7 −3,750.5 Figure 9. Molecular surface representation for the structure of the NUDT9 and NUDT9-H domains. (A) The side view of molecular surface representation for the NUDT9 structure. The previously proposed critical residues (W110, N168, M216, R273, M280, and D305) are highlighted in red. (B) The side view of molecular surface representation for the structure of NUDT9-H domain. The same critical residues shown in A are labeled in red. (C) Molecular surface representation for the structure of NUDT9 domain binding with R5P in the previously reported model. (D) Molecular surface representation for the structure of NUDT9-H domain binding with ADPR in our structure model. (E) D172, R204, E234, R273, and H324 residues labeled in red interact with R5P in the crystal structure of the NUDT9 protein. (F) Y1349, L1379, E1409, R1433, and H1488 residues that were reported to interact with R5P are labeled in in our NUDT9-H structure to interact with ADPR. To further verify our model, we also examined how compatible our ADPR-binding pocket is with the results from previously published studies. First, a recent study reported that when the 2’-hydroxyl group of ADPR was replaced with a phosphate group, it could still function as a ligand of TRPM2 (Tóth et al., 2015). This can be explained by our model: there is an open space at the vicinity of the 2’-hydroxyl of ADPR that can conceivably accommodate both the extra mass and the more polar phosphate group (Fig. 7 A). Second, Moreau et al. (2013) synthesized several ADPR analogues and examined their effects on the function of TRPM2. They found that a compound, as modified with a phenyl group at C8 of the adenine and simultaneously modified with a 2’-deoxy motif at the adenosine ribose (8-phenyl-2-deoxy-ADPR), became a highly potent and specific antagonist with high affinity for the NUDT9-H domain (half-maximal inhibitory concentration of 3 µM). Consistently, our model has a cavity close to the purine ring of ADPR in the binding pocket that can accommodate a phenyl group (Fig. 7 A). Because only the C8-position-modified ADPR analogues exhibit highly antagonistic activities, the adenosine base of ADPR may very well be a critical factor in ADPR binding and gating of the TRPM2 channel. Finally, most of the residues identified in the present study have not been studied previously. The only exception is E1409, which was tested in two studies using an E-to-K mutation (Perraud et al., 2005; Tóth et al., 2014). The E1409K mutation was found to have little impact on ADPR binding to the NUDT9-H domain. Our simulations predicted no electrostatic interaction between the terminal ribose of ADPR and the hydroxyl oxygen atom of E1409. This prediction was further validated by our results from the E1409Q and E1409A mutants. Although we did not test the E1409K mutant, the reported results supported our model: by reversing the charge carried by the side chain, the E1409K mutation retained the polar solvation interaction but reversed the electrostatic interaction, further indicating that the electronic properties of E1409 do not affect ADPR binding to the NUDT9-H domain. All of these results further confirm the power and effectiveness of the methodological strategy of combining computation with experimental assays. In summary, we revealed in this study the structural basis for ADPR binding to the NUDT9-H domain in the TRPM2 channel. Accumulating evidence has indicated TRPM2 is a potential therapeutic target for many diseases, and our findings will not only significantly advance the understanding of the molecular mechanisms of TRPM2 channel activation by ADPR, but will also provide the structural basis for future drug design efforts targeting the TRPM2 channel. Supplementary Material Supplemental Materials (PDF) Acknowledgments The authors thank Prof. Jie Zheng, Dr. Fan Yang, and Dr. John Hugh Snyder for critical comments and reading the manuscript. This work was supported by grants from the National Basic Research Program of China (2013CB910204 to W. Yang and 2014CB910300 to J. Luo) and the Natural Science Foundation of China (21402171 to P. Yu). The authors declare no competing financial interests. Sharona E. Gordon served as editor. Abbreviations used: ADPR adenosine diphosphate ribose EC50 half-maximal effective concentration GBSA generalized Born surface area HEK human embryonic kidney IFD induced-fit docking MD molecular dynamic MM molecular mechanics NUDT9-H NUDT9 homology VDW van der Waals ==== Refs Alim, I., L. Teves, R. Li, Y. Mori, and M. Tymianski. 2013. Modulation of NMDAR subunit expression by TRPM2 channels regulates neuronal vulnerability to ischemic cell death. J. Neurosci. 33 :17264–17277. 10.1523/JNEUROSCI.1729-13.2013 24174660 Ashby, J.A., I.V. McGonigle, K.L. Price, N. Cohen, F. Comitani, D.A. Dougherty, C. Molteni, and S.C. Lummis. 2012. GABA binding to an insect GABA receptor: A molecular dynamics and mutagenesis study. Biophys. J. 103 :2071–2081. 10.1016/j.bpj.2012.10.016 23200041 Csanády, L., and B. Törocsik. 2009. Four Ca2+ ions activate TRPM2 channels by binding in deep crevices near the pore but intracellularly of the gate. J. Gen. Physiol. 133 :189–203. 10.1085/jgp.200810109 19171771 Di, A., X.P. Gao, F. Qian, T. Kawamura, J. Han, C. Hecquet, R.D. Ye, S.M. Vogel, and A.B. Malik. 2012. The redox-sensitive cation channel TRPM2 modulates phagocyte ROS production and inflammation. Nat. Immunol. 13 :29–34. 10.1038/ni.2171 Du, J., J. Xie, and L. Yue. 2009. Intracellular calcium activates TRPM2 and its alternative spliced isoforms. Proc. Natl. Acad. Sci. USA. 106 :7239–7244. 10.1073/pnas.0811725106 19372375 Friesner, R.A., J.L. Banks, R.B. Murphy, T.A. Halgren, J.J. Klicic, D.T. Mainz, M.P. Repasky, E.H. Knoll, M. Shelley, J.K. Perry, 2004. Glide: A new approach for rapid, accurate docking and scoring. 1. Method and assessment of docking accuracy. J. Med. Chem. 47 :1739–1749. 10.1021/jm0306430 15027865 Gao, G., W. Wang, R.K. Tadagavadi, N.E. Briley, M.I. Love, B.A. Miller, and W.B. Reeves. 2014. TRPM2 mediates ischemic kidney injury and oxidant stress through RAC1. J. Clin. Invest. 124 :4989–5001. 10.1172/JCI76042 25295536 Iordanov, I., C. Mihályi, B. Tóth, and L. Csanády. 2016. The proposed channel-enzyme transient receptor potential melastatin 2 does not possess ADP ribose hydrolase activity. eLife. 5 :e17600. 10.7554/eLife.17600 27383051 Jiang, L.H., W. Yang, J. Zou, and D.J. Beech. 2010. TRPM2 channel properties, functions and therapeutic potentials. Expert Opin. Ther. Targets. 14 :973–988. 10.1517/14728222.2010.510135 20670202 Laskowski, R.A., M.W. Macarthur, D.S. Moss, and J.M. Thornton. 1993. PROCHECK: A program to check the stereochemical quality of protein structures. J. Appl. Cryst. 26 :283–291. 10.1107/S0021889892009944 Lu, W., W. Fang, J. Li, B. Zhang, Q. Yang, X. Yan, L. Peng, H. Ai, J.J. Wang, X. Liu, 2015. Phosphorylation of tyrosine 1070 at the GluN2B subunit is regulated by synaptic activity and critical for surface expression of N-methyl-D-aspartate (NMDA) receptors. J. Biol. Chem. 290 :22945–22954. 10.1074/jbc.M115.663450 26229100 Manna, P.T., T.S. Munsey, N. Abuarab, F. Li, A. Asipu, G. Howell, A. Sedo, W. Yang, J. Naylor, D.J. Beech, 2015. TRPM2-mediated intracellular Zn2+ release triggers pancreatic β-cell death. Biochem. J. 466 :537–546. 10.1042/BJ20140747 25562606 Massova, I., and P.A. Kollman. 2000. Combined molecular mechanical and continuum solvent approach (MM-PBSA/GBSA) to predict ligand binding. Perspect. Drug Discov. Des. 18 :113–135. 10.1023/A:1008763014207 Miller, B.A., N.E. Hoffman, S. Merali, X.Q. Zhang, J. Wang, S. Rajan, S. Shanmughapriya, E. Gao, C.A. Barrero, K. Mallilankaraman, 2014. TRPM2 channels protect against cardiac ischemia-reperfusion injury: Role of mitochondria. J. Biol. Chem. 289 :7615–7629. 10.1074/jbc.M113.533851 24492610 Moreau, C., T. Kirchberger, J.M. Swarbrick, S.J. Bartlett, R. Fliegert, T. Yorgan, A. Bauche, A. Harneit, A.H. Guse, and B.V. Potter. 2013. Structure-activity relationship of adenosine 5′-diphosphoribose at the transient receptor potential melastatin 2 (TRPM2) channel: Rational design of antagonists. J. Med. Chem. 56 :10079–10102. 10.1021/jm401497a 24304219 Perraud, A.L., A. Fleig, C.A. Dunn, L.A. Bagley, P. Launay, C. Schmitz, A.J. Stokes, Q. Zhu, M.J. Bessman, R. Penner, 2001. ADP-ribose gating of the calcium-permeable LTRPC2 channel revealed by Nudix motif homology. Nature. 411 :595–599. 10.1038/35079100 11385575 Perraud, A.L., B. Shen, C.A. Dunn, K. Rippe, M.K. Smith, M.J. Bessman, B.L. Stoddard, and A.M. Scharenberg. 2003. NUDT9, a member of the Nudix hydrolase family, is an evolutionarily conserved mitochondrial ADP-ribose pyrophosphatase. J. Biol. Chem. 278 :1794–1801. 10.1074/jbc.M205601200 12427752 Perraud, A.L., C.L. Takanishi, B. Shen, S. Kang, M.K. Smith, C. Schmitz, H.M. Knowles, D. Ferraris, W. Li, J. Zhang, 2005. Accumulation of free ADP-ribose from mitochondria mediates oxidative stress-induced gating of TRPM2 cation channels. J. Biol. Chem. 280 :6138–6148. 10.1074/jbc.M411446200 15561722 Qian, X., T. Numata, K. Zhang, C. Li, J. Hou, Y. Mori, and X. Fang. 2014. Transient receptor potential melastatin 2 protects mice against polymicrobial sepsis by enhancing bacterial clearance. Anesthesiology. 121 :336–351. 10.1097/ALN.0000000000000275 24781495 Sano, Y., K. Inamura, A. Miyake, S. Mochizuki, H. Yokoi, H. Matsushime, and K. Furuichi. 2001. Immunocyte Ca2+ influx system mediated by LTRPC2. Science. 293 :1327–1330. 10.1126/science.1062473 11509734 Shen, B.W., A.L. Perraud, A. Scharenberg, and B.L. Stoddard. 2003. The crystal structure and mutational analysis of human NUDT9. J. Mol. Biol. 332 :385–398. 10.1016/S0022-2836(03)00954-9 12948489 Song, K., H. Wang, G.B. Kamm, J. Pohle, F.C. Reis, P. Heppenstall, H. Wende, and J. Siemens. 2016. The TRPM2 channel is a hypothalamic heat sensor that limits fever and can drive hypothermia. Science. 353 :1393–1398 (P). 10.1126/science.aaf7537 27562954 Starkus, J., A. Beck, A. Fleig, and R. Penner. 2007. Regulation of TRPM2 by extra- and intracellular calcium. J. Gen. Physiol. 130 :427–440. 10.1085/jgp.200709836 17893195 Tan, C.H., and P.A. McNaughton. 2016. The TRPM2 ion channel is required for sensitivity to warmth. Nature. 536 :460–463. 10.1038/nature19074 27533035 Tóth, B., and L. Csanády. 2012. Pore collapse underlies irreversible inactivation of TRPM2 cation channel currents. Proc. Natl. Acad. Sci. USA. 109 :13440–13445. 10.1073/pnas.1204702109 22847436 Tóth, B., I. Iordanov, and L. Csanády. 2014. Putative chanzyme activity of TRPM2 cation channel is unrelated to pore gating. Proc. Natl. Acad. Sci. USA. 111 :16949–16954. 10.1073/pnas.1412449111 25385633 Tóth, B., I. Iordanov, and L. Csanády. 2015. Ruling out pyridine dinucleotides as true TRPM2 channel activators reveals novel direct agonist ADP-ribose-2′-phosphate. J. Gen. Physiol. 145 :419–430. 10.1085/jgp.201511377 25918360 Uchida, K., K. Dezaki, B. Damdindorj, H. Inada, T. Shiuchi, Y. Mori, T. Yada, Y. Minokoshi, and M. Tominaga. 2011. Lack of TRPM2 impaired insulin secretion and glucose metabolisms in mice. Diabetes. 60 :119–126. 10.2337/db10-0276 20921208 Weiss, J.N. 1997. The Hill equation revisited: Uses and misuses. FASEB J. 11 :835–841.9285481 Yamamoto, S., S. Shimizu, S. Kiyonaka, N. Takahashi, T. Wajima, Y. Hara, T. Negoro, T. Hiroi, Y. Kiuchi, T. Okada, 2008. TRPM2-mediated Ca2+influx induces chemokine production in monocytes that aggravates inflammatory neutrophil infiltration. Nat. Med. 14 :738–747. 10.1038/nm1758 18542050 Yang, F., X. Xiao, W. Cheng, W. Yang, P. Yu, Z. Song, V. Yarov-Yarovoy, and J. Zheng. 2015. Structural mechanism underlying capsaicin binding and activation of the TRPV1 ion channel. Nat. Chem. Biol. 11 :518–524. 10.1038/nchembio.1835 26053297 Yang, W., J. Zou, R. Xia, M.L. Vaal, V.A. Seymour, J. Luo, D.J. Beech, and L.H. Jiang. 2010. State-dependent inhibition of TRPM2 channel by acidic pH. J. Biol. Chem. 285 :30411–30418. 10.1074/jbc.M110.139774 20660597 Yang, W., P.T. Manna, J. Zou, J. Luo, D.J. Beech, A. Sivaprasadarao, and L.H. Jiang. 2011. Zinc inactivates melastatin transient receptor potential 2 channels via the outer pore. J. Biol. Chem. 286 :23789–23798. 10.1074/jbc.M111.247478 21602277 Ye, M., W. Yang, J.F. Ainscough, X.P. Hu, X. Li, A. Sedo, X.H. Zhang, X. Zhang, Z. Chen, X.M. Li, 2014. TRPM2 channel deficiency prevents delayed cytosolic Zn2+ accumulation and CA1 pyramidal neuronal death after transient global ischemia. Cell Death Dis. 5 :e1541. 10.1038/cddis.2014.494 25429618 Yu, W., L.H. Jiang, Y. Zheng, X. Hu, J. Luo, and W. Yang. 2014. Inactivation of TRPM2 channels by extracellular divalent copper. PLoS One. 9 :e112071. 10.1371/journal.pone.0112071 25386648 Zhang, X.M., X.Y. Yan, B. Zhang, Q. Yang, M. Ye, W. Cao, W.B. Qiang, L.J. Zhu, Y.L. Du, X.X. Xu, 2015. Activity-induced synaptic delivery of the GluN2A-containing NMDA receptor is dependent on endoplasmic reticulum chaperone Bip and involved in fear memory. Cell Res. 25 :818–836. 10.1038/cr.2015.75 26088419 Zhou, L., and S.A. Siegelbaum. 2007. Gating of HCN channels by cyclic nucleotides: Residue contacts that underlie ligand binding, selectivity, and efficacy. Structure. 15 :655–670. 10.1016/j.str.2007.04.012 17562313

PMC005xxxxxx/PMC5379919.txt

==== Front J Gen Physiol J Gen Physiol jgp jgp The Journal of General Physiology 0022-1295 1540-7748 The Rockefeller University Press 28330839 201611683 10.1085/jgp.201611683 Research Articles Research Article 509 501 Evolutionary insights into T-type Ca2+ channel structure, function, and ion selectivity from the Trichoplax adhaerens homologue The Trichoplax adhaerens T-type calcium channel Smith Carolyn L. 1* Abdallah Salsabil 2* http://orcid.org/0000-0002-9567-5304 Wong Yuen Yan 2* Le Phuong 2* Harracksingh Alicia N. 2 Artinian Liana 3 Tamvacakis Arianna N. 3 Rehder Vincent 3 Reese Thomas S. 1 http://orcid.org/0000-0001-5180-1180 Senatore Adriano 2 1 National Institute of Neurological Diseases and Stroke, National Institutes of Health, Bethesda, MD 20892 2 University of Toronto Mississauga, Mississauga, Ontario L5L 1C6, Canada 3 Georgia State University, Atlanta, GA 30302 Correspondence to Adriano Senatore: adriano.senatore@utoronto.ca * C.L. Smith, S. Abdallah, Y.Y. Wong, and P. Le contributed equally to this paper. 03 4 2017 149 4 483510 28 8 2016 07 2 2017 © 2017 Smith et al. 2017 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). The role of T-type calcium channels in animals without nervous systems is unknown. Smith et al. characterize TCav3 from Trichoplax adhaerens, finding expression in neurosecretory-like cells and preference for Ca2+ over Na+ via strong extracellular Ca2+ block, despite low selectivity for Ca2+ in the pore. Four-domain voltage-gated Ca2+ (Cav) channels play fundamental roles in the nervous system, but little is known about when or how their unique properties and cellular roles evolved. Of the three types of metazoan Cav channels, Cav1 (L-type), Cav2 (P/Q-, N- and R-type) and Cav3 (T-type), Cav3 channels are optimized for regulating cellular excitability because of their fast kinetics and low activation voltages. These same properties permit Cav3 channels to drive low-threshold exocytosis in select neurons and neurosecretory cells. Here, we characterize the single T-type calcium channel from Trichoplax adhaerens (TCav3), an early diverging animal that lacks muscle, neurons, and synapses. Co-immunolocalization using antibodies against TCav3 and neurosecretory cell marker complexin labeled gland cells, which are hypothesized to play roles in paracrine signaling. Cloning and in vitro expression of TCav3 reveals that, despite roughly 600 million years of divergence from other T-type channels, it bears the defining structural and biophysical features of the Cav3 family. We also characterize the channel’s cation permeation properties and find that its pore is less selective for Ca2+ over Na+ compared with the human homologue Cav3.1, yet it exhibits a similar potent block of inward Na+ current by low external Ca2+ concentrations (i.e., the Ca2+ block effect). A comparison of the permeability features of TCav3 with other cloned channels suggests that Ca2+ block is a locus of evolutionary change in T-type channel cation permeation properties and that mammalian channels distinguish themselves from invertebrate ones by bearing both stronger Ca2+ block and higher Ca2+ selectivity. TCav3 is the most divergent metazoan T-type calcium channel and thus provides an evolutionary perspective on Cav3 channel structure–function properties, ion selectivity, and cellular physiology. National Science and Engineering Research Council of Canada https://doi.org/10.13039/501100000046 RGPIN-2016-06023 Canadian Foundation for Innovation https://doi.org/10.13039/501100000196 CFI Project 35297 University of Toronto https://doi.org/10.13039/501100003579 NSERC https://doi.org/10.13039/501100000046 PDF-43851-2013 ==== Body pmcIntroduction Voltage-gated calcium (Cav) channels play fundamental roles in the physiology of neurons and muscle, by coupling electrical signals carried largely by voltage-gated sodium (Nav) and potassium (Kv) channels, with intracellular Ca2+-dependent processes (Clapham, 2007). Of the three classes of Cav channels, L-type/Cav1 channels are central for excitation-contraction coupling in muscle and excitation-transcription coupling in neurons and muscle, whereas N- and P-/Q-type (i.e., Cav2) channels are central for fast presynaptic excitation-secretion coupling (Catterall, 2011). T-type/Cav3 channels serve less obvious functions (Perez-Reyes, 2003; Senatore et al., 2012), but one clear contribution is their role in regulating cellular excitability, where their low voltages of activation and fast kinetics permit rapid depolarizing Ca2+ currents below the action potential threshold. T-type channels also play roles in driving low threshold exocytosis in both vertebrates and invertebrates, and in mammals have been shown to directly interact with presynaptic components of the vesicular SNARE complex (Weiss et al., 2012; Weiss and Zamponi, 2013). Notably, recent genomic studies indicate that T-type channels, and in fact the majority of genes with important roles in the nervous system, are present in primitive animals that lack nervous systems and single-celled organisms that predate animals (King et al., 2008; Srivastava et al., 2008, 2010; Steinmetz et al., 2012; Moran et al., 2015; Moroz and Kohn, 2015). We know little, however, about the function and properties of these extant gene homologues or about the functional or proteomic adaptations that were required to incorporate their primordial counterparts into nervous system function. One very intriguing early diverging animal is Trichoplax adhaerens (phylum Placozoa), which has only six cell types and lacks synaptically connected neurons and muscle (Schierwater, 2005; Smith et al., 2014). Despite these absences, Trichoplax is able to coordinate motile behavior such as feeding (Smith et al., 2015), chemotaxis, and phototaxis (Heyland et al., 2014), indicative of trans-cellular signaling and communication independent of both chemical and electrical synapses. Given that Cav channels play crucial roles in both intra- and intercellular signaling, it is intriguing that the Trichoplax genome bears a full complement of Cav channel genes: Cav1, Cav2, and Cav3 (Srivastava et al., 2008). Here, we sought to characterize the molecular properties of the most basal metazoan homologue of T-type channels from T. adhaerens. Co-immunolocalization of the channel, named TCav3, with neurosecretory cell marker complexin labeled gland cells, shown previously to resemble neurosecretory cells in their expression of SNARE proteins and the presence of membrane-apposed vesicles (Syed and Schierwater, 2002; Smith et al., 2014). We cloned and in vitro expressed TCav3, finding that despite its ancient divergence, it bears the hallmark structural and biophysical features of T-type channels, including a low voltage of activation, rapid and transient kinetics, and an apparent Ca2+ window current near resting membrane potential. We also characterized the permeation properties of TCav3, finding that the channel conducts moderately mixed inward Ca2+-Na+ currents, with a majority of current carried by Ca2+, similar to mammalian homologues (Shcheglovitov et al., 2007). Paradoxically, measuring Ca2+ over Na+ selectivity using bi-ionic reversal potential analysis (i.e., where inward Ca2+ ions compete with outward Na+ for permeation), revealed poor Ca2+ versus Na+ selectivity compared with human Cav3.1, similar to the cloned T-type channel from invertebrate snail Lymnaea stagnalis (Senatore and Spafford, 2010; Senatore et al., 2014). We attribute the relatively low Na+ permeation through TCav3, in spite of its poor Ca2+ over Na+ selectivity, to retention of a potent Ca2+ block. Based on comparative data, we suggest that Ca2+ block is more crucial for determining the degree of Na+ that permeates alongside Ca2+ compared with pore selectivity and is a locus for evolutionary change in T-type channel cation permeability. Materials and methods Cloning of the TCav3 channel cDNA Two cDNA libraries were made from Trichoplax whole-animal total RNA, one with an anchored oligo-dT18 primer, for PCR amplification and cloning of the C-terminal half of TCav3, and the other with a primer targeting a central region of the TCav3 coding sequence, for cloning the N terminus (Table 1). The TCav3 N- and C-terminal coding sequences were then independently amplified three times from the cDNA, via nested PCR using Pfu Turbo DNA polymerase (Agilent Technologies), with nested N- and C-terminal primer pairs containing NheI–XhoI and XhoI–XmaI sites, respectively. The nested NT primer (TCav3 NT 5′2) also contained a mammalian Kozak translation initiation site (Kozak, 1986; i.e., 5′-GCCACC-3′; Table 1) for effective expression of the TCav3 channel protein in mammalian cells. PCR-amplified DNA fragments were subcloned into pIRES2-IR–enhanced green fluorescent protein (EGFP), sequenced, and compared with each other plus the Trichoplax genome (JGI Genome Portal, Grell-BS-1999 v1.0, scaffold_2:6781672-6793175) to generate a consensus coding sequence. The full-length TCav3 clone was then prepared by inserting the XhoI–XmaI C-terminal subclone into the pIRES2 vector bearing the N-terminal TCav3 fragment, producing pTCav3-IR-EGFP. The full-length consensus coding sequence of TCav3 was submitted to GenBank (accession no. KJ466205). Table 1. Sequence of primers used for cloning TCav3 cDNA and semiquantitative RT-PCR Primer Sequence (5′ to 3′) TCav3 NT cDNA CTTTAGGTAGTATGAGCAAGGAATG TCav3 CT cDNA TTTTTTTTTTTTTTTTTTVN TCav3 NT 5′1 TGATGTTTTATTCAAGTCATGC TCav3 NT 5′2 TACTTAGCTAGCCGCGGGAGCCACC ATG GATATTTCGTTCGCTATCAT TCav3 NT 3′1 CATCAGGCGTCTTAACTTTGC TCav3 NT 3′2 TGCTCTGTGGTGGATCTCGAG TCav3 CT 3′1 CTTGATATTGATTAAATTTATCCAGATG TCav3 CT 3′2 ACTCATCCCGGG TTA TACTAATGTTTGAATCAT TCav3 CT 5′1 AAATCTGGTAGATCCTAATGAAGTC TCav3 CT 5′2 TCGCGTTCCAATTGGTCCTCGAG TCav1_F AGTCTTTACCGATTTGGCTCTTCTTTG TCav1_R CAATTATACGACTGACACTCTTTCAAAGC TCav2_F CGTTTACAAAAGTCTGGATCAAGTTGCG TCav2_R CTGCTGTACATTTTGATATGTTAGTAAGATCATCTG TCav3_F GTAGAAGATCAGATGGAATTAACGGCTTCC TCav3_R TACTAATGTTTGAATCATTTTTCGAGGTAATGTGACC TCavβ_F AGTAATAAATTCTCAGGATTAGTGGAAGTGTCC TCavβ_R AACATCATTACGTTTTATTACTAGAGGATCTTG TCavα2δ-a_F CTCAAGCAGTGGACTAACTAAACTTTACC TCavα2δ-a_R CATAGCTGGTTATATGTCACAAATTCCTCTC TCavα2δ-b_F TTTACCACATCCTTCTTCAGCTTCTTGAC TCavα2δ-b_R AGCATACACTGATAGCTTTACATAGCG TCavα2δ-c_F GATATTCGAAATTTCGAATGATACCCTTTGC TCavα2δ-c-R CAACGCGAGAAGTAGAAGTGTCCG Translation start and stop codons are in bold, and the Kozak sequences and restriction enzyme sites used for cloning are underlined. Reverse transcription (RT)–PCR amplification of Trichoplax Cav channel and accessory subunit mRNAs The Trichoplax genome encodes single gene homologues for each of the three metazoan Cav channel types (Cav1, Cav2, and Cav3; NCBI accession nos. XM_002108894.1, XM_002109739.1, and KJ466205, respectively), as well as a single Cavβ accessory subunit gene (XM_002110305.1) and three Cavα2δ Cav1/Cav2 accessory subunit genes (Cavα2δa, Cavα2δb, and Cavα2δc; NCBI accession nos. XM_002112625.1, XM_002112621.1, and XM_002111347.1). Primers were designed to amplify ∼500-bp cDNA fragments of each of these genes by RT–PCR (Table 1), using a cDNA library prepared by RT (SuperScript III Reverse Transcription; Thermo Fisher Scientific) with an anchored oligo-dT18 primer (Table 1) and whole-animal total RNA. PCR amplification was achieved in 25-µl reactions each containing 1.25 µl of 10 µM forward and reverse primers (Table 1); 0.125 µl of Taq DNA polymerase and 2.5 µl of corresponding 10× buffer (New England Biolabs, Inc.); 1 µl of 25 mM MgCl2; 0.5 µl of 10 mM dNTP mix (New England Biolabs, Inc.); and 0.5 µl of cDNA template. Thermocycling conditions were 95°C for 2 min, 30 cycles of 94°C for 1 min, 59°C for 45 s, and 72°C for 1 min, and a final 10-min extension at 72°C. Phylogenetic inference Maximum likelihood (ML) phylogeny of various Cav channels was inferred from a MUSCLE-alignment of select channel protein sequences (Edgar, 2004), generated with the program MEGA7 (Kumar et al., 2016). Alignments were timed with trimAl (Capella-Gutiérrez et al., 2009), followed by some minor manual trimming to remove highly heterogeneous regions (raw sequences and the trimmed alignment are provided in FASTA format as Supplementary Files 1 and 2, respectively). ML model selection was achieved with MEGA7, revealing that the LG+G model was most suitable under both the corrected Akaike’s information criterion (AICc) and the Bayesian information criterion (BIC). The ML phylogenetic tree presented in Fig. 1 B was thus inferred from the trimmed alignment using the LG+G model, with 1,000 bootstrap replicates to generate node support values. Protein accession numbers used in the analysis are as follows: Amphimedon Cav1/Cav2, Aqu2.38198_001 from published transcriptome (Fernandez-Valverde et al., 2015); Salpingoeca Cav1/Cav2, XP_004989719.1; Salpingoeca Cav3, XP_004995501.1; Trichoplax Cav1 and Cav2, unpublished transcriptome; Trichoplax Cav3, KJ466205; Nematostella Cav1, XP_001639054.1; Nematostella Cav2a, Cav2b, and Cav2c, NVE4667, NVE18768, and NVE1263, respectively, from published transcriptome (Fredman et al., 2013); Nematostella Cav3a and Cav3b, NVE5017 and NVE7616, respectively, from published transcriptome (Fredman et al., 2013); Caenorhabditis elegans Cav1 (egl-19), NP_001023079.1; C. elegans Cav2 (unc-2), NP_001123176.1; C. elegans Cav3 (cca-1), CCD68017.1; Drosophila Cav1 (α1-D), AAF53504.1; Drosophila Cav2 (cacophony), AFH07350.1; Drosophila Cav3 (Ca-α1T), ABW09342.1; Lymnaea Cav1, AAO83839.1; Lymnaea Cav2, AAO83841.1; Lymnaea Cav3, AAO83843.2; human Cav1.1, NP_000060.2; human Cav1.2, Q13936.4; human Cav1.3, NP_001122312.1; human Cav1.4, NP_005174.2; human Cav2.1, O00555.2; human Cav2.2, NP_000709.1; human Cav2.3, NP_001192222.1; human Cav3.1, NP_061496.2; human Cav3.2, NP_066921.2; human Cav3.3, NP_066919.2; Mnemiopsis Cav2, fragmented transcriptome sequences manually pieced together from published transcriptome (Ryan et al., 2013); Hormiphora Cav2, sequence extracted from a de novo assembly of RNA-Seq data (SRR1992642; Francis et al., 2015); Chlamydomonas CAV2, XP_001701475.1; Schizosaccharomyces pombe CCH1, NP_593894.1; and Saccharomyces cerevisiae CCH1, NP_011733.3. Immunostaining and confocal microscopy Trichoplax were frozen and freeze-substituted as described previously (Smith et al., 2014) with the following modifications. Coverslips (22 mm square, #1.5 thickness; ZEISS) were cleaned in nitric acid and treated with 3-aminopropyltriethoxysilane (#A3648; Sigma-Aldrich) to produce a positively charged surface. Trichoplax were transferred to a 500-µl drop of artificial seawater (ASW) placed in the center of the coverslips and left to adhere for 1–2 h. 300 µl of the ASW was removed and replaced with 500 µl of a 1:1 mixture of ASW and 1 M mannitol. After ∼5 min, the liquid was removed, and the coverslips were plunged into tetrahydrofuran at −80°C on dry ice and kept overnight. The coverslips were transferred to methanol with 1.6% paraformaldehyde on dry ice and then held at −20°C for 2–3 h and room temperature for 2 h. The specimens were rinsed in 100% ethanol (EtOH) and rehydrated gradually with 90%, 70%, and 50% EtOH (diluted with PBS) and PBS each for ∼10 min and blocking buffer (BB: 3% normal goat serum, 2% horse serum, 1% BSA in PBS) for 15 min. Then specimens were incubated in custom (Thermo Fisher Scientific) antiserum against the epitope ESRVNGNAKFTSDDQRLDR corresponding to the middle of the TCav3 I–II cytoplasmic linker (Fig. 1 A) or, to control for specificity, serum from the same rabbit before immunization both diluted 1:400 in BB overnight at 4°C. A custom (New England Peptide) chicken antibody against the epitope EATAPKKDSSKSNFSSR, found in the Trichoplax complexin protein, was added in some experiments to mark neurosecretory cells. After washing in PBS, the coverslips were incubated with Atto 488 goat anti–rabbit IgG (62197; Sigma-Aldrich) with/without Alexa Fluor 647 goat anti–chicken IgY (A-21449; Thermo Fisher Scientific) diluted 1:500 in BB for 2 h at room temperature. Nuclei were stained with Hoechst. Images of immunostaining in Trichoplax were captured on an LSM 880 confocal microscope (ZEISS) with a 63× 1.4-N.A. PlanApo objective and 488-nm illumination for Atto 488 and 405-nm for Hoechst. Overview image stacks (17 images, 0.7-µm interval) were captured with a Quasar spectral detector with emission windows at 415–480 nm (blue) and 490–588 nm (green). Enhanced resolution image stacks (36 images, 0.185-µm interval) were collected with an Airyscan detector and 420–480- and 495–550-nm filters. Image stacks were displayed as maximum-intensity projections. Immunodetection of TCav3 on Western blots Trichoplax whole-animal protein lysates were prepared from ∼30 specimens and lysed directly in 200 µl of reducing sample buffer preheated to 95°C (50 mM dithiothreitol, 1% wt/vol SDS, 7.5% glycerol, 0.003% bromophenol blue, and 40 mM Tris pH 6.8). Protein lysates from ectopically expressed TCav3 channels in HEK-293T cells were prepared as follows. The entire coding sequence of TCav3 was excised from the pTCav3-IR-EGFP vector with restriction enzymes SacII and XmaI and cloned into pEGFP-C1 with the same sites (Takara Bio Inc.). The resulting plasmid pEGFP-TCav3, pTCav3-IR-EGFP, or the empty fusion vector pEGFP-C1 was cotransfected into HEK cells with rat Cavβ1b and Cavα2δ1 subunits (as outlined in the culturing and transfection section of the Materials and methods below), and cells were incubated at 28°C for 4–5 d and then briefly washed with warm PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, and 1.8 mM KH2PO4, pH 7.4). Cells were then lysed with 600 µl of sample buffer (50 mM dithiothreitol, 1% wt/vol SDS, 7.5% glycerol, 0.003% bromophenol blue, and 40 mM Tris, pH 6.8). Equal volumes for each lysate were loaded on either freshly prepared 7.5% polyacrylamide gels or precast 4–20% polyacrylamide gradient gels (Invitrogen), and electrophoresis was performed in Invitrogen MES buffer using an XCell SureLock Mini-Cell Electrophoresis System (Invitrogen). For each experiment, paired gels were run: one was subjected to Coomassie staining to confirm equal protein content among samples, whereas the other was transferred to a nitrocellulose membrane using a Tris-glycine transfer buffer (25 mM Tris, 192 mM glycine, 20% vol/vol methanol, and 0.5% SDS, pH 8.3). Western blots performed using custom anti-TCav3 antibodies (rabbit polyclonal; Thermo Fisher Scientific) were done using (a) unpurified antibodies (terminal bleed serum, 1:1,000 dilution); (b) preimmune serum (1:1,000 dilution); or (c) affinity-purified antibodies (isolated by the manufacturer using a conjugated antigen peptide, 1:500 dilution). Western blots against EGFP epitopes (as an N-terminal fusion with TCav3 or alone) were performed using a rabbit polyclonal anti-GFP antibody (Sigma-Aldrich) at 1:5,000 dilution. Primary antibody incubations were performed overnight at 4°C, and secondary antibody incubations, washing, and detection were performed using standard chemiluminescent methods. Culturing and transfection of HEK-293T cells with TCav3 cDNAs The detailed methods used for culture and CaPO4 transfection of cloned Cav3 channels into HEK-293T cells, as well as techniques for whole-cell patch-clamp electrophysiology, have been previously documented in detail (Senatore et al., 2011, 2014; Senatore and Spafford, 2012). In brief, for electrophysiological experiments of in vitro–expressed TCav3, 6 µg of the pTCav3-IR-EGFP construct was transfected into HEK cells in 6-ml flasks, along with 3 µg of high voltage–activated calcium channel rat Cavβ1b and Cavα2δ1 subunit cDNAs cloned into mammalian expression vector pMT2 (Tomlinson et al., 1993). These Cav1/Cav2 subunits have been shown to not interact with or alter the biophysical properties of heterologously expressed mammalian and invertebrate T-type channels (Dubel et al., 2004; Dawson et al., 2014), but nevertheless boost membrane expression by some unknown mechanism (Dubel et al., 2004). CaPO4 transfections were done overnight at 37°C, after which cells were washed and transferred to 28°C for 2–3 d before recording. On the day of recording, cells were trypsinized (Sigma-Aldrich) and plated onto glass coverslips, which were then transferred into 2-ml culture dishes with appropriate extracellular recoding solutions. For experiments involving the quantification of EGFP fluorescence in transfected HEK cells, transfections were performed in quadruplicate, with 6 µg pTCav3-IR-EGFP, pEGFP-TCav3, or pEGFP-C1 cotransfected with 3-µg combinations of rat Cavβ1b and Cavα2δ1 subunits or the empty mammalian expression vector pCDNA 3.1. After incubation at 28°C for 3 d, the cells were imaged with transmitted and fluorescent light at 100 magnification, using a ZEISS AxioCam MRm Rev3 camera mounted on an AxioObserver A1 inverted microscope. All flasks were imaged with the same exposure settings, ensuring that the brightest cells did not saturate the pixels during acquisition with ZEN Lite software (ZEISS). Integrated density of the acquired fluorescence images was measured using the ImageJ software (Schneider et al., 2012), and values were normalized against the highest value for each replicate set, averaged, and plotted. Patch-clamp electrophysiology We used previously documented electrophysiological recording solutions to characterize the biophysical properties of TCav3 in 2 mM Ca2+ saline (Figs. 7 and 8; Senatore and Spafford, 2012), assess the Ni2+ block (Fig. 9; Senatore and Spafford, 2010), compare Ca2+ versus Ba2+ conductance (Fig. 10, C and D; Senatore and Spafford, 2012), and assess divalent versus monovalent cation selectivity and permeability features (Fig. 10, A, B, E, and F; Figs. 11 and 12; and Fig. S3; Senatore et al., 2014). Whole-cell patch voltage clamp recordings were performed using an Axopatch 200B amplifier and a Digidata 1440A digitizer controlled with pCLAMP 10 software (Molecular Devices). Pipettes were pulled using a Sutter P-97 micropipette puller and heat polished such that pipette resistance in the bath ranged from 2 to 5 MΩ, with access resistance after membrane breakthrough between 4 and 10 MΩ. Series resistance was not compensated for because we only kept data with minimal access resistance and tight capacitive transients upon voltage step. Only recordings in which leak current was <10% of the peak inward current were used, and offline leak subtraction was done using Clampfit software (Molecular Devices). Methods for Boltzmann transformation and curve fitting of electrophysiological data are described in previous publications (Senatore and Spafford, 2010, 2012). Relative permeabilities under bi-ionic conditions for TCav3 (i.e., PCa/PX, where X = Li+, Na+, K+, or Cs+) were calculated using the bi-ionic Nernst equation (Hille, 2001) as described previously (Senatore et al., 2014). Statistical analyses comparing electrophysiology data for TCav3 with data from other in vitro–expressed channels were done using one-way analysis of variance (ANOVA); p-values are presented in Table 2, where we provide citations for data from other studies. Table 2. Comparison of mean biophysical and permeability parameters of in vitro expressed Cav3 channels (±SE; n values in parentheses) Parameter TCav3 LCav3 (+8b-25c) P-value Ref. Cav3.1 P-value Ref. Cav3.2 P-value Ref. Cav3.3 P-value Ref. Activation Peak of IV (mV) −45 −40 1 −35 2 −35 2 −25 2 V1/2 −59.32 ± 0.9 (8) −53.48 ± 0.34 (13) *** 1 −49.3 ± 0.7 (26) *** 2 −48.4 ± 1.2 (10) *** 2 −41.5 ± 1.1 (17) *** 2 Slope (k, mV) 4.50 ± 0.25 (8) 5.46 ± 0.14 (13) ** 1 4.6 ± 0.1 (26) NS 2 5.2 ± 0.4 (10) NS 2 6.2 ± 0.2 (17) *** 2 Inactivation V1/2 −74.15 ± 0.90 (10) −70.89 ± 0.49 (16) ** 1 −74.2 ± 1.1 (8) NS 2 −75.6 ± 0.7 (19) NS 2 −69.8 ± 0.9 (17) ** 2 Slope (k, mV) 2.69 ± 0.11 (8) 2.93 ± 0.08 (16) NS 1 5.5 ± 0.3 (8) *** 2 6.2 ± 0.2 (19) *** 2 6.1 ± 0.1 (17) *** 2 Activation kinetics τ (−50 mV) 14.36 ± 0.31 (8) 3.26 ± 0.13 (16) *** 1 8.2 ± 0.9 (8) *** 3 9.9 ± 0.4 (10) *** 3 43 ± 3 (9) *** 3 τ (−10 mV) 3.48 ± 0.13 (8) 0.74 ± 0.03 (16) *** 1 1.1 ± 0.1 (8) *** 3 1.8 ± 0.1 (10) *** 3 5.9 ± 0.5 (9) *** 3 Fold change 4.1 4.4 7.5 5.5 7.3 Inactivation kinetics τ (−50 mV) 66.33 ± 3.39 (6) 27.51 ± 1.34 (16) *** 1 62 ± 23 (8) NS 3 28 ± 3 (10) *** 3 126 ± 22 (9) * 3 τ (−10 mV) 51.36 ± 1.53 (6) 15.67 ± 0.56 (16) *** 1 16 ± 1 (8) *** 3 15 ± 1 (10) *** 3 80 ± 5 (9) *** 3 Fold change 1.3 1.8 3.9 1.9 1.6 Deactivation −100 mV 1.46 ± 0.06 (12) 1.37 ± 0.05 (19) NS 1 2.6 ± 0.2 (9) *** 2 3.6 ± 0.4 (14) *** 2 1.12 ± 0.1 (31) * 2 −70 mV 9.15 ± 0.66 (12) 2.28 ± 0.18 (19) *** 1 6.2 ± 0.4 (9) ** 2 8.5 ± 1.1 (14) NS 2 2.1 ± 0.1 (30) *** 2 Recovery τ recovery (ms) 1275.05 ± 54.43 (6) 908.04 ± 27.05 (14) *** 1 137 ± 5 (12) *** 2 448 ± 36 (7) *** 2 260 ± 30 (18) *** 2 Nickel (IC50 µM) 335.0 ± 6.5 (9) 300.0 ± 29.2 (4) NS 4 304.8 ± 6.2 (5–6) * 5 4.9 ± 2.0 (5–6) *** 5 216 ± 9 (5-6) *** 5 Mixed Ca2+/Na+ currents % increase in Imax (Ca2+) 42.09 ± 3.19 (8) 12a: 1439.75 ± 82.50 (6) *** 6 26.64 ± 9.24 (4) * 31.96 ± 4.58 (11) NS 45.45 ± 7.59 (6) NS 12b: 153.02 ± 9.61 (7) *** 6 % increase in Imax (Ba2+) 51.19 ± 2.64 (4) 12a: 901.29 ± 120.33 (5) *** 6 27.01 ± 3.17 (9) *** 6 36.88 ± 3.50 (7) * 6 49.91 ± 6.23 (4) NS 6 12b: 121.07 ± 4.26 (3) *** 6 Ca2+ vs. X+ permeability pCa/pLi 19.87 ± 1.30 (8) 12a: 22.74 ± 0.41 (8) * 6 46.67 ± 2.32 (7) *** 6 12b: 27.40 ± 1.13 (7) *** 6 pCa/pNa 35.61 ± 1.52 (8) 12a: 33.06 ± 1.50 (6) NS 6 89.56 ± 8.21 (6) *** 6 80.58 ± 3.69 (4) *** 56.19 ± 2.99 (7) *** 12b: 41.49 ± 1.98 (8) * 6 pCa/pK 63.99 ± 4.19 (9) 12a: 50.25 ± 1.56 (12) ** 6 140.16 ± 12.02 (5) *** 6 12b: 78.93 ± 4.17 (11) * 6 pCa/pCs 114.83 ± 10.64 (7) 12a: 111.30 ± 8.57 (10) NS 6 154.65 ± 7.99 (5) * 6 12b: 113.76 ± 7.19 (11) NS 6 P-values for statistical comparisons for all channels with respect to TCav3 were generated using one-way ANOVA. *, P < 0.05; **, P < 0.005; ***, P < 0.0005; NS, not significantly different. Bold indicates the value most similar to TCav3, and bold italics indicates the value most different from TCav3. References: 1, Senatore and Spafford (2012); 2, Chemin et al. (2002); 3, Gomora et al. (2002); 4, Senatore and Spafford (2010); 5, Kang et al. (2006); 6, Senatore et al. (2014). Online supplemental material Fig. S1 shows the percentage minimum–maximum plot of the TCav3 channel protein coding sequence compared with the three human Cav3 channels (hCav3.1, hCav3.2, and hCav3.3). Fig. S2 is an alignment of the domain II P-loop of T-type channels corresponding to the exon 12 region. Fig. S3 shows superimposed current–voltage plots of various Cav3 channel homologues under different bi-ionic conditions. The Cav channel protein sequences used to generate the phylogenetic tree depicted in Fig. 1 B are provided in supplemental text file 1 (untrimmed, unaligned sequences) and supplemental text file 2 (trimmed, MUSCLE-aligned sequences), both in FASTA format. Results Identification and sequencing of a Cav3 channel homologue from T. adhaerens Various Cav3 channel protein sequences were blasted against the T. adhaerens genome (Srivastava et al., 2008; JGI Genome Portal), identifying a predicted Trichoplax T-type/Cav3 channel homologue (T. adhaerens Grell-BS-1999 v1.0, scaffold_2:6781672-6793175). The predicted coding sequence served as a reference for RT-PCR amplification and sequencing of the Trichoplax Cav3 (TCav3) cDNA, amplified in two large fragments, with primers listed in Table 1. To build a consensus, every nucleotide along the 6,192-bp coding sequence of TCav3 was validated with a minimum of three independently amplified sequences. The resulting full-length open reading frame (submitted to GenBank with accession no. KJ466205) predicts a channel protein of 2,063 aa, a molecular mass of ∼238 kD, and a Kyte–Doolittle hydrophobicity profile with hydrophobic peaks corresponding to transmembrane helices (i.e., segments 1 to 6 or S1–S6) within each of IV repeat domains, conserved for all four-domain channels (Fig. 1 A). An inferred ML phylogeny of various Cav channel proteins places TCav3 basal to the two cnidarian T-type channels from Nematostella vectensis (Cav3a and Cav3b), as well as bilaterian protostome homologues C. elegans cca-1, Drosophila Ca-α1T, and Lymnaea LCav3, and chordate deuterostome homologues (human Cav3 isotypes Cav3.1-Cav3.3; Fig. 1 B). Recent phylogenomic studies have placed Trichoplax and its phylum (Placozoa) as a sister clade to the bilaterians and cnidarians, and sponges (phylum Porifera) and comb jellies (phylum Ctenophora) as the most early diverging animals (Srivastava et al., 2008, 2010; Ryan et al., 2013; Moroz et al., 2014; Pisani et al., 2015). Based on this phylogeny, Trichoplax is the earliest diverging animal to possess all three types of bilaterian/cnidarian Cav channels (i.e., Cav1, Cav2, and Cav3). Instead, the marine sponge/poriferan Amphimedon queenslandica and the two ctenophores Mnemiopsis leidyi and Hormiphora californiensis have only single Cav channel genes, forming either a clade with Cav2 types (i.e., ctenophores) or a sister clade with Cav1 and Cav2 types (i.e., Amphimedon, hence dubbed Cav1/Cav2-like; Moran and Zakon, 2014; Senatore et al., 2016). Interestingly, recent genome sequencing of choanoflagellate Salpingoeca rosetta revealed the presence of a T-type channel, indicating that Cav3 types likely predate Metazoa (Fig. 1 B; Fairclough et al., 2013; Moran and Zakon, 2014), and hence were lost in Porifera and Ctenophora. Thus, TCav3 is the most divergent homologue of vertebrate/human T-type channels identified to date in animals. Figure 1. Hydropathy profile of TCav3 and protein phylogeny of various Cav channels. (A) The in silico–translated protein sequence of TCav3 produces hydrophobic peaks on a Kyte–Doolittle hydrophobicity plot corresponding to six transmembrane helices (S1–S6) present in each of four homologous domains (I–IV), conforming to the characteristic structure of four-domain channels. The vertical green bar illustrates the location of the peptide epitope used to generate polyclonal anti-TCav3 antibodies. (B) Protein phylogeny of various Cav channels, inferred by ML using the LG+G model. Node support values resulting from 1,000 bootstrap replicates are indicated, and the scale bar indicates the number of amino acid substitutions per site. Of select Cav3 channels with validated mRNA sequences (i.e., TCav3, C. elegans Cav3 channel cca-1, human Cav3.1 to Cav3.3 isotypes, L. stagnalis Cav3, and Drosophila melanogaster Ca-α1T), the Trichoplax channel protein is among the smallest, with shorter cytoplasmic N- and C-terminal regions, as well as linkers between domains I–IV (Fig. 2 A). Instead, transmembrane regions (S1–S6 helices and corresponding linkers) are much more similar in length (Fig. 2 B), and indeed carry most of the protein sequence homology between different channels, whereas the cytoplasmic linkers and N/C termini exhibit much more divergence (Senatore and Spafford, 2010). The distant TCav3 bears what are perhaps the most distinguishing features of T-type channels: (a) a “selectivity filter” motif of EEDD, made up of negatively charged glutamate (E) and aspartate (D) residues that project into the pore to govern ion selectivity (Talavera and Nilius, 2006), distinct from the more calcium-selective EEEE selectivity filters of Cav1 and Cav2 channels (Figs. 1 A and 2 C); and (b) a predicted helix-loop-helix motif in the cytoplasmic linker between domains I and II, dubbed the “gating brake,” which serves to prevent channel opening at hyperpolarized membrane voltages (Perez-Reyes, 2010a) and where mutations in human Cav3.2 are associated with childhood absence epilepsy (Figs. 1 A and 2 D; Arias-Olguín et al., 2008). Figure 2. Structural features of the TCav3 protein sequence compared to Cav1, Cav2, and Cav3 channels from other species. (A) TCav3 has short N and C termini, as well as cytoplasmic linkers joining the four domains similar to the channel from C. elegans and much smaller, with the largest known T-type channels from Lymnaea and Drosophila. (B) Length of transmembrane regions and loops between segments 1 and 6 of each domain (D1–D4) are highly conserved for T-type channels, with domain I being the longest in amino acid length and domain II the shortest. Depicted values for the three human Cav3 channels (hCav3.1–hCav3.3) represent mean ± SE. (C) Alignment of Cav channel selectivity filter and flanking amino acids, showing the conserved EEDD motif common to all Cav3 channels and distinct from the EEEE selectivity filters of high voltage–activated channels Cav1.2 and Cav2.2. (D) Prediction of secondary protein structure with PSIPRED (McGuffin et al., 2000) reveals a conserved helix-loop-helix gating brake motif in all T-type channels (green and yellow underlines depict predicted helices and regions where the prediction dropped below threshold, respectively) in the cytoplasmic I–II linker just proximal to the S6 pore-forming helix of domain I. Reduced genomic complexity and absence of alternative splicing of the TCav3 gene Most of the ∼11,500 genes in the Trichoplax genome bear genomic architectures similar to orthologous genes in other animals (Srivastava et al., 2008). Accordingly, of the 28 exons/27 introns that make up the TCav3 channel gene, 26 splice junctions have counterparts in the Cav3.1 channel gene from mouse (MusCav3.1), whose mRNA transcript sequence is encoded by 38 exons/37 introns (Fig. 3). The Trichoplax Cav3 channel gene is ∼10-fold shorter than MusCav3.1, attributable to much smaller intron sizes here and across the entire Trichoplax genome (Srivastava et al., 2008). To summarize major structural differences between the Trichoplax and mouse Cav3 genes (Fig. 3): (a) TCav3 lacks an intron separating exons 2 and 3, in the coding region for the domain I S1-S2 linker (Fig. 1 A), but retains N-terminal introns separating exons 1 and 2, as well as exons 3–4, 4–5, and 5–6, which are conserved in all metazoan four-domain channels including Cav and Nav channels (Spafford et al., 1999); (b) TCav3 also lacks an intron separating exons 11 and 12 encoding the domain II extracellular P-loop between S5 and S6, which is found in all Cav3 channels stemming from basal bilaterians (Senatore et al., 2014); (c) TCav3 lacks alternative donor splice sites at the 3′ ends of exons 8 and 25, which in snail LCav3 and MusCav3.1 create optional exons 8b and 25c that regulate channel surface expression and gating, respectively (Chemin et al., 2001; Emerick et al., 2006; Shcheglovitov et al., 2008; Senatore and Spafford, 2012); (d) TCav3 lacks exon 26 found in all mammalian Cav3.1 and Cav3.2 channel genes, as well as C. elegans cca-1, which produces similar but more slight gating effects compared with exon 25c (Chemin et al., 2001; Ohkubo et al., 2005; Steger et al., 2005; Zhong et al., 2006; Senatore and Spafford, 2012); and (e) TCav3 is missing four introns between exons 33 and 38 of MusCav3.1, corresponding to the channel C-terminal region. An overall reduction in genomic complexity for TCav3, especially in regions associated with alternative splicing, is consistent with reports that genes from basal metazoans generally undergo less alternative splicing (Pan et al., 2008; Wang et al., 2008; Gerstein et al., 2010; Graveley et al., 2011; Ramani et al., 2011). This was certainly evident during the sequencing and cloning of the TCav3 cDNA, in which we failed to identify a single alternatively spliced isoform, as well in an ongoing transcriptome analysis of Trichoplax whole-animal mRNAs (Senatore et al., 2016). Figure 3. Analysis of Cav3 channel genes from S. rosetta (SalpCav3), T. adhaerens (TCav3), N. vectensis (TCav3), and Mus musculus (mouse; MusCav3.1). Exons bearing start to stop codons are depicted by boxes, and introns are depicted by thin lines, with corresponding scale bars on the right indicating length in base pairs. Homologous protein sequences corresponding to mouse exons 1–38 are depicted by alternating blue and red background fills, with expanded mouse exons 14–17 colored in green and C-terminal exons 33–38 colored in gray. The TCav3 channel gene has a similar genomic structure as MusCav3.1; however, it is shorter in length (i.e., scale bars of 1,000 bp for TCav3 vs. 10,000 bp), and lacks nine introns (red arrowheads), resulting in 10 fewer exons (fused TCav3 exons are shown in red). TCav3 also lacks alternative splicing identified in channels from mammals and protostome invertebrates, including optional exons 8b, 25c, and 26, and mutually exclusive exons 12a and 12b (white arrow) found only in protostomes (red portions of the MusCav3.1 exons). Asterisks depict introns that are conserved among all metazoan four-domain channels. Adding to the analysis, Cav3 genes from premetazoan choanoflagellate S. rosetta and cnidarian N. vectensis reveal unique patterns in intron gain/loss across these different organismal lineages (Fig. 3). In keeping with the MusCav3.1 intro/exon numbering scheme, SalpCav3 exon 8 is fused with flanking exons 7 and 9, which are conserved as separate exons in mouse and Trichoplax, and bears an additional three internal exons. Notably, this region encodes the channel I–II linker protein sequence (Fig. 1 A), which tends to be highly divergent between different Cav3 channels. In this equivalent position, NemCav3a has an additional exon compared with mouse and Trichoplax, which, interestingly, overlaps with mouse optional exon 8b, which alters channel membrane expression. SalpCav3 notably lacks numerous introns between exons 12 and 33, a region encoding the C-terminal half of the channel protein (i.e., domains III and IV). Also evident is that MusCav3.1 exons 14–17 appear to have arisen via expansion from a single exon conserved in TCav3 and NemCav3a, in a region that corresponds the III–IV linker, which interestingly serves as a hotbed for modulation of mammalian T-type channels by kinases and G-proteins (Chemin et al., 2006; Perez-Reyes, 2010b; Senatore et al., 2012). Finally, NemCav3a and MusCav3.1 bear more exons/introns in the 3′ end of the gene, perhaps indicative of intron expansion in the cytoplasmic C terminus, a region with poor sequence homology where it is difficult to infer conserved intron/exon structure. TCav3 is expressed in neurosecretory-like gland cells Immunostaining Trichoplax with both crude and affinity-purified custom antibodies against a I–II linker epitope of TCav3 (site depicted in Fig. 1 A) outlined cells around the rim of the animal and, less intensely marked, scattered cells further in the interior (Fig. 4 A). Enhanced imaging with the Airyscan detector (1.7-fold improvement in resolution and improved signal-to-noise relative to conventional confocal [Huff, 2015]) revealed that TCav3 staining was near the surfaces of hourglass-shaped cells and concentrated at the side facing the exterior of the animal (Fig. 4 A, inset). Staining was not evident when using preimmune serum (Fig. 4 B). The cells that labeled for TCav3 also were stained by an antibody against Trichoplax complexin (Fig. 4 C), a regulator of SNARE secretory proteins in neurons. The distribution of the cells as well as their shapes closely matched those of gland cells labeled with antibodies against the SNARE proteins syntaxin-1, synaptobrevin, and SNAP-25 (Smith et al., 2014). No signal was apparent in specimens incubated with preimmune serum and imaged with the same parameters. Figure 4. Immunostaining for TCav3 in Trichoplax prepared by freezing and freeze substitution. (A) Large field shows a projection of 17 optical sections extending through the dorsoventral thickness of the animal but omitting the dorsal and ventral surfaces where there is nonspecific staining. Labeled cells are concentrated around the rim but also present in the interior, although too dim to see in this unenhanced image. (Inset) Projection of 36 optical sections through cells reaching the ventral surface at the exterior of the animal. Staining is concentrated near the surfaces of hourglass-shaped cells and is most intense on the side facing the exterior. (B) Comparison of staining for TCav3 with matched preimmune control showing lack of staining for TCav3. (A and B) Nuclei are blue. (C) Cells that immunostain for TCav3 also label for complexin, a marker for neurosecretory cells. Merged view of double stain (right) shows TCav3 in green and complexin in red. The left panel shows TCav3 (green), and the middle panel shows complexin (red). A (inset) and C were collected with an enhanced detector (Airyscan). Bars: (A–C) 10 µm; (A, inset) 5 µm. The scale bar for B is shared with C. We also confirmed expression of TCav3 at the mRNA level by RT-PCR, where gene-specific primers amplified an appropriate ∼500-bp fragment from cDNA reverse transcribed from whole-animal total RNA using an oligo-dT18 primer (Table 1 and Fig. 5). Similarly, primers targeting the other two Trichoplax Cav channels, Cav1 and Cav2, as well as the Cav1/Cav2 accessory subunit Cavβ and three Cavα2δ subunits (Cavα2δ-a, Cavα2δ-b, and Cavα2δ-c; Table 1), all produced expected bands of ∼500 bp, with no bands evident when the RTase enzyme was omitted from the reaction (Fig. 5). We also note that mRNA expression of these genes, as well as other genes homologous to those involved in cellular excitability (e.g., Nav2 and Kv channels, K+ leak channels), are expressed in the whole-animal transcriptome of Trichoplax (Senatore et al., 2016). Figure 5. RT-PCR amplification of various Trichoplax Cav channel subunits. RT-PCR amplification of cDNA corresponding to the three Trichoplax Cav channels (TCav1, TCav2, and TCav3), the single Cavβ subunit, and the three Cavα2δ subunits (Cavα2δ-a, Cavα2δ-b, and Cavα2δ-c) from a whole-animal total RNA cDNA library prepared with an anchored oligo-dT18 primer. Ectopic expression of TCav3 in HEK-293T cells The TCav3 open reading frame (NCBI accession no. KJ466205) was cloned into bicistronic expression vector pIRES2-EGFP, which enables identification of positively transfected mammalian cells separately expressing TCav3 plus the fluorescent marker EGFP. Transfection of this construct (pTCav3-IR-EGFP) into HEK-293T cells produced barely detectable EGFP fluorescence (Fig. 6 A), suggesting that the TCav3 insert was somehow inhibiting expression of bicistronic EGFP because the empty pIRES2-EGFP vector normally produces robust EGFP fluorescence in these cells. Attempts to record voltage-activated Ca2+ currents from HEK cells transfected with pTCav3-IR-EGFP via whole-cell patch-clamp were unsuccessful, suggesting that the channel protein was also not expressed at high enough levels to accumulate at the cell membrane. A possible reason for the apparent poor expression of TCav3 is the prevalence of tandem rare codons in its cDNA sequence with respect to humans, which is expected to decrease efficiency of ectopic protein translation (Gustafsson et al., 2004; Kobayashi, 2015; Presnyak et al., 2015). Specifically, TCav3 exhibits a low codon adaptation index of 0.61 with respect to human preferred codons (Sharp and Li, 1987), and percentage minimum–maximum analysis reveals rare codon clustering along the entire length of the channel coding sequence (Fig. S1; Clarke and Clark, 2008). Figure 6. In vitro expression of TCav3 in HEK-293T cells. (A) Bicistronic EGFP fluorescence images (top) and merged EGFP/transmitted light images (bottom) of HEK-293T cells cotransfected with pTCav3-IR-EGFP (vector map on left) and pCDNA-3.1 (+pC) and/or pMT2 constructs bearing the coding sequences of rat Cavβ1b and Cavα2δ1 accessory subunits. (B) Similar to A, but cells were transfected with the pEGFP-TCav3 construct, encoding a fusion protein of TCav3 tagged with N-terminal EGFP. (C) Similar to A, but cells were transfected with the empty EGFP fusion vector pEGFP-C1. Bar, 200 µm. (D) Bar graph depicting normalized mean fluorescence of imaged HEK cells from quadruplicate transfections (±SE). The purple asterisks denote statistically significant means (***, P < 0.005) comparing the +pC condition with all others via one-way ANOVA. (E) Western blot of HEK cell protein lysates with anti-EGFP antibody, prepared by cotransfection of the pEGFP-TCav3 vector with indicated combinations of pCDNA-3.1, rat Cavβ1b, and/or Cavα2δ1. (F) Western blot of HEK cell protein lysates with anti-EGFP antibody, prepared by cotransfection of the pEGFP-C1 vector (top) or pEGFP-TCav3 (bottom) with the indicated combinations of pCDNA-3.1, rat Cavβ1b, and/or Cavα2δ1 accessory subunits. Note that the three smaller bands near 33 kD visible in E converge into one band in the lower panel because of an increase in the percentage of polyacrylamide from 7.5% to ∼20% in the gels used for Western blotting. (G) Bar graph depicting the mean increase (±SE) in integrated density of HEK cells transfected with pEGFP-TCav3 and pCDNA-3.1 upon the addition of proteasome inhibitor MG-132 to the cultured cells 12 h before imaging. The purple asterisks denote statistically significant means with ***, P < 0.005. Interestingly, previous research on cloned mammalian T-type channels revealed that coexpression with Cavβ and Cavα2δ accessory subunits of high voltage–activated Cav1 and Cav2 channels increases Cav3 channel expression by an indirect mechanism (Dubel et al., 2004). Thus, we sought to increase TCav3 channel expression in HEK cells by cotransfecting with rat Cavβ1b and Cavα2δ1 subunit cDNAs cloned into the mammalian expression vector pMT2 (Tomlinson et al., 1993). Strikingly, cotransfection of pTCav3-IR-EGFP with rat Cavβ1b and Cavα2δ1 subunit vectors produced an ∼217-fold increase in bicistronic EGFP fluorescence compared with cotransfection with empty mammalian expression vector pCDNA-3.1 (Fig. 6 A), quantified as relative integrated density with ImageJ software (Fig. 6 D). Cotransfection with just Cavβ1b or Cavα2δ1 separately also increased fluorescence, but more moderately, with respective increases of ∼165-fold and 148-fold (Fig. 6 D). In addition, cotransfection of pTCav3-IR-EGFP with the rat Cavβ1b and Cavα2δ1 subunits allowed us to electrophysiologically record robust TCav3 Ca2+ currents in HEK cells. Because bicistronic expression and fluorescence of EGFP from pTCav3-IR-EGFP only indirectly implies TCav3 channel protein expression, we repeated the cotransfection experiment using a construct in which the channel coding sequence was cloned in frame with that of EGFP at the channel’s N terminus, in the EGFP fusion vector pEGFP-C1. Cotransfection of the resulting pEGFP-TCav3 construct with rat Cavβ1b and Cavα2δ1 subunit vectors increased EGFP fluorescence ∼75-fold, whereas Cavβ1b and Cavα2δ1 alone increased fluorescence ∼47-fold and ∼38-fold, respectively (Fig. 6, B and D). Interestingly, the Cavβ1b and Cavα2δ1 vectors also increased fluorescence of EGFP when transfected without TCav3, from the empty vector pEGFP-C1 (Fig. 6, C and D), and this corresponded with increased EGFP protein levels apparent on Western blots of corresponding HEK cell lysates probed with anti-EGFP antibodies (Fig. 6 F). We were unable to detect the endogenous TCav3 protein in Western blots of Trichoplax whole-animal lysates or blots of protein lysates from TCav3-transfected HEK cells when using the custom anti-TCav3 antibodies. Instead, blots of HEK cells transfected with pEGFP-TCav3 plus the Cavβ1b and Cavα2δ1 vectors produced appropriate bands of ∼270 kD when probed with anti-EGFP (Fig. 6 E), consistent with the predicted molecular weight of the TCav3 channel protein (238 kD) plus EGFP (32.7 kD). In accordance with our inability to record TCav3 Ca2+ currents when pTCav3-IR-EGFP was transfected without the Cavβ1b and Cavα2δ1 subunit vectors, we were unable to detect the EGFP-TCav3 fusion protein without cotransfection of Cavβ1b and Cavα2δ1 (Fig. 6 E, left lane). Thus, the rat Cavβ1b and Cavα2δ1 subunit vectors appear to dramatically boost protein expression of TCav3, as either an EGFP fusion protein or a separate protein from the pTCav3-IR-EGFP construct. However, the effect of Cavβ1b and Cavα2δ1 subunit vectors on ectopic protein expression appears to be at least in part nonspecific because they also boost expression of coexpressed EGFP in the absence of TCav3 (Fig. 6, D and F). Finally, in lanes on Western blots in which the subunits were included and EGFP-TCav3 protein expression was evident, several additional bands could be observed with molecular weights of ∼100 kD and a triplet of intense bands near 33 kD (Fig. 6, E and F), suggesting that the channel either is being degraded in HEK cells or is incompletely translated. Application of proteasome inhibitor MG-132 to cells transfected with pEGFP-TCav3 plus pCDNA-3.1 for 12 h before fluorescence imaging caused only a 7.2-fold increase in EGFP integrated density (Fig. 6 G), still below the detection limit of Western blotting (not depicted), suggesting that proteasomal degradation only partly accounts for the poor expression of TCav3. Notable is that coexpression of EGFP-TCav3 with the Cavβ1b and Cavα2δ1 subunits does not appear to increase channel protein expression by decreasing the amount of degradation, but rather, by increasing the total amount of protein, including the complete protein, plus all of the incomplete intermediates (Fig. 6, E and F). TCav3 conducts low voltage–activated calcium currents in vitro, characteristic of T-type channels Whole-cell voltage clamp recording of HEK cells cotransfected with pTCav3-IR-EGFP, rat Cavβ1b, and rat Cavα2δ1 produced low voltage–activated calcium currents in 2 mM external Ca2+ in response to depolarizing voltage steps from −110 mV to between −90 and 40 mV (Fig. 7 A). A plot of peak currents versus step potential (i.e., current–voltage or IV plot) reveals a slightly hyperpolarized maximal peak inward current of −45 mV (Fig. 7 B), which is between 5 and 20 mV more negative than that of other Cav3 channels (Table 2). Boltzmann transformation of the IV plot, which removes the influence of driving force to estimate the voltage dependence for channel activation, indicates that TCav3 activation begins at very low depolarizing potentials compared with published data for other Cav3 channels derived using similar methods (i.e., compare half-maximal activation [V1/2] of −59.32 ± 0.9 mV for TCav3 vs. −53.48 ± 0.34 for Lymnaea Cav3 [Senatore and Spafford, 2012]; −49.3 ± 0.7 for human Cav3.1 [Chemin et al., 2002]; −48.4 ± 1.2 for Cav3.2 [Chemin et al., 2002]; and −41.5 ± 1.1 for human Cav3.3 [Chemin et al., 2002]), reaching maximal activation near −40 mV (Fig. 7 C and Table 2). Conversely, steady-state inactivation of TCav3, determined by measuring peak residual current after exposure to prolonged steady-state voltages (Fig. 7 C, inset), is quite similar for TCav3 with respect to other channels, especially Cav3.1 and Cav3.2 (i.e., V1/2 for inactivation is −74.15 ± 0.90 for TCav3 and −74.2 ± 1.1 for human Cav3.1 [Chemin et al., 2002] and −75.6 ± 0.7 for human Cav3.2 [Chemin et al., 2002]; Table 2). Altogether, the voltage properties of TCav3 indicate that it is likely more active at threshold voltages compared with other Cav3 channels, where the channel is subject to roughly the same amount of inactivation, but is more readily activated by depolarization. Figure 7. Activation and inactivation properties of TCav3 Ca2+ currents. (A) Sample traces of recorded Ca2+ currents conducted by the TCav3 channel ectopically expressed in HEK-293T cells, in response to depolarizing voltage steps from −110 mV to between −90 and 40 mV. The scale base below the current traces reflects time in milliseconds and current amplitude in picoamperes. (B) Plot of mean peak current for different depolarizing voltages, relative to the maximal peak inward current (i.e., an IV plot), reveals a maximal peak inward current at −45 mV, indicating a low voltage of activation for TCav3. (C) Plot of mean voltage dependence for channel activation (filled black squares, conductance G/Gmax) and steady-state inactivation (white circles, I/Imax) reveals the presence of a window current for TCav3 (red fill). The inset shows the voltage protocol used to measure steady-state inactivation and sample elicited currents. (D) Mean τ values for mono-exponential curves fitted over the activation and inactivation phases of TCav3 macroscopic current waveforms reveals acceleration of channel kinetics upon stronger depolarization, measured as τact and τinact, respectively. (E) TCav3 is at the lower end of the scale with respect to acceleration of kinetics from −50 to −10 mV, with less pronounced fold decreases in τact and τinact compared with other in vitro–expressed channels, indicative of attenuated voltage dependency for activation and inactivation kinetics. Error bars indicate SE. An important and characteristic feature of T-type channels are their “window” currents, which occur at steady-state voltages near rest through a pool of constitutively open channels (Dreyfus et al., 2010), providing a constant influx of Ca2+ that depolarizes the cell membrane to alter cellular excitability (Cain and Snutch, 2010) and increases cytosolic Ca2+ to regulate cell growth and proliferation (Lory et al., 2006; Taylor et al., 2008; Senatore et al., 2012; Gackière et al., 2013). Such a window current is evident for TCav3, for which an overlap between channel activation and inactivation reveals a voltage range between −60 and −75 mV, where not all channels are inactivated and some degree of activation takes place (Fig. 7 C, red fill). Kinetic properties of TCav3 macroscopic currents are also characteristic of T-type channels In vitro TCav3 currents exhibit slow onset (activation) and attenuation (inactivation) at slight depolarizing voltage steps, which accelerate with stronger depolarization “tightening” current waveforms, a hallmark of Cav3 channels that produces a crossing over of inactivation curves toward peak inward current (Fig. 7 A). Such changes in current waveforms can be quantified with time constants (τ) for mono-exponential curve fits over the rise (τactivation) and decay (τinactivation) phases of channel currents, which for TCav3 results in a decrease in τ consistent with accelerating kinetics upon stronger depolarization (Fig. 7 D). The rate at which TCav3 activation accelerates through depolarization is lower compared with other in vitro–expressed channels, with only a 4.13-fold drop in τactivation from −50 to −10 mV versus 4.41-fold for Lymnaea Cav3 (Senatore and Spafford, 2010), 5.50-fold for human Cav3.2 (Gomora et al., 2002), 7.29-fold for human Cav3.3 (Gomora et al., 2002), and 7.45-fold for human Cav3.1 (Fig. 7 E and Table 2; Gomora et al., 2002). Likewise, acceleration of inactivation kinetics is slower, with only a 1.29-fold decrease in τinactivation for TCav3 versus 1.76-fold for LCav3, 1.87-fold for hCav3.2, 1.58-fold for hCav3.3, and 3.88-fold for hCav3.1 (Fig. 7 E and Table 2). From this data, it is clear that the two invertebrate channels compared here, TCav3 and the T-type channel from mollusk L. stagnalis (LCav3), have kinetics with an overall reduced voltage dependency, especially for current activation. TCav3 current kinetics are marginally slower across all negative voltages than other in vitro–expressed channels, with the exception of the slow Cav3.3 channel (Table 2). Similar to Lymnaea Cav3, and in contrast to mammalian Cav3 channels, TCav3 recovers slowly from inactivation (Fig. 8, A and B; and Table 2), indicating that prolonged hyperpolarization would be required to effectively recruit the channel from depolarized membrane voltages. TCav3 also has slow deactivation kinetics relative to mammalian channels at voltages near −70 mV (Fig. 8, C and D; and Table 2), which during action potential repolarization would result in increased inward Ca2+ influx. Overall, despite the noted differences, the voltage dependencies and kinetics of TCav3 currents are remarkably similar to those of Cav3 channels from animals that have neurons and muscle (Table 2), in particular where a low voltage of activation and rapid activation and inactivation kinetics (which allow T-type channels to contribute depolarizing currents near action potential threshold) are conserved. Figure 8. Recovery from inactivation and deactivation properties of TCav3 Ca2+ currents. (A) Illustration of the voltage protocol (top) used to measure recovery from inactivation for TCav3. A 1-s inactivating voltage pulse from −110 to −35 mV was followed by increasing interpulse interval durations of hyperpolarization at −110 mV to remove inactivation, followed by a short test pulse to −35 mV to record recovered peak inward Ca2+ current (sample traces, bottom). (B) Mean recovery from inactivation for TCav3 Ca2+ currents in HEK-293T cells, measured relative to the maximal peak inward current (I/Imax). A mono-exponential curve fitted over the data produces a τ time constant for recovered current (τrecov) of 1,275 ± 54 ms. (C) Sample traces of TCav3 deactivation currents resulting from rapid membrane repolarization to different deactivating potentials while channels are fully open. The inset illustrates the voltage protocol used for deactivation. (D) τ values (τdeact) for mono-exponential curve fitted over decaying inward currents at different deactivating voltages. Scale bars indicate time (milliseconds) and current amplitude (picoamperes). Plotted values in B and D represent mean ± SE. Ni2+ block of TCav3 Ca2+ currents Early electrophysiological experiments revealed that the divalent cation Ni2+ could potently block low voltage–activated Ca2+ currents in some vertebrate preparations (Perez-Reyes, 2003). This high-affinity block was subsequently attributed only to the Cav3.2 channel isotype, and specifically to a unique histidine residue in its domain I S3–S4 extracellular loop (His-191) that strongly binds Ni2+ to disrupt channel gating (Fig. 9 A; Kang et al., 2006, 2010). The recently cloned T-type channel from Drosophila also bears a histidine in this loop (albeit 2 aa positions upstream of Cav3.2 His-191) and, not surprisingly, is also highly sensitive to Ni2+ (Jeong et al., 2015). Instead, TCav3, mammalian Cav3.1 and Cav3.3 channels, and Lymnaea LCav3, all lack histidines in this region (Fig. 9 A). Accordingly, all of these channels have low and remarkably similar IC50 values for Ni2+ block: 335.0 ± 6.5 µM for TCav3 (Fig. 9, B–D); 300.0 ± 29.2 for LCav3 (Senatore and Spafford, 2010); 304.8 ± 6.2 for human Cav3.1 (Kang et al., 2006); and 216 ± 9 for human Cav3.3 (Table 2; Kang et al., 2006). The similarity in IC50 values for all of these channels suggests that low-affinity Ni2+ block occurs through a common mechanism, where perhaps, as has been suggested for Cav3.1, extracellular Ni2+ ions bind two distinct regions of the pore in a cooperative manner (Obejero-Paz et al., 2008). Such a model is perhaps applicable to TCav3, where the Hill coefficient for Ni2+ block is greater than 1 (i.e., 1.18 ± 0.03; Fig. 9 D), suggesting some degree of cooperative binding. Interestingly, washout of Ni2+ is particularly fast for TCav3 compared with other T-type channels (Kang et al., 2006; Senatore and Spafford, 2010), with a transient increase in peak current amplitude (I/Imax) after perfusion of extracellular Ni2+ is replaced with Ni2+-free saline (Fig. 9 C). Figure 9. Ni2+ block of TCav3 Ca2+ currents. (A) Alignment of the domain I S3–S4 extracellular loop of various T-type channels, showing the presence of a key histidine residue in vertebrate (human) Cav3.2 associated with high sensitivity to Ni2+ block (His-191). Drosophila Ca-α1T is also highly Ni2+ sensitive and contains a histidine residue in this same loop 2 aa upstream of Cav3.2 His-191. (B) Sample peak Ca2+ currents through TCav3 resulting from increased concentrations of externally perfused Ni2+. Applied Ni2+ concentrations, micromolar, are depicted on the left of the current waveforms. (C) Plot of sequential peak current responses of TCav3 as depicted in B resulting from the application of increasing concentrations of Ni2+ (mean ± SE; n = 9). Note the rapid washout of Ni2+ current inhibition and a transient augmented peak current the first few steps after washout. (D) Cumulative dose–response curve of TCav3 Ni2+ block, revealing an IC50 value of 335.0 ± 6.5 µM and a Hill coefficient of 1.18 ± 0.03. Ca2+ versus Na+ permeation properties of TCav3 Recently, protostome invertebrates were found to uniquely possess alternative exons 12a and 12b, encoding alternate turret and descending helices of the domain II pore-loop (P-loop). In the freshwater mollusk L. stagnalis, these exons were found to produce channels with extremely bifurcated permeability features: whereas channels with exon 12b (i.e., LCav3-12b) conduct moderately mixed Ca2+-Na+ currents under physiological conditions, LCav3-12a is extremely Na+ permeant, such that less than 10% of inward current is carried by Ca2+ (Senatore et al., 2014). Alignment of the domain II turrets of Cav3 channels from basal metazoans Trichoplax (TCav3) and N. vectensis (Cav3a) with channels from protostomes (i.e., L. stagnalis, Drosophila, and C. elegans), deuterostomes (Ciona intestinalis, human), and a premetazoan species (choanoflagellate S. rosetta) reveals that the TCav3 domain II P-loop resembles those of exon 12a–bearing channels, being smaller and containing fewer cysteines than exon 12b (Fig. S2). We sought to assess whether the exon 12a–like turret of TCav3 is associated with moderate Na+ permeability, similar to cnidarian (Lin and Spencer, 2001), mammalian (Shcheglovitov et al., 2007), and basal deuterostome (Hagiwara et al., 1975) channels, or instead associates with extreme Na+ permeability, similar to exon 12a variants of Lymnaea Cav3. For this, we assessed the degree of mixing of Ca2+ and Na+ currents through TCav3 ectopically expressed in HEK-293T cells by replacing a perfused extracellular recording solution containing 2 mM Ca2+ and 135 mM impermeant NMDG+, with one containing Na+ instead of NMDG+ (Fig. 10, A and B). Addition of Na+ resulted in a 42% increase in peak inward current elicited by stepping from −110 to −45 mV, which approximates the degree of Na+ expected to move through the channel alongside Ca2+ upon membrane depolarization (Shcheglovitov et al., 2007; Senatore et al., 2014). Comparing the current increase of TCav3 with previously published data of other cloned channels (Senatore et al., 2014; Stephens et al., 2015) reveals that the Trichoplax T-type is most similar to the least Ca2+-selective mammalian isotype Cav3.3 (i.e., 45%), whereas the exon 12b variant of LCav3 exhibits a larger increase of 153% in current amplitude and human Cav3.1 an increase of only 27%. In stark contrast, LCav3-12a undergoes a striking 1,440% increase in current amplitude upon perfusion of external Na+ (Fig. 10 B), reflecting its dramatic Na+ permeability. Thus, in the presence of Ca2+, TCav3 appears to conduct only moderately mixed Ca2+-Na+ currents, in a range similar to that of other in vitro–expressed T-type channels, with the exception of the highly Na+-permeant LCav3-12a. Figure 10. TCav3 conducts moderately mixed Na+/divalent cation currents in the presence of external Ca2+ or Ba2+. (A) Replacing 135 mM impermeant external cation NMDG+ with Na+, in the presence of invariant 2 mM Ca2+ produces a 42% increase in peak inward current through TCav3 elicited by a voltage step from −110 to −45 mV (black and red current traces, respectively), indicative of a moderate mixing of inward Ca2+ and Na+ ions in macroscopic currents. (B) Mean percent increase in peak inward current at −45 mV (±SE), upon replacement of 135 mM external NMDG+ with equimolar Na+ in the presence of 2 mM external Ca2+, for TCav3 compared with previously published data for other cloned Cav3 channels (Senatore et al., 2014; Stephens et al., 2015). (C) Macroscopic currents for TCav3 are larger in the presence of 2 mM external Ca2+ versus Ba2 (inset), where mono-exponential curves fitted over activation and inactivation phases of the current waveforms (τinact and τinact, respectively) reveal statistically indistinguishable kinetics under one-way ANOVA. Plotted values represent mean ± SE. (D) Corresponding IV plots for mean peak Ca2+ versus Ba2+ currents (±SE) reveal a 2.29-fold increase in maximal peak inward current when Ca2+ is present in the extracellular solution instead of Ba2+. (E) Replacing 100 mM impermeant external cation NMDG+ with Na+, in the presence of invariant 2 mM external Ba2+, produces a 51% increase in peak inward current through TCav3 elicited by a voltage ramp from −110 to 100 mV (black and red current traces, respectively), indicative of a moderate mixing of Ba2+ and Na+ ions in macroscopic currents. (F) Mean percent increase in peak inward current using the voltage ramp protocol (±SE), upon replacement of 135 mM external NMDG+ with equimolar Na+ in the presence of 2 mM external Ba2+, for TCav3 compared with previously published data for other cloned Cav3 channels (Stephens et al., 2015). Asterisks depict statistical significance (p-values) for comparisons of means for increased inward current with Na+ relative to TCav3 (generated by one-way ANOVA; *, P ≤ 0.05; ***, P ≤ 0.0005). Ca2+ versus Ba2+ permeability does not predict the amplitude of mixed divalent-monovalent cation currents through T-type channels T-type channels are highly permeable to divalent cation Ba2+, where for reasons that are unclear, the three rat Cav3 channel isotypes have dissimilarities with respect to macroscopic Ca2+ versus Ba2+currents in vitro (Talavera and Nilius, 2006): Cav3.1 conducts larger Ca2+ currents, Cav3.2 conducts larger Ba2+ currents, and Cav3.3 conducts equal Ca2+ and Ba2+ currents (McRory et al., 2001). Here, we found TCav3 to be similar to rat Cav3.1, where replacing 2 mM external Ba2+ with equimolar Ca2+ by perfusion produced an increase in current amplitude upon membrane depolarization from −110 mV (Fig. 10 C), with a 2.29-fold increase in maximal peak inward current visible on IV plots (Fig. 10 D). It appears as though the difference in macroscopic Ca2+ versus Ba2+ current amplitude for TCav3 is not attributable to differences in activation/inactivation kinetics in the presence of either ion, as was shown for rat Cav3.1 (Khan et al., 2008), where mono-exponential curves fitted over activation and inactivation phases of macroscopic currents for Ba2+ versus Ca2+ produce statistically indistinguishable τ time constants across all depolarizing voltages (Fig. 10 C). Interestingly, the Lymnaea T-type channel, which can alter its Ca2+ versus Na+ permeation properties via alternative splicing of exons 12a and 12b, nevertheless always conducts approximately twofold larger Ba2+ than Ca2+ currents in vitro, regardless of exon 12 splicing (Senatore and Spafford, 2010; Senatore et al., 2014). Differences in Ca2+ versus Ba2+ permeation among the various in vitro–expressed T-type channels have no consequence for the pattern in fold increases in peak inward current upon addition of external Na+. Comparing the fold increase in peak macroscopic current through TCav3 elicited by ramping the voltage from −110 to 100 mV in the presence of 2 mM Ba2+ and either 100 mM Na+ or NMDG+ (Fig. 10 E) with previously published data from other in vitro–expressed channels (Senatore et al., 2014; Stephens et al., 2015) reveals a similar ranking in Na+ permeation as observed in the presence of 2 mM external Ca2+: LCav3-12a >> LCav3-12b > TCav3 ∼ hCav3.3 > hCav3.2 > hCav3.1 (i.e., compare Fig. 10, B and F). Ca2+ block of TCav3 Na+ currents indicates high-affinity pore binding of external Ca2+, similar to human Cav3.1 At extremely low concentrations of external Ca2+, all Cav channels conduct prominent Na+ currents. Titrating increasing concentrations of extracellular Ca2+ leads to a block of Na+ current, with a sensitivity that reflects the affinity of Ca2+ to binding sites located in the pore. As [Ca2+]out increases, the blocking effect reaches saturation levels, then Ca2+ itself becomes abundant enough to occupy multiple sites within the pore, leading to Ca2+ permeation (Tsien et al., 1987; Sather and McCleskey, 2003; Cheng et al., 2010; Buraei et al., 2014; Tang et al., 2014). This property, of a decrease in Na+ current amplitude and a subsequent rise in Ca2+ current as [Ca2+]out increases (known as the Ca2+ block effect), is exemplified by peak currents through human Cav3.1 elicited by repeating voltage steps from −110 to −35 mV, whereas external Ca2+ is perfused from 1 nM to 10 mM in the presence of invariant 60 mM external Na+ (Fig. 11 A; Senatore et al., 2014). Notably, the TCav3 Ca2+-block data are similar to previously published data for human Cav3.1 (Senatore et al., 2014; Stephens et al., 2015), where at 10 µM [Ca2+]out, 97.3% and 96.1% of peak Na+ current elicited by voltage steps to −35 mV were blocked, respectively (Fig. 11, A and B). Human Cav3.1 Na+ currents appear to be slightly more sensitive to Ca2+, where at 0.1 µM [Ca2+]out, 34.1% of Na+ current was blocked, whereas TCav3 and the two exon 12 variants of LCav3 only exhibit between −5.6 and 4.6% block. Beyond 10 µM [Ca2+]out, TCav3 and Cav3.1 show 13.8- and 9.4-fold increases in peak current caused by an emerging Ca2+ conductance (Fig. 11, A and C). In contrast, LCav3-12b shows a moderate reduction in Ca2+ block (82.6%), whereas the highly sodium-permeant LCav3-12a has a dramatically right-shifted dose–response for Ca2+ (Fig. 11 A), with only 44.5% of current blocked at 10 µM Ca2+ (Fig. 11 B; Senatore et al., 2014). Also noticeable is a continued decline in current amplitude beyond 10 µM [Ca2+]out for the Lymnaea channel variants (Fig. 11 A), most marked for LCav3-12a, with a sevenfold continued decrease in current from 10 µM to 10 mM external Ca2+ (Fig. 11 C), further reflecting the reduced affinity for Ca2+ in the LCav3 pore. Collectively, the Ca2+ block data indicates that TCav3 exhibits a potent, but perhaps slightly diminished Ca2+-block of Na+ currents compared with human Cav3.1. Instead, the two exon 12 variants of Lymnaea Cav3 exhibit moderately attenuated and extremely attenuated Ca2+ block, indicative of lowered binding affinity for Ca2+ along the extracellular surface of the channel pore. Figure 11. Ca2+ block of inward Na+ current for TCav3 compared with other cloned T-type channels. (A) Ca2+ block of Na+ currents for TCav3 compared with previously published data for human Cav3.1 and the two exon 12 splice variants of the Lymnaea channel, LCav3-12a and LCav3-12b (Senatore et al., 2014). Increasing external [Ca2+] from 1 nM to 10 mM, in the presence of invariant external [Na+] of 60 mM, blocks Na+ currents through TCav3 with similar affinity as human Cav3.1, where at 10 µM external Ca2+ (i.e., 10−5 [Ca2+]out), mean peak inward currents (±SE) for Cav3.1and TCav3, elicited by voltage steps from −110 to −35 mV, show substantial attenuation from maximum (I/Imax). Instead, exon 12 variants of LCav3 show a reduced Ca2+ block, most dramatically for LCav3-12a. (B) Mean percent block of inward Na+ currents (±SE) for the various Cav3 channels at 10 and 0.1 µM [Ca2+]out. Asterisks depict statistical significance (p-values) for comparisons of means for percent block relative to TCav3 (generated by one-way ANOVA; *, P ≤ 0.05; **, P ≤ 0.005; ***, P ≤ 0.0005). (C) Mean fold increase/decrease in I/Imax for the various cloned channels (±SE) as [Ca2+]out is increased from 10 µM to 10 mM. Asterisks depict statistical significance (p-values) for comparisons of means for fold change in current relative to TCav3 (generated by one-way ANOVA; *, P ≤ 0.05; ***, P ≤ 0.0005). Ca2+ block, and not Ca2+ selectivity, determines the degree of Na+ permeation through T-type channels To further understand the Na+ permeation properties of TCav3, we characterized its Ca2+ versus monovalent cation selectivity by measuring zero-current reversal potentials (ERev) under bi-ionic conditions (i.e., 4 mM Ca2+out and 100 mM Li+in, Na+in, K+in, or Cs+in). The reversal potential of peak bi-ionic currents is determined by the pore’s preference for inward-permeating Ca2+, which pulls ERev toward more positive voltages, versus outward-permeating monovalent cations (i.e., X+, where X = Li, Na, K, or Cs), which pull ERev toward more negative voltages. Depolarizing voltage steps from −90 to 70 mV under the four bi-ionic conditions (i.e., Ca2+out-Li+in, Ca2+out-Na+in, Ca2+out-K+in, and Ca2+out-Cs+in) produce markedly different outward current components for TCav3, with decreasing amplitudes from Li+ to Na+ to K+ to Cs+ (Fig. 12 A), reflecting the pore’s decreasing permeability to monovalent cation flow according to the Eisenman selectivity model (i.e., ions with smaller radii are more permeable through a narrow pore: Li+ < Na+ < K+ < Cs+ with respect to radius; Eisenman et al., 1967; Eisenman and Horn, 1983). Correspondingly, ERev values for the four bi-ionic IV plots for TCav3 have leftward shifts corresponding to increased monovalent permeation from Cs+ to Li+ (Fig. 12 B). Converting ERev values to permeability ratios PCa/PX using the bi-ionic Nernst equation (Senatore et al., 2014), which reflect the pore’s preference for Ca2+ over monovalent X+, reveals that the Trichoplax channel is poorly selective for Ca2+ over X+ monovalents compared with human Cav3.1 (Stephens et al., 2015), falling between previously published values for Lymnaea LCav3-12a and LCav3-12b variants for PCa/PNa and PCa/PK (Fig. 12 C; Senatore et al., 2014). Interestingly, TCav3 is the most Li+-permeable channel, whereas the Cs+ permeability for all non-deuterostome channels is roughly equal and significantly higher than human Cav3.1. Figure 12. Bi-ionic reversal potential analysis of TCav3 Ca2+ versus monovalent cation permeability and comparison with other cloned T-type channels. (A) Sample current traces recorded for TCav3 recorded under different bi-ionic conditions, with 4 mM [Ca2+]out and 100 mM [X+]in (X = Li, Na, K, or Cs), reveal increasing outward current amplitudes from Cs+ to K+ to Na+ to Li+, reflecting increasing permeability with decreasing ionic radius, consistent with the Eisenman model of permeability through a narrow pore (Eisenman et al., 1967; Eisenman and Horn, 1983). (B) Decreasing permeability for Ca2+ over X+, from Cs+ to Li+, is measured by zero-current reversal potentials (ERev) on IV plots, which represent voltages at which inward-flowing Ca2+ and outward-flowing X+ ions are at equilibrium. (C) Converting ERev values to relative permeabilities for Ca2+ over X+ (i.e., PCa/PX), using the bi-ionic Nernst equation, reveals that TCav3 is poorly selective for Ca2+ over X+ ions, with a Ca2+-selectivity profile that falls between the extremely Na+-permeable LCav3-12a and the moderately Na+-permeable LCav3-12b. (D) A zoomed-out view of the IV plot in B reveals three conductance states for TCav3 (i.e., the slope of current I/Imax as a function of depolarizing voltage), with an inward conductance (GIn) at voltages that favor inward Ca2+ flow, a rectifying conductance across the reversal potential (GRev) where Ca2+ and X+ ions are competing for permeation, and an outward conductance (GOut) where the voltage favors outward flow of monovalent cations. (E) Comparing the three different conductance states for TCav3 with those of the two exon 12 variants of LCav3 and the human Cav3.1 channel places TCav3 somewhere between Cav3.1 and LCav3-12b. Asterisks depict statistical significance (p-values) for differences between mean conductance relative to TCav3 (one-way ANOVA; *, P ≤ 0.05; **, P ≤ 0.005; ***, P ≤ 0.0005). All values for B–E represent mean ± SE. Ca2+ versus X+ permeability features of T-type channels are also reflected in the rectification of macroscopic bi-ionic currents, where three conductance states are evident on IV plots: (1) an inward conductance at voltages where the driving force favors inward flow of Ca2+ (GIn); (2) a conductance through ERev where a transition between inward Ca2+ and outward X+ occurs, (GRev); and (3) an outward conductance where driving force favors outward X+ flow (GOut; Fig. 12 D and Fig. S3). Comparison of the three conductance states of TCav3 with previously published data for other in vitro–expressed T-type channels (Senatore et al., 2014; Stephens et al., 2015) corroborates a low Ca2+ selectivity for the Trichoplax Cav3 channel compared with human Cav3.1. Both GRev and GOut values for TCav3, which reflect permeability to monovalent cations, are significantly larger than those of Cav3.1, regardless of monovalent cation type (Fig. 12 E). Indeed, Cav3.1 is very ineffective at conducting outward monovalent currents, even at extremely depolarized potentials >60 mV (Fig. 12 D and Fig. S3). Cav3.1 also has the largest conductance for inward Ca2+ flow (GIn), but noticeably, under the most physiological conditions with K+ present in the internal saline, the GIn Ca2+ conductance for TCav3 encroaches on that of Cav3.1 (i.e., means are not statistically different; Fig. 12 E). At the other extreme, the Lymnaea channel LCav3-12a has the lowest conductance for inward Ca2+ (GIn) and the largest monovalent conductances through ERev and outwards (GRev and GOut, respectively; Fig. 12 D and Fig. S3). Collectively, the bi-ionic reversal potential data indicate that the TCav3 channel pore is poorly selective for Ca2+ over monovalents compared with human Cav3.1, with bi-ionic selectivity properties more similar to those of highly Na+-permeant LCav3-12a and the more Ca2+-selective LCav3-12b channels (Table 2). In light of these data, it is apparent that Ca2+ block, and not Ca2+ selectivity, is the major determinant for the degree of Na+ that permeates through TCav3 and other T-type channels. Indeed, the efficiency of Ca2+ block for the various cloned T-type channels (Cav3.1 > TCav3 > LCav3-12b > LCav3-12a; Fig. 11) correlates with their degree of Na+ permeation (e.g., 21.0%, 29.6%, 60.5%, and 93.5% of total current carried by Na+, respectively; measured by dividing percentage increases in peak current shown in Fig. 10 B by that same value plus 100%). Discussion TCav3 is the most divergent metazoan T-type calcium channel Cav3 channels appear to have emerged upwards of a billion years ago in a eukaryotic ancestor of choanoflagellates and metazoans (Morris, 1998), being present in the genome of S. rosetta (Fairclough et al., 2013; Moran and Zakon, 2014). We were unable to find a Cav3 channel homologue in the genome of choanoflagellate Monosiga brevicollis (King et al., 2008), indicative of either gene loss or an incomplete genome sequence. We also failed to identify Cav3 channel genes in various genomes and transcriptomes available for sponges and ctenophores (Srivastava et al., 2010; Ryan et al., 2013; Moroz et al., 2014; Fernandez-Valverde et al., 2015), suggesting that these two phyla lost Cav3 channels. Analysis of various Cav3 channel gene exon/intron structures revealed a general trend toward increased intron/exon number from choanoflagellates to vertebrates, with the Salpingoeca Cav3 gene bearing 13 exons from start codon to stop codon, Trichoplax Cav3 bearing 28 exons, Nematostella Cav3a bearing 30 exons, and mouse Cav3.1 bearing 38 exons (Fig. 3). Such an increase in intron number from premetazoans, to early diverging metazoans, to bilaterians is consistent with observed trends at the whole-genome level (King et al., 2008). Interestingly, Cav3 channel genes from bilaterians appear to have undergone the most significant changes in intron/exon structure within coding sequences for channel structures associated with modulation of function, such as optional exon 8b found in mouse Cav3.1 and Lymnaea Cav3, which regulates channel membrane expression (Senatore and Spafford, 2012); mouse exons 14–17 in the II–III linker, which serve as a hotbed for modulation by kinases and G-proteins (Chemin et al., 2006; Perez-Reyes, 2010b; Senatore et al., 2012); exons 25c and 26 in the III–IV linker, which in both vertebrates and protostome invertebrates cause alterations in channel voltage-gating and kinetics (Senatore and Spafford, 2012); the C terminus, where vertebrate Cav3 channels physically interact with other ion channel types such as Ca2+-modulated A-type K+ channels (Anderson et al., 2010); and SNARE proteins, which facilitate T-type channel involvement in low-threshold exocytosis (Weiss et al., 2012). Indeed, if a similar pattern extends to numerous other gene orthologues in the metazoan genome, it might account for some of the apparent increase in anatomical, cellular, and molecular complexity of vertebrates and bilaterians compared with more early diverging animals (Valentine et al., 1994), in spite of comparable total gene counts. Despite its comparative simplicity at the gene structure level, the TCav3 protein sequence retains all of the hallmark structural features of four-domain channels, with four homologous repeat domains, each containing extracellular turret-containing P-loops separated by pore-forming S5 and S6 helices, and voltage-sensor modules (S1 to S4 helices), with S4 helices packed with positively charged lysine (K) and arginine (R) residues critical for voltage sensing (Fig. 1 A; Catterall, 2010). In addition, TCav3 bears the three most prominent structural features that distinguish Cav3 channels from Cav1 and Cav2 types: (1) a selectivity filter motif with two aspartates (D) instead of two glutamates (E) in the P-loops of domains III and IV (i.e., EEDD vs. EEEE); (2) the absence of a calmodulin-binding isoleucine-glutamine (IQ) motif in the C terminus, which permits modulation of Cav1 and Cav2 channels by cytoplasmic Ca2+ influx (Ben-Johny et al., 2014); and (3) a predicted helix-turn-helix gating brake structure in the domain I–II intracellular linker, in an analogous region where Cav1 and Cav2 channels bind accessory Cavβ subunits via a distinct structure called the α interaction domain (AID; Perez-Reyes, 2010a).The absence of an AID in T-type channels highlights another distinguishing feature, which is a lack of dependence on Cavβ subunits. For Cav1 and Cav2 channels, Cavβ as well as Cavα2δ subunits are obligate counterparts that complex with the channels to regulate gating, trafficking, and proteolytic turnover (Arikkath and Campbell, 2003; Richards et al., 2004; Altier et al., 2011). Instead, T-type channels function as separate entities, with an autonomous gating brake serving in lieu of the AID/Cavβ subunit, which nevertheless also regulates channel gating (Perez-Reyes, 2010a). Even the most basal of all known T-type channels, from S. rosetta, bears a predicted gating brake motif in its I–II linker (Fig. 2 D). The structural distinction between AIDs of high voltage–activated Cav1/Cav2 channels and gating brakes of low voltage–activated Cav3 channels thus appears to have ancient origins predating Metazoa. However, given the similar helical arrangement that extends into the cytoplasm from the domain I S6 helix, it is conceivable that both structures evolved via divergence from a helical structure present in an ancient Cav channel ancestor of Cav1/Cav2 and Cav3 channels. Localization of TCav3 channel in gland cells In bilaterians, the three types of Cav channels are specialized to carry out distinct and pivotal roles in neurons and muscle, and other excitable cell types, where they translate electrical signals of Nav, Kv, and synaptic ligand-gated cation channels into cellular events by coupling with Ca2+-sensitive cytoplasmic proteins (Berridge, 2006; Rizzuto and Pozzan, 2006). Cav1 channels are most classically associated with excitation-contraction coupling in muscle, as well as excitation-transcription coupling in neurons and muscle, and Cav2 channels are associated with fast presynaptic exocytosis of neurotransmitters (excitation-secretion coupling; Catterall, 2011). Instead, Cav3 channels have long eluded such a stereotyped and ubiquitous classification, where their roles tend to vary depending on cell type, ranging from regulating cellular excitability in select neurons and other excitable cells, to driving low-threshold exocytosis in select neurons and neurosecretory cells, to regulating tone and contraction of various muscle cell types (Perez-Reyes, 2003; Senatore et al., 2012). Trichoplax is very interesting in its highly simplified cellular body plan, bearing only six cell types: ciliated dorsal and ventral epithelial cells, with ventral epithelial cells responsible for ciliary locomotion (i.e., gliding along hard surfaces); crystal cells, which bear internal birefringent crystals with unknown function (Smith et al., 2014); fiber cells, positioned between the epithelial cell layers and proposed to be contractile in nature (Behrendt and Ruthmann, 1986; Smith et al., 2014); gland cells, proposed to play roles in paracrine signaling and which resemble neurons and neurosecretory cells in their expression of exocytotic SNARE proteins and membrane-apposed vesicles (Grell and Ruthmann, 1991; Smith et al., 2014); and ventral lipophil cells, also apparently exocytotic in nature but specialized for secretion of hydrolytic enzymes for external digestion of algae during feeding (Smith et al., 2014, 2015). The apparent morphological, anatomical, and ultrastructural similarly of Trichoplax cell types with those from more complex animals (i.e., epithelial, neuron/neuroendocrine, muscle, and digestive) is suggestive of cellular homology at the level of genes and proteins. In accordance, the presence of numerous genes in the Trichoplax genome crucial for cell-specific functioning (Srivastava et al., 2008), including those for generating and packaging neurotransmitters and neuropeptides (Nikitin, 2015), cellular excitability (e.g., Nav2 and Kv channels, K+ leak channels; Senatore et al., 2016), and muscle contraction (Steinmetz et al., 2012), suggests that homologous molecular processes are taking place in some of these cell types. Here, we extend the apparent homology of gland cells with neurons and neuroendocrine cells in our localization of TCav3 to this cell type (Fig. 4), suggesting that the cells undergo rapid fluxes in membrane voltage and transient rises in cytoplasmic [Ca2+], perhaps for engaging Ca2+ sensitive elements of the exocytotic machinery. A Ca2+ dependence for gland cell exocytosis is further suggested by their expression of complexin, a Ca2+-responsive regulator of the SNARE complex and vesicle fusion (Yoon et al., 2008), and further, by the presence of synaptotagmin in the genome (Srivastava et al., 2008), a key Ca2+ sensor for exocytotic release which interacts with complexin during fusion (Tang et al., 2006). We note from our RT-PCR experiment that beyond the Cav3 channel, Trichoplax expresses Cav1 and Cav2 channels at the whole-animal mRNA level, as well as the single Cavβ and three Cavα2δ accessory subunit genes (Fig. 5). If the Cav1 and/or Cav2 channels are similarly localized to gland cells, TCav3 might thus serve to activate the high voltage–activated channels that in turn drive vesicle fusion and exocytosis, as is common in neuroendocrine cells (Mansvelder and Kits, 2000). Notably, the apparent enrichment of TCav3 along the outward-facing area of gland cells (Fig. 4), where the SNARE proteins and membrane-apposed vesicles were previously shown to reside (Smith et al., 2014), is suggestive of a more direct involvement in exocytosis, perhaps driving graded exocytosis before action potential threshold as occurs in select vertebrate and invertebrate neurons and neurosecretory cells (Carbone et al., 2006; Weiss and Zamponi, 2013; Senatore et al., 2016). In vertebrates, a direct interaction between Cav3 channels and SNARE proteins syntaxin-1A and SNAP-25 has been documented (Weiss et al., 2012), tethering the channels close to Ca2+-sensitive elements of the exocytotic machinery. The expression of TCav3 in gland cells implies that the use of T-type channels in regulating excitation and secretion of neuron/neuroendocrine-like cells might have evolved very early on, at least in the ancestor of placozoans and cnidarians/bilaterians, but perhaps even further back, in the single-celled ancestor of choanoflagellates and metazoans, which possessed a T-type channel gene as well as the core elements of the exocytotic apparatus (King et al., 2008; Fairclough et al., 2013; Moran and Zakon, 2014). Difficult expression of TCav3 in HEK-293T cells Our first efforts to express the cloned TCav3 channel in HEK-293T cells for electrophysiological recording were unsuccessful. Instead, we found that coexpression with rat Cavβ1b and Cavα2δ1 accessory subunits of high voltage–activated Cav1 and Cav2 channels dramatically increased channel expression either as a fusion protein with EGFP, or as a separate protein recordable via whole-cell patch clamp. Interestingly, Cavβ and Cavα2δ subunits increase membrane expression of high voltage-activated Cav channels in part by blocking their internalization and proteasomal degradation (Bernstein and Jones, 2007; Altier et al., 2011; Dolphin, 2012), although the process seems to depend on direct interactions between the subunits and the channel proteins. We do not necessarily expect direct protein–protein interaction between TCav3 and the Cavβ1b and Cavα2δ1 accessory subunits because such an interaction has yet to be reported for any other cloned vertebrate or invertebrate T-type channel (Dubel et al., 2004; Senatore and Spafford, 2010; Dawson et al., 2014; Cens et al., 2015; Jeong et al., 2015). However, given the overlapping emergence of Cav1/Cav2 channels, Cav3 channels, and the Cavβ subunit in a premetazoan ancestor (Dawson et al., 2014; Moran and Zakon, 2014), the possibility exists that ancestral T-type channels and those from extant basal organisms such as Trichoplax physically interact with high voltage-activated Cavβ1b and Cavα2δ1 subunits. Here, we found that rat Cavβ1b and Cavα2δ1 subunits significantly increased ectopic expression of GFP expressed in the absence of TCav3 (Fig. 6), indicating that their effect on ectopic protein expression might at least in part be a result of nonspecific processes. The biophysical properties of TCav3 are consistent with roles in regulating gland cell excitability One of the most clear cellular functions of Cav3 channels is regulating excitability (Perez-Reyes, 2003; Senatore et al., 2012), invoking their low voltages of activation and fast, transient kinetics. For example, mammalian Cav3.2 calcium channels are enriched in pain-sensing neurons, where they amplify depolarizing sensory inputs to increase nociceptive signaling to the spinal cord (Bourinet et al., 2005; Rose et al., 2013). Similarly, in the brain, neuronal Cav3 channels are enriched along dendrites (McKay et al., 2006), where they boost postsynaptic excitatory potentials to increase the likelihood of eliciting action potentials (Perez-Reyes, 2003; Senatore et al., 2012). In some cases, the rapid kinetics and low voltages of activation of Cav3 channels enable them to function in lieu of Nav channels where they can drive Ca2+ action potentials, as occurs in striated muscle cells from jellyfish (Lin and Spencer, 2001), snail cardiomyocytes (Yeoman et al., 1999; Senatore et al., 2014), and C. elegans pharyngeal muscle (Steger et al., 2005). Here, we show that the biophysical properties of the Trichoplax Cav3 channel are consistent with a role in regulating excitability in gland cells. Inward Ca2+ currents recorded from HEK cells expressing recombinant TCav3 emerge upon slight membrane depolarization from a holding potential of −110 mV (Fig. 7, A and B), indicating that like all other in vitro–expressed Cav3 channels, TCav3 is low voltage activated. In fact, TCav3 has the lowest voltage dependency for activation of any cloned T-type channel, with a 5 mV more negative maximal peak inward current than Lymnaea Cav3, and 10 to 20 mV more negative than the three mammalian isotypes, Cav3.1 to Cav3.3 (Fig. 7 B and Table 2). Accordingly, Boltzmann transformation of the peak current–voltage (IV) plot for TCav3, into an activation curve (Fig. 7 C), reveals a half-maximal activation considerably left-shifted compared with other Cav3 channels. Instead, the channel’s half-maximal steady-state inactivation is roughly similar to those of other channels (Table 2). TCav3 currents also have reasonably fast activation and inactivation kinetics, which although marginally slower than Lymnaea Cav3 and the mammalian Cav3.1/Cav3.2 isotypes, are considerably faster than those of Cav3.3, the slowest of the vertebrate T-type channels (Table 2). TCav3 is thus capable of conducting fast inward Ca2+ currents upon slight membrane depolarization, with a lower voltage threshold than other T-type channels, while being equally subject to voltage-dependent inactivation. These features indicate that TCav3 is poised to be more active at threshold voltages compared with other T-type channels. Furthermore, the particularly slow deactivation kinetics for TCav3, at voltages near −70 mV (Table 2), would serve to counter Kv channel–driven action potential repolarization, effectively widening action potentials and increasing net Ca2+ influx. Overall, the voltage dependencies and kinetic properties of TCav3 are most similar to those of fellow invertebrate T-type, Lymnaea Cav3, and least to those of mammalian Cav3.3 (Table 2). What is striking is that upon side-by-side comparison of biophysical properties of various cloned Cav3 channels, the structurally divergent Trichoplax homologue, which is >600 million years separated from mammalian isotypes, is more similar to mammalian Cav3.1 and Cav3.2 than is Cav3.3. Indeed, there appear to have been strong evolutionary constraints on the TCav3 channel to retain a core set of biophysical properties, suggesting that the need for its cellular contributions are conserved even in the absence of neurons and muscle. An important caveat toward speculation about the physiological roles for TCav3 in vivo is that its contributions would ultimately depend on the membrane potential, which is controlled by a milieu of electrogenic proteins. Although Trichoplax has the majority of these electrogenic genes (Srivastava et al., 2008; Senatore et al., 2016), suggesting that some of its cells have polarized resting membrane potentials and exhibit rapid fluxes in membrane voltage such as action potentials, the membrane properties of Trichoplax cells have yet to be reported. We and others have attempted intracellular recording of isolated Trichoplax cells, but their small size (<10-µm diameter) and particular membrane features make obtaining a gigaohm seal during patch-clamp recording particularly difficult. Based on the biophysical properties of TCav3, we can speculate that if the resting membrane potential of a typical Trichoplax cell sits above −60 mV, the channel would not contribute to excitability because of inactivation (Fig. 7 C). However, transient hyperpolarization from such potentials could recruit the channel by removing inactivation, where it would contribute to postinhibitory rebound (PIR) excitation. In mammals, T-type channel-mediated PIR excitation plays an important role in certain neuronal circuits, such as the thalamus, where postinhibitory Ca2+ spikes support rhythmic bursts of action potentials that project to the cortex and gate sensory information during non-REM sleep (Lee et al., 2004; Anderson et al., 2005; Crunelli et al., 2006). Notably, TCav3 and the Cav3 channel from Lymnaea have a slower recovery from inactivation than mammalian channels (Table 2), so they would require more prolonged hyperpolarization to be recruited for PIR excitation. Finally, even if Trichoplax cells do not undergo rapid changes in membrane voltage, T-type channels could nevertheless contribute a consistent steam of Ca2+ into the cytosol through a window current (Fig. 7 C), which can be used by cells to transition between bimodal resting membrane potentials (Dreyfus et al., 2010) and are associated with cellular proliferation during development and cancer (Lory et al., 2006; Senatore et al., 2012; Gackière et al., 2013). TCav3 resembles mammalian Cav3 channels with respect to Na+ permeation Previous work examining altered Na+ permeation of the Lymnaea T-type channel, caused by alternative splicing of exons 12a and 12b in a region of the domain II P-loop called the turret, revealed that factors outside of the selectivity filter can nevertheless have important consequences for defining Ca2+ versus Na+ permeation. However, based on these studies, it is difficult to reconcile differences in permeation properties among Cav3 channels purely on the structure of exon 12, where in Lymnaea, the smaller exon 12a imposes extreme Na+ permeability (i.e., 93.5% of current carried by Na+; Fig. 10 B: 1,440 ÷ 1,540% = 93.5%), whereas in mammalian channels, homologous 12a-like exons produce only moderate Na+ permeability (∼21.0%, 24.3%, and 31.3% for Cav3.1, Cav3.2, and Cav3.3, respectively). We found the basal TCav3 channel to resemble mammalian channels, with ∼29.6% of inward current carried by Na+, making it statistically indistinguishable from Cav3.2 and Cav3.3 (Table 2). Indeed, a low Na+ permeability for T-type channels bearing exon 12a–like turrets extends to other non-protostome channels that span the lineages between placozoans and mammals, including those from cnidarians (i.e., jellyfish; Lin and Spencer, 2001) and deuterostomes (i.e., echinoderm starfish eggs; Hagiwara et al., 1975), where T-type currents were reported to be carried mostly by Ca2+. Based on the available data, one explanation for the emergence of altered Na+ permeation in the snail and other protostome T-type channels via exon 12 splicing is that after dupliction of exon 12, structural alterations took place outside of the exon 12 region, rendering 12a-bearing channels more Na+ permeable. The duplicated exon 12b, unique to protostomes, was possibly adaptated to retain Ca2+ conducting channels via enlargement relative to exon 12a by ∼11 aa and increase in cysteine content from 0–3 to ∼5. In Lymnaea, exon 12b produces a channel more in line with non-protostome 12a-like channels, from Trichoplax through to mammals, with only ∼60.5% of current carried by Na+. Interestingly, the degree of co-permeation of Na+ alongside Ca2+ for the various cloned T-type channels seems to hold true even in the presence of Ba2+, which can exhibit increased or decreased permeation relative to Ca2+ in a channel-dependent manner. Like rat Cav3.1, TCav3 conducts larger Ca2+ currents than Ba2+ currents in vitro, whereas rat Cav3.2, LCav3-12a, and LCav3-12b all conduct larger Ba2+ currents and rat Cav3.3 conducts equal Ca2+ and Ba2+ currents (McRory et al., 2001). Despite these differences, the pattern of fold increases in peak inward current for the various channels upon replacement of impermeant external NMDG+ with Na+ is consistent regardless of whether Ca2+ or Ba2+ is present in the extracellular solution (Fig. 10). As such, it appears as though the factors that determine macroscopic conduction preference among divalent cations are different from those that determine preference between divalent versus monovalent cations. Notably, a previous study found that NMDG+ might directly block inward current through T-type channels (Khan et al., 2008), potentially confounding loss of current amplitude caused by replacement of Na+ in our experiments. However, for the snail T-type channel exon 12a and 12b variants, we previously found that replacement of Na+ with impermeant Tris+ resulted in a similar pattern of current attenuation compared with NMDG+ (Senatore, 2012), suggesting that the major effect on T-type channel current amplitude in these experiments is caused by Na+ depletion, and not NMDG+ block. Ca2+ block versus Ca2+ over Na+ selectivity in defining T-type channel cation permeability We sought to identify different aspects of cation permeation through T-type channels that could account for their varying Na+ permeabilities. For high voltage–activated Cav1 and Cav2 channels, which are considerably better than T-types at selecting for Ca2+ (e.g., compare 21–90% Na+ current for Cav3 channels with <0.1% for Cav1/Cav2; Tsien et al., 1987; Sather and McCleskey, 2003; Shcheglovitov et al., 2007; Cheng et al., 2010; Buraei et al., 2014; Tang et al., 2014), selectivity for Ca2+ is attributed to the ability of the ion to associate with a high-affinity binding site at the extracellular surface of the pore to repel and block inward Na+ flux (i.e., Ca2+ block). T-type channels are also expected to bind Ca2+, but their reduced selectivity is attributed to a ∼10-fold lowered binding affinity and reduced Ca2+ block (Shcheglovitov et al., 2007). We compared the Ca2+ block properties of various Cav3 channels, revealing that TCav3 shares with mammalian channels a potent Ca2+ block, indicated by rapidly attenuating Na+ currents as Ca2+ is incrementally added to the extracellular solution (Fig. 11). By stark contrast, exon 12 variants of LCav3 exhibit a reduction in Ca2+ block that is most extreme for LCav3-12a, for which the Ca2+ titration curve does not exhibit the classic U shape (where decreasing Ca2+-blocked Na+ currents gradually give way to increasing Ca2+ currents as [Ca2+]out increases; Fig. 11 A). Instead, the channels exhibit continued decline through to 10 mM [Ca2+]out, reflecting dramatically reduced Ca2+ binding affinity in the pore. Altogether, the Ca2+ block properties of the different Cav3 channels (Fig. 11) correlate with their respective Na+ permeabilities, apparent in Na+/NMDG+ replacement experiments (Fig. 10 B): TCav3 and human Cav3.1 have the most potent Ca2+ block and lowest Na+ permeability, and LCav3-12a and LCav-12b have reduced Ca2+ block proportional to their respective increases in Na+ permeability. However, we point out a minor inconsistency between our Ca2+-Na+ permeation data (Fig. 10) and our Ca2+ block data (Fig. 11). Whereas the Ca2+ block data suggests that at near-physiological external Ca2+ concentrations (i.e., 1 mM [Ca2+]out), most of the Na+ current has been blocked for all T-type channels (Fig. 11, A and B), replacement of 135 mM NMDG+ with Na+ in the presence of 2 mM [Ca2+]out causes significant increases in peak inward current (most marked for LCav3-12a), reflecting additive Na+ currents over already present Ca2+ currents (Fig. 10, A and B). So a question arises: why in 2 mM Ca2+ can you observe a considerable additive Na+ current via NMDG+ replacement, whereas 1–3 mM Ca2+ seems to mostly block the Na+ currents? We explain this inconsistency by noting that in Ca2+ block experiments, there are considerable residual currents, ranging in amplitudes from 100–200 pA (i.e., for LCav3-12b) to 100–800 pA (for LCav3-12a), which persist despite saturation in the Ca2+ block effect. We suggest that these residual currents represent combined Ca2+/Na+ currents, which would be differentially attenuated if the Na+ ions were to be replaced with impermeant NMDG+, consistent with the Na+/NMDG+ substitution data (Fig. 10, A, B, E, and F). TCav3 appears to be slightly more Na+ permeable than human Cav3.1, in part because of a slightly reduced Ca2+-block and pore Ca2+-binding affinity (Fig. 11, A and B). However, other aspects of the pore might influence the degree of Na+ permeation. Our bi-ionic reversal potential experiments, which approximate the preference of a given channel pore for inward-flowing Ca2+ ions versus outward-flowing monovalents (i.e., Li+, Na+, K+, and Cs+), revealed that TCav3 was considerably less selective for Ca2+ than human Cav3.1 (e.g., PCa/PNa = 35.61 ± 1.52 vs. 89.56 ± 8.21, respectively), falling between the more Na+ permeable channels LCav3-12a (33.06 ± 1.50) and LCav3-12b (41.49 ± 1.98). Indeed, if preference for Ca2+ over monovalents measured under bi-ionic conditions was the major determinant for Na+ permeation, then we would expect TCav3 to conduct a much higher proportion of inward Na+ upon depolarization, most similar to the LCav3-12a variant (Fig. 10 B). Instead, the selectivity for Ca2+ over monovalents in the TCav3 channel pore seems somewhat inconsequential, where 29.6% of the total current is carried by Na+, which is more similar to Cav3.1 (21.0%) than either LCav3-12b (60.5%) or LCav3-12a (93.5%). As such, Ca2+ block appears to be more consequential for determining Na+ permeation than is Ca2+ over Na+ selectivity. Instead, Ca2+ selectivity appears to play a more marginal role, possibly accounting for the 8.6% increase in Na+ permeability for TCav3 relative to Cav3.1. Finally, we also find an interesting inverse correlation between conductance values across bi-ionic reversal potential (GRev) and outward monovalent current flow (GOut) versus the potency of the Ca2+-block effect: whereas LCav3-12a has the largest GRev and GOut values and the weakest Ca2+ block, Cav3.1 has the smallest GRev and GOut values and the strongest Ca2+ block (compare Fig. 11 B with the top two panels of Fig. 12 E). This is not surprising because GRev and GOut should reflect the pore’s permeability to outward-flowing monovalents, against external Ca2+ which seeks to bind the extracellular high-affinity site. However, there is a slight inconsistency in this correlation, where GRev and GOut conductance values for TCav3 are significantly larger than Cav3.1, despite both channels bearing similarly potent Ca2+ block properties. Instead, the conductance values for TCav3 lie between those of Cav3.1 and the two exon 12 splice variants of LCav3. Thus, conductance across the reversal potential and during outward current flow though T-type channels might depend on a combination of both affinity for Ca2+ at the external pore surface (where TCav3 and Cav3.1 are similar), and pore Ca2+ versus Na+ permeability (where TCav3 and LCav3 are more similar). In conclusion, we find evidence that for T-type channels, Ca2+ block—and not Ca2+ versus Na+ selectivity—best correlates with the degree of inward Na+ permeation under simulated physiological conditions. For Trichoplax, which lives in seawater, the abundance of external Ca2+ must ensure efficient saturation of the Ca2+ block effect, rendering the channels mostly permeable to Ca2+. Here, our studies were performed using salines that are compatible with HEK-293T cells, with reduced ionic concentrations and osmolarity across both sides of the membrane compared with seawater. In future studies, it will be interesting to evaluate the consequences of altered ion concentrations on TCav3 biophysical, permeation, and pharmacological properties (and indeed other ion channel types), where evolutionary transitions from seawater to land/freshwater environments, and perhaps back again, would likely require some level of adaptation in channel function, such that contributions to cellular excitability and Ca2+ influx remain within acceptable parameters. Supplementary Material Supplemental Materials (PDF) Text S1-S2 (zipped TXT files) Acknowledgments We thank Drs. David Spafford and Paul S. Katz for providing support to A. Senatore during the preliminary stages of this research, Dr. Arnaud Monteil for preliminary Western blotting of the TCav3 channel protein, Farid R. Ahmadli for help analyzing Cav3 channel gene sequences, and Dr. Andreas Heyland for providing the Trichoplax specimens used for sequencing and cloning of the TCav3 channel cDNA. This work was funded by a National Science and Engineering Research Council of Canada (NSERC) Discovery Grant (RGPIN-2016-06023), a Canadian Foundation for Innovation Grant (CFI Project 35297), University of Toronto Mississauga start-up funds, and an NSERC postdoctoral fellowship (PDF-43851-2013) to A. Senatore. The authors declare no competing financial interests. Richard W. Aldrich served as editor. Abbreviations used: AID α interaction domain EGFP enhanced green fluorescent protein ML maximum likelihood RT reverse transcription ==== Refs Altier, C., A. Garcia-Caballero, B. Simms, H. You, L. Chen, J. Walcher, H.W. Tedford, T. Hermosilla, and G.W. Zamponi. 2011. The Cavβ subunit prevents RFP2-mediated ubiquitination and proteasomal degradation of L-type channels. Nat. Neurosci. 14 :173–180. 10.1038/nn.2712 21186355 Anderson, D., W.H. Mehaffey, M. Iftinca, R. Rehak, J.D. Engbers, S. Hameed, G.W. Zamponi, and R.W. Turner. 2010. Regulation of neuronal activity by Cav3-Kv4 channel signaling complexes. Nat. Neurosci. 13 :333–337. 10.1038/nn.2493 20154682 Anderson, M.P., T. Mochizuki, J. Xie, W. Fischler, J.P. Manger, E.M. Talley, T.E. Scammell, and S. Tonegawa. 2005. Thalamic Cav3.1 T-type Ca2+ channel plays a crucial role in stabilizing sleep. Proc. Natl. Acad. Sci. USA. 102 :1743–1748. 10.1073/pnas.0409644102 15677322 Arias-Olguín, I.I., I. Vitko, M. Fortuna, J.P. Baumgart, S. Sokolova, I.A. Shumilin, A. Van Deusen, M. Soriano-García, J.C. Gomora, and E. Perez-Reyes. 2008. Characterization of the gating brake in the I-II loop of Cav3.2 T-type Ca2+ channels. J. Biol. Chem. 283 :8136–8144. 10.1074/jbc.M708761200 18218623 Arikkath, J., and K.P. Campbell. 2003. Auxiliary subunits: Essential components of the voltage-gated calcium channel complex. Curr. Opin. Neurobiol. 13 :298–307. 10.1016/S0959-4388(03)00066-7 12850214 Behrendt, G., and A. Ruthmann. 1986. The cytoskeleton of the fiber cells of Trichoplax adhaerens (Placozoa). Zoomorphology. 106 :123–130. 10.1007/BF00312114 Ben-Johny, M., P.S. Yang, J. Niu, W. Yang, R. Joshi-Mukherjee, and D.T. Yue. 2014. Conservation of Ca2+/calmodulin regulation across Na and Ca2+ channels. Cell. 157 :1657–1670. 10.1016/j.cell.2014.04.035 24949975 Bernstein, G.M., and O.T. Jones. 2007. Kinetics of internalization and degradation of N-type voltage-gated calcium channels: Role of the α2/δ subunit. Cell Calcium. 41 :27–40. 10.1016/j.ceca.2006.04.010 16759698 Berridge, M.J. 2006. Calcium microdomains: Organization and function. Cell Calcium. 40 :405–412. 10.1016/j.ceca.2006.09.002 17030366 Bourinet, E., A. Alloui, A. Monteil, C. Barrère, B. Couette, O. Poirot, A. Pages, J. McRory, T.P. Snutch, A. Eschalier, and J. Nargeot. 2005. Silencing of the Cav3.2 T-type calcium channel gene in sensory neurons demonstrates its major role in nociception. EMBO J. 24 :315–324. 10.1038/sj.emboj.7600515 15616581 Buraei, Z., H. Liang, and K.S. Elmslie. 2014. Voltage control of Ca2+ permeation through N-type calcium (CaV2.2) channels. J. Gen. Physiol. 144 :207–220. 10.1085/jgp.201411201 25114024 Cain, S.M., and T.P. Snutch. 2010. Contributions of T-type calcium channel isoforms to neuronal firing. Channels (Austin). 4 :475–482. 10.4161/chan.4.6.14106 21139420 Capella-Gutiérrez, S., J.M. Silla-Martínez, and T. Gabaldón. 2009. trimAl: A tool for automated alignment trimming in large-scale phylogenetic analyses. Bioinformatics. 25 :1972–1973. 10.1093/bioinformatics/btp348 19505945 Carbone, E., A. Giancippoli, A. Marcantoni, D. Guido, and V. Carabelli. 2006. A new role for T-type channels in fast “low-threshold” exocytosis. Cell Calcium. 40 :147–154. 10.1016/j.ceca.2006.04.019 16759700 Catterall, W.A. 2010. Ion channel voltage sensors: Structure, function, and pathophysiology. Neuron. 67 :915–928. 10.1016/j.neuron.2010.08.021 20869590 Catterall, W.A. 2011. Voltage-gated calcium channels. Cold Spring Harb. Perspect. Biol. 3 :a003947. 10.1101/cshperspect.a003947 21746798 Cens, T., M. Rousset, C. Collet, M. Charreton, L. Garnery, Y. Le Conte, M. Chahine, J.-C. Sandoz, and P. Charnet. 2015. Molecular characterization and functional expression of the Apis mellifera voltage-dependent Ca2+ channels. Insect Biochem. Mol. Biol. 58 :12–27. 10.1016/j.ibmb.2015.01.005 25602183 Chemin, J., A. Monteil, E. Bourinet, J. Nargeot, and P. Lory. 2001. Alternatively spliced α(1G) (CaV3.1) intracellular loops promote specific T-type Ca2+ channel gating properties. Biophys. J. 80 :1238–1250. 10.1016/S0006-3495(01)76100-0 11222288 Chemin, J., A. Monteil, E. Perez-Reyes, E. Bourinet, J. Nargeot, and P. Lory. 2002. Specific contribution of human T-type calcium channel isotypes (α1G, α1H and α1I) to neuronal excitability. J. Physiol. 540 :3–14. 10.1113/jphysiol.2001.013269 11927664 Chemin, J., A. Traboulsie, and P. Lory. 2006. Molecular pathways underlying the modulation of T-type calcium channels by neurotransmitters and hormones. Cell Calcium. 40 :121–134. 10.1016/j.ceca.2006.04.015 16797700 Cheng, R.C., D.B. Tikhonov, and B.S. Zhorov. 2010. Structural modeling of calcium binding in the selectivity filter of the L-type calcium channel. Eur. Biophys. J. 39 :839–853. 10.1007/s00249-009-0574-2 20054687 Clapham, D.E. 2007. Calcium signaling. Cell. 131 :1047–1058. 10.1016/j.cell.2007.11.028 18083096 Clarke, T.F. IV, and P.L. Clark. 2008. Rare codons cluster. PLoS One. 3 :e3412. 10.1371/journal.pone.0003412 18923675 Crunelli, V., D.W. Cope, and S.W. Hughes. 2006. Thalamic T-type Ca2+ channels and NREM sleep. Cell Calcium. 40 :175–190. 10.1016/j.ceca.2006.04.022 16777223 Dawson, T.F., A.N. Boone, A. Senatore, J. Piticaru, S. Thiyagalingam, D. Jackson, A. Davison, and J.D. Spafford. 2014. Gene splicing of an invertebrate beta subunit (LCavβ) in the N-terminal and HOOK domains and its regulation of LCav1 and LCav2 calcium channels. PLoS One. 9 :e92941. 10.1371/journal.pone.0092941 24690951 Dolphin, A.C. 2012. Calcium channel auxiliary α2δ and β subunits: trafficking and one step beyond. Nat. Rev. Neurosci. 13 :542–555 (published erratum appears in Nat. Rev. Neurosci. 2012. 13:664). 10.1038/nrn3317 22805911 Dreyfus, F.M., A. Tscherter, A.C. Errington, J.J. Renger, H.S. Shin, V.N. Uebele, V. Crunelli, R.C. Lambert, and N. Leresche. 2010. Selective T-type calcium channel block in thalamic neurons reveals channel redundancy and physiological impact of I(T)window. J. Neurosci. 30 :99–109. 10.1523/JNEUROSCI.4305-09.2010 20053892 Dubel, S.J., C. Altier, S. Chaumont, P. Lory, E. Bourinet, and J. Nargeot. 2004. Plasma membrane expression of T-type calcium channel α1 subunits is modulated by high voltage-activated auxiliary subunits. J. Biol. Chem. 279 :29263–29269. 10.1074/jbc.M313450200 15123697 Edgar, R.C. 2004. MUSCLE: Multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res. 32 :1792–1797. 10.1093/nar/gkh340 15034147 Eisenman, G., and R. Horn. 1983. Ionic selectivity revisited: The role of kinetic and equilibrium processes in ion permeation through channels. J. Membr. Biol. 76 :197–225. 10.1007/BF01870364 6100862 Eisenman, G., J.P. Sandblom, and J.L. Walker Jr. 1967. Membrane structure and ion permeation. Study of ion exchange membrane structure and function is relevant to analysis of biological ion permeation. Science. 155 :965–974. 10.1126/science.155.3765.965 5334938 Emerick, M.C., R. Stein, R. Kunze, M.M. McNulty, M.R. Regan, D.A. Hanck, and W.S. Agnew. 2006. Profiling the array of Cav3.1 variants from the human T-type calcium channel gene CACNA1G: alternative structures, developmental expression, and biophysical variations. Proteins. 64 :320–342. 10.1002/prot.20877 16671074 Fairclough, S.R., Z. Chen, E. Kramer, Q. Zeng, S. Young, H.M. Robertson, E. Begovic, D.J. Richter, C. Russ, M.J. Westbrook, 2013. Premetazoan genome evolution and the regulation of cell differentiation in the choanoflagellate Salpingoeca rosetta. Genome Biol. 14 :R15. 10.1186/gb-2013-14-2-r15 23419129 Fernandez-Valverde, S.L., A.D. Calcino, and B.M. Degnan. 2015. Deep developmental transcriptome sequencing uncovers numerous new genes and enhances gene annotation in the sponge Amphimedon queenslandica. BMC Genomics. 16 :387. 10.1186/s12864-015-1588-z 25975661 Francis, W.R., N.C. Shaner, L.M. Christianson, M.L. Powers, and S.H. Haddock. 2015. Occurrence of isopenicillin-N-synthase homologs in bioluminescent ctenophores and implications for coelenterazine biosynthesis. PLoS One. 10 :e0128742. 10.1371/journal.pone.0128742 26125183 Fredman, D., M. Schwaiger, F. Rentzsch, and U. Technau. 2013. Nematostella vectensis transcriptome and gene models v2.0. Available at https://figshare.com/articles/Nematostella_vectensis_transcriptome_and_gene_models_v2_0/807696. Gackière, F., M. Warnier, M. Katsogiannou, S. Derouiche, P. Delcourt, E. Dewailly, C. Slomianny, S. Humez, N. Prevarskaya, M. Roudbaraki, and P. Mariot. 2013. Functional coupling between large-conductance potassium channels and Cav3.2 voltage-dependent calcium channels participates in prostate cancer cell growth. Biol. Open. 2 :941–951. 10.1242/bio.20135215 24143281 Gerstein, M.B., Z.J. Lu, E.L. Van Nostrand, C. Cheng, B.I. Arshinoff, T. Liu, K.Y. Yip, R. Robilotto, A. Rechtsteiner, K. Ikegami, modENCODE Consortium. 2010. Integrative analysis of the Caenorhabditis elegans genome by the modENCODE project. Science. 330 :1775–1787. 10.1126/science.1196914 21177976 Gomora, J.C., J. Murbartián, J.M. Arias, J.H. Lee, and E. Perez-Reyes. 2002. Cloning and expression of the human T-type channel Cav3.3: insights into prepulse facilitation. Biophys. J. 83 :229–241. 10.1016/S0006-3495(02)75164-3 12080115 Graveley, B.R., A.N. Brooks, J.W. Carlson, M.O. Duff, J.M. Landolin, L. Yang, C.G. Artieri, M.J. van Baren, N. Boley, B.W. Booth, 2011. The developmental transcriptome of Drosophila melanogaster. Nature. 471 :473–479. 10.1038/nature09715 21179090 Grell, K.G., and A. Ruthmann. 1991. Placozoa. In Microscopic Anatomy of Invertebrates. Volume 2: Placozoa, Porifera, Cnidaria, and Ctenophora. F.W. Harrison and J.A. Westfall, editors. Wiley-Liss, Inc., Hoboken, NJ. 13–27. Gustafsson, C., S. Govindarajan, and J. Minshull. 2004. Codon bias and heterologous protein expression. Trends Biotechnol. 22 :346–353. 10.1016/j.tibtech.2004.04.006 15245907 Hagiwara, S., S. Ozawa, and O. Sand. 1975. Voltage clamp analysis of two inward current mechanisms in the egg cell membrane of a starfish. J. Gen. Physiol. 65 :617–644. 10.1085/jgp.65.5.617 240906 Heyland, A., R. Croll, S. Goodall, J. Kranyak, and R. Wyeth. 2014. Trichoplax adhaerens, an enigmatic basal metazoan with potential. In Developmental Biology of the Sea Urchin and Other Marine Invertebrates: Methods and Protocols. D.J. Carroll, and S.A. Stricker, editors. Humana Press, New York. 45–61. 10.1007/978-1-62703-974-1_4 Hille, B. 2001. Ion channels of excitable membranes. Third edition. Sinauer, Sunderland, MA. 814 pp. Huff, J. 2015. The Airyscan detector from ZEISS: Confocal imaging with improved signal-to-noise ratio and super-resolution. Nat. Methods. 12 . Available at: http://www.nature.com/nmeth/journal/v12/n12/full/nmeth.f.388.html Jeong, K., S. Lee, H. Seo, Y. Oh, D. Jang, J. Choe, D. Kim, J.-H. Lee, and W.D. Jones. 2015. Ca-α1T, a fly T-type Ca2+ channel, negatively modulates sleep. Sci. Rep. 5 :17893. 10.1038/srep17893 26647714 Kang, H.W., J.Y. Park, S.W. Jeong, J.A. Kim, H.J. Moon, E. Perez-Reyes, and J.H. Lee. 2006. A molecular determinant of nickel inhibition in Cav3.2 T-type calcium channels. J. Biol. Chem. 281 :4823–4830. 10.1074/jbc.M510197200 16377633 Kang, H.W., I. Vitko, S.S. Lee, E. Perez-Reyes, and J.H. Lee. 2010. Structural determinants of the high affinity extracellular zinc binding site on Cav3.2 T-type calcium channels. J. Biol. Chem. 285 :3271–3281. 10.1074/jbc.M109.067660 19940152 Khan, N., I.P. Gray, C.A. Obejero-Paz, and S.W. Jones. 2008. Permeation and gating in CaV3.1 (α1G) T-type calcium channels effects of Ca2+, Ba2+, Mg2+, and Na+. J. Gen. Physiol. 132 :223–238. 10.1085/jgp.200809986 18663131 King, N., M.J. Westbrook, S.L. Young, A. Kuo, M. Abedin, J. Chapman, S. Fairclough, U. Hellsten, Y. Isogai, I. Letunic, 2008. The genome of the choanoflagellate Monosiga brevicollis and the origin of metazoans. Nature. 451 :783–788. 10.1038/nature06617 18273011 Kobayashi, H. 2015. Inducible suppression of global translation by overuse of rare codons. Appl. Environ. Microbiol. 81 :2544–2553. 10.1128/AEM.03708-14 25636849 Kozak, M. 1986. Point mutations define a sequence flanking the AUG initiator codon that modulates translation by eukaryotic ribosomes. Cell. 44 :283–292. 10.1016/0092-8674(86)90762-2 3943125 Kumar, S., G. Stecher, and K. Tamura. 2016. MEGA7: Molecular Evolutionary Genetics Analysis version 7.0 for bigger datasets. Mol. Biol. Evol. 33 :1870–1874. 10.1093/molbev/msw054 27004904 Lee, J., D. Kim, and H.-S. Shin. 2004. Lack of delta waves and sleep disturbances during non-rapid eye movement sleep in mice lacking α1G-subunit of T-type calcium channels. Proc. Natl. Acad. Sci. USA. 101 :18195–18199. 10.1073/pnas.0408089101 15601764 Lin, Y.C., and A.N. Spencer. 2001. Calcium currents from jellyfish striated muscle cells: Preservation of phenotype, characterisation of currents and channel localisation. J. Exp. Biol. 204 :3717–3726.11719535 Lory, P., I. Bidaud, and J. Chemin. 2006. T-type calcium channels in differentiation and proliferation. Cell Calcium. 40 :135–146. 10.1016/j.ceca.2006.04.017 16797068 Mansvelder, H.D., and K.S. Kits. 2000. Calcium channels and the release of large dense core vesicles from neuroendocrine cells: Spatial organization and functional coupling. Prog. Neurobiol. 62 :427–441. 10.1016/S0301-0082(00)00003-4 10856612 McGuffin, L.J., K. Bryson, and D.T. Jones. 2000. The PSIPRED protein structure prediction server. Bioinformatics. 16 :404–405. 10.1093/bioinformatics/16.4.404 10869041 McKay, B.E., J.E. McRory, M.L. Molineux, J. Hamid, T.P. Snutch, G.W. Zamponi, and R.W. Turner. 2006. CaV3 T-type calcium channel isoforms differentially distribute to somatic and dendritic compartments in rat central neurons. Eur. J. Neurosci. 24 :2581–2594. 10.1111/j.1460-9568.2006.05136.x 17100846 McRory, J.E., C.M. Santi, K.S. Hamming, J. Mezeyova, K.G. Sutton, D.L. Baillie, A. Stea, and T.P. Snutch. 2001. Molecular and functional characterization of a family of rat brain T-type calcium channels. J. Biol. Chem. 276 :3999–4011. 10.1074/jbc.M008215200 11073957 Moran, Y., and H.H. Zakon. 2014. The evolution of the four subunits of voltage-gated calcium channels: Ancient roots, increasing complexity, and multiple losses. Genome Biol. Evol. 6 :2210–2217. 10.1093/gbe/evu177 25146647 Moran, Y., M.G. Barzilai, B.J. Liebeskind, and H.H. Zakon. 2015. Evolution of voltage-gated ion channels at the emergence of metazoa. J. Exp. Biol. 218 :515–525. 10.1242/jeb.110270 25696815 Moroz, L.L., and A.B. Kohn. 2015. Unbiased view of synaptic and neuronal gene complement in ctenophores: Are there pan-neuronal and pan-synaptic genes across metazoa? Integr. Comp. Biol. 55 :1028–1049.26454853 Moroz, L.L., K.M. Kocot, M.R. Citarella, S. Dosung, T.P. Norekian, I.S. Povolotskaya, A.P. Grigorenko, C. Dailey, E. Berezikov, K.M. Buckley, 2014. The ctenophore genome and the evolutionary origins of neural systems. Nature. 510 :109–114. 10.1038/nature13400 24847885 Morris, S.C. 1998. Early metazoan evolution: Reconciling paleontology and molecular biology. Am. Zool. 38 :867–877. 10.1093/icb/38.6.867 Nikitin, M. 2015. Bioinformatic prediction of Trichoplax adhaerens regulatory peptides. Gen. Comp. Endocrinol. 212 :145–155. 10.1016/j.ygcen.2014.03.049 24747483 Obejero-Paz, C.A., I.P. Gray, and S.W. Jones. 2008. Ni2+ block of CaV3.1 (α1G) T-type calcium channels. J. Gen. Physiol. 132 :239–250. 10.1085/jgp.200809988 18663132 Ohkubo, T., Y. Inoue, T. Kawarabayashi, and K. Kitamura. 2005. Identification and electrophysiological characteristics of isoforms of T-type calcium channel Cav3.2 expressed in pregnant human uterus. Cell. Physiol. Biochem. 16 :245–254. 10.1159/000089850 16301824 Pan, Q., O. Shai, L.J. Lee, B.J. Frey, and B.J. Blencowe. 2008. Deep surveying of alternative splicing complexity in the human transcriptome by high-throughput sequencing. Nat. Genet. 40 :1413–1415. 10.1038/ng.259 18978789 Perez-Reyes, E. 2003. Molecular physiology of low-voltage-activated t-type calcium channels. Physiol. Rev. 83 :117–161. 10.1152/physrev.00018.2002 12506128 Perez-Reyes, E. 2010 a. Characterization of the gating brake in the I-II loop of CaV3 T-type calcium channels. Channels (Austin). 4 :453–458. 10.4161/chan.4.6.12889 21099341 Perez-Reyes, E. 2010 b. G protein-mediated inhibition of Cav3.2 T-type channels revisited. Mol. Pharmacol. 77 :136–138. 10.1124/mol.109.062133 19903827 Pisani, D., W. Pett, M. Dohrmann, R. Feuda, O. Rota-Stabelli, H. Philippe, N. Lartillot, and G. Wörheide. 2015. Genomic data do not support comb jellies as the sister group to all other animals. Proc. Natl. Acad. Sci. USA. 112 :15402–15407. 10.1073/pnas.1518127112 26621703 Presnyak, V., N. Alhusaini, Y.-H. Chen, S. Martin, N. Morris, N. Kline, S. Olson, D. Weinberg, K.E. Baker, B.R. Graveley, and J. Coller. 2015. Codon optimality is a major determinant of mRNA stability. Cell. 160 :1111–1124. 10.1016/j.cell.2015.02.029 25768907 Ramani, A.K., J.A. Calarco, Q. Pan, S. Mavandadi, Y. Wang, A.C. Nelson, L.J. Lee, Q. Morris, B.J. Blencowe, M. Zhen, and A.G. Fraser. 2011. Genome-wide analysis of alternative splicing in Caenorhabditis elegans. Genome Res. 21 :342–348. 10.1101/gr.114645.110 21177968 Richards, M.W., A.J. Butcher, and A.C. Dolphin. 2004. Ca2+ channel β-subunits: Structural insights AID our understanding. Trends Pharmacol. Sci. 25 :626–632. 10.1016/j.tips.2004.10.008 15530640 Rizzuto, R., and T. Pozzan. 2006. Microdomains of intracellular Ca2+: Molecular determinants and functional consequences. Physiol. Rev. 86 :369–408. 10.1152/physrev.00004.2005 16371601 Rose, K.E., N. Lunardi, A. Boscolo, X. Dong, A. Erisir, V. Jevtovic-Todorovic, and S.M. Todorovic. 2013. Immunohistological demonstration of CaV3.2 T-type voltage-gated calcium channel expression in soma of dorsal root ganglion neurons and peripheral axons of rat and mouse. Neuroscience. 250 :263–274. 10.1016/j.neuroscience.2013.07.005 23867767 Ryan, J.F., K. Pang, C.E. Schnitzler, A.-D. Nguyen, R.T. Moreland, D.K. Simmons, B.J. Koch, W.R. Francis, P. Havlak, S.A. Smith, NISC Comparative Sequencing Program. 2013. The genome of the ctenophore Mnemiopsis leidyi and its implications for cell type evolution. Science. 342 :1242592. 10.1126/science.1242592 24337300 Sather, W.A., and E.W. McCleskey. 2003. Permeation and selectivity in calcium channels. Annu. Rev. Physiol. 65 :133–159. 10.1146/annurev.physiol.65.092101.142345 12471162 Schierwater, B. 2005. My favorite animal, Trichoplax adhaerens. BioEssays. 27 :1294–1302. 10.1002/bies.20320 16299758 Schneider, C.A., W.S. Rasband, and K.W. Eliceiri. 2012. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods. 9 :671–675. 10.1038/nmeth.2089 22930834 Senatore, A. 2012. Alternative splicing of Lymnaea Cav3 and NALCN ion channel genes serves to alter biophysical properties, membrane expression, and ion selectivity. UWSpace. Available at https://uwspace.uwaterloo.ca/handle/10012/6926 Senatore, A., and J.D. Spafford. 2010. Transient and big are key features of an invertebrate T-type channel (LCav3) from the central nervous system of Lymnaea stagnalis. J. Biol. Chem. 285 :7447–7458. 10.1074/jbc.M109.090753 20056611 Senatore, A., and J.D. Spafford. 2012. Gene transcription and splicing of T-type channels are evolutionarily-conserved strategies for regulating channel expression and gating. PLoS One. 7 :e37409 (published erratum appears in PLoS One. 2013. 8. https://doi.org/10.1371/annotation/9d69fd63-4378-4e05-b2ad-2a5c98561223). 10.1371/journal.pone.0037409 22719839 Senatore, A., A.N. Boone, and J.D. Spafford. 2011. Optimized transfection strategy for expression and electrophysiological recording of recombinant voltage-gated ion channels in HEK-293T cells. J. Vis. Exp. (47 ):2314.21304463 Senatore, A., B.S. Zhorov, and J.D. Spafford. 2012. CaV3 T-type calcium channels. WIREs Membrane Transport and Signaling. 1 :467–491. 10.1002/wmts.41 Senatore, A., W. Guan, A.N. Boone, and J.D. Spafford. 2014. T-type channels become highly permeable to sodium ions using an alternative extracellular turret region (S5-P) outside the selectivity filter. J. Biol. Chem. 289 :11952–11969. 10.1074/jbc.M114.551473 24596098 Senatore, A., H. Raiss, and P. Le. 2016. Physiology and evolution of voltage-gated calcium channels in early diverging animal phyla: Cnidaria, placozoa, porifera and ctenophora. Front. Physiol. 7 :481. 10.3389/fphys.2016.00481 27867359 Sharp, P.M., and W.-H. Li. 1987. The codon Adaptation Index--a measure of directional synonymous codon usage bias, and its potential applications. Nucleic Acids Res. 15 :1281–1295. 10.1093/nar/15.3.1281 3547335 Shcheglovitov, A., P. Kostyuk, and Y. Shuba. 2007. Selectivity signatures of three isoforms of recombinant T-type Ca2+ channels. Biochim. Biophys. Acta. 1768 :1406–1419. 10.1016/j.bbamem.2007.02.017 17400181 Shcheglovitov, A., I. Vitko, I. Bidaud, J.P. Baumgart, M.F. Navarro-Gonzalez, T.H. Grayson, P. Lory, C.E. Hill, and E. Perez-Reyes. 2008. Alternative splicing within the I-II loop controls surface expression of T-type Cav3.1 calcium channels. FEBS Lett. 582 :3765–3770. 10.1016/j.febslet.2008.10.013 18930057 Smith, C.L., F. Varoqueaux, M. Kittelmann, R.N. Azzam, B. Cooper, C.A. Winters, M. Eitel, D. Fasshauer, and T.S. Reese. 2014. Novel cell types, neurosecretory cells, and body plan of the early-diverging metazoan Trichoplax adhaerens. Curr. Biol. 24 :1565–1572. 10.1016/j.cub.2014.05.046 24954051 Smith, C.L., N. Pivovarova, and T.S. Reese. 2015. Coordinated feeding behavior in Trichoplax, an animal without snapses. PLoS One. 10 :e0136098. 10.1371/journal.pone.0136098 26333190 Spafford, J.D., A.N. Spencer, and W.J. Gallin. 1999. Genomic organization of a voltage-gated Na+ channel in a hydrozoan jellyfish: Insights into the evolution of voltage-gated Na+ channel genes. Receptors Channels. 6 :493–506.10635065 Srivastava, M., E. Begovic, J. Chapman, N.H. Putnam, U. Hellsten, T. Kawashima, A. Kuo, T. Mitros, A. Salamov, M.L. Carpenter, 2008. The Trichoplax genome and the nature of placozoans. Nature. 454 :955–960. 10.1038/nature07191 18719581 Srivastava, M., O. Simakov, J. Chapman, B. Fahey, M.E.A. Gauthier, T. Mitros, G.S. Richards, C. Conaco, M. Dacre, U. Hellsten, 2010. The Amphimedon queenslandica genome and the evolution of animal complexity. Nature. 466 :720–726. 10.1038/nature09201 20686567 Steger, K.A., B.B. Shtonda, C. Thacker, T.P. Snutch, and L. Avery. 2005. The C. elegans T-type calcium channel CCA-1 boosts neuromuscular transmission. J. Exp. Biol. 208 :2191–2203. 10.1242/jeb.01616 15914662 Steinmetz, P.R., J.E. Kraus, C. Larroux, J.U. Hammel, A. Amon-Hassenzahl, E. Houliston, G. Wörheide, M. Nickel, B.M. Degnan, and U. Technau. 2012. Independent evolution of striated muscles in cnidarians and bilaterians. Nature. 487 :231–234. 10.1038/nature11180 22763458 Stephens, R.F., W. Guan, B.S. Zhorov, and J.D. Spafford. 2015. Selectivity filters and cysteine-rich extracellular loops in voltage-gated sodium, calcium, and NALCN channels. Front. Physiol. 6 :153. 10.3389/fphys.2015.00153 26042044 Syed, T., and B. Schierwater. 2002. The evolution of the Placozoa: a new morphological model. Senckenbergiana lethaea. 82 :315–324. 10.1007/BF03043791 Talavera, K., and B. Nilius. 2006. Biophysics and structure-function relationship of T-type Ca2+ channels. Cell Calcium. 40 :97–114. 10.1016/j.ceca.2006.04.013 16777221 Tang, J., A. Maximov, O.-H. Shin, H. Dai, J. Rizo, and T.C. Südhof. 2006. A complexin/synaptotagmin 1 switch controls fast synaptic vesicle exocytosis. Cell. 126 :1175–1187. 10.1016/j.cell.2006.08.030 16990140 Tang, L., T.M. Gamal El-Din, J. Payandeh, G.Q. Martinez, T.M. Heard, T. Scheuer, N. Zheng, and W.A. Catterall. 2014. Structural basis for Ca2+ selectivity of a voltage-gated calcium channel. Nature. 505 :56–61. 10.1038/nature12775 24270805 Taylor, J.T., X.B. Zeng, J.E. Pottle, K. Lee, A.R. Wang, S.G. Yi, J.A. Scruggs, S.S. Sikka, and M. Li. 2008. Calcium signaling and T-type calcium channels in cancer cell cycling. World J. Gastroenterol. 14 :4984–4991. 10.3748/wjg.14.4984 18763278 Tomlinson, W.J., A. Stea, E. Bourinet, P. Charnet, J. Nargeot, and T.P. Snutch. 1993. Functional properties of a neuronal class C L-type calcium channel. Neuropharmacology. 32 :1117–1126. 10.1016/0028-3908(93)90006-O 8107966 Tsien, R.W., P. Hess, E.W. McCleskey, and R.L. Rosenberg. 1987. Calcium channels: Mechanisms of selectivity, permeation, and block. Annu. Rev. Biophys. Biophys. Chem. 16 :265–290. 10.1146/annurev.bb.16.060187.001405 2439098 Valentine, J.W., A.G. Collins, and C.P. Meyer. 1994. Morphological complexity increase in metazoans. Paleobiology. 20 :131–142. 10.1017/S0094837300012641 Wang, E.T., R. Sandberg, S. Luo, I. Khrebtukova, L. Zhang, C. Mayr, S.F. Kingsmore, G.P. Schroth, and C.B. Burge. 2008. Alternative isoform regulation in human tissue transcriptomes. Nature. 456 :470–476. 10.1038/nature07509 18978772 Weiss, N., and G.W. Zamponi. 2013. Control of low-threshold exocytosis by T-type calcium channels. Biochimica et Biophysica Acta (BBA) - Biomembranes. 1828 :1579–1586. 10.1016/j.bbamem.2012.07.031 22885170 Weiss, N., S. Hameed, J.M. Fernández-Fernández, K. Fablet, M. Karmazinova, C. Poillot, J. Proft, L. Chen, I. Bidaud, A. Monteil, 2012. A Cav3.2/syntaxin-1A signaling complex controls T-type channel activity and low-threshold exocytosis. J. Biol. Chem. 287 :2810–2818. 10.1074/jbc.M111.290882 22130660 Yeoman, M.S., B.L. Brezden, and P.R. Benjamin. 1999. LVA and HVA Ca2+ currents in ventricular muscle cells of the Lymnaea heart. J. Neurophysiol. 82 :2428–2440.10561416 Yoon, T.-Y., X. Lu, J. Diao, S.-M. Lee, T. Ha, and Y.-K. Shin. 2008. Complexin and Ca2+ stimulate SNARE-mediated membrane fusion. Nat. Struct. Mol. Biol. 15 :707–713. 10.1038/nsmb.1446 18552825 Zhong, X., J.R. Liu, J.W. Kyle, D.A. Hanck, and W.S. Agnew. 2006. A profile of alternative RNA splicing and transcript variation of CACNA1H, a human T-channel gene candidate for idiopathic generalized epilepsies. Hum. Mol. Genet. 15 :1497–1512. 10.1093/hmg/ddl068 16565161

PMC005xxxxxx/PMC5460951.txt

==== Front J Gen Physiol J Gen Physiol jgp jgp The Journal of General Physiology 0022-1295 1540-7748 The Rockefeller University Press 28507080 201711762 10.1085/jgp.201711762 Research Articles Research Article 509 503 501 Divergent roles of a peripheral transmembrane segment in AMPA and NMDA receptors M4 segment in iGluR function Amin Johansen B. 138* http://orcid.org/0000-0003-0168-3205 Salussolia Catherine L. 48* Chan Kelvin 238 Regan Michael C. 9 Dai Jian 1011 http://orcid.org/0000-0001-9020-0302 Zhou Huan-Xiang 1011 Furukawa Hiro 9 Bowen Mark E. 5 http://orcid.org/0000-0002-8179-1259 Wollmuth Lonnie P. 678 1 Graduate Program in Cellular and Molecular Pharmacology, Stony Brook University, Stony Brook, NY 11794 2 Graduate Program in Neuroscience, Stony Brook University, Stony Brook, NY 11794 3 Medical Scientist Training Program (MSTP), Stony Brook University, Stony Brook, NY 11794 4 Department of Pediatrics, Pediatrics Residency Program, Stony Brook University, Stony Brook, NY 11794 5 Department of Physiology and Biophysics, Stony Brook University, Stony Brook, NY 11794 6 Department of Neurobiology and Behavior, Stony Brook University, Stony Brook, NY 11794 7 Department of Biochemistry and Cell Biology, Stony Brook University, Stony Brook, NY 11794 8 Center for Nervous System Disorders, Stony Brook University, Stony Brook, NY 11794 9 Cold Spring Harbor Laboratory, Cold Spring Harbor, NY 11724 10 Department of Physics, Florida State University, Tallahassee, FL 32306 11 Institute of Molecular Biophysics, Florida State University, Tallahassee, FL 32306 Correspondence to Lonnie P. Wollmuth: lonnie.wollmuth@stonybrook.edu * J.B. Amin and C.L. Salussolia contributed equally to this paper. 05 6 2017 149 6 661680 21 1 2017 01 5 2017 © 2017 Amin et al. 2017 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). AMPA and NMDA receptors are ionotropic glutamate receptors that make fundamental contributions to synaptic activity in the brain in different ways. Amin et al. show that their respective M4 segments, located on the periphery of their pore domains, contribute to their functional diversity. Ionotropic glutamate receptors (iGluRs), including AMPA receptor (AMPAR) and NMDA receptor (NMDAR) subtypes, are ligand-gated ion channels that mediate signaling at the majority of excitatory synapses in the nervous system. The iGluR pore domain is structurally and evolutionarily related to an inverted two-transmembrane K+ channel. Peripheral to the pore domain in eukaryotic iGluRs is an additional transmembrane helix, the M4 segment, which interacts with the pore domain of a neighboring subunit. In AMPARs, the integrity of the alignment of a specific face of M4 with the adjacent pore domain is essential for receptor oligomerization. In contrast to AMPARs, NMDARs are obligate heterotetramers composed of two GluN1 and typically two GluN2 subunits. Here, to address the function of the M4 segments in NMDARs, we carry out a tryptophan scan of M4 in GluN1 and GluN2A subunits. Unlike AMPARs, the M4 segments in NMDAR subunits makes only a limited contribution to their biogenesis. However, the M4 segments in both NMDAR subunits are critical for receptor activation, with mutations at some positions, most notably at the extreme extracellular end, completely halting the gating process. Furthermore, although the AMPAR M4 makes a minimal contribution to receptor desensitization, the NMDAR M4 segments have robust and subunit-specific effects on desensitization. These findings reveal that the functional roles of the M4 segments in AMPARs and NMDARs have diverged in the course of their evolution and that the M4 segments in NMDARs may act as a transduction pathway for receptor modulation at synapses. National Institutes of Health http://dx.doi.org/10.13039/100000002 NS088479 MH081923 NS093753 MH085926 GM105730 GM118091 ==== Body pmcIntroduction Ionotropic glutamate receptors (iGluR) are ligand-gated ion channels that mediate fast excitatory synaptic transmission in the mammalian nervous system. AMPA and NMDA receptors (AMPAR and NMDAR, respectively) are the two main postsynaptic iGluR subtypes. These receptors carry out different functions (Traynelis et al., 2010). AMPARs mediate fast synaptic currents and participate in rapidly translating presynaptic glutamate release into a postsynaptic signal. AMPARs also show strong desensitization in the continual presence of glutamate. NMDARs mediate a slower synaptic component impacting the electrical and biochemical activity of the postsynaptic neuron. The basis for these differences between AMPARs and NMDARs is incompletely understood (Mayer, 2016; Plested, 2016). iGluRs are members of the pore loop superfamily of ion channels, which includes two-transmembrane K+ channels, voltage-gated K+, Na+, and Ca2+ channels, and transient receptor potential channels. This superfamily is defined by a pore domain, consisting of two membrane-spanning segments joined by a nonmembrane-spanning pore loop. Upon tetramerization, the pore domains from individual subunits form the ion channel. In iGluRs, the pore domain consists of the M1 and M3 transmembrane segments and an M2 pore loop (Sobolevsky et al., 2009; Chen et al., 2014; Karakas and Furukawa, 2014; Huettner, 2015). Homologous to TM2 in two-transmembrane K+ channels, M3 contains elements of the activation gate (Chang and Kuo, 2008; Sobolevsky et al., 2009) and undergoes extensive rearrangement upon pore opening (Jones et al., 2002; Sobolevsky et al., 2002). In addition to a pore domain, all eukaryotic iGluRs have an additional transmembrane helix, the M4 segment (Fig. 1). For AMPARs, the M4 segment is critical to the transition from dimers to tetramers (Salussolia et al., 2011, 2013; Gan et al., 2015), in part by overcoming energy constraints imposed by other domains (Gan et al., 2016). In contrast to AMPARs, NMDARs are obligate heterotetramers composed of two GluN1 and typically two GluN2 subunits (Traynelis et al., 2010; Glasgow et al., 2015). Any role of the M4 segments in NMDAR assembly is uncertain (Meddows et al., 2001; Cao et al., 2011). In contrast, the NMDAR M4 segments may contribute to receptor gating. Deletion of the NMDAR M4 segments resulted in nonfunctional channels that could be rescued by coexpression of the M4 segments (Schorge and Colquhoun, 2003). Furthermore, the NMDAR M4 segments contain sites for ethanol modulation (Honse et al., 2004; Ren et al., 2012) and elements important for pore opening (Talukder et al., 2010). Highlighting the possible significance of the M4 segments to NMDAR function, numerous de novo missense mutations identified in the M4 segments of NMDAR subunits are implicated in neurological disorders (Lemke et al., 2013; Hamdan et al., 2014; Yuan et al., 2015; Hardingham and Do, 2016; Chen et al., 2017). Still, the functional role of the M4 segments in NMDARs and their relation to that in AMPARs is unknown. Figure 1. Structural features of AMPAR and NMDAR TMDs. (A) Comparison of AMPARs (GluA2, 3KG2, Sobolevsky et al., 2009) and NMDARs (model structure based on 4TLM of GluN1/GluN2B, see Materials and methods) lacking the intracellular CTD. Subunits are colored light orange (GluA2 A and C, GluN1) and gray 60% (GluA2 B and D, GluN2A). For iGluRs, individual subunits as well as the oligomeric complexes are composed of four highly modular domains. Two of these domains are positioned on the extracellular side of the membrane: the amino-terminal domain (ATD) and the LBD. The TMD spans the lipid bilayer and forms the ion channel; in both receptor subtypes at the level of the TMD, the M4 segment of one subunit is associated with the pore domain or ion channel core (M1–M3) of a neighboring subunit (cartoon, right). The fourth domain is the intracellular CTD. (B) View of the TMD for the model NMDAR structure from either the intracellular (bottom-up view) or extracellular (top-down view) side illustrating the association of M4 with an adjacent pore domain. For clarity, one GluN1 and one GluN2A M4 segment are represented as spheres, with VVLGAVE face positions highlighted in red. The center of the pore is indicated by a black dot. (far right) Illustration showing that residues occupying the VVLGAVE face in NMDAR subunits are aligned quite closely, mainly with the M3 segment of an adjacent subunit. The S1-M1 linker of the same subunit is also positioned closely to the extracellular portion of its M4 segments (not depicted). (C) Alignment of the M4 segments and residues on the N- and C-terminal sides for rat AMPAR and NMDAR subunits. Only three residues, a glycine (G), phenylalanine (F), and a glutamate (E), are completely conserved across all subunits (asterisks). Still, residues occupying the VVLGAVE face (boxed) tend to have comparable noncharged (valine [V], leucine [L], methionine [M]) or small (glycine [G], alanine [A], serine [S]) side chains with the exception of a threonine (T) at the VVLGAVE position. Structurally, the M4 segment of one subunit interfaces with the pore domain of a neighboring subunit (Fig. 1; Sobolevsky et al., 2009; Chen et al., 2014; Karakas and Furukawa, 2014). A specific face of the AMPAR M4 segment, termed VVLGAVE for the residues occupying it (Fig. 1, B and C), interacts with the pore domain, mainly the M3 transmembrane segment (Sobolevsky et al., 2009). The integrity of the entire extend of the VVLGAVE face is required for receptor tetramerization (Salussolia et al., 2013). Here, we make a direct comparison of the roles of the M4 segments in AMPAR and NMDAR function by contrasting the effects of a tryptophan scan in both subtypes. In AMPARs, such a scan identified the importance of the VVLGAVE face to receptor tetramerization (Salussolia et al., 2011, 2013). In contrast, we find that the M4 segments in NMDARs make a highly significant contribution to receptor gating, while any contribution to biogenesis (i.e., assembly and/or trafficking) is limited. Notably, the M4 segments in both iGluR subtypes are split structurally in terms of their functional impact. For AMPARs, although the M4 segment contributes to receptor assembly, this role is dramatically reduced at the extreme extracellular end of M4. For NMDARs, the extreme extracellular ends of the M4s in both GluN1 and GluN2A also show a functional distinction, having the most dramatic impact on receptor gating possibly reflecting their interaction with their own S1-M1 (Ogden et al., 2017). Materials and methods Molecular biology and cell culture All manipulations were made in rat GluN1 (GluN1a; NCBI Protein database accession no. P35439), GluN2A (Q00959), or GluA2 (P19491) subunits. In all instances, numbering included the signal peptide (GluN1, 18 residues; GluN2A, 19 residues; GluA2, 21 residues). In previous publications, we typically used numbering for the mature protein. Individual mutations were generated via QuikChange site-directed mutagenesis (Agilent Technologies) with XL1-Blue super-competent cells. pHmystik-tagged constructs pHmystik-pRK7 was a gift from M. Aurousseau and D. Bowie (Aurousseau, 2015). pHmystik generates a monomeric fluorescent protein that fluoresces in the blue-green spectrum. Purified pHmystik-pRK7 DNA was isolated via Mini Prep (QIAGEN) and confirmed by sequencing. Nhe1 restriction enzyme cloning sites were introduced via site-directed mutagenesis into GluN1 and GluN2A after the fourth amino acid of the mature protein (not including signal peptide). Nhe1 restriction enzyme sites were introduced at the 5′ and 3′ ends of the pHmystik insert in the pRK7 backbone. pHmystik-5′,3′ Nhe1, GluN1-Nhe1, and GluN2A-Nhe1 constructs were digested with Nhe1 in the presence of BSA, dephosphorylated with rSAP (NEB M0371S) and then separated on 0.8% agarose gel (1× TAE). Digested pHmystik and linearized GluN1 and GluN2A were extracted via Zymoclean gel-purification (Zymo D4001S). Purified pHmystik was inserted into the vector, with either GluN1 or GluN2A, using Roche rapid DNA ligation kit (04898117001; Roche). Ligation reaction of pHmystik-GluN1 and pHmystik-GluN2A was transformed into XL1-Blue super-competent cells (Agilent Technologies), and isolated colonies were picked for Mini Prep. Purified DNA of desired constructs were confirmed by DNA sequencing. Cell culture and transfection Human embryonic kidney 293 (HEK 293) or HEK 293T cells were grown in Dulbecco’s modified Eagle’s medium (DMEM), supplemented with 10% FBS, for 24 h before transfection. Non-tagged cDNA constructs were cotransfected into HEK 293 cells along with a separate pEGFP-Cl vector at a ratio of 4.5:4.5:1 (N1/N2/EGFP for NMDARs) or 9:1 (for AMPARs) using X-tremeGENE HP (Roche). pEGFP-Cl was not included in the transfection mix with GFP-tagged constructs, which were transfected at a 1:1 ratio. To improve cell survivability, HEK 293 cells transfected with NMDAR subunits were bathed in a media containing the NMDAR competitive antagonist APV (100 µM) and Mg2+ (100 µM; single-channel recording experiments) or the transfection mixture was replaced 4 h after transfection with fresh 5% FBS-DMEM culture media containing APV (100 µM) and Mg2+ (1 mM; whole-cell and imaging experiments). Cells transfected with AMPAR subunits were bathed in the competitive antagonist CNQX (10 µM). All experiments were performed 18–48 h after transfection. Molecular modeling The structural model for GluN1/GluN2B was modified from PDB 4TLM (Lee et al., 2014) by restoring the sequences of the N1 and N2B chains to the native Xenopus laevis sequences, containing residues 23–834 for N1 (chains A and C) and residues 26–839 for N2B (chains B and D). Specifically, mutations introduced for the crystallographic study were reverted to the native residues, and missing loop residues and missing atoms on other residues were added. These were all done by using MODELLER v9.10 (Šali and Blundell, 1993), except for one missing stretch (residues 482–502) in chain C, which was grafted from chain A. After adding all hydrogens, the model was slightly refined by 5,000 steps of energy minimization with backbone fixed and 2 ns of Langevin dynamics with backbone constrained. The refinement procedure was performed in NAMD 2.9 (Phillips et al., 2005) using the CHARMM36 force field (Mackerell et al., 2004). Fluorescence-detection size exclusion chromatography (FSEC) For FSEC experiments, we used GluA2(Q) tagged with EGFP at the C terminus (Sobolevsky et al., 2009; Salussolia et al., 2013). In brief, transfected cells were rinsed with PBS, pelleted, and resuspended in 300 µl solubilization buffer containing 20 mM Tris-HCl, 200 mM NaCl (TBS), 1% DDM (Affymetrix), and 1 mM PMSF. Cells were lysed using Misonix Sonicator 3000 (2 min cycle of 30 s at power level 2 followed by 30 s resting) and rotated for 2 h at 4°C before ultracentrifugation (TLA110 rotor) at 70,000 rpm for 10 min. A fraction of supernatant (100 µl) was loaded onto a Superose 6 column (10/300 GL; GE Healthcare), preequilibrated with TBS buffer containing 0.05% DDM and run at a flow rate of 0.4 ml/min. The eluent from the Superose 6 column was passed through a fluorometer with the following settings: excitation, 475 nm; emission, 507 nm; time increment, 0.5 s; integration time, 1 s; recording time, 0–4,500 s. Chromatograms for a given transfection cycle were normalized to the tetramer peak (occurring between 1,700 and 2,000 s) for wild type collected during the same run and were analyzed over ∼1,700–2,400 s using the multi-peak fitting routine in Igor Pro (WaveMetrics; Salussolia et al., 2015). An area under the curve (AUC) was defined for the tetramer (AUCT), dimer (AUCD), and monomer (AUCM) fractions, from which we derived the %tetramer = 100 × AUCT/(AUCT + AUCD + AUCM) (see Gan et al., 2016). Macroscopic current recordings Macroscopic currents in the whole-cell mode or outside-out patches, isolated from HEK 293 cells, were recorded at room temperature (20–23°C) using an EPC-9 amplifier with PatchMaster software (HEKA), digitized at 10 kHz and low-pass filtered at 2.9 kHz (−3 dB) using an 8 pole low pass Bessel filter (Yelshansky et al., 2004). Patch microelectrodes were filled with our standard intracellular solution (mM): 140 KCl, 10 HEPES, and 1 BAPTA, pH 7.2 (KOH). Our standard extracellular solution consisted of (mM) 140 NaCl, 1 CaCl2, and 10 HEPES, pH 7.2 (NaOH). For experiments examining the number of channels expressing in whole-cell patches, 0.01 mM EDTA was added to our standard extracellular solution to approximate conditions used in single channels. Pipettes had resistances of 2–6 MΩ when filled with the pipette solution and measured in the standard Na+ external solution. We did not use series resistance compensation nor did we correct for junction potentials. Currents were measured within 15 min of going whole cell. External solutions were applied using a piezo-driven double barrel application system. For NMDARs, one barrel contained the external solution +0.1 mM glycine, whereas the other barrel contained the same solution +1 mM glutamate. For AMPARs, we did not include glycine and 3 mM glutamate was applied instead of 1 mM. The open tip response (10–90% rise time) of the application system was between 400 and 600 µs. For display, NMDAR currents were digitally refiltered at 500 Hz and resampled at 1 kHz, and AMPARs at 1 kHz and at 2 kHz. Assaying surface expression using whole-cell currents We recorded whole-cell currents for each construct a minimum of two different transfection cycles (Salussolia et al., 2011). Constructs where we could not detect glutamate-activated currents were recorded on at least one additional transfection cycle. Wild type was recorded at minimum every other transfection cycle. A caveat of using whole-cell currents to assay surface expression is a bias in cell selection: cells that express a high number of normal or high functioning NMDARs are more likely to die. Hence, when selecting cells for recording or imaging, we may be selecting cells that on average have fewer receptors expressed in the membrane. This issue will be most challenging for those constructs that show increased surface expression and/or show strongly enhanced gating properties. Most of our efforts focused on constructs that showed reduced current amplitudes. Determining desensitization To determine the extent and rate of desensitization, we applied glutamate at −70 mV for 100 ms to outside-out patches (AMPARs) or 2.5 s in the whole-cell mode (NMDARs). Percent desensitization (%des) was calculated from the ratio of peak (Ipeak) and steady-state (Iss) current amplitudes: %des = 100 × (1 − Iss/Ipeak). Time constants of desensitization (τdes) were determined by fitting the decaying phase of currents to either a single (AMPARs) or double (NMDARs) exponential function. In some instances when current amplitudes were small, we averaged three to five records. Single-channel recordings Single-channel recordings were performed in the on-cell configuration at steady-state using transfected HEK 293 cells at room temperature. The pipette solution mimicked extracellular conditions and contained (mM) 150 NaCl, 10 HEPES, 0.05 EDTA, 1 glutamate, and 0.1 glycine, pH 8.0 (NaOH). High pH and EDTA were used to minimize proton and divalent mediated inhibitory effects (Popescu and Auerbach, 2003). Recording pipettes were pulled from thick-wall borosilicate capillary glass (Sutter Instrument) and fire polished to final pipette resistances ranging from 5 to 27 MΩ. Cells were patched to resistances exceeding 1.5 GΩ. To elicit distinct inward current amplitudes, we held the electrode voltage at 100 mV. Currents were recorded using a patch clamp amplifier (Axopatch 200B; Molecular Devices), filtered at 10 kHz (four-pole Bessel filter), and digitized at 40 kHz (ITC-16 interfaced with PatchMaster). Experiments ran for ∼3–20 min to ensure a significant number of events for analysis. Analysis of single-channel records was comparable with Talukder and Wollmuth (2011). In brief, recordings were exported from PatchMaster to QuB (http://www.qub.buffalo.edu). Processed data were idealized using the segmental k-means (SKM) algorithm in QuB with a dead time of 40 µs (Qin, 2004). From the idealization, we derived single-channel current amplitudes, mean closed time (MCT) and mean open time (MOT) as well as an equilibrium open probability (eq. Po). Eq. Po is the fractional occupancy of the open states in the entire single-channel recording, including long-lived closed states. For recordings of wild-type GluN1/GluN2A, detection of single-channel patches was straightforward because this construct has a relatively high Po and recordings were made for a long duration (∼10,000–500,000 events). Certain constructs showed an extremely low Po and MOT. Although we did not detect any overlap in channel openings, we cannot rule out that these patches did not contain more than one channel. However, this issue is not critical to the present study because we did not do kinetic analysis on these patches. Assaying surface expression using pH-sensitive GFP HEK 293T cells were plated at a density of 1.2 × 105 on 35-mm glass-bottom dishes and were transfected as described in Cell culture and transfection section above. Samples were imaged 24–48 h after transfection using a Nikon Ti Eclipse microscope equipped with the Nikon TIRF slider. Excitation used the 488-nm line from a fiber-coupled argon Ion laser (Lasos). The fluorescence emission was collected with a Nikon 60× 1.45 NA oil-immersion TIRF objective and relayed to an iXon DU897 emCCD camera (Andor Technologies). The fluorescence emission was transmitted using a dual band 488/561 TIRF filter cube with a GFP band pass of 525/50 (Chroma). Cells were observed at 5-s intervals with 30–50 exposures collected for each field of view, which contained one to four fluorescent cells. Cells were imaged in either a bath solution at pH 7.4, consisting of (mM) 140 NaCl and 10 HEPES, pH 7.4 (NaOH); or a bath solution at pH 5.5, consisting of (mM) 140 NaCl and 30 Mes, pH 5.5 (HCl). These bathing solutions were exchanged using a continuous flow perfusion system. Fluorescence intensity was quantified in ImageJ (National Institutes of Health; Schneider et al., 2012). Cells were selected for analysis that were at least 50% isolated from other cells. The fluorescence intensity (F) of a cell of interest was calculated for each frame: F = Fcell − Fbackgnd, where Fcell is the mean fluorescence of the cell, defined by a polygon circumscribing it, and Fbackgnd is the mean fluorescence of an acellular region directly adjacent to the cell of interest. For display and analysis, we normalized fluorescence intensity to a baseline fluorescence (Fo), which was the mean F just before the solution was changed from pH 7.4 to 5.5. The change in fluorescence (ΔF) was defined as ΔF = Fo − Ftest, where Ftest is the fluorescence intensity taken 15–30 s after the solution was switched from pH 7.4 to 5.5. We defined “detectable” surface expression as occurring when the fluorescence intensity decreased rapidly upon switching to pH 5.5 and returned to baseline with a return to pH 7.4. For some constructs, we detected both positive and negative (no detectable ΔF); in these instances, we averaged surface expression only for those cells that were positive. Statistics Data analysis was performed using Igor Pro, QuB, Excel, and MiniTab 17. All average values are presented as mean ± SEM. The number of replicates are indicated in the figure legend or in a table associated with the figure. In instances where we were only interested in whether outcomes were statistically different from that for wild type, we used an unpaired two-tailed Student’s t test to test for significant differences. Statistical significance was typically set at P < 0.05. In instances where we were interested in how constructs varied from each other, we used an ANOVA and followed with Tukey’s test (P < 0.05). Online supplemental material Three supplemental figures and two supplemental tables are included. Fig. S1 gives additional example FSEC chromatographs. Fig. S2 shows additional analysis of the interaction of the NMDAR M4 segments with other transmembrane segments. Fig. S3 shows control experiments with pHmystick constructs. Table S1 contrasts side chain properties of amino acids substituted in the GluA2 M4 segment. Table S2 shows single-channel data for positions that showed significant current potentiation, an issue we do not explore in the main text. Results Subtle mutations in the AMPAR M4 segment highlight its key role in receptor assembly Tryptophan has a large and bulky side chain, which can disrupt transmembrane interactions (Soler-Llavina et al., 2006). For AMPARs, tryptophan substitutions of the M4 segment identified that a face (VVLGAVE) aligned with the pore domain (M1–M3) of an adjacent subunit (Fig. 1, B and C; Sobolevsky et al., 2009) disrupts receptor tetramerization (Salussolia et al., 2011, 2013). To further define the significance of the M4/ion channel core interactions to AMPAR function, we made more subtle mutations in the VVLGAVE face in the unedited GluA2 subunit tagged on the C terminus with EGFP (GluA2(Q)-EGFP). These subtle substitutions, rather than drastically changing the volume of the side chain as done by tryptophan, typically preserved the nature of the side chain while adding or removing a methyl group and hence changed the side chain volume in a more limited fashion (Table S1). Further, as we will show later, receptors containing these subtle substitutions still functioned, permitting us to test whether the VVLGAVE face has an impact on receptor gating, specifically desensitization. We quantified the effect of these subtle mutations on receptor assembly using FSEC (Fig. 2; Kawate and Gouaux, 2006; Salussolia et al., 2013). Figure 2. FSEC to assay AMPAR oligomerization. (A and B) Raw FSEC chromatographs of wild-type GluA2(Q)-EGFP (A) or the same construct containing an alanine substitution at G823 (GluA2(Q)(G823A); B). For quantification, we normalized all chromatographs to the tetramer peak in the wild type for that transfection cycle. For the top panel in each plot, the black line is the original data, the green line the baseline, and the red line is the sum of the individual fits derived from a multi-peak fitting routine (see Materials and methods); the bottom panels show the fraction of the total chromatograph corresponding to, from left to right, tetramer, dimer, or monomer. (C) Mean (±SEM) of the %tetramer (see Materials and methods) for various substitutions of VVLGAVE face positions of the M4 segment in GluA2(Q). Black bars indicate values significantly different from wild type (P < 0.05, ANOVA); asterisks indicate values significantly different from subtle mutations at V813 and V816. Number of independent FSEC runs: wt, 16; V813W, 3; V813A, 4; V813L, 4; V813I, 3; V816A, 4; V816L, 3; V816I, 4; L820A, 4; L820F, 6; G823A, 5; A827S, 3; A827V, 4; V830A, 3; V830L, 3; V830I, 3; E834Q, 6; E834D, 7; E834N, 3. (D) GluA2 M4 segment with VVLGAVE positions shown in red space–filled balls. On FSEC, wild-type GluA2 receptors predominantly give tetramers with a small fraction of dimers and monomers (Fig. 2 A). To quantify these results, we fit these curves with a polynomial to derive the tetramer, dimer, and monomer fraction (Fig. 2 A, bottom trace) from which we calculated the %tetramer (see Materials and methods; Salussolia et al., 2015; Gan et al., 2016). For GluA2(Q) homomers, this approach yielded a %tetramer of 63 ± 3%, n = 16 (mean ± SEM, n = number of independent runs), a result comparable with that published previously for homomeric AMPARs (Greger et al., 2003; Salussolia et al., 2013; Gan et al., 2016). At the core of the VVLGAVE face is a GxxxG transmembrane–interacting motif (Moore et al., 2008). Substitution of alanine (A) for the glycine (G; G823) dramatically attenuates the extent of tetramerization (Fig. 2 B), reducing the %tetramer to 8 ± 1%, n = 5. Indeed, subtle substitutions at many of the positions in the VVLGAVE face, specifically from L820 inward, significantly attenuated tetramerization (Fig. S1 and Fig. 2 C, solid bars indicate significance relative to wild type). For example, changing side chains at the most intracellular position, a glutamate (E) at 834, with either a smaller, charged side chain (aspartate, D) or an uncharged side chain (glutamine, Q; or asparagine, N) in all instances significantly reduced the %tetramer. In contrast, subtle substitutions at the extreme extracellular positions, V813 and V816, had no significant effects on receptor assembly, though tryptophan substitutions at these same positions essentially eliminated tetramerization (for V813, Fig. 2 C; Salussolia et al., 2013). Thus, although all positions in the VVLGAVE face impact receptor assembly in AMPARs based on tryptophan substitutions, these actions based on subtle substitutions are more critical at positions intracellular to V816, encompassing L820 and inward. Tryptophan scan of the NMDAR M4 segments indicates differences with AMPARs Because of the robust effects of tryptophan substitutions in AMPARs (Salussolia et al., 2011), we performed a similar scan in NMDARs, substituting individual positions in the GluN1 and GluN2A M4 segments with tryptophan. As an initial means to assay the functional role of the M4 segments in NMDARs, we measured whole-cell currents for these constructs (Fig. 3 and Table 1) in a manner comparable with what was done previously (Salussolia et al., 2011) to facilitate direct comparisons (see Materials and methods for limitations of this approach). Figure 3. Tryptophan mutagenesis scan of the M4 segments in the GluN1 or GluN2A subunits. (A) Representative whole-cell recordings of current through wild-type GluN1/GluN2A, GluN1(M813W)/GluN2A, or GluN1/GluN2A(S831W). Glutamate (1 mM, shaded box) was applied for 2.5 s. Cells were continuously bathed in glycine (0.1 mM). Holding potential, −70 mV. (B) Mean current amplitudes (±SEM) at −70 mV normalized to current amplitudes for wild-type GluN1/GluN2A (−750 ± 50, n = 18; raw values shown in Table 1). Positions that did not show detectable glutamate-activated currents are demarcated by an “X” (see Materials and methods). Positions that showed current amplitudes significantly less or greater than wild-type are colored blue and green, respectively (P < 0.05, two-tailed Student’s t test, unpaired). The red dots highlight positions homologous to the VVLGAVE face in AMPARs. (C) Orientation of the M3 segments (shadowed), viewed from the center of the pore, to the M4 segments in either AMPARs (left) or NMDARs either GluN2B M3 relative to GluN1 M4 (middle) or GluN1 M3 relative to GluN2B M4 (right). Positions that showed significant changes in current amplitudes are highlighted in blue (reduced), green (greater), or red (no detectable current). For AMPARs, only those positions that showed no detectable current (Salussolia et al., 2011) are indicated, which include the VVLGAVE positions as well as I819. Table 1. Raw macroscopic peak current amplitudes, shown as normalized values in Fig. 3 B, for wild-type GluN1/GluN2A or GluN1/GluN2A containing tryptophan substitutions in either GluN1 or GluN2A M4 segments Tryptophans in GluN1 subunit Tryptophans in GluN2A subunit Construct Ipeak n Construct Ipeak n pA pA N1/N2A −750 ± 50 18 F810W −830 ± 290 4 I814W −680 ± 125 4 E811W −720 ± 95 4 D815W −580 ± 110 4 N812W −28 ± 6* 4 N816W ND 9 M813W −65 ± 9* 5 M817W −17 ± 3* 6 A814W −530 ± 130 4 A818W −1,160 ± 175 5 G815W −330 ± 90 4 G819W −11 ± 2* 6 V816W −250 ± 30* 4 V820W ND 8 F817W −1,150 ± 240 5 F821W −370 ± 35 5 M818W −355 ± 110 6 Y822W −1,500 ± 120 4 L819W −2,010 ± 210^ 4 M823W −2,090 ± 100^ 5 V820W −1,830 ± 280 4 L824W −1,030 ± 150 4 A821W −760 ± 110 4 A825W −860 ± 120 4 G822W −1,050 ± 220 4 A826W −560 ± 65 5 G823W −730 ± 110 4 A827W −1,680 ± 205 5 I824W −730 ± 150 5 M828W −1,110 ± 150 5 V825W −330 ± 20 4 A829W −890 ± 130 6 A826W −880 ± 170 4 L830W −1,830 ± 470 5 G827W −78 ± 8* 4 S831W ND 9 I828W −720 ± 90 4 L832W −760 ± 90 5 F829W −960 ± 130 5 I833W −1,340 ± 100 4 L830W −410 ± 45 4 T834W −1,000 ± 160 4 I831W −670 ± 60 4 F835W −1,020 ± 190 4 F832W −1,290 ± 220 5 I836W −1,320 ± 160 5 I833W −1,110 ± 130 4 W837 −750 ± 50 18 E834W −100 ± 10* 6 E838W −265 ± 40* 6 Values shown are mean ± SEM. n indicates the number of whole-cell recordings. Peak current amplitudes were recorded in the whole-cell mode at −70 mV (e.g., Fig. 3 A). GluN1 constructs were co-expressed with wild-type GluN2A and vice-versa. Tagged values are significantly less (*) or greater (^) than wild-type (P < 0.05, two-tailed Student’s t test, unpaired). ND, no glutamate-activated currents detected. As done previously (Salussolia et al., 2011), constructs were tested on at least two different transfection cycles with wild type tested every other transfection cycle. Constructs that showed no detectable glutamate-activated current were tested on at least one additional transfection cycle. Wild-type GluN1/GluN2A shows robust glutamate-activated currents (Fig. 3 A, left; −750 ± 50 pA, n = 18; n = number of whole-cell recordings). Most other tryptophan-substituted constructs also showed robust glutamate-activated current, though certain constructs showed either significantly reduced peak current amplitudes (e.g., GluN1(M813W)/GluN2A; Fig. 3 A, middle) or no detectable glutamate-activated membrane current (e.g., GluN1/GluN2A(S831W); Fig. 3 A, right). To compare current amplitudes, we normalized current amplitudes to those for wild type (Fig. 3 B and Table 1). Constructs are highlighted as to whether they showed significantly reduced current amplitudes (blue), significantly greater current amplitudes (green) or no detectable current (X, red) relative to wild type (P < 0.05, t test). This experiment suggests several initial conclusions. First, compared with similar experiments for AMPARs, where 9 out of 23 tested positions in the M4 segment showed no detectable current (Salussolia et al., 2011), the number of positions with no detectable current in NMDARs is notably less, only 3 (N816, V820, and S831, all within the GluN2A subunit) out of a total of 25 homologous positioned tested on either GluN1 or GluN2A. Second, when significantly reduced or no detectable current did occur in NMDAR subunits, it almost always occurred (4 of 5, GluN1; 4 of 6, GluN2A) at positions homologous to the AMPAR VVLGAVE face (red dots). One possible explanation for the reduced number of positions showing no detectable current in NMDARs is that the tested constructs had substituted tryptophans in only two of the four subunits, whereas in homomeric AMPARs all four subunits possessed the substitution. In terms of transmembrane interactions, substitutions of M4 positions pointing toward M1 had no significant effect on current amplitudes (Fig. S2). In contrast, the majority of M4 positions that showed significant changes in current amplitudes are homologous to the AMPAR VVLGAVE and like the AMPAR VVLGAVE face point toward the M3 segments (Fig. S2). We therefore focused on positions homologous to the AMPAR VVLGAVE face and coexpressed GluN1 and GluN2A subunits with tryptophans substituted at homologous positions, corresponding to the VVLGAVE face, and again tested whole-cell current amplitudes (Table 2). We did not test two combinations in which any individual tryptophan substitution yielded no detectable current. Of the other five mutation pairs tested, three again showed high current amplitudes. The combination between the most extracellular and intracellular positions, GluN1(M813W)/GluN2A(M817W) and GluN1(E834W)/GluN2A(E838W), respectively, showed no detectable current. Thus, even when both subunits have tryptophans substituted for positions homologous to the AMPAR VVLGAVE face, any effect on whole cell is less notable than that for AMPARs. Table 2. Macroscopic current amplitudes for GluN1/GluN2A containing tryptophan substitutions in the VVLGAVE face in the GluN1 and GluN2A M4 segments Construct Ipeak n Comments pA N1/N2A −750 ± 50 18 V N1(M813W)/N2A(M817W) ND 9 V N1(V816W)/N2A(V820W) NT - N1/N2A(V820W) showed no current L N1(V820W)/N2A(L824W) −340 ± 35 4 G N1(G823W)/N2A(A827W) −535 ± 60 4 A N1(G827W)/N2A(S831W) NT - N1/N2A(S831W) showed no current V N1(L830W)/N2A(T834W) −570 ± 140 4 E N1(E834W)/N2A(E838W) ND 6 Values shown are mean ± SEM. Currents recorded as in Fig. 1. n indicates the number of whole-cell recordings. None of the constructs that showed current were significantly different from wild type (P < 0.05, two-tailed Student’s t test, unpaired). NT, not tested. ND, no glutamate-activated currents detected. In AMPAR subunits, the “VVLGAVE” face positions are aligned with the adjacent M3 segment (Fig. 3 C, left; Sobolevsky et al., 2009). For NMDAR M4 segments, the homologous positions are also generally aligned with the adjacent M3 segments (Fig. 3 C, middle and right). Positions L830 in GluN1 and T834 in GluN2A (VVLGAVE) appear oriented to the side of M3, possibly accounting for the lack of an effect at this position. However, G823 in GluN1 and A827 in GluN2A (VVLGAVE) as well as V820 in GluN1 and L824 in GluN2A (VVLGAVE) are oriented toward M3, yet show no change in function when both are substituted with tryptophan (Table 2). Three positions in NMDAR subunits, L819 in GluN1 and Y822 and M823 in GluN2A, which point away from the M3 segment, showed significant current potentiation when substituted with tryptophan (Fig. 3 B). The homologous position to L819 and M823 in AMPARs, I819, showed no detectable current when substituted with tryptophan (Salussolia et al., 2011). In summary, assuming that a major effect of a tryptophan substitution is to disrupt transmembrane interactions, the interaction of the M4 segments in NMDAR subunits with other transmembrane segments affects receptor function, though this action is considerably less extensive than in AMPARs. Many nonfunctional tryptophan-substituted NMDARs express robustly on the cell surface For AMPARs, the lack of membrane currents for receptors containing tryptophan substitutions of the VVLGAVE face reflects an inability to tetramerize (Salussolia et al., 2011, 2013), thereby preventing surface expression. To discern the mechanistic basis of significantly reduced current amplitudes in tryptophan-substituted NMDARs, we initially assayed surface expression of these constructs. To do so, we tagged NMDAR subunits at the extracellular N-terminal end with a pH-sensitive GFP, pHmystik (Aurousseau, 2015). We used total internal reflection fluorescence (TIRF) to assay cellular fluorescence while changing the extracellular pH from 7.4 to 5.5. At pH 5.5, the fluorescence of surface-expressed pHmystik is nearly fully quenched in contrast to other pH-sensitive GFP constructs (Aurousseau, 2015). Wild-type receptors, either pHmystik-GluN1/GluN2A (Fig. S3 A) or GluN1/pHmystik-GluN2A (Fig. 4 A), showed robust and rapid changes in fluorescence when the external solution was switched from pH 7.4 to 5.5. This fluorescence change was reversible upon switching back to pH 7.4. No rapid change in fluorescence occurred when pHmystik-GluN1 (−0.02 ± −0.02, n = 16) or pHmystik-GluN2A (−0.02 ± −0.01, n = 8) was expressed alone (Fig. S3 B), though a slow change did occur with very long exposures (>50 s) to pH 5.5 (Fig. S3 C). Thus, we used the rapid and reversible change in fluorescence (ΔF; e.g., points i and ii in Fig. 4 A) as an index of surface-expressed NMDARs. Figure 4. Assaying surface expression of NMDARs using pHmystik, a pH-sensitive GFP. (A–D) Fluorescence time trace of GFP intensity as the extracellular bath solution pH was changed from 7.4 to 5.5 (gray bar, 30-s duration) and back again. Shown are HEK293T cells expressing GluN1/pHmystick-GluN2A (A), pHmystik-GluN1/GluN2A(N816W) (B), GluN1(G827)/pHmystick-GluN2A (C), or pHmystik-GluN1/GluN2A(S831W) (D). The representative cell images are from the time-points labeled in panel A; the images were taken at the approximate time points when the baseline fluorescence (Fo), corresponding to image (i), and the test fluorescence (Ftest), corresponding to image (ii), were measured as well as an image (iii) after return to pH 7.4. The change in fluorescence (ΔF = Fo − Ftest) was used as an index of surface expression. Sampling rate, 5 s. (E) Changes in cell fluorescence (±SEM) at low pH (ΔF, see Materials and methods) for positions in either GluN1 (left) or GluN2A (right). Solid bars indicate values significantly different from their respective wild type either N1/pHmystik-N2A (left) or pHmystik-N1/N2A (right; P < 0.05, t test). The numbers (far right in each plot) indicate the number of cells that showed detectable changes in surface expression relative to the total number of cells tested (see Materials and methods). Only cells that showed detectable changes in fluorescence were included in statistical analysis. (F) Relationship between surface expression, assayed by changes in fluorescence, and whole-cell current amplitudes. For comparison, values (±SEM) are normalized to their respective wild type, either for surface expression (ΔFnorm) or for peak whole-cell current amplitudes (Ipeak norm; Fig. 3 B). For this ratio of normalized values, a number close to unity implies a correlation between the two parameters, whereas a number much greater than unity indicates surface expression is much greater than expected from whole-cell currents. Although both pHmystik-GluN1/GluN2A and GluN1/pHmystik-GluN2A showed robust surface expression, the expression was significantly greater for GluN1/pHmystik-GluN2A (0.63 ± 0.01, n = 17) than for pHmystik-GluN1/GluN2A (0.48 ± 0.01, n = 28). GluN1/pHmystik-GluN2A also showed considerably higher whole-cell current amplitudes (−2,050 ± 310 pA, n = 5) than pHmystik-GluN1/GluN2A (−930 ± 100 pA, n = 7). Compared with wild-type constructs, surface expression for NMDARs containing tryptophan substitutions that were associated with reduced or no detectable glutamate-activated current fell into three categories (Fig. 4, B–E): (1) Surprisingly, despite showing either reduced or no whole-cell current, several positions in the extreme extracellular portion of the M4 (N1(N812W), N1(M813W), N1(V816W), N2A(N816W) [Fig. 4 B], N2A(M817W), N2A(G819W)) showed surface expression comparable with their respective wild-type construct (Fig. 4 E). In contrast, positions in the intracellular portion of the M4 showed (2) significantly reduced surface expression, including N1(G827W) (Fig. 4 C), N1(E834W), and N2A(E838W) or (3) no detectable surface expression, including N2A(S831W) (Fig. 4 D) and pHmystik-N1(E834W)/N2A(E838W) (−0.02 ± −0.04, n = 20). Distinct features of NMDAR M4 extreme extracellular and intracellular portions In contrast to AMPARs containing tryptophan substitutions in the VVLGAVE face, NMDARs containing comparable substitutions in the extreme extracellular portion of the M4s (at or near VVLGAVE) showed robust surface expression even when they displayed reduced or no detectable membrane currents (Fig. 3 B, Table 2, and Fig. 4 E). To define the relationship between surface expression as assayed by pHmystik (Fig. 4 E) and whole-cell current amplitudes (Fig. 3 B), we plotted a ratio of responses normalized to that for wild type (Fig. 4 F): surface expression normalized to the respective wild type (ΔFnorm) over whole-cell current amplitudes normalized to wild type (Ipeak norm; Fig. 3 B). A number close to unity indicates a good correspondence between surface expression and whole-cell current amplitudes; such a correspondence would suggest that the underlying single-channel activity of the construct is comparable with that for wild type. A ratio of normalized responses much greater than unity reflects that there is more surface expression than anticipated based on whole-cell currents, suggesting impaired channel function. Positions on the extreme extracellular portion of M4 in both GluN1 and GluN2A showed values much greater than unity, whereas intracellular positions showed values close or less than unity. There are two conclusions from this plot of the ratio of normalized responses. First, tryptophan substitutions in the extreme extracellular portion of M4 still show robust surface expression (Fig. 4 E), much greater than that anticipated for whole-cell currents (Fig. 4 F). The most likely explanation for this is that these constructs can assemble and traffic to the membrane, like wild type, with the reduced or no detectable glutamate-activated currents probably caused by changes in receptor gating. Second, there is good correspondence between surface expression and whole-cell currents for intracellular positions (N1(G827), N1(E834), N2A(S831), N2A(E838); Fig. 4 F); because both surface expression and whole-cell currents show reduced or no expression, there appears some restriction for these constructs to get to the membrane, either a reduced receptor assembly and/or membrane trafficking. Thus, based on these considerations, there appears to be a dichotomy in the M4 segments in NMDARs with the extreme extracellular (N-terminal) and intracellular (C-terminal) portions taking on different functional roles. Extreme extracellular M4 positions in NMDAR subunits impact receptor gating The data in Fig. 4 based on tryptophan substitutions suggest that rather than participating in receptor assembly, as for AMPARs, the extreme extracellular portions of the M4 segments in NMDARs contribute to ion channel gating. To test this idea directly, we used single-channel recordings of tryptophan substitutions in either GluN1 or GluN2A that had significantly reduced whole-cell current amplitudes. We made these recordings in the on-cell mode at pH 8.0 and in the absence of divalents (see Materials and methods; Popescu and Auerbach, 2003; Talukder and Wollmuth, 2011). The single-channel profile for wild-type GluN1/GluN2A receptors, under the present recording conditions, showed robust activity (Fig. 5 A, top), with a mean single-channel current amplitude (i) around −7.2 pA (−7.2 ± 0.1 pA, n = 10) and an equilibrium open probability (eq. Po) of ∼0.7 (0.69 ± 0.04; Fig. 5 B and Table 3). Tryptophan substitutions in the extreme extracellular portion of M4, for example GluN1(M813W)/GluN2A (Fig. 5 A, middle), showed dramatic effects on receptor gating. These extracellular positions, N812, M813, and V816 in GluN1 and M817 and G819 in GluN2A, all showed significantly reduced eq. Po (Fig. 5 B and Table 3). The most consistent basis for this reduced open probability was a dramatic reduction in the MOT (Table 3). In wild type, MOT is ∼5.5 ms (5.5 ± 0.5 ms), whereas in these extracellular tryptophan substitutions it was reduced to <0.4 ms (Fig. 5 B and Table 3). For most of these constructs, there was also a trend for an increased MCT, though this was only consistently significant in the GluN2A positions (Table 3). Many of these same constructs also showed significantly reduced single-channel current amplitudes (Fig. 5 B and Table 3), though because of the extremely brief openings these amplitudes may reflect incomplete openings. In contrast, tryptophan substitutions of intracellular M4 positions, for example GluN1(E834W)/GluN2A (Fig. 5 A, bottom), had typically only weak or no effect on single-channel activity (Fig. 5 B and Table 3). Figure 5. Single-channel recordings of wild-type NMDARs and NMDARs with tryptophan substitutions that showed significantly reduced whole-cell current amplitudes. (A) Example single-channel recordings of GluN1/GluN2A, GluN1(M813W)/GluN2A, or GluN1(E834W)/GluN2A. Recordings were performed in the cell-attached configuration with a pipette potential of 100 mV. Downward deflections reflect inward currents. For each construct, the top half shows a low-resolution example (filtered at 1 kHz), and the bottom half shows a higher-resolution portion of the same record (filtered at 3 kHz). (B) Single-channel current amplitudes (top) and equilibrium open probability (eq. Po; bottom) shown as mean ± SEM (Table 3). Solid bars indicate values significantly different from wild type (P < 0.05, t test). (C) Approximate number of ion channels in the membrane mediating steady-state whole-cell current amplitudes (Table 4). Table 3. Single-channel properties of wild-type GluN1/GluN2A or tryptophan-substituted GluN1 or GluN2A NMDARs where whole-cell currents were significantly changed Construct Total events (# of patches) i eq. Po MCT MOT pA ms ms N1/N2A 1,531,680 (10) −7.2 ± 0.1 0.69 ± 0.04 2.3 ± 0.2 5.5 ± 0.5 N1(N812W) 147,374 (5) −4.1 ± 0.5* 0.03 ± 0.01* 31.8 ± 10.3* 0.40 ± 0.07* N1(M813W) 484,988 (4) −4.4 ± 0.4* 0.04 ± 0.01* 7.3 ± 1.5* 0.28 ± 0.05* N1(V816W) 475,752 (4) −5.3 ± 0.7* 0.02 ± 0.01* 7.2 ± 2.2 0.13 ± 0.01* N1(G827W) 218,498 (4) −7.6 ± 0.5 0.30 ± 0.05* 11.6 ± 1.9* 4.7 ± 0.8 N1(E834W) 198,394 (4) −7.0 ± 0.3 0.64 ± 0.07 3.8 ± 0.8 6.5 ± 0.6 N2A(M817W) 20,985 (5) −8.0 ± 0.2 0.001 ± 0.0005* 213 ± 73* 0.20 ± 0.10* N2A(G819W) 27,892 (5) −5.4 ± 0.3* 0.005 ± 0.001* 73.8 ± 7.1* 0.33 ± 0.03* N2A(E838W) 81,550 (4) −6.6 ± 0.4 0.54 ± 0.08 9.1 ± 3.4 8.8 ± 0.6* Values shown are mean ± SEM for single-channel current amplitude (i), equilibrium open probability (eq. Po), MCT, and MOT. Single-channel currents were recorded in the on-cell mode at approximately −100 mV and analyzed in QuB (see Materials and methods). Number of patches is in parenthesis to the right of total events. Eq. Po is the fractional occupancy of the open states in the entire single-channel recording, including long-lived closed states. All data were idealized and fit at a dead time of 40 µs. Asterisks indicate values significantly different from wild type (P < 0.05, t test). The results in Figs. 4 F and 5 B suggest that the major effect of tryptophan substitutions at extreme extracellular M4 positions is to disrupt gating, whereas for intracellular positions it disrupts receptor biogenesis. To further test this idea, we estimated the number of ion channels in the membrane mediating whole-cell currents. To do so, we measured steady-state whole-cell current amplitudes under extracellular conditions that approximated those for which single-channel activities were recorded: no added divalents and 0.01 mM EDTA though at pH 7.2 (Table 4). As for peak current amplitudes (Fig. 3 B), steady-state current amplitudes measured under these conditions and relative to wild type were also significantly reduced (Table 4). Table 4. Whole-cell current amplitudes in EDTA for wild-type GluN1/GluN2A or tryptophan-substituted GluN1 or GluN2A NMDARs where whole-cell currents were significantly reduced Construct Ipeak Isteady-state n Estimated channel number (N) pA pA N1/N2A −1,170 ± 70 −760 ± 50 13 220 ± 20 N1(N812W) −350 ± 40* −96 ± 12* 7 1,340 ± 390 N1(M813W) −590 ± 90 −220 ± 30* 7 1,810 ± 340 N1(V816W) −120 ± 20* −80 ± 12* 5 920 ± 210 N1(G827W) −150 ± 30* −54 ± 13* 5 34 ± 7 N1(E834W) −270 ± 40* −150 ± 30* 5 48 ± 7 N2A(M817W) −35 ± 4* −27 ± 3* 5 3,830 ± 330 N2A(G819W) −26 ± 6* −5 ± 1* 5 280 ± 60 N2A(E838W) −420 ± 50* −230 ± 30* 4 33 ± 5 Values shown are mean ± SEM. Tagged values are significantly less (*) or greater (^) than wild type (P < 0.05, t test). n indicates the number of whole-cell recordings. Currents were recorded in the whole-cell mode at a holding potential of −70 mV, as in Table 1, but with our standard extracellular solution (pH 7.2) also containing 0.01 mM EDTA. These conditions were aligned to those for on-cell single-channel recording except for the pH. The estimated channel number mediating the observed steady-state whole-cell current amplitudes (Isteady-state) was derived from: Isteady-state = N * Po * i, where N is the approximate number of channels on the membrane, Po is the equilibrium Popen, and i is single-channel current amplitude corrected for difference in membrane potential assuming Ohmic behavior (Table 3). We estimated the approximate number of ion channels in the membrane mediating steady-state whole-cell current amplitudes (Isteady-state; Table 4) using the following relationship: Isteady-state = N * Po * i, where N is the approximate number of channels on the membrane, Po is the equilibrium open probability, and i is single-channel current amplitude (Table 3). For wild type, there was on average ∼200 channels on the membrane (Fig. 5 C and Table 4). For extreme extracellular positions in GluN1 (N812, M813, and V816) or GluN2A (M817 and G819), the estimated number of ion channels on the membrane was at least as high as wild type, if not considerably higher, indicating that reduced whole-cell currents is completely caused by a gating deficit rather than a disruption in receptor biogenesis. In contrast, for intracellular positions, N1(G827), N1(E834W), and N2A(E838W), the estimated number of ion channels was consistently lower than that for wild type. In summary, these results strongly support the distinction between the extreme extracellular and intracellular portions of the NMDAR M4 segments. The extreme extracellular M4 in both GluN1 and GluN2A modifies receptor gating with little or no contribution to biogenesis, whereas the intracellular M4 appears to have a role in biogenesis, either assembly and/or trafficking to the membrane. Recovery of function in pore-dead constructs Tryptophan substitutions of three positions in the GluN2A subunit yielded receptors that showed no current (Fig. 3 B). Two of these positions, N816 and V820, are expressed on the membrane (Fig. 4 E) suggesting that they are “pore-dead”: they can assemble and traffic to the membrane, but agonists are unable to open the ion channel. Hence, the free energy generated by agonist binding is no longer able to overcome the stability of the closed conformation of the ion channel. To test this idea, we introduced in these backgrounds a missense mutation, GluN2A(L812M), that dramatically enhances receptor activation presumably by altering the energetics of the closed and open conformations (Yuan et al., 2014). Indeed, N816W and V820W, when combined with the L812M mutation, now showed detectable glutamate-activated currents (Fig. 6 A), though the amplitude remained significantly reduced relative to wild type (Fig. 6 B). In contrast, S831W, which does not express on the membrane (Fig. 4 F), still showed no detectable glutamate-activated current when combined with the L812M mutation, consistent with this position involved in biogenesis rather than gating. Figure 6. Recovery of function in “pore-dead” constructs. (A) Example whole-cell recordings, displayed as in Fig. 3 A, for NMDARs containing double mutations in the GluN2A subunit, L812M, and a tryptophan substituted at either N816W, V820W, or S831W. Receptors containing single tryptophan substitutions of N816, V820, or S831 showed no detectable glutamate-activated whole-cell currents (Fig. 3 B and Table 1). (B) Mean current amplitudes (±SEM) at −70 mV for single or double mutation constructs. Positions that did not show detectable glutamate-activated currents are demarcated by an “X.” Number of whole-cell current recordings for each construct: GluN1/GluN2A(N816W), 9; GluN1/GluN2A(L812M/N816W), 5; GluN1/GluN2A(V820W), 8; GluN1/GluN2A(L812M/V820W), 4; GluN1/GluN2A(S831W), 9; GluN1/GluN2A(L812M/S831W), 6. The M4 segments in NMDAR subunits have strong effects on receptor desensitization In the continual presence of glutamate, iGluRs can enter into a nonconducting, desensitized state. For non-NMDARs, the process of desensitization is fairly well understood being driven by rearrangements of the ligand-binding domain (LBD; Meyerson et al., 2016; Plested, 2016). In contrast to non-NMDARs, rearrangement of the LBD in NMDARs make only a limited contribution to desensitization (Borschel et al., 2011) and the energetics of domains in addition to the LBD strongly contribute to desensitization including the transmembrane domain (TMD; Krupp et al., 1998; Villarroel et al., 1998; Alsaloum et al., 2016). To directly contrast the contribution of the M4 segments to desensitization in iGluR subtypes, we characterized desensitization in these receptors, initially focusing on NMDARs containing tryptophan substitutions in the GluN1 or GluN2A M4 segment (Fig. 7). Figure 7. Impact of M4 substitutions on NMDAR desensitization. (A) Representative whole-cell recordings of current through wild-type GluN1/GluN2A (gray traces) or NMDARs containing a tryptophan substitution in the M4 segment (black traces). Currents were recorded as in Fig. 3. (B) Mean (±SEM) of the %Des or normalized weighted τ for wild-type GluN1/GluN2A or GluN1/GluN2A containing tryptophan substitutions in either the GluN1 (left) or GluN2A (right) M4 segments. Raw values and details of number of recordings made are shown in Table 5. Solid bars indicate values significantly different from wild type (P < 0.01, t test). We used a more stringent level of significance for this analysis to focus on only those positions with prominent effects on desensitization. Dots highlight positions homologous to the VVLGAVE face in AMPARs. Wild-type GluN1/GluN2A (Fig. 7 A, gray traces) desensitize incompletely to ∼60% (62 ± 1%, n = 18) with a time course consisting of two components (weighted τ, 500 ± 20 ms). Tryptophan substitutions in the M4 segments of either the GluN1 or GluN2A subunit strongly altered receptor desensitization (Fig. 7 A and Table 5), increasing (e.g., GluN1(G815W)/GluN2A) or decreasing (GluN1/GluN2A(L830W)) the rate of desensitization as well as decreasing (e.g., GluN1(L830W)/GluN2A) or increasing (e.g., GluN1/GluN2A(M823W)) the extent of desensitization. Table 5. Desensitization properties of wild-type GluN1/GluN2A or GluN1/GluN2A containing tryptophan substitutions in either GluN1 or GluN2A M4 segment Tryptophans in GluN1 subunit Tryptophans in Glu2A subunit Construct %des τweighted n Construct %des τweighted n ms ms N1/N2A 62 ± 1 500 ± 20 18 18 F810W 60 ± 3 510 ± 30 4 I814W 68 ± 3 290 ± 30 4 E811W 55 ± 2 300 ± 30 4 D815W 82 ± 3 230 ± 20* 4 N812W 68 ± 2 88 ± 3* 4 N816W – – M813W 77 ± 1^ 220 ± 10* 5 M817W 51 ± 1* 990 ± 100^ 4 A814W 64 ± 1 220 ± 25* 4 A818W 39 ± 3 450 ± 70 4 G815W 74 ± 1^ 48 ± 3* 4 G819W 84 ± 2^ 310 ± 40 5 V816W 67 ± 3 170 ± 20* 4 V820W – – F817W 68 ± 2 220 ± 30* 5 F821W 52 ± 3 420 ± 60 5 M818W 56 ± 2 200 ± 20* 5 Y822W 48 ± 2 1,010 ± 145 4 L819W 63 ± 3 380 ± 60 4 M823W 93 ± 5^ 350 ± 10 5 V820W 45 ± 3 210 ± 10* 4 L824W 34 ± 4 1,350 ± 70^ 4 A821W 53 ± 3 390 ± 50 4 A825W 77 ± 3 300 ± 20 4 G822W 68 ± 2 280 ± 20* 4 A826W 74 ± 2 540 ± 30 5 G823W 80 ± 2^ 310 ± 30 4 A827W 95 ± 1^ 210 ± 10* 5 I824W 60 ± 1 460 ± 30 5 M828W 64 ± 2 500 ± 30 5 V825W 47 ± 1* 370 ± 80 4 A829W 44 ± 3 850 ± 160 5 A826W 65 ± 4 460 ± 20 4 L830W 52 ± 2 990 ± 40^ 5 G827W 74 ± 3 150 ± 20* 4 S831W – – I828W 45 ± 3 450 ± 40 4 L832W 54 ± 3 670 ± 50 5 F829W 59 ± 2 390 ± 20 5 I833W 59 ± 5 420 ± 30 4 L830W 40 ± 2* 500 ± 70 4 T834W 55 ± 3 620 ± 60 4 I831W 63 ± 4 290 ± 25 4 F835W 53 ± 6 420 ± 50 4 F832W 49 ± 1* 580 ± 40 5 I836W 50 ± 2 720 ± 90 5 I833W 67 ± 2 340 ± 40 4 W837 62 ± 1 500 ± 20 18 E834W 65 ± 1 300 ± 20 6 E838W 68 ± 2 360 ± 30 5 Values shown are mean ± SEM. n indicates the number of whole-cell recordings. Tagged values are significantly less or slower (*) or greater or faster (^) than wild type (P < 0.01, t test). Tryptophan substitutions of the M4 segments had strong effects in both the GluN1 (Fig. 7 B, left) and GluN2A (Fig. 7 B, right) subunits. There are several notable conclusions from these experiments. First, significant effects on desensitization occurred throughout both M4 segments, but such effects tended to be more limited at extreme intracellular positons, encompassing and C-terminal to G827 in GluN1 and S831 in GluN2A. Second, there is a strong subunit-specific difference in terms of the kinetics of desensitization: tryptophan substitutions in GluN1 consistently increase the rate of desensitization, whereas in GluN2A such substitutions preferentially decrease the rate of desensitization. Finally, substitutions at positions homologous to the AMPAR VVLGAVE face (Fig. 7 B, dots) consistently altered some aspect of NMDAR desensitization, suggesting that the interaction of M4 with the M3 segment impacts NMDAR desensitization. The M4 segment in AMPARs has weak effects on receptor desensitization Many of the subtle substitutions of the AMPAR VVLGAVE face yielded tetramers (Fig. 2 C), and if they showed detectable tetramers, also showed membrane currents. We therefore characterized the effect of these subtle substitutions on AMPAR desensitization (Fig. 8). For these experiments, we used outside-out patches and fast agonist solution exchange. We did not test some constructs either because they showed extremely low whole-cell current amplitudes (e.g., L820A and E834N) or they had no effect on receptor assembly (e.g., V816A). Figure 8. Substitutions of the VVLGAVE face in GluA2 have no or only weak effects on receptor desensitization. (A) Example outside-out recordings from wild-type GluA2(Q)-EGFP (top trace) or V816I (bottom trace) in response to 100-ms pulses of glutamate. (B) Mean (±SEM) %Des (left) and rate of desensitization (right). Black bars indicate values significantly different from wild type (P < 0.05, t test). Many of the tested positions had strong effects on receptor tetramerization (Fig. 2). We also tested GluA2(L832W), a position on the side of the helix facing away from the pore domain. Number of outside-out patches for each construct: wt, 9; V813I, 4; V816I, 4; L820F, 5; G823A, 4; A827S, 4; V830L, 4; L832W, 4; E834Q, 4; E834D, 4; E834N, 3. GluA2(Q)-EGFP showed rapid (5.5 ± 0.1 ms, n = 9) and strong (98.4 ± 0.1%Des) desensitization (Fig. 8, A and B) comparable with wild-type GluA2(Q) (Yelshansky et al., 2004). Receptors containing subtle substitutions of the VVLGAVE face, which often had dramatic effects on receptor assembly (Fig. 2), had no significant effect on desensitization with the exception of L820F, which showed a statistically significant but modest increase in %Des from ∼98.4% (wild type) to 99.1 ± 0.1%Des (Fig. 8 B). Thus, in contrast to what is observed for NMDARs (Fig. 7), the M4 segment in AMPARs, at least that component interacting with the M3 segment (VVLGAVE face), makes no notable contribution to receptor desensitization. Discussion In all iGluRs subtypes, an additional eukaryotic-specific transmembrane segment, the M4 segment, is associated with the pore domain of a neighboring subunit (Fig. 1). To test the functional role of the M4 segments to receptor function, we introduced tryptophans either in AMPAR (Salussolia et al., 2011, 2013) or NMDAR (present study) subunits. The M4 segments in AMPARs and NMDARs share similar sequences (Fig. 1 C) and structural arrangements (Figs. 1 and 3 C). Still, their functional roles are divergent. For AMPARs and based on tryptophan substitutions, the interaction along the entire extent of the M4 segment with the neighboring pore domain is required for receptor assembly. Even subtle mutations in the VVLGAVE face of AMPARs, except for two extreme extracellular positons, strongly disrupt tetramerization (Fig. 2). In contrast, and again based on tryptophan substitutions, the extracellular portion of the M4 segments in NMDAR subunits makes no obvious contribution to receptor assembly, but rather plays a dominant role in receptor gating, including generating “pore-dead” constructs—receptors that assemble and get to the cell surface efficiently but are completely unable to open their pore in response to agonist. Still, this distinction is not absolute: the M4 segment in AMPARs can impact receptor gating (Fig. 8; Terhag et al., 2010), and the extreme intracellular portion of the M4 in NMDARs can alter their surface expression (Fig. 4). Impact of M4 segment on iGluR gating In iGluRs, as well as other pore loop family members, the M3 transmembrane segment (homologous to TM2 or S6 in K+ channels) is the main pore-lining segment (Sobolevsky et al., 2009; Karakas and Furukawa, 2014; Lee et al., 2014). Surrounding the M3 segment are the outer transmembrane segments, M1 and M4. In NMDAR subunits, these outer transmembrane segment, M1 and M4, must be displaced for efficient pore opening to occur (Kazi et al., 2013) and contain missense mutations that alter receptor gating (Yuan et al., 2014; Chen et al., 2017; Ogden et al., 2017). The GluN1 and GluN2A M4 segments, at least in the closed state, interact extensively with the M3 segments of GluN2A and GluN1, respectively (Figs. 1 and 3 C). In addition, the M4 segments interact with the respective adjacent subunits’ M1 segments (Figs. 1 B and 9). Notable is the strong interaction at the extracellular end of each M4 segment with the S1-M1 of the same subunit (Fig. 9 B), an interaction that is much less extensive in AMPARs (Fig. 9 A). This interaction may be part of the critical difference between AMPAR and NMDAR gating, as the S1-M1 of the same subunit most closely interacts with pore-dead residues identified in the tryptophan scan. Figure 9. Differential roles of M4 segments in AMPAR and NMDAR function. (A and B) The M4 segments, viewed from the center of the pore, in either AMPARs (A) or NMDARs (B) either GluN1 M4 (left) or GluN2B M4 (right). The M3 segment from the adjacent subunit (see Fig. 3 C) is hidden to emphasis other M4 interactions, including the S1-M1 LBD-TMD linker from the same subunit and the M1 transmembrane segment from the neighboring subunit. Positions that showed significant changes in current amplitudes are highlighted in blue (reduced), green (greater), or red (no detectable current). For AMPARs, only those positions that showed no detectable current (Salussolia et al., 2011) are indicated, which include the VVLGAVE positions as well as I819. Subunits are colored light orange (GluA2 A and C, GluN1) and gray 60% (GluA2 B and D, GluN2B). The M4 is represented as a cartoon, whereas the S1-M1 and the adjacent M1 are shown as dots. A surprising feature is that tryptophan substitutions at extracellular positions, rather than destabilizing the closed state (i.e., reducing the MCT), as one might expect from disrupting transmembrane interactions observed in our closed state model, actually increased the MCT and at the same time also consistently decreased the MOT (Fig. 5 and Table 3). The lack of an available open state NMDAR structure makes it difficult to identify or hypothesize why this observation may occur. Further study will be needed to reveal how shifting interactions during channel opening facilitate efficient gating. Differential contribution of the M4 segments to iGluR desensitization For NMDARs, receptors containing tryptophan substitutions at 12 VVLGAVE face positions, 7 in GluN1 and 5 in GluN2A, showed whole-cell currents. Of these positions that showed current, 9 showed some form of altered desensitization, either a change in the extent and/or rate of desensitization (Fig. 7). Those that did not show any significant change, GluN1(L830W), GluN1(E834W), and GluN2A(E838W), are all located at the extreme intracellular end of the M4 segments, emphasizing that this region only has weak effects on NMDAR gating. In contrast, for AMPAR desensitization, substitutions at VVLGAVE positions had no effect on desensitization except for one L820 (VVLGAVE; Fig. 8), whereas these same substitutions had strong effects on receptor assembly (Fig. 2). These experiments highlight the differential roles of the M4 segments in iGluRs. In addition, these results also support the idea that desensitization of AMPARs and NMDARs are controlled by different processes, likely in different domains. For AMPARs, it is mainly regulated by the conformation of the LBD (Sun et al., 2002; Carbone and Plested, 2012), whereas for NMDARs, the TMD makes a more significant contribution (Krupp et al., 1998; Villarroel et al., 1998; Alsaloum et al., 2016; present study). Nevertheless, the structural basis for desensitization in NMDARs remains poorly defined. For NMDARs, the two different subunits, GluN1 and GluN2A, also take on divergent roles. The GluN1 M4 seems to be highly important in regulating the rate of desensitization with 10 out 25 tryptophan substitutions drastically increasing the rate of desensitization and none decreasing the rate. Of these, nearly all fall within a tight seven-residue portion of extracellular region from N812 to M818 (Fig. 7 B). Interestingly, this region of the GluN1 M4 interacts strongly with a phenylalanine (F) previously identified to impact NMDAR desensitization (Alsaloum et al., 2016). In contrast, the interaction between similar regions in GluN2A is less notable; also, tryptophan substitutions in the GluN2A M4 often slowed the rate of desensitization. Still, the mechanistic and structural basis for the contribution of the GluN1 and GluN2A M4 segments to desensitization remains unknown. Role of the M4 segments to receptor assembly Except for the two valines located at the extreme extracellular end of M4, subtle substitutions in the VVLGAVE face in AMPARs can have profound effects on receptor tetramerization (Fig. 2 C), highlighting the important role of the M4 segment to AMPAR oligomerization. The exact mechanistic role of the AMPAR M4 segment in assembly however remains unclear because substitutions with both larger-sized as well as smaller-sized side chains disrupt tetramerization. Hence, factors in addition to steric hindrances between side chains such as van der Waals interactions may contribute to the role of the AMPAR M4 segment in receptor assembly. Subtle substitutions at the two extreme extracellular positions in the AMPAR M4 segment, V813 and V816, had no significant effect on receptor oligomerization (Fig. 2 C) or receptor desensitization (Fig. 8). Interestingly, this extreme extracellular end corresponds to those positions in the GluN1 and GluN2A M4 segments that had extremely strong effects on receptor gating (Fig. 5) and also altered receptor desensitization (Fig. 7). For NMDARs, the M4 segment plays an uneven role in receptor biogenesis. In contrast to AMPARs, the extracellular M4 of GluN1 and GluN2A makes no obvious contribution to receptor biogenesis. In addition, only two of the more extreme intracellular positions, GluN2A(S831) and the negative charges, GluN1(E834) and GluN2A(E838), contribute in some fashion to biogenesis, either receptor assembly or forward trafficking to the membrane. We were unable to test any contribution to receptor assembly using BN-PAGE or FSEC because the mammalian GluN1/GluN2A NMDAR has a strong tendency to dissociate into a dimer in detergents, making them difficult to be analyzed precisely. Biology of the impact of the M4 segment on receptor gating In iGluRs, the M4 segment is the most peripheral transmembrane segment, interacting extensively with phospholipids, and is attached to the intracellular C-terminal domain (CTD). NMDAR function is strongly modulated by lipids (Casado and Ascher, 1998; Korinek et al., 2015). Further, posttranslational modification of the CTDs in NMDAR (Salter and Kalia, 2004; Tu et al., 2010) and AMPAR (Lu and Roche, 2012) can alter their gating properties. How these factors mediate changes in receptor gating is unknown. One possibility is that the M4 segments act as transduction pathways for these effects, which may occur for non-NMDARs (Wilding et al., 2016). The lipid composition in contact with the M4 segment, for example, might alter the positioning or orientation of the M4 segments, which in turn alters how they interact with the pore domain, thereby impacting gating. The posttranslational status of the CTD might similarly impact the M4 positioning and orientation. Still, the importance of the M4 segment to these key biological pathways remains to be elucidated. Supplementary Material Supplemental Materials (PDF) Acknowledgments We thank Drs. Quan Gan and Kasper Hansen for helpful discussions and/or comments on the manuscript. We also thank Drs. Mark Aurousseau and Derek Bowie for their generous gift of pHmystik. This work was supported by National Institutes of Health RO1 grants NS088479 (to L.P. Wollmuth), including a minority supplement (to J.B. Amin), MH081923 (to M.E. Bowen), NS093753 (to M.C. Regan), MH085926 and GM105730 (to H. Furukawa), and GM118091 (to H.-X. Zhou). The authors declare no competing financial interests. Author contributions: J.B. Amin and L.P. Wollmuth designed experiments, executed experiments, analyzed data, and wrote the paper. C.L. Salussolia and M.E. Bowen designed experiments, executed experiments, and wrote the paper. K. Chan and M.C. Regan executed experiments, analyzed data, and wrote the paper. J. Dai designed experiments, executed experiments, and analyzed data. H.-X. Zhou designed experiments and wrote the paper. H. Furukawa designed experiments and analyzed data. Angus C. Nairn served as editor. Abbreviations used: AMPAR AMPA receptor CTD C-terminal domain FSEC fluorescence-detection size exclusion chromatography iGluR ionotropic glutamate receptor LBD ligand-binding domain MCT mean closed time MOT mean open time NMDAR NMDA receptor TMD transmembrane domain ==== Refs Alsaloum, M., R. Kazi, Q. Gan, J. Amin, and L.P. Wollmuth. 2016. A molecular determinant of subtype-specific desensitization in ionotropic glutamate receptors. J. Neurosci. 36 :2617–2622. 10.1523/JNEUROSCI.2667-15.2016 26937003 Aurousseau, M. 2015. Novel fluorescence-based methods to study the stoichiometric and surface expression properties of ionotropic glutamate receptors. McGill University. Available at: http://digitool.Library.McGill.CA:80/R/-?func=dbin-jump-full&object_id=135326&silo_library=GEN01 Borschel, W.F., S.E. Murthy, E.M. Kasperek, and G.K. Popescu. 2011. NMDA receptor activation requires remodelling of intersubunit contacts within ligand-binding heterodimers. Nat. Commun. 2 :498. 10.1038/ncomms1512 21988914 Cao, J.Y., S. Qiu, J. Zhang, J.J. Wang, X.M. Zhang, and J.H. Luo. 2011. Transmembrane region of N-methyl-d-aspartate receptor (NMDAR) subunit is required for receptor subunit assembly. J. Biol. Chem. 286 :27698–27705. 10.1074/jbc.M111.235333 21659529 Carbone, A.L., and A.J. Plested. 2012. Coupled control of desensitization and gating by the ligand binding domain of glutamate receptors. Neuron. 74 :845–857. 10.1016/j.neuron.2012.04.020 22681689 Casado, M., and P. Ascher. 1998. Opposite modulation of NMDA receptors by lysophospholipids and arachidonic acid: common features with mechanosensitivity. J. Physiol. 513 :317–330. 10.1111/j.1469-7793.1998.317bb.x 9806985 Chang, H.R., and C.C. Kuo. 2008. The activation gate and gating mechanism of the NMDA receptor. J. Neurosci. 28 :1546–1556. 10.1523/JNEUROSCI.3485-07.2008 18272676 Chen, L., K.L. Dürr, and E. Gouaux. 2014. X-ray structures of AMPA receptor-cone snail toxin complexes illuminate activation mechanism. Science. 345 :1021–1026. 10.1126/science.1258409 25103405 Chen, W., A. Tankovic, P.B. Burger, H. Kusumoto, S.F. Traynelis, and H. Yuan. 2017. Functional evaluation of a de novo GRIN2A mutation identified in a patient with profound global developmental delay and refractory epilepsy. Mol. Pharmacol. 91 :317–330. 10.1124/mol.116.106781 28126851 Gan, Q., C.L. Salussolia, and L.P. Wollmuth. 2015. Assembly of AMPA receptors: mechanisms and regulation. J. Physiol. 593 :39–48. 10.1113/jphysiol.2014.273755 25556786 Gan, Q., J. Dai, H.X. Zhou, and L.P. Wollmuth. 2016. The transmembrane domain mediates tetramerization of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors. J. Biol. Chem. 291 :6595–6606. 10.1074/jbc.M115.686246 26839312 Glasgow, N.G., B. Siegler Retchless, and J.W. Johnson. 2015. Molecular bases of NMDA receptor subtype-dependent properties. J. Physiol. 593 :83–95. 10.1113/jphysiol.2014.273763 25556790 Greger, I.H., L. Khatri, X. Kong, and E.B. Ziff. 2003. AMPA receptor tetramerization is mediated by Q/R editing. Neuron. 40 :763–774. 10.1016/S0896-6273(03)00668-8 14622580 Hamdan, F.F., M. Srour, J.M. Capo-Chichi, H. Daoud, C. Nassif, L. Patry, C. Massicotte, A. Ambalavanan, D. Spiegelman, O. Diallo, 2014. De novo mutations in moderate or severe intellectual disability. PLoS Genet. 10 :e1004772. 10.1371/journal.pgen.1004772 25356899 Hardingham, G.E., and K.Q. Do. 2016. Linking early-life NMDAR hypofunction and oxidative stress in schizophrenia pathogenesis. Nat. Rev. Neurosci. 17 :125–134. 10.1038/nrn.2015.19 26763624 Honse, Y., H. Ren, R.H. Lipsky, and R.W. Peoples. 2004. Sites in the fourth membrane-associated domain regulate alcohol sensitivity of the NMDA receptor. Neuropharmacology. 46 :647–654. 10.1016/j.neuropharm.2003.11.006 14996542 Huettner, J.E. 2015. Glutamate receptor pores. J. Physiol. 593 :49–59. 10.1113/jphysiol.2014.272724 25556787 Jones, K.S., H.M. VanDongen, and A.M. VanDongen. 2002. The NMDA receptor M3 segment is a conserved transduction element coupling ligand binding to channel opening. J. Neurosci. 22 :2044–2053.11896144 Karakas, E., and H. Furukawa. 2014. Crystal structure of a heterotetrameric NMDA receptor ion channel. Science. 344 :992–997. 10.1126/science.1251915 24876489 Kawate, T., and E. Gouaux. 2006. Fluorescence-detection size-exclusion chromatography for precrystallization screening of integral membrane proteins. Structure. 14 :673–681. 10.1016/j.str.2006.01.013 16615909 Kazi, R., Q. Gan, I. Talukder, M. Markowitz, C.L. Salussolia, and L.P. Wollmuth. 2013. Asynchronous movements prior to pore opening in NMDA receptors. J. Neurosci. 33 :12052–12066. 10.1523/JNEUROSCI.5780-12.2013 23864691 Korinek, M., V. Vyklicky, J. Borovska, K. Lichnerova, M. Kaniakova, B. Krausova, J. Krusek, A. Balik, T. Smejkalova, M. Horak, and L. Vyklicky. 2015. Cholesterol modulates open probability and desensitization of NMDA receptors. J. Physiol. 593 :2279–2293. 10.1113/jphysiol.2014.288209 25651798 Krupp, J.J., B. Vissel, S.F. Heinemann, and G.L. Westbrook. 1998. N-terminal domains in the NR2 subunit control desensitization of NMDA receptors. Neuron. 20 :317–327. 10.1016/S0896-6273(00)80459-6 9491992 Lee, C.H., W. Lü, J.C. Michel, A. Goehring, J. Du, X. Song, and E. Gouaux. 2014. NMDA receptor structures reveal subunit arrangement and pore architecture. Nature. 511 :191–197. 10.1038/nature13548 25008524 Lemke, J.R., D. Lal, E.M. Reinthaler, I. Steiner, M. Nothnagel, M. Alber, K. Geider, B. Laube, M. Schwake, K. Finsterwalder, 2013. Mutations in GRIN2A cause idiopathic focal epilepsy with rolandic spikes. Nat. Genet. 45 :1067–1072. 10.1038/ng.2728 23933819 Lu, W., and K.W. Roche. 2012. Posttranslational regulation of AMPA receptor trafficking and function. Curr. Opin. Neurobiol. 22 :470–479. 10.1016/j.conb.2011.09.008 22000952 Mackerell, A.D. Jr., M. Feig, and C.L. Brooks III. 2004. Extending the treatment of backbone energetics in protein force fields: limitations of gas-phase quantum mechanics in reproducing protein conformational distributions in molecular dynamics simulations. J. Comput. Chem. 25 :1400–1415. 10.1002/jcc.20065 15185334 Mayer, M.L. 2016. Structural biology of glutamate receptor ion channel complexes. Curr. Opin. Struct. Biol. 41 :119–127. 10.1016/j.sbi.2016.07.002 27454049 Meddows, E., B. Le Bourdelles, S. Grimwood, K. Wafford, S. Sandhu, P. Whiting, and R.A. McIlhinney. 2001. Identification of molecular determinants that are important in the assembly of N-methyl-d-aspartate receptors. J. Biol. Chem. 276 :18795–18803. 10.1074/jbc.M101382200 11279200 Meyerson, J.R., S. Chittori, A. Merk, P. Rao, T.H. Han, M. Serpe, M.L. Mayer, and S. Subramaniam. 2016. Structural basis of kainate subtype glutamate receptor desensitization. Nature. 537 :567–571. 10.1038/nature19352 27580033 Moore, D.T., B.W. Berger, and W.F. DeGrado. 2008. Protein-protein interactions in the membrane: sequence, structural, and biological motifs. Structure. 16 :991–1001. 10.1016/j.str.2008.05.007 18611372 Ogden, K.K., W. Chen, S.A. Swanger, M.J. McDaniel, L.Z. Fan, C. Hu, A. Tankovic, H. Kusumoto, G.J. Kosobucki, A.J. Schulien, 2017. Molecular mechanism of disease-associated mutations in the pre-M1 helix of NMDA receptors and potential rescue pharmacology. PLoS Genet. 13 :e1006536. 10.1371/journal.pgen.1006536 28095420 Phillips, J.C., R. Braun, W. Wang, J. Gumbart, E. Tajkhorshid, E. Villa, C. Chipot, R.D. Skeel, L. Kalé, and K. Schulten. 2005. Scalable molecular dynamics with NAMD. J. Comput. Chem. 26 :1781–1802. 10.1002/jcc.20289 16222654 Plested, A.J. 2016. Structural mechanisms of activation and desensitization in neurotransmitter-gated ion channels. Nat. Struct. Mol. Biol. 23 :494–502. 10.1038/nsmb.3214 27273633 Popescu, G., and A. Auerbach. 2003. Modal gating of NMDA receptors and the shape of their synaptic response. Nat. Neurosci. 6 :476–483.12679783 Qin, F. 2004. Restoration of single-channel currents using the segmental k-means method based on hidden Markov modeling. Biophys. J. 86 :1488–1501. 10.1016/S0006-3495(04)74217-4 14990476 Ren, H., Y. Zhao, D.S. Dwyer, and R.W. Peoples. 2012. Interactions among positions in the third and fourth membrane-associated domains at the intersubunit interface of the N-methyl-d-aspartate receptor forming sites of alcohol action. J. Biol. Chem. 287 :27302–27312. 10.1074/jbc.M111.338921 22715100 Šali, A., and T.L. Blundell. 1993. Comparative protein modelling by satisfaction of spatial restraints. J. Mol. Biol. 234 :779–815. 10.1006/jmbi.1993.1626 8254673 Salter, M.W., and L.V. Kalia. 2004. Src kinases: a hub for NMDA receptor regulation. Nat. Rev. Neurosci. 5 :317–328. 10.1038/nrn1368 15034556 Salussolia, C.L., A. Corrales, I. Talukder, R. Kazi, G. Akgul, M. Bowen, and L.P. Wollmuth. 2011. Interaction of the M4 segment with other transmembrane segments is required for surface expression of mammalian α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors. J. Biol. Chem. 286 :40205–40218. 10.1074/jbc.M111.268839 21930708 Salussolia, C.L., Q. Gan, R. Kazi, P. Singh, J. Allopenna, H. Furukawa, and L.P. Wollmuth. 2013. A eukaryotic specific transmembrane segment is required for tetramerization in AMPA receptors. J. Neurosci. 33 :9840–9845. 10.1523/JNEUROSCI.2626-12.2013 23739980 Salussolia, C.L., Q. Gan, and L.P. Wollmuth. 2015. Assaying AMPA receptor oligomerization. In Ionotropic Glutamate Receptor Technologies. G.K. Popescu, editor. Humana Press, New York. 3–14. Schneider, C.A., W.S. Rasband, and K.W. Eliceiri. 2012. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods. 9 :671–675. 10.1038/nmeth.2089 22930834 Schorge, S., and D. Colquhoun. 2003. Studies of NMDA receptor function and stoichiometry with truncated and tandem subunits. J. Neurosci. 23 :1151–1158.12598603 Sobolevsky, A.I., C. Beck, and L.P. Wollmuth. 2002. Molecular rearrangements of the extracellular vestibule in NMDAR channels during gating. Neuron. 33 :75–85. 10.1016/S0896-6273(01)00560-8 11779481 Sobolevsky, A.I., M.P. Rosconi, and E. Gouaux. 2009. X-ray structure, symmetry and mechanism of an AMPA-subtype glutamate receptor. Nature. 462 :745–756. 10.1038/nature08624 19946266 Soler-Llavina, G.J., T.H. Chang, and K.J. Swartz. 2006. Functional interactions at the interface between voltage-sensing and pore domains in the Shaker Kv channel. Neuron. 52 :623–634. 10.1016/j.neuron.2006.10.005 17114047 Sun, Y., R. Olson, M. Horning, N. Armstrong, M. Mayer, and E. Gouaux. 2002. Mechanism of glutamate receptor desensitization. Nature. 417 :245–253. 10.1038/417245a 12015593 Talukder, I., and L.P. Wollmuth. 2011. Local constraints in either the GluN1 or GluN2 subunit equally impair NMDA receptor pore opening. J. Gen. Physiol. 138 :179–194. 10.1085/jgp.201110623 21746848 Talukder, I., P. Borker, and L.P. Wollmuth. 2010. Specific sites within the ligand-binding domain and ion channel linkers modulate NMDA receptor gating. J. Neurosci. 30 :11792–11804. 10.1523/JNEUROSCI.5382-09.2010 20810899 Terhag, J., K. Gottschling, and M. Hollmann. 2010. The transmembrane domain C of AMPA receptors is critically involved in receptor function and modulation. Front. Mol. Neurosci. 3 :117. 10.3389/fnmol.2010.00117 21206529 Traynelis, S.F., L.P. Wollmuth, C.J. McBain, F.S. Menniti, K.M. Vance, K.K. Ogden, K.B. Hansen, H. Yuan, S.J. Myers, and R. Dingledine. 2010. Glutamate receptor ion channels: structure, regulation, and function. Pharmacol. Rev. 62 :405–496. 10.1124/pr.109.002451 20716669 Tu, W., X. Xu, L. Peng, X. Zhong, W. Zhang, M.M. Soundarapandian, C. Balel, M. Wang, N. Jia, W. Zhang, 2010. DAPK1 interaction with NMDA receptor NR2B subunits mediates brain damage in stroke. Cell. 140 :222–234. 10.1016/j.cell.2009.12.055 20141836 Villarroel, A., M.P. Regalado, and J. Lerma. 1998. Glycine-independent NMDA receptor desensitization: localization of structural determinants. Neuron. 20 :329–339. 10.1016/S0896-6273(00)80460-2 9491993 Wilding, T.J., M.N. Lopez, and J.E. Huettner. 2016. Chimeric glutamate receptor subunits reveal the transmembrane domain is sufficient for NMDA receptor pore properties but some positive allosteric modulators require additional domains. J. Neurosci. 36 :8815–8825. 10.1523/JNEUROSCI.0345-16.2016 27559165 Yelshansky, M.V., A.I. Sobolevsky, C. Jatzke, and L.P. Wollmuth. 2004. Block of AMPA receptor desensitization by a point mutation outside the ligand-binding domain. J. Neurosci. 24 :4728–4736. 10.1523/JNEUROSCI.0757-04.2004 15152033 Yuan, H., K.B. Hansen, J. Zhang, T.M. Pierson, T.C. Markello, K.V. Fajardo, C.M. Holloman, G. Golas, D.R. Adams, C.F. Boerkoel, 2014. Functional analysis of a de novo GRIN2A missense mutation associated with early-onset epileptic encephalopathy. Nat. Commun. 5 :3251. 10.1038/ncomms4251 24504326 Yuan, H., C.M. Low, O.A. Moody, A. Jenkins, and S.F. Traynelis. 2015. Ionotropic GABA and glutamate receptor mutations and human neurologic diseases. Mol. Pharmacol. 88 :203–217. 10.1124/mol.115.097998 25904555

PMC005xxxxxx/PMC5496621.txt

==== Front J Cell Biol J Cell Biol jcb jcb The Journal of Cell Biology 0021-9525 1540-8140 The Rockefeller University Press 28592443 201612040 10.1083/jcb.201612040 Research Articles Article 19 45 ESCRT-mediated vesicle concatenation in plant endosomes ESCRT-mediated vesicle concatenation in plant endosomes http://orcid.org/0000-0002-6675-3836 Buono Rafael Andrade 12* http://orcid.org/0000-0002-2647-2693 Leier André 78* Paez-Valencia Julio 12* http://orcid.org/0000-0001-9744-3526 Pennington Janice 9 Goodman Kaija 12 Miller Nathan 1 Ahlquist Paul 345910 Marquez-Lago Tatiana T. 78 http://orcid.org/0000-0003-4699-6950 Otegui Marisa S. 126 1 Department of Botany, University of Wisconsin-Madison, Madison, WI 2 R.M. Bock Laboratories of Cell and Molecular Biology, University of Wisconsin-Madison, Madison, WI 3 Institute for Molecular Virology, University of Wisconsin-Madison, Madison, WI 4 Departments of Oncology, University of Wisconsin-Madison, Madison, WI 5 Department of Plant Pathology, University of Wisconsin-Madison, Madison, WI 6 Department of Genetics, University of Wisconsin-Madison, Madison, WI 7 Informatics Institute, School of Medicine, University of Alabama at Birmingham, Birmingham, AL 8 Department of Genetics, School of Medicine, University of Alabama at Birmingham, Birmingham, AL 9 Howard Hughes Medical Institute, Chevy Chase, MD 10 Morgridge Institute for Research, Madison, WI Correspondence to Marisa S. Otegui: otegui@wisc.edu * R.A. Buono, A. Leier, and J. Paez-Valencia contributed equally to this paper. R.A. Buono’s present address is Dept. of Plant Systems Biology, Vlaams Instituut voor Biotechnologie, Ghent University, Ghent, Belgium. 03 7 2017 216 7 21672177 16 12 2016 06 4 2017 01 5 2017 © 2017 Buono et al. 2017 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). ESCRT proteins play essential functions by remodeling cellular membranes. Buono et al. report on a novel ESCRT-dependent mechanism in plant endosomes that leads to sequential concatenation of vesicle buds by temporally uncoupling membrane constriction from membrane fission. During this process, ESCRT-III proteins remain inside endosomes after intralumenal vesicle release. Ubiquitinated plasma membrane proteins (cargo) are delivered to endosomes and sorted by endosomal sorting complex required for transport (ESCRT) machinery into endosome intralumenal vesicles (ILVs) for degradation. In contrast to the current model that postulates that ILVs form individually from inward budding of the endosomal limiting membrane, plant ILVs form as networks of concatenated vesicle buds by a novel vesiculation mechanism. We ran computational simulations based on experimentally derived diffusion coefficients of an ESCRT cargo protein and electron tomograms of Arabidopsis thaliana endosomes to measure cargo escape from budding ILVs. We found that 50% of the ESCRT cargo would escape from a single budding profile in 5–20 ms and from three concatenated ILVs in 80–200 ms. These short cargo escape times predict the need for strong diffusion barriers in ILVs. Consistent with a potential role as a diffusion barrier, we find that the ESCRT-III protein SNF7 remains associated with ILVs and is delivered to the vacuole for degradation. Howard Hughes Medical Institute http://dx.doi.org/10.13039/100000011 National Science Foundation http://dx.doi.org/10.13039/100000001 MCB1157824 MCB1614965 University of Wisconsin-Madison http://dx.doi.org/10.13039/100007015 ==== Body pmcIntroduction Vesicle trafficking pathways are highly regulated processes that require specific machineries to select vesicle cargo, deform and fission membranes, and mediate the transport and fusion of the resulting vesicle with the acceptor compartment. The initial sorting and clustering of cargo macromolecules in a vesicle require their concentration in a small region of the donor membrane. In the case of clathrin-mediated endocytosis and the anterograde and retrograde trafficking between the endoplasmic reticulum and Golgi via COP-II and COP-I vesicles, respectively, the cargo molecules directly or through adapters recruit a protein coat that aids in the local deformation of the membrane into a bud (Traub and Bonifacino, 2013; Gomez-Navarro and Miller, 2016). The cargo remains physically associated with these coats after the vesicle is free in the cytoplasm and until the coat disassembles, greatly limiting the possibility of cargo escape back to the donor membrane by diffusion during vesicle formation. The degradation of plasma membrane proteins depends on different membrane vesiculation events along the endocytic and endosomal pathways. At the plasma membrane, linear amino acid motifs, conformational motifs, and/or posttranslational modifications such as phosphorylation and ubiquitination of plasma membrane proteins are recognized by adapter and accessory proteins (Traub and Bonifacino, 2013) that also facilitate the assembly of the clathrin coat and the formation of endocytic vesicles. Once released from the plasma membrane, clathrin-coated vesicles lose their coats and fuse with endosomes. At endosomes, the endocytosed cargo proteins can be recycled back to the plasma membrane or internalized into endosomal intralumenal vesicles (ILVs) for their degradation upon the fusion of multivesicular endosomes (MVEs) with vacuoles/lysosomes (Henne et al., 2011). The major sorting signal for endosomal-mediated vesiculation is ubiquitin (MacDonald et al., 2012). Ubiquitinated cargo proteins on the endosomal limiting membrane are recognized and sorted into ILVs by endosomal sorting complex required for transport (ESCRT) proteins (Schuh and Audhya, 2014; Paez Valencia et al., 2016). The ESCRT machinery mediates negative membrane bending (away from the cytoplasm), leading to a vesiculation process in the reverse topology to most membrane-deformation events, including clathrin-coated endocytosis. In fungi and metazoans, five multimeric ESCRT complexes have been identified (ESCRT-0, -I, -II, and -III and VPS4-VTA1; Henne et al., 2011; Hurley, 2015). ESCRT-0, -I, and -II contain ubiquitin-binding domains that interact with the ubiquitinated cargo proteins, contributing to the clustering of cargo on endosomal membranes. The ESCRT-III subunit VPS32/Snf7p is known to assemble into long spiral filaments and induce membrane curvature (Shen et al., 2014; Chiaruttini et al., 2015). Although ESCRT-III does not seem to bind ubiquitin, the ESCRT-III spirals could have a critical role in corralling the ESCRT cargo at the endosomal membrane (Henne et al., 2012; Chiaruttini et al., 2015). However, whether the membrane-associated ESCRT-III spirals would serve as barriers, crowding cargo molecules and preventing their escape by lateral diffusion, has not been demonstrated. Moreover, deubiquitinating enzymes remove the ubiquitin on cargo proteins before the ILVs are released into the endosomal lumen (Richter et al., 2007, 2013; Johnson et al., 2017). This means that, in contrast with other types of coat-mediated vesiculation (e.g., clathrin-mediated endocytosis), the ESCRT coat loses direct contact with the cargo proteins before completion of vesicle formation. In addition, the ESCRTs are assumed to be removed from the ILV neck before membrane fission and recycled back to the cytoplasm (Babst et al., 2002). This raises the question of what prevents cargo proteins from diffusing away from the ILV buds. Once the ESCRT-III complex is recruited by ESCRT-II onto the endosomal membranes, the ESCRT-III subunit VPS2 recruits the AAA ATPase VPS4, leading to the constriction of the budding ILV neck into a 17-nm-wide tubule, ESCRT disassembly, membrane fission, and ILV release into the endosomal lumen (Obita et al., 2007; Adell et al., 2014). Electron microscopy and electron tomographic analyses of yeast and mammalian MVEs have repeatedly shown single ILVs inside the MVE lumen (Murk et al., 2003a; Wemmer et al., 2011; Adell et al., 2014), supporting the notion that ILVs form one by one from single budding events by the sequential action of multiple ESCRT components. However, exactly how ESCRT proteins mediate ILV formation is unclear. A study of SNF7 polymers on membranes suggests that ESCRT-III spirals become loaded springs because of lateral compression from adjacent filaments, and their relaxation could lead to membrane deformation and ILV bud formation (Chiaruttini et al., 2015). VPS4 could further modify the architecture of the ESCRT-III polymers assisting in membrane constriction (Adell et al., 2014; Shen et al., 2014). Interestingly, although the SNF7 filaments are highly flexible and can grow at different radii, the smallest radius reported for SNF7 spirals is 18 nm (Chiaruttini et al., 2015), which is 10 times larger than the radius of the superconstricted, prefission state of dynamin polymers (Sundborger et al., 2014) that lead to membrane fission. This indicates that although SNF7 could trigger ILV bud formation, it may not by itself mediate pore closure and ILV release. ESCRT proteins are present from archaea to eukaryotes. Plants contain orthologues for most of the ESCRT proteins, with the exception of the canonical ESCRT-0 (Winter and Hauser, 2006; Leung et al., 2008) together with plant-specific ESCRT components (Paez Valencia et al., 2016). In this study, we report on the occurrence of ILV concatenation in plant endosomes as a novel mechanism of vesicle formation as well as on the participation of ESCRT-III proteins in this process. We also analyze the effect of prerelease ILV membrane geometry on the potential escape time of deubiquitinated ESCRT cargo during the formation on ILVs by computational simulations based on electron tomograms of plant MVEs. Finally, and consistent with the presence of a persistent diffusion barrier in ILV networks predicted by our simulations, we show that SNF7 remains associated with internalized endosomal membranes after ILV release and fusion of MVEs with the vacuole. Results ILVs form as part of concatenated networks within MVEs To understand changes in membrane geometry during endosomal vesiculation, we analyzed dual-axis electron tomograms of plant MVEs (32 MVEs from wild-type Arabidopsis thaliana root epidermal and cortical cells) and found that most ILVs (71%; sample size, 183 ILVs) do not form individually but as large networks of concatenated budding profiles (Fig. 1, A–F; and Video 1). We observed linear chains of vesicle buds interconnected through axial narrow membrane bridges or more complex networks of up to 12 ILV buds connected through both lateral and axial bridges (Fig. 1, C–F; and Video 1). Most of these networks were still connected to the limiting membrane by a nascent ILV bud with a narrow neck similar in size to that of inter-ILV bud bridges (mean width, 17 nm; SD, 1.5 nm; sample size, 25 bridges/necks; Fig. 1, A–F). We also obtained electron tomograms of MVEs from another plant cell type, root meristematic cells, and found similar networks of concatenated ILVs (Fig. S1, A and A’), suggesting that this is a common structural feature of plant MVEs. These observations are hard to reconcile with the current model that postulates that all ILVs are free in the MVE lumen (Murk et al., 2003b) and form one at a time when ESCRT-III and VPS4 pinch off the membrane at the bud neck. How, then, do these concatenated vesicles form? For the linear vesicle chain configuration, we considered two possible models. Once the first necked membrane bud is established, the second one forms either by initiation of further budding from the distal end of the same bud or by deformation of the limiting membrane above the first. We found multiple intermediates consistent with the second model (Fig. 1, G and H; and Fig. S1 C), suggesting that the first ILV bud to form in a linear concatenated chain is the one furthest away from the limiting membrane. Figure 1. Structural analysis of ILV formation. (A and A’) Tomographic slice (A) and corresponding tomographic reconstruction (A’) of a wild-type MVE from a root cell. (B and B’) Tomographic slice (B) and tomographic reconstruction (B’) of the MVE depicted in A and A’, showing a detail of the concatenated network of ILV buds connected through narrow membrane bridges. (C–F) Tomographic slices from plant MVEs showing examples of concatenated ILV bud networks connected to the endosomal limiting membrane. (G) Tomographic slices showing intermediates in the process of ILV bud concatenation. Shallow invaginations corresponding with early stages in ILV bud formation are marked by asterisks; “a” indicates the first ILV to form, and “b” the second one. (H) Model of ILV bud concatenation based on the images depicted in G. (I–I”) Tomographic slice (I) and corresponding tomographic reconstruction (I’ and I”) of three ILV buds connected by lateral membrane bridges. Red arrowheads indicate interconnecting membrane bridges. Bars: (A and A’) 50 nm; (B–I”) 20 nm. We also found geometries showing that single ILV buds in close proximity to each other were connected laterally through narrow membrane bridges (Fig. 1, I–I”). We were unable to find early intermediates showing whether these membrane connections derive from initial fusion of ILV buds followed by membrane bridge stabilization or if they are established simultaneously with ILV bud formation. To determine whether ILV bud concatenation is also observed in other organisms besides plants, we obtained dual-axis electron tomograms of MVEs from high-pressure–frozen freeze-substituted Saccharomyces cerevisiae cells. Interconnected ILVs were detected in very low frequencies (6% concatenated ILVs; sample size, 65 ILVs) in yeast cells (Fig. S1, C–E), suggesting that either ILV networks are very short-lived or that ILVs do not form from a process of membrane bud concatenation as frequently as observed in plant cells. To test whether ESCRT-III and other late ESCRT components are required for ILV concatenation, we obtained dual-axis electron tomograms of MVEs from Arabidopsis mutants lacking either the ESCRT-III subunits CHMP1A and B (Did2p in yeast) or the positive VPS4 regulator LIP5 (Vta1p in yeast). These mutants, chmp1 and lip5, are unable to efficiently sequester cargo into ILVs (Spitzer et al., 2009; Buono et al., 2016). We found that, compared with wild type, both chmp1 and lip5 MVEs had a lower density of ILVs, with a smaller proportion of concatenated ILV buds and significantly smaller ILV bud networks (Fig. 2, A–H; and Videos 2 and 3). In addition, we found that in some chmp1 MVEs, concatenated ILV buds were connected to the limiting membrane by very long (∼45 nm) necks (Fig. 2 D), suggesting a defect in the sequential formation of ILV buds that leads to ILV bud concatenation. Thus, the analysis of plant ESCRT mutant MVEs showed a correlation between ILV bud concatenation and efficient ILV formation. Figure 2. Electron tomography of wild-type and ESCRT mutant MVEs from Arabidopsis. (A) Tomographic reconstructions of dual-axis tomograms of wild-type, chmp1, and lip5 mutant MVEs. Red spheres were placed inside ILV and ILV buds to facilitate their identification. Note the reduced number of ILVs and the limited ILV concatenation in mutant MVEs. (B–D) Tomographic slices of WT, lip5, and chmp1 MVEs. The red arrowhead in D indicates a long neck connecting two concatenated ILV buds with the limiting membrane. Bars: (A) 50 nm; (B–D) 20 nm. (E–H) Quantitative analyses of MVE structural features. Error bars indicate SEM. Letters above bars represent statistical significance (one-way ANOVA followed by Tukey’s test: P < 0.05); bars sharing a letter are not significantly different from one another. n = 11 (wild type), 7 (chmp1), and 11 (lip5) MVEs; n = 183 (wild type), 27 (chmp1), and 162 (lip5) ILVs. Simulation of ESCRT cargo diffusion during ILV formation ESCRT proteins are known to bind ubiquitin on cargo proteins. We have very little information about the extent to which ESCRT-III proteins can bind cargo directly and whether other proteins, such as the yeast tetraspan Cos proteins (MacDonald et al., 2015), can assist in ESCRT cargo sorting in plant cells. However, it is reasonable to assume that most of the deubiquitinated cargo molecules that arise after ESCRT-III assembles and before ILV release (Richter et al., 2007, 2013; Johnson et al., 2017) may not be physically connected to the ESCRT coat, and thus could diffuse away from the forming vesicle. Moreover, the escaped cargo molecules cannot be retrieved by the ESCRTs unless they are ubiquitinated again. This means that cargo sequestration into ILVs needs to be extremely efficient and that ILV release needs to be completed before cargo molecules can escape by diffusion. To determine whether the novel vesiculation mechanism that operates in plant MVEs could affect endosomal cargo entrapment, we performed computational simulations of cargo escape in single and concatenated ILV buds based solely on cargo lateral diffusion and membrane geometry. We first calculated the diffusion coefficient of a known ESCRT cargo in plants, the auxin efflux facilitator PIN2 (Spitzer et al., 2009), by two imaging methodologies, single-particle tracking by total internal reflection fluorescence microscopy (TIRFM) and FRAP. Because it is extremely challenging to measure the diffusion of membrane proteins in the highly mobile endosomes, we measured diffusion of PIN2 fused to the photoconvertable tag EosFP (for TIRFM measurements) or to GFP (for FRAP measurements) in the plasma membrane of plant cells devoid of cell walls (protoplasts) as a proxy for PIN2 diffusion on MVE membranes (Videos 4 and 5). Based on the tracking of photoconverted PIN2-EosFP particles in protoplasts, we estimated their diffusion coefficient to be 0.06 µm2/s (Figs. 3 A and S2). Using FRAP of protoplasts expressing PIN2-GFP, we estimated the diffusion coefficient of PIN2-GFP to be 0.17 µm2/s (Fig. 3 B). Similar discrepancies in diffusion coefficients calculated by FRAP or TIRFM have been reported previously for other plasma membrane proteins and may be explained by the fact that FRAP measurements can overestimate lateral diffusion by not considering that part of the fluorescence recovery is because of secretion (Ghosh et al., 2014). Regardless of this discrepancy, the diffusion coefficients obtained for fluorescently tagged PIN2 are within the range of what has been reported for other multispanning membrane proteins in animal and plant cells (Martinière et al., 2012; Oyola-Cintrón et al., 2015; Wang et al., 2016) and are useful estimates of the lateral diffusion of a potentially wide range of ESCRT cargo proteins. Figure 3. Calculation of diffusion coefficients and escape times for PIN2. (A) Distribution of diffusion coefficients (D) for PIN2-EosFP on the plasma membrane of fresh protoplasts (n = 648 tracks from four protoplasts) in a histogram with logarithmically spaced bins. The data were fitted using a Gaussian distribution (red curve). Single-particle tracks of PIN2-EosFP on the plasma membrane of protoplasts were visualized by TIRFM and used to calculate MSD. D values were calculated by linear fit to MSD versus time. (B) Quantitative FRAP experiments of 16 protoplasts to determine diffusion coefficient of PIN2-GFP. Fluorescence intensities during recovery after photobleaching are plotted versus time. (C) Electron tomography slice of a single ILV bud from an Arabidopsis root cell. The yellow line indicates the position of the contour used to segment the ILV bud in IMOD. Bar, 10 nm. (D) Extracted 3D geometry from the segmented ILV bud shown in C and used to run the simulations shown in E–G. Green indicates the lower ILV hemisphere, orange indicates the ILV bud neck, and blue indicates the rest of the ILV bud membrane. Numbers shown in nanometers. (E–G) Simulated escape times of 40 PIN2 particles placed on the lower hemisphere of the ILV bud using the two D values experimentally calculated by TIRFM (0.06 µm2) and by FRAP (0.17 µm2). The colored lines indicate the mean numbers of remaining cargo over time calculated from >300 simulations, and the shaded regions correspond with the SD. Gray vertical lines indicate the mean time required for 50% of the PIN2 particle to escape the ILV bud. Simulations were run either without (E) or with (F) volume exclusion and with a fivefold decrease in D when PIN2 particles entered the neck region (G). VE, volume exclusion. To ensure that the model’s physical properties of the simulated diffusion mimic the endosomal environment, we obtained dual-axis electron tomograms from Arabidopsis MVEs and identified multiple membrane-inward ILV budding events. As cargo deubiquitination is assumed to happen after full assembly of ESCRT-III but before membrane scission (Johnson et al., 2017), we chose those geometries that corresponded with late ILV formation stages (i.e., narrow necks), segmented the cytoplasmic leaflet of the endosomal membrane (Figs. 3 C and S3), and adopted a diffusion model using a particle-based spatial stochastic simulator. Aside from the vesicle geometry, the basic model required specification of the size of the PIN2 molecules, their diffusion constant, the time step of the simulator, and a criterion for measuring exit times. We used three vesicle geometries from tomography reconstructions (Figs. 3 D and S3). Each geometry was divided into four regions: the lower and upper hemisphere of the vesicle, the endosomal limiting membrane, and the neck that connects the upper hemisphere with the endosomal limiting membrane (Figs. 3 D and S3). PIN2 molecules were initially placed at the lower hemisphere, creating a worst-case scenario for each molecule to exit the vesicle. As we were interested in escape times to the endosomal limiting membrane, PIN2 reentering the vesicle membrane from the endosomal surface was not possible in the model; once a PIN2 molecule moved to the endosomal limiting membrane, it was removed via a location-specific degradation reaction with probability 1. This is further justified by the consideration that the area of the limiting endosomal membrane is so large in comparison to that of the vesicle that the chance of the protein diffusing back once it escaped is very low. Consequently, reducing the endosomal surface to a small environment around the vesicle neck was sufficient for our modeling goal and reduced computing time. Escape times were then obtained from PIN2 decay over time. For obtaining a reasonable, maximal number of PIN2 in the vesicle membrane, we considered that PIN2 is predicted to contain 10 α helices as transmembrane domains (UniProt database accession no. Q9LU77; Liu et al., 2014). The diameter of a transmembrane domain is ∼1.5 nm (Rosenberg et al., 2003), and thus the overall area covered by all PIN2 transmembrane domains was calculated to be 17.67 nm2. Therefore, we roughly estimated the membrane area effectively covered by a PIN2 molecule at 20 nm2. Based on the dimension of budding profiles (inner diameter, 23 nm; SD, 3 nm; sample size, 183 ILVs) in tomographic reconstructions of plant MVEs, we estimated that 41 PIN2 molecules could fit in the lower hemispheric region of the forming ILV (∼830 nm2). We modeled the diffusion of 40 PIN2 molecules covering a circular area (Video 6). To estimate changes in escape time according to variations in protein diffusion, we ran simulations for the two experimentally calculated diffusion coefficients, 0.17 µm2/s (D1) and 0.06 µm2/s (D2). The spatial resolution was initially set to 0.715 nm, the pixel size of the tomography dataset. The simulation time step that guarantees such spatial resolution was then calculated as Δt = s2/4D. Because the spatial unit of the imported geometries was in px, we chose to convert all parameters into the unit system px-ms. Simulations stopped when >10 PIN2 molecules were left in the vesicle. To account for steric effects in particle movement such as crowding effects and collisions, we also simulated PIN2 diffusion with volume exclusion. Based solely on diffusion, the 50% PIN2 escape time in simulation with no steric effects was between 10 ms (D1) and 28 ms (D2) (Figs. 3 E and S3), whereas with excluded-volume effects, PIN2 escape times decreased to 5 ms (D1) and 15 ms (D2; Fig. 3 F). This is consistent with theoretical models showing that lateral diffusion of particles with no electrostatic or hydrodynamic interactions is enhanced when steric effects are included (Bruna and Chapman, 2012). We also ran simulations for two additional ILV bud geometries at spatial resolutions of 0.715 nm or 1.43 nm (Fig. S3). Simulations in all ILV profiles yielded similar PIN2 escape dynamics (Fig. 3, E and F; and Fig. S3), confirming that minor geometric differences or different resolutions do not significantly alter the simulations of diffusive behavior and escape times of PIN2. To account more accurately for local changes in particle mobility caused by membrane geometry, we considered the narrow 17-nm-wide neck connecting the body of the vesicle bud to the endosomal limiting membrane. The mobility of the potassium-gated channel KvAP with six transmembrane domains (Lee et al., 2005) was shown to be reduced fivefold in ∼10-nm-wide membrane tubes compared with the diffusion in relatively flat membranes (Domanov et al., 2011). Therefore, we applied a fivefold reduction in the diffusion coefficient to all PIN2 particles entering the neck region in our simulations. We found that under these conditions, the 50% PIN2 escape time was between 7 ms (D1) and 18 ms (D2; Fig. 3 G). It is important to note, however, that the effect of the neck region on cargo diffusion depends on the size, geometry, and mechanical properties of the cargo proteins. Thus, the diffusion retardation effect in the neck regions could be more than fivefold for PIN2 but also less than fivefold for smaller cargo proteins with only one transmembrane-spanning region. Vesicle concatenation and ESCRT cargo diffusion Although ILVs in plant MVEs should be ultimately released into the endosomal or vacuolar lumen for degradation, early ILV bud concatenation could contribute to cargo entrapment while the forming ILV buds remain connected to the limiting membrane. For example, the concatenated ILV bud membranes would be still in contact with the cytoplasm, allowing for cargo deubiquitination, and therefore free cargo diffusion could only start once large and complex ILV networks have formed, greatly reducing the chances of cargo escape. To test this hypothesis in silico, we ran simulations of cargo escape through geometries of two or three concatenated ILV buds (Fig. 4, A and B). Because our electron tomography analysis indicated that the ILV furthest away from the limiting membrane is the first one to form, we placed 40 PIN2 molecules in the hemisphere for the ILVs deepest into the endosomal lumen. Most likely, ESCRT cargo is also found in the other regions of internalized membranes, but we restricted the simulation to the PIN2 pool that would be sequestered earlier when the first ILV bud formed. Because the ILV bud necks and interconnecting bridges between ILVs were similar in diameter, we applied the same fivefold reduction to the PIN2 particles moving into both regions. Compared with simulations on a single ILV bud, PIN2 escape times increased approximately fivefold for two concatenated ILVs and 10-fold for three concatenated ILVs (Fig. 3 B). Figure 4. Simulation of cargo escape in concatenated ILV geometries. (A) Geometries derived from electron tomograms of plant MVEs used for the cargo diffusion simulations depicted in B. Green indicates the lower ILV hemisphere, where the PIN2 molecules were placed to start the simulation, orange indicates the ILV bud neck and interconnecting membrane bridges, blue indicates the rest of the internalized membrane, and white indicates the endosomal limiting membrane. Bars, 10 nm. (B) Simulated escape times of 40 PIN2 particles placed on the lower hemisphere of the ILV bud using two D values (0.06 and 0.17 µm2). The colored lines indicate the mean numbers of remaining cargo over time calculated from >300 simulations, and the shaded regions correspond with the SD. Gray vertical lines indicate the mean times required for 50% of the PIN2 particle to escape the ILV bud. Simulations were run incorporating volume exclusion and a fivefold decrease in D when PIN2 particles entered neck or bridge regions (orange in A). To fully appreciate the meaning of these calculated escape times, it would be critical to know how fast ILVs form. Unfortunately, because endosomes are highly mobile within cells, it is very challenging to measure the timing of ILV formation. However, we know that in animal cells, conventional clathrin-mediated vesiculation at the plasma membrane takes typically 10–20 s (Heuser and Reese, 1973), the entry of Singapore grouper iridovirus virions by endocytosis takes 1 s (Pan et al., 2015), and ultrafast endocytosis at hippocampal synapses proceeds in 50–100 ms (Watanabe et al., 2013). Given the relatively slow speed of such vesiculation events, the increased complexity of the ILV bud network geometry could delay cargo escape, but it does not seem to be sufficient to trap the cargo efficiently. Thus, even if endosome intralumenal vesiculation is as fast as the fastest vesiculation event ever recorded (ultrafast endocytosis), regardless of the mechanism by which ILVs forms (either by single or concatenated budding events), they would have lost most of their cargo by the time they detached from the endosomal membrane. Thus, our simulations predict the presence of additional diffusion barriers within endosomes to trap cargo within ILVs. SNF7 as a component of a persistent diffusion barrier on MVE membranes SNF7 is considered to be the most abundant protein within the ESCRT-III complex and a key component in membrane constriction and ILV formation (Shen et al., 2014; Chiaruttini et al., 2015; Lee et al., 2015). To test whether SNF7 is associated with concatenated ILV buds, we developed antibodies against Arabidopsis SNF7 (SNF7.1; Fig. S4 A) and immunolabeled high-pressure–frozen/freeze-substituted roots. We found that SNF7 not only localized to patches on the surface of the endosomal limiting membrane (Fig. S4 B) but also on interconnected ILV buds (Fig. 5, A–C), indicating that ESCRT-III components are associated with ILV bud networks, most likely stabilizing the narrow membrane bridges between vesicles. Figure 5. Detection of SNF7 in MVEs and isolated vacuoles. (A and B) Immunolabeling of SNF7 on MVEs from high-pressure–frozen/freeze-substituted root cells. (C) The area indicated by a yellow box in B shown at higher magnification. Red arrowheads indicate gold particles. (D) Immunoblot detection of SNF7, cBPPase (cytoplasmic control), and H+PPase (vacuolar membrane control) in protein extracts from isolated protoplasts and vacuoles of wild-type, lip5, and atg7 mutant seedlings. Images of isolated protoplasts and vacuoles are shown at the top. Bars: (A–C) 50 nm; (D) 5 µm. (E) Protease protection assay. Isolated vacuoles were incubated in proteinase K (PK) for 1 h with or without 1% Triton X-100 (TX). SNF7 and H+PPase were detected by immunoblotting. Molecular masses are indicated in kilodaltons. ESCRT-III, including SNF7, has been proposed to corral cargo during MVE sorting (Nickerson et al., 2007; Teis et al., 2008; Henne et al., 2012; Chiaruttini et al., 2015) and to be recycled back to the cytoplasm by the time the ILVs are released into the MVE lumen (Babst et al., 2002). Its presence in narrow necks of ILV buds and inter-ILV bridges is consistent with a potential role of SNF7 as a component of a physical barrier for the lateral diffusion of cargo molecules. However, our computational simulations of cargo escape predicted that upon the removal of such a barrier, half of the cargo molecules would escape in 5–20 ms. This prediction, together with our immunogold labeling in Fig. 5 (A–C), strongly argues for the persistence of the hypothetical barrier imposed by SNF7 at the neck and bridge membranes until the physical separation of the ILVs from the limiting membrane. This situation could lead to the complete or partial entrapment of SNF7 in concatenated ILVs and its delivery to the vacuole upon MVE–vacuole fusion. To test this prediction, we isolated protoplasts and vacuoles from wild-type seedlings. We used antibodies against an Arabidopsis H+PPase to detect vacuolar membranes (Paez-Valencia et al., 2011) and the cytoplasmic FBPase to assess any potential contamination of cytoplasm in the vacuolar fraction (Fig. 5 D). We detected monomeric SNF7 (∼25 KD) in both wild-type protoplasts and vacuoles. To rule out that SNF7 had been delivered to the vacuole by autophagy in an MVE-independent fashion, we isolated vacuoles from the atg7 mutant, which is unable to undergo autophagy of cytoplasmic material (Thompson et al., 2005). We detected comparable amounts of SNF7 in wild-type and atg7 vacuoles (Fig. 5 D). In contrast, we reasoned that if SNF7 actually remains associated with ILV necks and inter-ILV bridges, as shown by our immunogold labeling and inferred from our modeling, mutants with reduced ILV concatenation would trap less SNF7 in MVEs and therefore deliver less SNF7 to vacuoles. Indeed, as predicted, we detected significantly less SNF7 in vacuoles from lip5 mutant seedlings compared with those from wild-type control (Fig. 5 D). This observation is consistent with the hypothesis that SNF7 remains associated with ILVs and is trapped inside MVEs once ILVs are released from the limiting membrane. Finally, to confirm that SNF7 was inside the vacuolar lumen and not associated with the vacuolar surface as a cytoplasmic contaminant, we performed a protease protection assay. Isolated vacuoles were incubated with proteinase K with or without 1% Triton X-100. The anti-H+PPase antibody detects an epitope at the lumenal C-terminal tail (Paez-Valencia et al., 2011) that should be protected from the proteinase K treatment in intact vacuoles. We found both SNF7 and the relevant H+PPase epitope to be degraded by proteinase K only when Triton X-100 was added (Fig. 5 E), demonstrating that the SNF7 pool detected in the vacuolar fraction is indeed located in the vacuolar lumen. To explore the effects of a potential ESCRT-imposed cargo diffusion barrier at the membrane-constricted regions of the ILV networks (i.e., necks and inter-ILV bridges), we ran mathematical simulations using the same geometries as in Fig. 4 but applying a retardation factor of 50× and 500× to the two calculated PIN2 diffusion coefficients at the neck- and ILV-interconnecting bridges (Fig. S5). As expected, decreased diffusion at neck and bridge regions led to a delay in cargo escape that was more pronounced for two and three concatenated vesicles than for a one-vesicle geometry, highlighting that multiple diffusion barriers at bridges/necks of an ILV network could be more efficient in trapping cargo than a single diffusion barrier at the neck of a single ILV bud. Discussion Our analysis of plant MVEs highlights two important aspects of the robustness and plasticity of the ESCRT machinery across eukaryotes. First, plant MVEs are characterized by a drastically different endosomal membrane geometry and degree of ILV concatenation, which must reflect variation in the relative timing of ESCRT-mediated ILV vesicle formation and membrane fission. Second, this ancestral vesiculation mechanism seems to require that some ESCRT components remain associated with ILVs even after membrane fission, which leads to their entrapment within MVEs and vacuolar degradation. During ILV formation, the bending of endosomal membranes starts with the recruitment of ESCRT-II and the assembly of the ESCRT-III complex (Fyfe et al., 2011; Adell et al., 2014; Shen et al., 2014; Chiaruttini et al., 2015). As predicted by their behavior on artificial membranes, SNF7 polymers can form spirals of different radii, but they are not small enough to drive membrane fission (Chiaruttini et al., 2015). The binding of VPS4 to ESCRT-III would lead to a remodeling of the ESCRT-III filaments, driving the constriction of the neck of the budding ILV and the subsequent ILV release (Shen et al., 2014; Chiaruttini et al., 2015). However, how neck constriction and membrane scission are temporally regulated remains unknown. The presence of SNF7 in the plant-concatenated ILV networks further supports the idea that stabilization of narrow ILV necks and bridges by SNF7 (and likely other ESCRT components) can be temporally uncoupled from the membrane fission step. This temporal uncoupling leads to a very different geometry of the internalized endosomal membranes that are, nevertheless, able to sequester cargo efficiently. Other tomographic studies have shown that in yeast and cultured animal cells, all ILVs are free in the MVE lumen (Murk et al., 2003b; Nickerson et al., 2010; Adell et al., 2014). Plants, however, seem to form stable ILV bud networks during MVE sorting. Although plants have overall conserved ESCRT functions, they have also evolved multiple isoforms of conserved ESCRT components and even plant-specific ESCRT proteins (Paez Valencia et al., 2016), which could be responsible for this evolutionary specialization in ILV formation. The capability to form concatenated ILV networks could be related to systems where single ILV release does not proceed fast enough, possibly because of a slower remodeling of the ESCRT-III coat and/or differences in VPS4 kinetics and regulation. It will be interesting to perform structural analyses on MVEs from other multicellular organisms or in different tissues or developmental stages to determine the occurrence and relevance of ILV bud concatenation in other eukaryotic groups. It is reasonable to speculate that ILV bud concatenation allows endosomes to internalize cargo-containing membrane domains faster than a system in which ILVs are formed one by one. Although our computational simulations suggest a delay in cargo escape as the ILV network becomes more complex, membrane geometry by itself does not seem to impose a diffusion barrier strong enough to prevent cargo escape. Thus, it is reasonable to predict that an additional diffusion barrier must remain in place until membrane fission and ILV release is completed. If ESCRT-III components were to be part of these diffusion barriers at necks and inter-ILV bridges, an immediate corollary is that part of them would be trapped inside MVEs and delivered to vacuoles for degradation. A previous study in yeast proposed that ESCRT-III is recycled back to the cytoplasm and not delivered to the vacuole together with the MVE cargo (Babst et al., 2002). However, we have observed in this study that in plants, SNF7 is detected in ILV networks of MVEs and in the vacuolar lumen. Moreover, as shown by a lip5 mutation (Fig. 5 D), the amount of SNF7 delivered to the vacuole appears to be correlated with the extent of ILV concatenation. Additionally, other ESCRT-III subunits, such as CHMP1A, VPS24.1, VPS2.1, and ISTL1, have been detected in isolated vacuoles by proteomic approaches (Yoshida et al., 2013), suggesting that SNF7 is not the only ESCRT-III subunit trapped within MVEs and delivered to vacuoles. Thus, we propose that at least in systems with concatenated ILVs, a fraction of SNF7 and likely other ESCRT-III components remain associated with ILV necks and bridges acting as cargo diffusion barriers even after ILVs lose physical continuity with the limiting membrane. Materials and methods Calculations of PIN2 diffusion coefficients For the FRAP measurements, protoplasts of root tips of 7-d-old seedlings expressing ProPIN2::PIN2:GFP were obtained with a modification of a previously described method (Yoo et al., 2007). In brief, root tips were incubated in enzyme solution (20 mM MES, pH 5.7, 0.4 M mannitol, 20 mM KCl, 10 mM CaCl2 containing 1.5% [wt/vol] cellulose R10, and 0.4% [wt/vol] macerozyme R-10) in light for 3 h with mild agitation. Protoplasts were washed in W5 solution (2 mM MES, pH 5.7, 154 mM NaCl, 125 mM CaCl2, and 5 mM KCl) and resuspended in WI solution (4 mM MES, pH 5.7, 0.5 M mannitol, and 20 mM KCl). Protoplasts were visually screened for fluorescence signal. Experiments were performed on an LSM 780 microscope (ZEISS) with a Plan Apochromat 100× 1.46 NA oil differential interference contrast M27 objective with excitation at 488 nm and emission detected between 490–553 nm. To establish the prebleach intensity of the GFP, five images were taken immediately before the bleaching step. A region of interest with a width of 5 µm was bleached on the plasma membrane. Recovery of fluorescence was recorded during 3 min with a delay of 1.40 s between frames. Normalization to prebleach intensity and for loss of fluorescence during imaging was performed as previously described (Phair et al., 2004). The experimental data were analyzed by the equation I(t) = I(final)(1 − (w2(w2 + 4π Dt)−1)1/2, described by Ellenberg et al. (1997) for strip regions of interest, in which I(t) is intensity as a function of time, I(final) is final intensity reached after complete recovery, w is strip width, and D is diffusion coefficient. TIRFM imaging was performed using an Elyra PS.1 microscope (ZEISS) with a 100× objective (α Plan Apochromat; 1.46 NA). The gain of the electron-multiplying charge-coupled device camera was set to 300 for all experiments, the setting was within the linear dynamic range of the camera, and images were acquired at 30-ms exposure times. Protoplasts of root tips of plants expressing Pro35S::PIN2:EosFP (Dhonukshe et al., 2007) were visually screened for fluorescent signal. To achieve photoconversion at a density compatible with single-particle tracking, a 405-nm laser was used at 0.1% while imaging. The red form of PIN2-EosFP was imaged with 561-nm laser excitation, and fluorescence was collected using a band-pass 570–650 + long-pass 750 filter. Single-particle track segmentation was performed using the TrackMate plugin for Fiji (ImageJ; National Institutes of Health). A total of 648 tracks with a minimum trajectory length of 150 ms and collected from four protoplasts were exported and analyzed with MATLAB using a previously described package for mean square displacement (MSD) analysis (Tarantino et al., 2014) and available at MathWorks. According to this method, we calculated the MSD for each individual trajectory and fitted its log–log representation with a linear function. When the MSD curve can be modeled by ρ(r) = r2 = Γta, then the log(r2) = f(log(t)) is fitted with Γ + α log(t). Those individual MSD curves with an R2 coefficient <0.8 were discarded. Modeling and simulation of PIN2 diffusion on the MVE membrane The PIN2-associated diffusion constant D was directly obtained from experimental measurements. The simulation time step dt depended on the pixel size of the imported MVE geometry and was chosen such that the spatial resolution s = √(4Ddt) of the simulation was guaranteed to be either 0.715 nm or ∼0.226 nm for all values of D. The latter value corresponds to a 10× smaller dt. Based on the presence of 10 predicted transmembrane domains and an intracellular domain, we estimated that a single PIN2 occupies an area of 20 nm2. Assuming an (idealized) spherical shape for all PIN2 molecules, we obtained a radius r of ∼2.5 nm. This value was required for incorporating volume exclusion effects into PIN2 interaction simulations similar to hard-sphere collision models. As a unit system, we chose px and ms, yielding different parameter values for geometries with pixel size 0.715 nm and 1.43 nm. Fig. S5 summarizes the chosen time step dt in ms for any given D in µm2/s and px2/ms. All simulations of PIN2 diffusion were performed using the computer program Smoldyn (Andrews and Bray, 2004), a particle-based spatial-stochastic simulator. To import the geometries of MVEs (see Figs. 3 and S5), we generated segmentation models of endosomal and ILV membranes using IMOD (Kremer et al., 1996). The files containing meshed objects were first converted into vrml2 format with imod2vrml2 and then into Smoldyn-readable format using wrl2smol from the Smoldyn package. The resulting file was imported into MATLAB (MathWorks) to manually correct missing triangles and other flaws. Groups of surface triangles within the individual geometries were defined according to their location in the “vesicle bottom” (membrane region furthest away from the endosomal limiting membrane), “neck or bridge” (used for applying a 5× reduction in PIN2 diffusion coefficients), “vesicle” (the remaining surface of the vesicles), and “endosomal limiting membrane” (needed to identify the exit of PIN2) to be used as different Smoldyn surfaces. Simulations were initialized by uniformly placing 40 PIN2 molecules into the lower hemisphere of the vesicle (“vesicle bottom”). As soon as a PIN2 molecule entered the “endosomal limiting membrane” region, it was permanently removed from the system. To investigate the impact of volume exclusion, we ran simulations using so-called “excluded volume reactions” or “bounce reactions” that efficiently mimic excluded volume behavior by pushing molecules apart that are closer than 2r. We further tested that PIN2 molecules cannot pass each other when diffusing on cylindrical shapes with a radius smaller than 2r. This was to be expected because we had chosen a time step dt that yielded a spatial resolution s that was significantly smaller than 2r. Lastly, numbers of PIN2 were saved at every time step in every simulation. The output file was then imported into and analyzed with MATLAB. Preparation of anti-SNF7 antibodies Escherichia coli BL21 cells expressing Arabidopsis SNF7.1 fused to GST were grown in 3 ml of Luria-Bertani broth containing kanamycin (50 µg/ml) at 37°C for 3 h, diluted with fresh broth (1:100), and, after 3 h of shaking at 37°C, protein expression was induced by addition of IPTG (0.1 mM final concentration). After 3 h, inclusion bodies containing GST-SNF7 were purified as described previously (Rodríguez-Carmona et al., 2010). In brief, 20 ml of bacterial culture were centrifuged at 5,000 g and 4°C for 5 min and resuspended in 20 ml of lysis buffer (50 mM Tris-HCl, pH 8.1, 100 mM NaCl, and 1 mM EDTA) and frozen at −80°C o/n. After thawing, 100 µl of 100-mM PMSF and 400 µl of 50-mg/ml lysozyme were added. After 2 h incubation at 37°C, 100 µl of 0.5% Triton X-100 was added and incubated at room temperature for 1 h. The mixture was then ice jacketed and sonicated between 4 and 10 cycles of 10 min at 40% amplitude under 0.5-s cycles. Then, 5 µl of NP-40 was added to the rest of the suspension, and samples were incubated at 4°C for 1 h. DNA was removed with 15 µl of 1% DNase and 15 µl of 1-M MgSO4 for 45 min at 37°C. Finally, samples were centrifuged at 4°C at 15,000 g for 15 min, and the pellet containing inclusion bodies was washed once with 1 ml of lysis buffer containing 0.5% Triton X-100. After a final centrifugation at 15,000 g for 15 min at 4°C, pellets were stored at −80°C until analysis. All incubations were done under agitation. The bacterial lysate and purified inclusion bodies were characterized by electrophoresis in 10% SDS polyacrylamide gel. Proteins were visualized by staining with Coomassie brilliant blue (Bio-Rad Laboratories). Isolated protein bodies were used for production of polyclonal antibodies in rabbits. For detection of SNF7 proteins in plant extracts, Arabidopsis roots were frozen in liquid nitrogen and homogenized in a buffer containing 250 mM sorbitol, 50 mM Hepes–bis-Tris propane (BTP), pH 7.4, 4-(2-hydroxyethyl) piperazine-1-ethanesulfonic acid, N-(2-hydroxyethyl) piperazine-N′-(2-ethanesulfonic acid) (Sigma-Aldrich), 25 mM ascorbic acid, 1 mM DTT (Fluka), 6 mM EGTA (Sigma-Aldrich), 1.2% (wt/vol) polyvinyl porrolidone-40 (Sigma-Aldrich) and cOmplete protease inhibitor cocktail (Roche). A 1:2 (wt/vol) ratio of tissue and homogenization media was used. The homogenization media was filtered through Miracloth (EMD Millipore) and centrifuged at 10,000 g for 15 min. The supernatant was recovered and kept aside. Cleared supernatants were mixed with acetone (1 ml of acetone per 200 µl of cleared homogenate), incubated for 10 min at −20°C, and centrifuged at 20,000 g for 10 min. The supernatant was discarded, and the pellet was suspended in 50 µl of a resuspension buffer containing 250 mM sorbitol, 25 mM Hepes-BTP, pH 7.4, 1 mM DTT, and cOmplete EDTA-free protease inhibitor cocktail. Samples were resolved by SDS-PAGE, transferred to nitrocellulose membranes, and analyzed by immunoblotting with the affinity-purified rabbit anti-SNF7.1 via a chemiluminescence detection system (ECL Western blotting substrate; Thermo Fisher Scientific). Vacuole isolation and protease protection assay Protoplasts from wild-type Col-0, atg7, and lip5-1 were isolated from a 14-d-old seedling according to Zhai et al. (2009). Intact vacuoles were removed and isolated from protoplasts using thermal and osmotic disruption of the plasma membrane followed by fractionation in a Ficol density gradient (Robert et al., 2007). Isolated vacuoles (50 µg of protein) were resuspended in 100 µl of 10 mM Tris-MES, pH 6.9, and 50 µg of proteinase K with or without 1% Triton X-100 and then were incubated on ice for 1 h (Yamaguchi et al., 2003). TCA was added to terminate the reaction and precipitate proteins. Samples were resolved by SDS-PAGE, transferred to nitrocellulose membranes, and analyzed by immunoblotting with the affinity-purified rabbit anti-SNF7.1, anti-H+PPase (Paez-Valencia et al., 2011), and cFBPase (AS04043; Agrisera). Electron microscopy and electron tomography Roots from 7-d-old Arabidopsis seedlings were high-pressure frozen in a BAL-TEC HPM 010 high-pressure freezer. Part of the high-pressure–frozen samples were freeze substituted in 2% OsO4 in acetone for 12 h at −80°C followed by infiltration in EPON resin (Electron Microscopy Sciences) for structural analysis. Another part of the samples were freeze-substituted in 0.2% glutaraldehyde plus 0.2% uranyl acetate in acetone at −90°C for 4 d in an automated freeze-substitution device (Leica Microsystems) and embedded in Lowicryl HM20 (Electron Microscopy Sciences) for immunolabeling. HM20 sections of HM20-embedded roots were mounted on formvar-coated nickel grids and blocked for 20 min with a 5% (wt/vol) solution of nonfat milk in TBS containing 0.1% Tween-20. The sections were incubated in the primary polyclonal anti-SNF7.1 antibodies for 1 h, rinsed in TBS containing 0.5% Tween-20, and transferred to the secondary antibody (1:10 anti–rabbit IgG) conjugated to either 10- or 6-nm gold particles for 1 h. Controls lacked the primary antibodies. Sections (300 nm thick) of EPON-embedded roots were mounted on formvar-coated copper slot grids and stained with 2% uranyl acetate in 70% methanol and Reynold’s lead citrate (2.6% lead nitrate and 3.5% sodium citrate, pH 12). Colloidal gold particles 15 nm in diameter were used as fiducial markers to align the series of tilted images. The sections were imaged in a Tecnai TF30 intermediate voltage electron microscope (FEI) operated at 300 kV. The images were taken from +60° to –60° at 1.0° intervals about two orthogonal axes (Mastronarde, 1997) and collected in a US1000 camera (Gatan) at a pixel size of 0.715 or 1.43 nm. The images were aligned as described by Ladinsky et al. (1999). Tomograms were computed using simultaneous iterative reconstruction technique (Gilbert, 1972). Merging of the two single-axis tomograms was done as previously described (Mastronarde, 1997). Tomograms were displayed and analyzed with 3Dmod, the graphic component of the IMOD software package (Kremer et al., 1996). The thinning factor for each tomogram was calculated and corrected in the models. High-pressure–frozen S. cerevisiae cells were freeze substituted in 0.25% glutaraldehyde and 0.05% uranyl acetate in acetone at −80°C for 3 d and then at −20°C for one day and then were infiltrated with HM20. Samples were sectioned, and 200-nm-thick sections were placed on formvar-coated copper slot grids and stained with Reynold’s lead citrate for 10 min. Fiducial markers of 10 nm gold were placed on both sides of the grid. Tomograms were collected in the same manner as for the Arabidopsis roots. Online supplemental material Fig. S1 shows Arabidopsis MVEs. Fig. S2 shows single-particle tracks collected by TIRFM from a protoplast expressing PIN2-EosFP and used to calculate MSD. Fig. S3 shows cargo diffusion simulations. Fig. S4 shows how SNF7 was detected inside MVEs and vacuoles. Fig. S5 shows simulated escape times of 40 PIN2 particles placed on the lower hemisphere of the ILV bud using two D values. Video 1 is a Dual-axis electron tomogram of a wild-type Arabidopsis MVE in a cortical root cell. Video 2 is a dual-axis electron tomogram of an MVE in a lip5 Arabidopsis root cell. Video 3 is a dual-axis electron tomogram of two chmp1a chmp1b Arabidopsis MVEs from a cortical root cell. Video 4 shows photoconverted PIN2-EosFP particles on Arabidopsis protoplasts from root tips imaged by TIRFM. Video 5 shows FRAP of PIN2-GFP in Arabidopsis protoplasts imaged by confocal microscopy. Video 6 shows simulation of PIN2 particle diffusion in a single ILV budding profile. Supplementary Material Supplemental Materials (PDF) Video 1 Video 2 Video 3 Video 4 Video 5 Video 6 Acknowledgments We thank Bryan Sibert and Christoph Spitzer for processing yeast and Arabidopsis chmp1 material by high-pressure freezing/freeze substitution, and we thank Anjon Audhya and Gabriele Monshausen for critical reading of the manuscript. P. Ahlquist is an investigator at the Howard Hughes Medical Institute and the Morgridge Institute for Research. This work was supported by National Science Foundation grants MCB1157824 and MCB1614965 to M.S. Otegui and funds from the University of Wisconsin-Madison Department of Botany to R.A. Buono. The authors declare no competing financial interests. Author contributions: R.A. Buono, A. Leier, J. Paez-Valencia, T.T. Marquez-Lago, J. Pennington, K. Goodman, N. Miller, P. Ahlquist, and M.S. Otegui designed research; R.A. Buono, A. Leier, T.T. Marquez-Lago, J. Pennington, and K. Goodman performed research; A. Leier and T.T. Marquez-Lago performed cargo diffusion simulations; R.A. Buono, A. Leier, J. Paez-Valencia, T.T. Marquez-Lago, N. Miller, P. Ahlquist, and M.S. Otegui analyzed data; and A. Leier, T.T. Marquez-Lago, P. Ahlquist, and M.S. Otegui wrote the paper with advice from the other authors. Abbreviations used: ESCRT endosomal sorting complex required for transport ILV intralumenal vesicle MSD mean square displacement MVE multivesicular endosome TIRFM total internal reflection fluorescence microscopy ==== Refs Adell, M.A.Y., G.F. Vogel, M. Pakdel, M. Müller, H. Lindner, M.W. Hess, and D. Teis. 2014. Coordinated binding of Vps4 to ESCRT-III drives membrane neck constriction during MVB vesicle formation. J. Cell Biol. 205 :33–49. 10.1083/jcb.201310114 24711499 Andrews, S.S., and D. Bray. 2004. Stochastic simulation of chemical reactions with spatial resolution and single molecule detail. Phys. Biol. 1 :137–151. 10.1088/1478-3967/1/3/001 16204833 Babst, M., D.J. Katzmann, E.J. Estepa-Sabal, T. Meerloo, and S.D. Emr. 2002. Escrt-III: An endosome-associated heterooligomeric protein complex required for mvb sorting. Dev. Cell. 3 :271–282. 10.1016/S1534-5807(02)00220-4 12194857 Bruna, M., and S.J. Chapman. 2012. Excluded-volume effects in the diffusion of hard spheres. Phys. Rev. E Stat. Nonlin. Soft Matter Phys. 85 :011103. 10.1103/PhysRevE.85.011103 22400508 Buono, R.A., J. Paez-Valencia, N.D. Miller, K. Goodman, C. Spitzer, E.P. Spalding, and M.S. Otegui. 2016. Role of SKD1 regulators LIP5 and IST1-LIKE1 in endosomal sorting and plant development. Plant Physiol. 171 :251–264. 10.1104/pp.16.00240 26983994 Chiaruttini, N., L. Redondo-Morata, A. Colom, F. Humbert, M. Lenz, S. Scheuring, and A. Roux. 2015. Relaxation of loaded ESCRT-III spiral springs drives membrane deformation. Cell. 163 :866–879. 10.1016/j.cell.2015.10.017 26522593 Dhonukshe, P., F. Aniento, I. Hwang, D.G. Robinson, J. Mravec, Y.D. Stierhof, and J. Friml. 2007. Clathrin-mediated constitutive endocytosis of PIN auxin efflux carriers in Arabidopsis. Curr. Biol. 17 :520–527. 10.1016/j.cub.2007.01.052 17306539 Domanov, Y.A., S. Aimon, G.E. Toombes, M. Renner, F. Quemeneur, A. Triller, M.S. Turner, and P. Bassereau. 2011. Mobility in geometrically confined membranes. Proc. Natl. Acad. Sci. USA. 108 :12605–12610. 10.1073/pnas.1102646108 21768336 Ellenberg, J., E.D. Siggia, J.E. Moreira, C.L. Smith, J.F. Presley, H.J. Worman, and J. Lippincott-Schwartz. 1997. Nuclear membrane dynamics and reassembly in living cells: Targeting of an inner nuclear membrane protein in interphase and mitosis. J. Cell Biol. 138 :1193–1206. 10.1083/jcb.138.6.1193 9298976 Fyfe, I., A.L. Schuh, J.M. Edwardson, and A. Audhya. 2011. Association of ESCRT-II with VPS20 generates a curvature-sensitive protein complex capable of nucleating filaments of ESCRT-III. J. Biol. Chem. 286 :34262–34270. 10.1074/jbc.M111.266411 21835927 Ghosh, D., A. Segal, and T. Voets. 2014. Distinct modes of perimembrane TRP channel turnover revealed by TIR-FRAP. Sci. Rep. 4 :7111. 10.1038/srep07111 25407951 Gilbert, P. 1972. Iterative methods for the three-dimensional reconstruction of an object from projections. J. Theor. Biol. 36 :105–117. 10.1016/0022-5193(72)90180-4 5070894 Gomez-Navarro, N., and E.A. Miller. 2016. COP-coated vesicles. Curr. Biol. 26 :R54–R57. 10.1016/j.cub.2015.12.017 26811885 Henne, W.M., N.J. Buchkovich, and S.D. Emr. 2011. The ESCRT pathway. Dev. Cell. 21 :77–91. 10.1016/j.devcel.2011.05.015 21763610 Henne, W.M., N.J. Buchkovich, Y. Zhao, and S.D. Emr. 2012. The endosomal sorting complex ESCRT-II mediates the assembly and architecture of ESCRT-III helices. Cell. 151 :356–371. 10.1016/j.cell.2012.08.039 23063125 Heuser, J.E., and T.S. Reese. 1973. Evidence for recycling of synaptic vesicle membrane during transmitter release at the frog neuromuscular junction. J. Cell Biol. 57 :315–344. 10.1083/jcb.57.2.315 4348786 Hurley, J.H. 2015. ESCRTs are everywhere. EMBO J. 34 :2398–2407. 10.15252/embj.201592484 26311197 Johnson, N., M. West, and G. Odorizzi. 2017. Regulation of yeast ESCRT-III membrane scission activity by the Doa4 ubiquitin hydrolase. Mol. Biol. Cell. 28 :661–672. 10.1091/mbc.E16-11-0761 28057764 Kremer, J.R., D.N. Mastronarde, and J.R. McIntosh. 1996. Computer visualization of three-dimensional image data using IMOD. J. Struct. Biol. 116 :71–76. 10.1006/jsbi.1996.0013 8742726 Ladinsky, M.S., D.N. Mastronarde, J.R. McIntosh, K.E. Howell, and L.A. Staehelin. 1999. Golgi structure in three dimensions: Functional insights from the normal rat kidney cell. J. Cell Biol. 144 :1135–1149. 10.1083/jcb.144.6.1135 10087259 Lee, I.H., H. Kai, L.A. Carlson, J.T. Groves, and J.H. Hurley. 2015. Negative membrane curvature catalyzes nucleation of endosomal sorting complex required for transport (ESCRT)-III assembly. Proc. Natl. Acad. Sci. USA. 112 :15892–15897. 10.1073/pnas.1518765113 26668364 Lee, S.Y., A. Lee, J. Chen, and R. MacKinnon. 2005. Structure of the KvAP voltage-dependent K+ channel and its dependence on the lipid membrane. Proc. Natl. Acad. Sci. USA. 102 :15441–15446. 10.1073/pnas.0507651102 16223877 Leung, K.F., J.B. Dacks, and M.C. Field. 2008. Evolution of the multivesicular body ESCRT machinery; Retention across the eukaryotic lineage. Traffic. 9 :1698–1716. 10.1111/j.1600-0854.2008.00797.x 18637903 Liu, B., J. Zhang, L. Wang, J. Li, H. Zheng, J. Chen, and M. Lu. 2014. A survey of Populus PIN-FORMED family genes reveals their diversified expression patterns. J. Exp. Bot. 65 :2437–2448. 10.1093/jxb/eru129 24663343 MacDonald, C., N.J. Buchkovich, D.K. Stringer, S.D. Emr, and R.C. Piper. 2012. Cargo ubiquitination is essential for multivesicular body intralumenal vesicle formation. EMBO Rep. 13 :331–338. 10.1038/embor.2012.18 22370727 MacDonald, C., J.A. Payne, M. Aboian, W. Smith, D.J. Katzmann, and R.C. Piper. 2015. A family of tetraspans organizes cargo for sorting into multivesicular bodies. Dev. Cell. 33 :328–342. 10.1016/j.devcel.2015.03.007 25942624 Martinière, A., I. Lavagi, G. Nageswaran, D.J. Rolfe, L. Maneta-Peyret, D.-T. Luu, S.W. Botchway, S.E.D. Webb, S. Mongrand, C. Maurel, 2012. Cell wall constrains lateral diffusion of plant plasma-membrane proteins. Proc. Natl. Acad. Sci. USA. 109 :12805–12810. 10.1073/pnas.1202040109 22689944 Mastronarde, D.N. 1997. Dual-axis tomography: An approach with alignment methods that preserve resolution. J. Struct. Biol. 120 :343–352. 10.1006/jsbi.1997.3919 9441937 Murk, J.L.A.N., B.M. Humbel, U. Ziese, J.M. Griffith, G. Posthuma, J.W. Slot, A.J. Koster, A.J. Verkleij, H.J. Geuze, and M.J. Kleijmeer. 2003 b. Endosomal compartmentalization in three dimensions: Implications for membrane fusion. Proc. Natl. Acad. Sci. USA. 100 :13332–13337. 10.1073/pnas.2232379100 14597718 Nickerson, D.P., M.R. Russell, and G. Odorizzi. 2007. A concentric circle model of multivesicular body cargo sorting. EMBO Rep. 8 :644–650. 10.1038/sj.embor.7401004 17603537 Nickerson, D.P., M. West, R. Henry, and G. Odorizzi. 2010. Regulators of Vps4 ATPase activity at endosomes differentially influence the size and rate of formation of intralumenal vesicles. Mol. Biol. Cell. 21 :1023–1032. 10.1091/mbc.E09-09-0776 20089837 Obita, T., S. Saksena, S. Ghazi-Tabatabai, D.J. Gill, O. Perisic, S.D. Emr, and R.L. Williams. 2007. Structural basis for selective recognition of ESCRT-III by the AAA ATPase Vps4. Nature. 449 :735–739. 10.1038/nature06171 17928861 Oyola-Cintrón, J., D. Caballero-Rivera, L. Ballester, C.A. Baéz-Pagán, H.L. Martínez, K.P. Vélez-Arroyo, O. Quesada, and J.A. Lasalde-Dominicci. 2015. Lateral diffusion, function, and expression of the slow channel congenital myasthenia syndrome αC418W nicotinic receptor mutation with changes in lipid raft components. J. Biol. Chem. 290 :26790–26800. 10.1074/jbc.M115.678573 26354438 Paez Valencia, J., K. Goodman, and M.S. Otegui. 2016. Endocytosis and endosomal trafficking in plants. Annu. Rev. Plant Biol. 67 :309–335. 10.1146/annurev-arplant-043015-112242 27128466 Paez-Valencia, J., A. Patron-Soberano, A. Rodriguez-Leviz, J. Sanchez-Lares, C. Sanchez-Gomez, P. Valencia-Mayoral, G. Diaz-Rosas, and R. Gaxiola. 2011. Plasma membrane localization of the type I H+-PPase AVP1 in sieve element–companion cell complexes from Arabidopsis thaliana. Plant Sci. 181 :23–30. 10.1016/j.plantsci.2011.03.008 21600394 Pan, Y., S. Wang, Y. Shan, D. Zhang, J. Gao, M. Zhang, S. Liu, M. Cai, H. Xu, G. Li, 2015. Ultrafast tracking of a single live virion during the invagination of a cell membrane. Small. 11 :2782–2788. 10.1002/smll.201403491 25689837 Phair, R.D., S.A. Gorski, and T. Misteli. 2004. Measurement of dynamic protein binding to chromatin in vivo, using photobleaching microscopy. Methods Enzymol. 375 :393–414. 10.1016/S0076-6879(03)75025-3 14870680 Richter, C., M. West, and G. Odorizzi. 2007. Dual mechanisms specify Doa4-mediated deubiquitination at multivesicular bodies. EMBO J. 26 :2454–2464. 10.1038/sj.emboj.7601692 17446860 Richter, C.M., M. West, and G. Odorizzi. 2013. Doa4 function in ILV budding is restricted through its interaction with the Vps20 subunit of ESCRT-III. J. Cell Sci. 126 :1881–1890. 10.1242/jcs.122499 23444383 Robert, S., J. Zouhar, C. Carter, and N. Raikhel. 2007. Isolation of intact vacuoles from Arabidopsis rosette leaf–derived protoplasts. Nat. Protoc. 2 :259–262. 10.1038/nprot.2007.26 17406583 Rodríguez-Carmona, E., O. Cano-Garrido, J. Seras-Franzoso, A. Villaverde, and E. García-Fruitós. 2010. Isolation of cell-free bacterial inclusion bodies. Microb. Cell Fact. 9 :71. 10.1186/1475-2859-9-71 20849629 Rosenberg, M.F., A.B. Kamis, R. Callaghan, C.F. Higgins, and R.C. Ford. 2003. Three-dimensional structures of the mammalian multidrug resistance P-glycoprotein demonstrate major conformational changes in the transmembrane domains upon nucleotide binding. J. Biol. Chem. 278 :8294–8299. 10.1074/jbc.M211758200 12501241 Schuh, A.L., and A. Audhya. 2014. The ESCRT machinery: From the plasma membrane to endosomes and back again. Crit. Rev. Biochem. Mol. Biol. 49 :242–261. 10.3109/10409238.2014.881777 24456136 Shen, Q.T., A.L. Schuh, Y. Zheng, K. Quinney, L. Wang, M. Hanna, J.C. Mitchell, M.S. Otegui, P. Ahlquist, Q. Cui, and A. Audhya. 2014. Structural analysis and modeling reveals new mechanisms governing ESCRT-III spiral filament assembly. J. Cell Biol. 206 :763–777. 10.1083/jcb.201403108 25202029 Spitzer, C., F.C. Reyes, R. Buono, M.K. Sliwinski, T.J. Haas, and M.S. Otegui. 2009. The ESCRT-related CHMP1A and B proteins mediate multivesicular body sorting of auxin carriers in Arabidopsis and are required for plant development. Plant Cell. 21 :749–766. 10.1105/tpc.108.064865 19304934 Sundborger, A.C., S. Fang, J.A. Heymann, P. Ray, J.S. Chappie, and J.E. Hinshaw. 2014. A dynamin mutant defines a superconstricted prefission state. Cell Reports. 8 :734–742. 10.1016/j.celrep.2014.06.054 25088425 Tarantino, N., J.Y. Tinevez, E.F. Crowell, B. Boisson, R. Henriques, M. Mhlanga, F. Agou, A. Israël, and E. Laplantine. 2014. TNF and IL-1 exhibit distinct ubiquitin requirements for inducing NEMO-IKK supramolecular structures. J. Cell Biol. 204 :231–245. 10.1083/jcb.201307172 24446482 Teis, D., S. Saksena, and S.D. Emr. 2008. Ordered assembly of the ESCRT-III complex on endosomes is required to sequester cargo during MVB formation. Dev. Cell. 15 :578–589. 10.1016/j.devcel.2008.08.013 18854142 Thompson, A.R., J.H. Doelling, A. Suttangkakul, and R.D. Vierstra. 2005. Autophagic nutrient recycling in Arabidopsis directed by the ATG8 and ATG12 conjugation pathways. Plant Physiol. 138 :2097–2110. 10.1104/pp.105.060673 16040659 Traub, L.M., and J.S. Bonifacino. 2013. Cargo recognition in clathrin-mediated endocytosis. Cold Spring Harb. Perspect. Biol. 5 :a016790. 10.1101/cshperspect.a016790 24186068 Wang, L., A. Dumoulin, M. Renner, A. Triller, and C.G. Specht. 2016. The role of synaptopodin in membrane protein diffusion in the dendritic spine neck. PLoS One. 11 :e0148310. 10.1371/journal.pone.0148310 26840625 Watanabe, S., B.R. Rost, M. Camacho-Pérez, M.W. Davis, B. Söhl-Kielczynski, C. Rosenmund, and E.M. Jorgensen. 2013. Ultrafast endocytosis at mouse hippocampal synapses. Nature. 504 :242–247. 10.1038/nature12809 24305055 Wemmer, M., I. Azmi, M. West, B. Davies, D. Katzmann, and G. Odorizzi. 2011. Bro1 binding to Snf7 regulates ESCRT-III membrane scission activity in yeast. J. Cell Biol. 192 :295–306. 10.1083/jcb.201007018 21263029 Winter, V., and M.-T. Hauser. 2006. Exploring the ESCRTing machinery in eukaryotes. Trends Plant Sci. 11 :115–123. 10.1016/j.tplants.2006.01.008 16488176 Yamaguchi, T., M.P. Apse, H. Shi, and E. Blumwald. 2003. Topological analysis of a plant vacuolar Na+/H+ antiporter reveals a luminal C terminus that regulates antiporter cation selectivity. Proc. Natl. Acad. Sci. USA. 100 :12510–12515. 10.1073/pnas.2034966100 14530406 Yoo, S.-D., Y.-H. Cho, and J. Sheen. 2007. Arabidopsis mesophyll protoplasts: a versatile cell system for transient gene expression analysis. Nat. Protoc. 2 :1565–1572. 10.1038/nprot.2007.199 17585298 Yoshida, K., M. Ohnishi, Y. Fukao, Y. Okazaki, M. Fujiwara, C. Song, Y. Nakanishi, K. Saito, T. Shimmen, T. Suzaki, 2013. Studies on vacuolar membrane microdomains isolated from Arabidopsis suspension-cultured cells: Local distribution of vacuolar membrane proteins. Plant Cell Physiol. 54 :1571–1584. 10.1093/pcp/pct107 23903016 Zhai, Z., H.I. Jung, and O.K. Vatamaniuk. 2009. Isolation of protoplasts from tissues of 14-day-old seedlings of Arabidopsis thaliana. J. Vis. Exp. 30 :1149. 10.3791/1149

PMC005xxxxxx/PMC5551575.txt

==== Front J Exp Med J Exp Med jem jem The Journal of Experimental Medicine 0022-1007 1540-9538 The Rockefeller University Press 28663435 20170229 10.1084/jem.20170229 Research Articles Brief Definitive Report 311 307 314 Critical role for Sec22b-dependent antigen cross-presentation in antitumor immunity Cross-presentation in antitumor immunity http://orcid.org/0000-0003-0555-0653 Alloatti Andrés 1 Rookhuizen Derek C. 1* Joannas Leonel 1* Carpier Jean-Marie 1 http://orcid.org/0000-0002-1607-1749 Iborra Salvador 2 http://orcid.org/0000-0002-2424-7097 Magalhaes Joao G. 1 Yatim Nader 3 http://orcid.org/0000-0001-8596-843X Kozik Patrycja 1 http://orcid.org/0000-0003-2890-3984 Sancho David 2 Albert Matthew L. 34 http://orcid.org/0000-0001-8583-8416 Amigorena Sebastian 1 1 INSERM U932, PSL Research University, Institut Curie, Paris, France 2 Centro Nacional de Investigaciones Cardiovasculares Carlos III (CNIC), Madrid, Spain 3 INSERM U1223, Institut Pasteur, Paris, France 4 Department of Cancer Immunology, Genentech, San Francisco, CA Correspondence to Sebastian Amigorena: sebastian.amigorena@curie.fr * D.C. Rookhuizen and L. Joannas contributed equally to this paper. P. Kozik’s present address is MRC Laboratory of Molecular Biology, Cambridge, England, UK. 07 8 2017 214 8 22312241 02 2 2017 17 5 2017 14 6 2017 © 2017 Alloatti et al. 2017 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). Alloatti et al. show that Sec22b-dependent antigen cross-presentation is critical to developing effective antitumor CD8+ T cell responses. Conditional deletion of Sec22b in dendritic cells decreases immune response against dead cells and promotes resistance to immunotherapy with anti–PD-1. CD8+ T cells mediate antigen-specific immune responses that can induce rejection of solid tumors. In this process, dendritic cells (DCs) are thought to take up tumor antigens, which are processed into peptides and loaded onto MHC-I molecules, a process called “cross-presentation.” Neither the actual contribution of cross-presentation to antitumor immune responses nor the intracellular pathways involved in vivo are clearly established because of the lack of experimental tools to manipulate this process. To develop such tools, we generated mice bearing a conditional DC-specific mutation in the sec22b gene, a critical regulator of endoplasmic reticulum–phagosome traffic required for cross-presentation. DCs from these mice show impaired cross-presentation ex vivo and defective cross-priming of CD8+ T cell responses in vivo. These mice are also defective for antitumor immune responses and are resistant to treatment with anti–PD-1. We conclude that Sec22b-dependent cross-presentation in DCs is required to initiate CD8+ T cell responses to dead cells and to induce effective antitumor immune responses during anti–PD-1 treatment in mice. Institut Curie Institut National de la Santé et de la Recherche Médicale http://dx.doi.org/10.13039/501100001677 Centre National de la Recherche Scientifique http://dx.doi.org/10.13039/501100004794 la Ligue Contre le Cancer http://dx.doi.org/10.13039/501100004099 EL2014.LNCC/SA Association de Recherche Contre le Cancer H2020 European Research Council http://dx.doi.org/10.13039/100010663 2013-AdG 340046 DCBIOX Institut National Du Cancer http://dx.doi.org/10.13039/501100006364 PLBIO13-057 Agence Nationale de la Recherche http://dx.doi.org/10.13039/501100001665 ANR-11-LABX-0043 ANR-10-IDEX-0001-02 PSL ANR-16-CE15001801 ANR-16-CE18002003 European Molecular Biology Organization http://dx.doi.org/10.13039/100004410 ALTF 883-2011 Agence Nationale de la Recherche http://dx.doi.org/10.13039/501100001665 ANR-15-CHIN-0002-01 ==== Body pmcIntroduction DCs are a specialized population of immune cells that excel in antigen presentation and induce adaptive immune responses (Mellman and Steinman, 2001). Like other cells, DCs can present peptides derived from cytosolic antigens loaded on MHC class I to CD8+ T cells and to both endogenous and exogenous antigens bound to MHC class II molecules for recognition by CD4+ T cells. In addition, DCs can take up exogenous antigens and process and load them onto MHC class I molecules to be presented to CD8+ T cells, a process called antigen “cross-presentation” (the resulting induction of a CD8+ T cell response is referred to as “cross-priming”; Joffre et al., 2012). Several pathways of antigen cross-presentation that involve membrane trafficking through different intracellular compartments were reported in cultured DCs (Savina et al., 2006, 2009; Jancic et al., 2007; Cebrian et al., 2011; Nair-Gupta et al., 2014; Alloatti et al., 2015). One of the described cross-presentation pathways requires transfer of ER resident proteins, including the machinery for MHC class I loading with peptides (TAP1/2 transporters, tapasin, calreticulin, etc.), to the endocytic and phagocytic pathways, a traffic step controlled by the SNARE family member Sec22b (Cebrian et al., 2011). The actual contribution of different antigen cross-presentation pathways to immune responses in vivo remains unclear. The K. Murphy group (Hildner et al., 2008) has shown that certain subsets of cross-presenting DCs (i.e., Batf3-dependent DCs) have a critical role in antiviral immune responses and in the rejection of established solid tumors by CD8+ T cells. Recently, the R. Germain group (Castellino et al., 2006; Eickhoff et al., 2015) showed that CD8+ DCs act as “cellular platforms” to support CD4+ T cell help for CD8+ responses, a role that goes beyond their cross-presentation capacities. In contrast, increasing examples of CD8− DCs cross-presenting antigen in vivo are being reported (den Haan et al., 2000; Kamphorst et al., 2010). The actual contribution of antigen cross-presentation by DCs to specific immune responses is, therefore, a critical unknown. This is particularly true in the context of immunotherapies that attempt to harness the immune system to treat cancer, including those using checkpoint inhibitors. Expression of programmed cell death protein-1 (PD-1) on the surface of tumor-specific lymphocytes, and interaction with its corresponding ligands (PD-L1 and PD-L2, respectively) on the tumor- or antigen-presenting target cells is a key immune checkpoint that inhibits T cell function. Seminal studies in mouse models of cancer and diverse clinical studies have established that mAbs blocking the PD-1/PD-L1 pathway, as well as other checkpoints, such as CTLA-4, can unleash the immune system to fight cancer (Leach et al., 1996; Iwai et al., 2002). These therapies can mediate tumor regression in patients with metastatic melanoma, non–small cell lung cancer and renal cell carcinoma, among others (Hodi et al., 2010; Topalian et al., 2012; Lebbé et al., 2014). In mice, anti-immune, checkpoint-based treatments have been analyzed with success in several tumor models. The Melero laboratory (Sánchez-Paulete et al., 2016) has shown recently that Batf3-dependent DCs actively contribute to rejection of tumors during anti–PD-1 and anti-CD137 immunotherapies. To define the contribution of antigen cross-presentation to CD8+ T cell responses, we generated a mouse line in which the expression of Sec22b was conditionally depleted in DCs. Reduced Sec22b expression in DCs impairs antigen cross-presentation and cross-priming of cell-associated antigens in vivo. Sec22b-defective mice also failed to mount effective antitumor immune responses, to control the growth of immunogenic tumors, and to respond to anti–PD-1–based immunotherapy. These results show that Sec22b-dependent antigen cross-presentation is required during cross-priming of CD8+ T cell responses with dead cell–derived antigens and for anti–checkpoint-tumor immunotherapy in mice. Results and discussion To investigate the role of Sec22b-dependent cross-presentation in vivo, we generated floxed sec22b knock-in mice and crossed them to CD11c-specific Cre-deleter mice (Caton et al., 2007). We thus obtained mice bearing a selective deletion of the sec22b gene in DCs (Sec22b−/−). As controls, throughout the study, we used littermates expressing the Cre recombinase, and WT alleles of the sec22b gene (Sec22b+/+). Western blot analysis of splenic CD11c+ cells isolated by two rounds of selection (negative and then positive) confirmed that Sec22b expression was reduced in primary DCs purified from Sec22b−/−, but not Sec22b+/+ mice (Fig. 1 A, top). Sec22b expression in peritoneal macrophages (Fig. 1, top), as well as in splenic B and T cells (Fig. 1 A, bottom) was not affected, confirming that Sec22b−/− mice bear a conditional defect in Sec22b expression in DCs. Figure 1. Phenotypic analysis of Sec22b−/− mice. (A) Western Blotting of Sec22b expression in peritoneal macrophages and CD11c+ splenic DCs (top: purified by negative (neg. sel.) and positive (pos. sel.) selection for CD11c+ cells), as well as B and T cells (bottom), isolated from Sec22b+/+ and Sec22b−/− mice. Shown is one representative experiment of three independent experiments. (B) Phenotypic characterization of DC subsets in spleens and lungs from Sec22b−/− and Sec22b+/+ mice. Gating strategy are shown for splenic DCs (left), percentages of CD8+ and CD11b+ DCs subpopulations in spleen (middle), and cell numbers (right). Data shown are means of four independent biological replicates. DC populations were analyzed by gating on CD11chigh, MHC class IIhigh/mid and CD8/CD11b. (C) Gating strategy are shown for lung DCs (left), percentages of CD103+ and CD11b+ DCs subpopulations in spleen (middle), and cell counting (right). Data shown are means of four independent biological replicates and each value. DC populations were analyzed by gating on CD11chigh, MHC class IIhigh, then in Siglec Flow (to discriminate from alveolar macrophages), and finally in CD103/CD11b. (D) Phenotypic analysis of BMDCs. Western blotting analysis of Sec22b expression in BMDCs generated from Sec22b+/+ and Sec22b−/− mice. (E) Shown are percentages of CD11chighCD11bhigh cells (left) and expression of the costimulatory molecules CD40, CD80, CD86, and MHC class II upon TLR4 engagement with LPS (right). (F) Expression of MHC class I H2-Kb and Db. For all of the BMDCs results, shown are the pooled data from at least three independent experiments (with the exception of LPS histograms showing one representative experiment of four independent experiments). All results in this figure were analyzed by two-way ANOVA with Bonferroni’s multiple comparisons test. Phenotypic analysis of DC subsets in spleen (Fig. 1 B) and lungs (Fig. 1 C) from Sec22b−/− and Sec22b+/+ mice did not show any significant differences, neither in the composition of cell subpopulations nor in total cells numbers. Neither the percentages nor the numbers of other immune cell subpopulations from spleen, thymus, blood, lungs, and different lymph nodes were affected in Sec22b-defective mice (Fig. S1, A–F). Sec22b depletion did not modify the capacity of splenic CD8+ DCs to respond to different stimuli, such as LPS, IFN-γ, and TNF, as measured by expression of MHC II and CD86 (Fig. S1 G). We also confirmed by Western blot that Sec22b expression was reduced in BMDCs generated from Sec22b−/− mice, as compared with littermates (Fig. 1 D). Sec22b depletion did not affect the percentages of CD11chigh CD11bhigh cells in culture (Fig. 1 E, left) or LPS-mediated activation, as detected by expression of costimulatory molecules, (CD40, CD80, and CD86 and MHC class II; Fig. 1 E, right). IFN-γ and TNF induced an equivalent augment in the expression of CD86 and MHC II in both Sec22b+/+ and Sec22b−/− mice (Fig. S1 H). Finally, levels of MHC class I expression (both H2-Kb and H2-Db) were not affected by Sec22b depletion (Fig. 1 F). Therefore, defective Sec22b expression in DCs does not affect their development, survival, or activation capacity in vitro or in vivo. As a first analysis pertaining to DC function, we assessed the phagocytic and endocytic capacities of BMDCs. Both endocytosis and phagocytosis were similar in bone marrow-derived DCs (BMDCs) generated from Sec22b+/+ or Sec22b−/− mice (Fig. 2, A and B, respectively). To investigate whether Sec22b-defective DCs bear a defect in antigen cross-presentation, we first generated BMDCs from Sec22b−/− and control littermates. As shown in Fig. 2 C, BMDCs from Sec22b-deficient mice bear a partial defect in cross-presentation of soluble and bead-bound OVA, as detected by B3Z activation (top; Kurts et al., 1996), CD69 and CD25 expression (middle), and proliferation of OT-I CD8+ T cells, measured by dilution of carboxyfluorescein succinimidyl ester (CFSE; bottom). The synthetic MHC class I–restricted OVA peptide (SIINFEKL) was presented with equal efficacy by BMDCs generated from Sec22b−/− and Sec22b+/+ mice (Fig. 2 C, left). Figure 2. In vitro functional analysis of BMDCs generated from Sec22b−/− mice. Analysis of OVA–Alexa Fluor488 endocytosis (A) and bead-bound OVA phagocytosis by BMDCs (B); MFI, mean fluorescence intensity. Data shown are means of three independent biological replicates and each value. The results were analyzed by paired t test (P > 0.05). (C) Antigen cross-presentation capacity of BMDCs generated from Sec22b+/+ and Sec22−/− mice. (top) B3Z hybridoma T cell activation for (from left to right) MHC I–restricted peptides, soluble OVA, and bead-bound OVA. (Middle) OT-I T cell activation measured by CD69 and CD25 expression. (Bottom) OT-I T cell proliferation followed by dilution of CFSE dye. Shown are the pooled data of at least three independent experiments, with each experiment measured in triplicate. Results were analyzed by two-way ANOVA with Bonferroni’s multiple comparisons test for statistical significance. Shown are the means ± SEM. (D) Direct MHC class I presentation (left) and cross-presentation (right) capacity of BMDCs generated from Sec22b+/+ and Sec22b−/− mice, measured by IFN-γ production of CD44+ CD8+ effector T cells obtained from OT-I mice (top) and VACV WT-immunized mice (bottom) after co-culture for 4 h. Shown are the pooled data of three independent experiments. Results were analyzed by two-way ANOVA with Bonferroni’s multiple comparisons for statistical significance. (E) MHC class II antigen presentation efficacy for BMDCs obtained from Sec22b+/+ and Sec22b−/− mice, analyzed by OT-II T cell activation (top) and proliferation (bottom). Shown are the pooled data of three independent experiments, with each experiment analyzed in triplicate. Results were analyzed by two-way ANOVA as stated in C for statistical significance. For all analyses, *, 0.01 < P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. In addition, as an alternative source of antigen, we used vaccinia virus (VACV)–OVA–infected RAW macrophages (H2-Kd; RAW-VACV), which transmit virus and infect DCs that can direct present antigens in H2-Kb MHC class I molecules. Alternatively, infected RAW cells were treated with UV light (RAW-VACV-UV) to inactivate the virus, blocking direct infection of DCs and leaving available only the cross-presentation route, as previously described (Iborra et al., 2012). We also used uninfected RAW cells treated with UV (RAW-UV) to test antigen specificity. BMDCs generated from Sec22b+/+ and Sec22b−/− mice were exposed to VACV-infected or control cells for 4 h and used to stimulate preactivated OT-I T cells or CD8+ T cells purified from WT mice previously infected with VACV WT. Production of IFN-γ by OT-I (Fig. 2 D, top) or vaccinia-specific effector CD8+ T cells (Fig. 2 D, bottom) was not affected by Sec22b-depletion in DCs directly infected with the virus (Fig. 2 D, left) but was impaired in DCs co-cultured with RAW-VACV-UV (Fig. 2 D, right). Moreover, MHC class II–restricted antigen presentation, as measured by activation and proliferation of OT-II CD4+ OVA-specific TCR transgenic T cells, was not affected in Sec22b-deficient BMDCs (Fig. 2 E). Therefore, in agreement with our previous results in Sec22b knockdown BMDCs (Cebrian et al., 2011), BMDCs generated from Sec22b-defective mice display impaired cross-presentation, but conventional MHC class I and II antigen presentation is not affected. Similar to BMDCs, CD11c+ splenic DCs isolated from Sec22b−/− mice also cross-present OVA to CD8+ T cells less efficiently than do those purified from littermates (Fig. 3 A), whereas MHC class II–restricted presentation to CD4+ T cells is unaffected (Fig. 3 B). To investigate the role of Sec22b depletion in DCs in cross-priming, we first tested whether CD8+ T cells responses were normal in Sec22b−/− mice. We immunized Sec22b+/+ and Sec22b−/− mice with IFA-CpG-SIINFEKL in the footpad and analyzed anti-OVA and anti-SIINFEKL responses by IFN-γ ELISPOT in draining popliteal and inguinal lymph nodes. CD8+ T cells from Sec22b+/+ and Sec22b−/− were equally restimulated with SIINFEKL (Fig. S2 A), whereas the restimulation with OVA was impaired in Sec22b−/− mice. These results indicate that Sec22b−/− mice have a functional CD8+ T cells compartment and that the SIINFEKL-specific repertoire is normal. Figure 3. Relevance of antigen cross-presentation to elicit CD8+ T cell responses against dead cell-derived antigen in vivo. Antigen cross-presentation (A) and MHC II presentation by splenic DCs (B) isolated from Sec22b+/+ and Sec22b−/− mice, as well as their peptide controls (right). Shown are the means ± SEM of three independent experiments. Statistical analysis was performed using two-way ANOVA with Bonferroni’s multiple comparisons test. **, P < 0.01; ****, P < 0.0001. (C) Schematic representation of the protocol for the in vivo experiments with necroptotic 3T3-RIPK3-OVA cells. (D) Gating strategy to measure endogenous CD8+ T cell responses with OVA-specific tetramers. (E) Analysis of endogenous OVA-specific CD8+ T cell responses generated against necroptotic 3T3-RIPK3-OVA cells in Sec22b+/+ and Sec22b−/− mice; shown are pooled data from three independent experiments (n = 11–12). ***, P = 0.0001 for the Welch’s t test used herein. (F) Gating strategy for lymph nodes (LN) DCs (left), percentages of migratory DCs (migDCs) and resident DCs (resDCs) subpopulations in LN (middle), and cell counting (right). Data shown are means of four independent biological replicates. DC populations were analyzed by gating on CD11chigh, MHC class IIhigh/mid, and finally in CD103/CD11b. sOVA, soluble OVA. To determine whether the defect in antigen cross-presentation in Sec22b-depleted DCs results in defective cross-priming in vivo, we injected necroptotic, OVA-expressing 3T3-RIPK3-OVA cells (Yatim et al., 2015) subcutaneously. The endogenous CD8+ T cell response against OVA was analyzed using MHC-I multimers (Yatim et al., 2015) after 9 d (Fig. 3 C). In Sec22b−/− mice, the endogenous cytotoxic CD8+ T cell response was significantly decreased from 2.134 ± 0.2835 to 0.6381 ± 0.1887%, as compared with control littermates (Fig. 3 D for gating strategy; Fig. 3 E). The proportion of migratory and resident DCs in lymph nodes was not affected by Sec22 depletion (Fig. 3 F, left, for gating; middle, for percentages; right, for cell numbers), indicating that the differences in cross-priming capacity are not the consequence of impaired migration capacity of Sec22b−/− DCs from the periphery. In addition, the number and percentages of the putative cross-presenting subsets CD8αα+ resident DCs and CD103+ migratory DCs were not modified by depletion of Sec22b (Fig. S2, B and C). We conclude that cross-priming of dead cell–associated antigens in vivo requires Sec22b expression in DCs. The results presented thus far indicate that Sec22b−/− mice represent a suitable model to investigate the contribution of antigen cross-presentation to immune responses. To evaluate the role of cross-presentation in antitumor immunity, we first analyzed the growth of a highly immunogenic OVA-expressing tumor cell line, EG7 (Moore et al., 1988; Helmich and Dutton, 2001). EG7 tumors grew faster in Sec22b−/− mice (Fig. 4, A and B), causing decreased survival (Fig. 4 C), as compared with littermates. That insufficiency to control tumors correlated with impaired OVA-specific CD8+ T cell responses (Fig. 4 D and Fig. S3 A). These results suggest that impaired cross-presentation of tumor antigens by Sec22-deficient DCs causes reduced antitumor immune responses and exacerbated tumor growth. Figure 4. Relevance of antigen cross-presentation to elicit CD8+ T cell responses against tumor-derived antigens in vivo. (A) Tumor growth curves for EG7-OVA injected subcutaneously in Sec22b+/+ (n = 10) and Sec22b−/− (n = 10) mice. (B) Tumor volumes in mm3 (on the day of sacrifice; see Materials and methods for sacrifice criteria). (C) Survival curves for age- and sex-matched Sec22b+/+ and Sec22b−/− mice injected with the EG7-OVA tumor cell line. (D) Endogenous CD8+ T cell response measured with OVA-specific tetramers in total blood cells on d 10 after tumor injection. (E) Tumor growth curves for the tumor cell line MCA101-OVA in Sec22b+/+ and Sec22b−/− mice treated or not treated with anti–PD-1 (αPD-1; n = 10–13). Pooled results from three independent experiments are shown. Statistical significance was analyzed by Welch’s t tests. Numbers refer to rejected tumors out of total mice analyzed. (F) Tumor sizes (on the day of sacrifice). (G) Survival of age- and sex-matched Sec22b+/+ and Sec22b−/− mice injected with the MCA101-OVA–expressing tumor cell line, treated or not treated with αPD-1. (H) Blood cells from Sec22b+/+ and Sec22b−/− mice were restimulated ex vivo with OVA MHC class I–restricted peptide or with nonrelated protein human serum albumin as control. The percentage of OVA-specific T cells producing IFN-γ per 5 × 104 blood cells was determined by ELISPOT analysis. **, P < 0.01; ***, P < 0.001; ****, P < 0.0001 for Welch’s t tests (B, D, F, and H) or log-rank test (C and G). Data in B, D, F, and H are from three independent experiments with at least three independent biological replicates (means ± SEM). Data in A, C, E, and G are pooled from three independent experiment. To address the possible role of Sec22b-dependent antigen cross-presentation in tumor immunotherapy by checkpoint blocking antibodies, we used a less-immunogenic tumor, OVA-secreting MCA-101 (Zeelenberg et al., 2008; Sedlik et al., 2014). This tumor is well controlled by antibodies against CTLA-4 or PD-1 (Gubin et al., 2014). Sec22b−/− and Sec22b+/+ mice were injected with tumor cells and received or did not receive anti–PD-1 treatment, as described previously (Gubin et al., 2014). As expected, treatment with anti–PD-1 induced tumor rejection in Sec22b+/+ littermates (Fig. 4 E). In contrast, Sec22b−/− mice failed to respond to anti–PD-1 treatment, both in terms of tumor growth and survival (Fig. 4, F and G). Consistently, treatment with anti–PD-1 induced an anti-OVA immune response in littermates, but not in Sec22b−/− mice, when analyzed by T cell restimulation and subsequent secretion of IFN-γ (Fig. 4 H and Fig. S3 B). To further investigate the mechanism of resistance to anti–PD-1, we measured IFN-γ production and proliferation capacity of CD8+ T cells isolated from spleen (Fig. S3 C) and tumor (Fig. S3 D) of mice bearing MCA101-OVA tumors and treated with anti–PD-1. CD8+ T cells from Sec22b−/− produce IFN-γ and proliferate to the same extent as CD8+ T cells from control mice. Expression of PD-1 in T cells from Sec22b+/+ and Sec22b−/− was similar (Fig. S3 E). The expression of PD-L1 and PD-L2 in CD8+ and CD8−/CD11b+ DCs from spleen was not reduced in Sec22b-deficient DCs (Fig. S3 F). These results show a critical, nonredundant role for Sec22b-dependent antigen cross-presentation in the onset of efficient antitumor immune responses induced by checkpoint blockade. Although cross-priming of allogeneic antigens in vivo was reported >30 yr ago (Bevan, 1976), the actual contribution of antigen cross-presentation (vs. other mechanisms such as cross-dressing (Wakim and Bevan, 2011; Yewdell and Dolan, 2011) or Gap junction–mediated peptide transfer (Neijssen et al., 2005) to cross-priming is still unclear. In the case of cross-presentation, multiple intracellular pathways have been reported in vitro. Each of those pathways is under the control of master regulators that control key intracellular traffic steps specific for each pathway. Here, we developed tools to perform loss of function experiments for Sec22b-dependent antigen cross-presentation in vivo. Our results show that Sec22b-dependent antigen cross-presentation by DCs has a critical role in the cross-priming of cell-associated antigens and antitumor immune responses in vivo (even though we did not exhaustively investigate all of the cell populations that might express CD11c, notably activated B cells that can help priming CD8+ T cell responses). It is possible that alternative cross-presentation pathways independent of Sec22b exist, but their contribution to the priming of CD8+ T cell responses in vivo needs to be addressed. Our results show that the induction of tumor-specific CD8+ T cells by anti–PD-1 requires Sec22b expression in DCs, suggesting that, in addition to acting on effector cytotoxic T cells by releasing tumor-imposed immunosuppression, anti–PD-1 may also facilitate the priming of CD8+ T cells. Whether Sec22b turns out to be a suitable target to modulate antigen cross-presentation therapeutically will need to be addressed in future studies. Materials and methods Compounds and antibodies For flow cytometry, the following antibodies were used: anti–CD69-eFluor450 (clone H1.2F3, 48-0691-82; eBioscience), anti–CD25-FITC (clone 7D4, 553072; BD), anti–CD8a-PerCP-Cy5.5 (clone 53–6.7, 45-0081-82; eBioscience), anti–TCR vβ 5.1-PE (clone MR9-4, 553190; BD), anti–CD4-PE-Cy7 (clone RM4-5, 552775; BD), anti–CD19-eFluor450 (clone 1D3, 48-0193; eBioscience), anti–CD3-eFluor450 (clone 17A2, 48-0032-80; eBioscience), anti–CD11c-FITC (clone HL3, 553801; BD), anti–CD11b-PerCPCyanine5.5 (clone M1/70, eBioscience #45-0112-80), anti-F4/80-FITC (clone BM8, eBioscience #11-4801-81), anti-NKp46-PE (clone 29A1.4, 12-3351-80; eBioscience), anti–MHC I (H-2Kb)-FITC (clone AF6-88.5.5.3, 11-5958-80; eBioscience), anti–MHC II (Iab)-eFluor450 (clone AF120.1, 48-5320-80; eBioscience). For Western blot analysis, the anti-Sec22b antibody was purchased from Santa Cruz Biotechnology, Inc. (29-F7, sc-101267), anti-α-actin from Sigma-Aldrich (A2066), and anti-gp96 from Enzo Life Sciences (spa850). For phenotypic analysis of immune cell populations CD4, CD8, CD25, CD44, CD62L, TCRb, and CD3 were used as T cell markers (in different fluorophores, depending on the tissue analyzed and the panel of antibodies used). B220 and CD19 were used as B cell markers. TCR-γδ was used as the γδ T cell marker. NK1.1 and NKp46 were used as natural killer markers. Gr-1 and Ly6G were used as granulocyte/neutrophil markers. CD11c was generally used as a DC marker, with the exception of lungs: alveolar macrophages (CD11chigh) were discriminated from DCs by analyzing Siglec F expression. CD103, I-Ab, H2-Kb, CD11b, CD8, CD40, and CD86 were used to analyze DCs phenotype (in different fluorophores). F4/80 and Ly6C were used as macrophage markers. As for the direct MHC I antigen-presentation analysis, anti-mouse antibodies to CD8α, CD44, and IFN-γ were used as conjugates to PE, FITC, and APC, and were obtained from eBioscience. Anti–PD-L1-PE and anti–PD-L2-APC were purchased from eBioscience. Cell lines and cell culture RAW264.7 macrophages (RAW) were grown in DMEM supplemented with 5 mM glutamine, penicillin, streptomycin, β-mercaptoethanol (all from Invitrogen) and 10% heat-inactivated FBS (Sigma-Aldrich). CD8+ T cells were negatively selected using a cocktail of biotin-conjugated antibodies (anti-CD11c, B220, MHC-II, CD4, NK1.1), followed by incubation with streptavidin microbeads (Miltenyi Biotec). Typical yields by FACS staining were >95% pure. Preactivated OT-I cells were obtained by culturing spleen cells from OT-I transgenic mice (C57BL/6-Tg(TcraTcrb)1100Mjb/J) with 10−9 M of the peptide 257SIINFEKL264 from OVA for 5 d. At that time, ∼90% of the cells were CD8+CD44hi. The 3T3-RIPK3-OVA cells, expressing a nonsecretable form of OVA cells, were obtained from the Matthew Albert laboratory (Yatim et al., 2015) and were cultured in DMEM (Thermo Fisher Scientific) supplemented with 10% FBS (Biowest), 0.1 mM nonessential aa, 1 mM sodium pyruvate, 10 mM Hepes, and 50 µM β-mercaptoethanol (all from Thermo Fisher Scientific). Necroptosis was induced by treatment with a specific drug ligand (AP20187, BB homodimerizer; Takara Bio Inc.). EG7 tumor cells were cultured in DMEM supplemented with 10% FBS (Lonza), as well as 100 IU/ml penicillin, 100 µg/ml streptomycin, 2 mM GlutaMAX, and 50 µM β-mercaptoethanol (all from Thermo Fisher Scientific). MCA-101 OVA-secreting cells were grown in RPMI-1640 containing 10% heat-inactivated FBS (Biowest), 100 IU/ml penicillin, 100 µg/ml streptomycin, 2 mM GlutaMAX, and 50 µM β-mercaptoethanol (all from Thermo Fisher Scientific), as well as hygromycin B 1 mg/ml (Gibco) for selection. All cell lines were tested as mycoplasma-negative by PCR. DC activation Maturation of BMDCs was induced by a 16-h treatment with 100 ng/ml of ultrapure LPS from Escherichia coli 0111:B4 (InvivoGen) or, alternatively, with 100 ng/ml of TNF (Protein Service Facility, Inflammation Research Center) or IFN-γ (Invitrogen). DC maturation was controlled by cell-surface expression of costimulatory molecules and MHC class II molecules using specific antibodies. Maturation of splenic CD11c+ DCs was induced by a 4-h treatment with 100 ng/ml of LPS, TNF, or IFN-γ. Virus strains The rVACV-OVA and VACV WR strains were gifts from Jonathan W. Yewdell and Jack R. Bennink (National Institutes of Health, Bethesda, MD). Stocks were grown in CV-1 monolayers and used as clarified sonicated cell extracts. Primary cell isolation and culture BMDCs were produced by culturing the cells for 10 d in GM-CSF–containing medium (as described in Alloatti et al., 2015), with IMDM (Sigma-Aldrich and Thermo Fisher Scientific) containing 10% heat-inactivated FBS (Biowest), 100 IU/ml penicillin, 100 µg/ml streptomycin, 2 mM GlutaMAX, and 50 µM β-mercaptoethanol (all from Thermo Fisher Scientific). Supernatant from J558 plasmacytoma cells was used as the granulocyte macrophage colony-stimulating factor source (Winzler et al., 1997). To obtain splenic CD11c+ DCs, each mouse spleen was injected with 2 ml of a digestion solution with 0.1 mg/ml Liberase (Roche), 0.1 mg/ml DNase I (Roche), 1× PenStrep in RPMI-1640 (both from Thermo Fisher Scientific), following the Roche instructions, and was cut into small pieces and incubated for 25 min at 37°C. Reaction was stopped by addition of complete medium. Spleens were passed through a cell strainer and centrifuged at 200× g for 5 min. After red blood cell lysis (Sigma-Aldrich), splenic DCs were isolated by performing one round of negative selection for CD11c+ cells (using EasySep Mouse Pan-DC enrichment kit from STEMCELL Technologies), followed by a round of positive selection using CD11c microbeads (Miltenyi Biotec), according to the manufacturer’s instructions. In all experiments, purity of cells was >90%. For phenotypic characterization of immune cells, thymus and lungs were processed by digestion with Liberase/DNase I as stated above and was subsequently analyzed by flow cytometry. Animals C57BL/6J mice and C57BL/6J recombination activating gene 1-deficient OT-I and OT-II TCR (Vα2, Vβ5.1) transgenic mice were obtained from Charles River, Janvier, and Centre de Distribution, Typage et Archive Animal. Mice were between 5 and 12 wk old. WT C57BL/6J mice for tumor experiments were always obtained from the aforementioned sources. Sec22bFlox/Flox (see below) and control mice were originally produced at the Centre d’Immunologie de Marseille (the Malissen group) and were subsequently bred in the animal facility of Institut Curie. CD11c-Cre transgenic mice were obtained from The Jackson Laboratory. All animal procedures were in accordance with the guidelines and regulations of the Institut Curie veterinary department, and all mice used were <6 mo old. Targeting strategy for generation of Sec22b mice A genomic fragment encompassing exon 2 and 4 of the sec22b gene was isolated from a BAC clone of B6 origin. Using ET recombination, loxP sites were introduced in the introns flanking the 5′ and 3′ ends of exon 2. The loxP site in the intron found at the 3′ end of exon 2 was abutted to a ftr-neoR-frt cassette. The final targeting construct was abutted to a cassette coding for thymidine kinase and linearized with Pme1. ES clone selection JM8.F6 C57BL/6N ES cells (Pettitt et al., 2009) were electroporated with the targeting vector. After selection in G418 and ganciclovir, ES cell clones were screened for proper homologous recombination by Southern blot and PCR analysis. A neomycin-specific probe was used to ensure that adventitious, nonhomologous recombination events had not occurred in the selected ES clones. Properly recombined ES cells were injected into FVB blastocysts. Upon germline transmission, mice were then crossed to flipper mice to delete the frt-flanked neoR cassette, and the resulting floxed Sec22b allele (official name B6-Sec22btm1Ciphe, called here Sec22bfl was identified by PCR of tail DNA. The pair of primers: sense 5′-ATGGTAAAAAGCACACCAAATACTTTGC-3′ and antisense 5′-TGAGGTAACCTTGAAGGCTAGAAGA-3′ amplified a 600-bp band in the case of the Sec22bWT allele and a 900-bp band in the case of the Sec22bfl allele. The excised Sec22bfl allele generated a band of 300 bp (animals were screened for breeding and experiments based on the presence of this 300-bp band). Considering that the phase of the introns flanking exon 2 is asymmetrical, deletion of exon 2 results in an out-of-frame Sec22b allele. Western blotting Total cell lysates from BMDCs and splenic DCs were subjected to 4–12% gradient gels (Thermo Fisher Scientific) or 10% and 12% gels and were separated by SDS-PAGE. After transfer to nitrocellulose membranes, they were blocked and incubated with primary antibodies and peroxidase-conjugated secondary antibodies. Bound antibodies were revealed using the BM chemiluminescence blotting substrate (POD) from Roche or GE Healthcare, according to the manufacturers’ directions. The intensity of the bands was quantified by densitometry using Quantity One 4.6.6 software (Bio-Rad Laboratories) and was expressed as arbitrary units. Antigen uptake assay Phagocytosis and the endocytic capacity of BMDCs were assessed using 3-µm blue latex beads (Polysciences) and OVA Alexa Fluor 488 conjugate (O34781; Thermo Fisher Scientific), respectively. Viral infections and virus titration. Mice were infected intradermally in the ears with 5 × 104 PFU of the required VACV WR strain, as previously described (Iborra et al., 2012). Antigen presentation assays Proliferation assay DCs were incubated with Low Endo soluble OVA from Worthington Biochemical Corporation (LS003061) or with 3-µm beads coated with different ratios of OVA and BSA proteins (OVA 10 mg/ml alone; OVA 2.5 mg/ml–BSA 7.5 mg/ml; OVA 5 mg/ml–BSA 5 mg/ml; and BSA 10 mg/ml alone) or different concentrations of the control minimal peptide (OVA SIINFEKL for cross-presentation and OVA323–339 for MHC II presentation). After 5 h, DCs were washed three times with PBS containing 0.1% (vol/vol) BSA and co-cultured with purified CFSE-OT-I CD8+ for 3 d. For monitoring T cell proliferation, diminution of CFSE staining on TCR+ CD8+ populations was measured by flow cytometry. T cell activation assay DCs were incubated for 5 h with different concentrations of Low Endo soluble OVA from Worthington Biochemical Corporation (LS003061) or with 3-µm beads coated with different ratios of OVA and BSA proteins (as mentioned above). Minimal peptide OVA257–264 was used as a control for the capacity of DCs to activate T cells. Next, DCs were washed three times with 0.1% (vol/vol) PBS/BSA, fixed with 0.008% (vol/vol) glutaraldehyde during 10 min at 4°C, washed twice with 0.2 M glycine and once with 0.1% (vol/vol) PBS/BSA, and finally, B3Z hybrid T cells were added. After 16 h, T cell activation was measured by detecting β-galactosidase activity by OD at 590 nm using chlorophenol red-β-d-galactopyranoside as the substrate for the reaction. The efficiency of antigen presentation on MHC I and MHC II was also evaluated using OT-I (CD8+) and OT-II (CD4+ T) cells, respectively. Where indicated, DCs were fixed with 0.008% of glutaraldehyde before T cell addition. The percentage of CD25+CD69+ T cells was measured by flow cytometry after 16 h of co-culture (MHC I and II–restricted peptides, OVA257–264 and OVA323–339, respectively, were used as controls). Direct MHC class I antigen presentation and cross-presentation assay with VACV DCs were stimulated by co-culture with VACV-OVA–infected RAW cells treated with or without UV irradiation to inactivate the virus. To test DC cross-presenting ability, RAW cells were irradiated with UVC (240 mJ/cm2) either without exposure to VACV-OVA (RAW UV) or after incubation with VACV for 4 h (RAW-VACV-UV). Alternatively, infected RAW cells were left unirradiated (RAW-VACV) to allow direct infection of DCs. 16 h after UV irradiation, RAW cells were co-cultured for 4 h with BMDCs generated from Sec22b+/+ and Sec22b−/− mice. To the co-cultures, we then added CD8+ T cells purified from splenocytes of mice intradermally injected 7 d earlier with WR VACV, or preactivated OT-I cells were added to the cultures for 6 h; then, Brefeldin A (5 µg/ml; Sigma-Aldrich) was added for the last 4 h of culture. Cells were then stained with PE-anti-CD8α and FITC-CD44, fixed in 4% PFA, and incubated with APC-anti-IFN-γ during permeabilization with 0.1% saponin. A mean of 10,000 of each T cell subset was analyzed in each sample. Background activation obtained with CD8+ T cells nonpulsed with any peptide (0–0.3%) was subtracted. In vivo cross-priming assay with 3T3-RIPK3-OVA necroptotic cells The cross-priming assay was performed as described in (Yatim et al., 2015). In brief, 3T3-RIPK3-OVA cells were harvested and resuspended in complete media at 5 × 106 cell/ml. Dimerizer was added at 1 µM, and cells were incubated for 10 min at 37°C, gently flicking the tube every 2 min. Ice-cold PBS was added, and cells were washed, counted, and resuspended in cold PBS at 107 cells/ml and were kept on ice until injections. 100 µl of cells (106 cells) were injected intradermally in the flanks of the mice. 9 d later, the spleen and the draining lymph node (inguinal lymph node) were harvested, pooled, and stained for surface markers and Kb-MHC I tetramers specific for OVA presented in the context of MHC class I molecules. Tumor growth experiments EG7 tumor assay Sec22b+/+ and Sec22b−/− mice were injected intradermally with 106 EG7 tumor cells in 100 µl of cold PBS. Tumor cells for injection were recovered from log phase in vitro growth and were injected into the right flank skin of recipient mice. Tumors were clearly visible after 7 d and grew progressively, in an encapsulated fashion. Tumor growth was measured each day and followed until d 20 or when the size reached 1,000 mm3. Afterward, mice were euthanized. 10 mice were used for both Sec22b+/+ and Sec22b−/− groups. To measure CD8+ T cell responses, blood samples were collected from mice 7 d after tumor injections, RBCs were lysed three times with RBC lysis buffer (Sigma-Aldrich), and the nonlysed cells were stained with a couple of Kb-MHC I tetramers specific for OVA presented in the context of MHC class I molecules (provided by the Albert group). MCA101 OVA-secreting tumor assay and immunotherapy Sec22b+/+ and Sec22b−/− mice were injected intradermally with 5 × 105 MCA-101 OVA-secreting tumor cells in 100 µl of cold PBS. Tumor cells for injection were recovered from log phase in vitro growth and were injected into the right flank skin of recipient mice. Tumors were clearly visible after 4–5 d and grew progressively, in an encapsulated fashion. Tumor growth was measured every 2 d and was followed until d 30 or until the size reached 1,000 mm3. Afterward, mice were euthanized. 10–15 mice were used for each condition. Anti–PD-L1 treatment consisted of five injections of 200 µg of antibody, delivered on d 0, 3, 6, 9, and 12, i.p. Cold PBS was injected to control groups. Mice were bled on d 13 after tumor injections, and RBCs were lysed as described in the previous section. The unrelated protein human serum albumin, the peptide SIINFEKL, and the PMA/ionomycin were given to nonlysed cells and loaded in MHC I by blood antigen-presenting cells. Subsequent restimulation of OVA-specific T cells was analyzed by IFN-γ ELISpot kit, following the manufacturer’s instructions. IFN-γ production by CD8+ T cells from spleen and tumors obtained from MCA101-OVA–injected mice Mice were inoculated with MCA101-OVA tumors and received anti–PD-1 treatment as mentioned above. At d 13, mice were sacrificed, and spleens and tumors were collected and processed with Liberase and DNase I, as described. CD8+ T cells from spleens were isolated with EasySep Mouse Naive CD8+ T Cell isolation kit, following manufacturer’s instruction, and were loaded with Cell Trace Violet (Invitrogen). In addition, total tumor cells were loaded with Cell Trace Violet. 500,000 cells from both spleens and tumors were plated per well in a flat, 96-well plate coated with CD3 (Miltenyi Biotec) at 1 mg/ml with increasing concentrations of CD28 (0, 0.25, 0.5, and 1 mg/ml). After 72 h, plated T cells were treated with Brefeldin A (5 µg/ml; BioLegend) for 3 h to enhance intracellular IFN-γ accumulation. Subsequently, intracellular staining for IFN-γ, as well as CD3, CD4, CD8, and PD-1 staining, was performed. Analysis of OVA-specific CD8+ T cells in Sec22b+/+ and Sec22b−/− mice Mice were immunized in the footpad with 50 μl of IFA-CpG (25 µg per mouse)-SIINFEKL (100 µg per mouse). 8 d later, draining popliteal and inguinal lymph nodes were collected and smashed with a cell strainer and subsequently filtered. 500,000 cells were plated per well, and IFN-γ ELISpot analysis was performed as described earlier using OVA and SIINFEKL as antigens for restimulation. Statistical analysis All statistical analyses were performed with Prism 7 (GraphPad Software). P-values, as well as statistical tests, are detailed in the figure legends. Online supplemental material Fig. S1 shows the phenotypic characterization of immune cell subsets from Sec22b+/+ and Sec22b−/− mice, as well as the expression of costimulatory molecules upon treatment with LPS, TNF, and IFN-γ in DCs from the aforementioned mice. It relates to Fig. 1. Fig. S2 comprises information regarding the immunization of Sec22b+/+ and Sec22b−/− mice to analyze endogenous responses, as well as the phenotypic analysis of putative cross-presenting, migratory and resident DCs. It relates to Fig. 3. Fig. S3 includes a correlative analysis of tumor growth and T cell responses and of the different controls regarding the functionality of CD8+ T cells from Sec22b+/+ and Sec22b−/− mice. Supplementary Material Supplemental Materials (PDF) Acknowledgments We would like to thank B. Malissen (Centre d’Immunologie de Marseille, Marseille, France) and Centre d’Immunophénomique for providing us with the Sec22bFloxFlox mice. We would also like to thank I. Cebrian, C. Sedlik, and J. Denizeau for technical assistance. S. Amigorena received funding from Institut Curie, Institut National de la Santé et de la Recherche Médicale; Centre National de la Recherche Scientifique, la Ligue Contre le Cancer (Equipe labelisée Ligue, grant EL2014.LNCC/SA), Association de Recherche Contre le Cancer, the H2020 European Research Council (grant 2013-AdG 340046 DCBIOX), Institut National Du Cancer (grant PLBIO13-057), and Agence Nationale de la Recherche (grants ANR-11-LABX-0043, ANR-10-IDEX-0001-02 PSL, ANR-16-CE15001801, and ANR-16-CE18002003). A. Alloatti was supported by the European Molecular Biology Organization (grant ALTF 883-2011) and the Agence Nationale de la Recherche (grant ANR-15-CHIN-0002-01). The authors declare no competing financial interests. Author contributions: A. Alloatti designed, carried out, and analyzed all experiments, except those detailed below. D.C. Rookhuizen assisted with all T cell assays. L. Joannas performed tumor experiments and assisted with all animal work. J.-M. Carpier assisted with tumor experiments and Western blotting. S. Iborra carried out all vaccinia virus assays. J.G. Magalhaes assisted in assay development and the EG7 tumor experiment. P. Kozik assisted with tumor experiments and cross-presentation assays. N. Yatim assisted with necroptotic cells assays. N. Yatim and M.L. Albert assisted with experimental design on the RipK3 experiments. D. Sancho conceived the vaccinia virus experiments. S. Amigorena conceived and supervised the study, and A. Alloatti and S. Amigorena wrote the manuscript. Abbreviations used: CFSE carboxyfluorescein succinimidyl ester VACV vaccinia virus ==== Refs Alloatti, A., F. Kotsias, A.-M. Pauwels, J.-M. Carpier, M. Jouve, E. Timmerman, L. Pace, P. Vargas, M. Maurin, U. Gehrmann, 2015. Toll-like receptor 4 engagement on dendritic cells restrains phago-lysosome fusion and promotes cross-presentation of antigens. Immunity. 43 :1087–1100. 10.1016/j.immuni.2015.11.006 26682983 Bevan, M.J. 1976. Cross-priming for a secondary cytotoxic response to minor H antigens with H-2 congenic cells which do not cross-react in the cytotoxic assay. J. Exp. Med. 143 :1283–1288. 10.1084/jem.143.5.1283 1083422 Castellino, F., A.Y. Huang, G. Altan-Bonnet, S. Stoll, C. Scheinecker, and R.N. Germain. 2006. Chemokines enhance immunity by guiding naive CD8+ T cells to sites of CD4+ T cell-dendritic cell interaction. Nature. 440 :890–895. 10.1038/nature04651 16612374 Caton, M.L., M.R. Smith-Raska, and B. Reizis. 2007. Notch-RBP-J signaling controls the homeostasis of CD8− dendritic cells in the spleen. J. Exp. Med. 204 :1653–1664. 10.1084/jem.20062648 17591855 Cebrian, I., G. Visentin, N. Blanchard, M. Jouve, A. Bobard, C. Moita, J. Enninga, L.F. Moita, S. Amigorena, and A. Savina. 2011. Sec22b regulates phagosomal maturation and antigen crosspresentation by dendritic cells. Cell. 147 :1355–1368. 10.1016/j.cell.2011.11.021 22153078 den Haan, J.M., S.M. Lehar, and M.J. Bevan. 2000. CD8+ but not CD8− dendritic cells cross-prime cytotoxic T cells in vivo. J. Exp. Med. 192 :1685–1696. 10.1084/jem.192.12.1685 11120766 Eickhoff, S., A. Brewitz, M.Y. Gerner, F. Klauschen, K. Komander, H. Hemmi, N. Garbi, T. Kaisho, R.N. Germain, and W. Kastenmüller. 2015. Robust anti-viral immunity requires multiple distinct T cell-dendritic cell interactions. Cell. 162 :1322–1337. 10.1016/j.cell.2015.08.004 26296422 Gubin, M.M., X. Zhang, H. Schuster, E. Caron, J.P. Ward, T. Noguchi, Y. Ivanova, J. Hundal, C.D. Arthur, W.-J. Krebber, 2014. Checkpoint blockade cancer immunotherapy targets tumour-specific mutant antigens. Nature. 515 :577–581. 10.1038/nature13988 25428507 Helmich, B.K., and R.W. Dutton. 2001. The role of adoptively transferred CD8 T cells and host cells in the control of the growth of the EG7 thymoma: factors that determine the relative effectiveness and homing properties of Tc1 and Tc2 effectors. J. Immunol. 166 :6500–6508. 10.4049/jimmunol.166.11.6500 11359800 Hildner, K., B.T. Edelson, W.E. Purtha, M. Diamond, H. Matsushita, M. Kohyama, B. Calderon, B.U. Schraml, E.R. Unanue, M.S. Diamond, 2008. Batf3 deficiency reveals a critical role for CD8α+ dendritic cells in cytotoxic T cell immunity. Science. 322 :1097–1100. 10.1126/science.1164206 19008445 Hodi, F.S., S.J. O’Day, D.F. McDermott, R.W. Weber, J.A. Sosman, J.B. Haanen, R. Gonzalez, C. Robert, D. Schadendorf, J.C. Hassel, 2010. Improved survival with ipilimumab in patients with metastatic melanoma. N. Engl. J. Med. 363 :711–723. 10.1056/NEJMoa1003466 20525992 Iborra, S., H.M. Izquierdo, M. Martínez-López, N. Blanco-Menéndez, C. Reis e Sousa, and D. Sancho. 2012. The DC receptor DNGR-1 mediates cross-priming of CTLs during vaccinia virus infection in mice. J. Clin. Invest. 122 :1628–1643. 10.1172/JCI60660 22505455 Iwai, Y., M. Ishida, Y. Tanaka, T. Okazaki, T. Honjo, and N. Minato. 2002. Involvement of PD-L1 on tumor cells in the escape from host immune system and tumor immunotherapy by PD-L1 blockade. Proc. Natl. Acad. Sci. USA. 99 :12293–12297. 10.1073/pnas.192461099 12218188 Jancic, C., A. Savina, C. Wasmeier, T. Tolmachova, J. El-Benna, P.M.-C. Dang, S. Pascolo, M.-A. Gougerot-Pocidalo, G. Raposo, M.C. Seabra, and S. Amigorena. 2007. Rab27a regulates phagosomal pH and NADPH oxidase recruitment to dendritic cell phagosomes. Nat. Cell Biol. 9 :367–378. 10.1038/ncb1552 17351642 Joffre, O.P., E. Segura, A. Savina, and S. Amigorena. 2012. Cross-presentation by dendritic cells. Nat. Rev. Immunol. 12 :557–569. 10.1038/nri3254 22790179 Kamphorst, A.O., P. Guermonprez, D. Dudziak, and M.C. Nussenzweig. 2010. Route of antigen uptake differentially impacts presentation by dendritic cells and activated monocytes. J. Immunol. 185 :3426–3435. 10.4049/jimmunol.1001205 20729332 Kurts, C., W.R. Heath, F.R. Carbone, J. Allison, J.F. Miller, and H. Kosaka. 1996. Constitutive class I-restricted exogenous presentation of self antigens in vivo. J. Exp. Med. 184 :923–930. 10.1084/jem.184.3.923 9064352 Leach, D.R., M.F. Krummel, and J.P. Allison. 1996. Enhancement of antitumor immunity by CTLA-4 blockade. Science. 271 :1734–1736. 10.1126/science.271.5256.1734 8596936 Lebbé, C., J.S. Weber, M. Maio, B. Neyns, K. Harmankaya, O. Hamid, S.J. O’Day, C. Konto, L. Cykowski, M.B. McHenry, and J.D. Wolchok. 2014. Survival follow-up and ipilimumab retreatment of patients with advanced melanoma who received ipilimumab in prior phase II studies. Ann. Oncol. 25 :2277–2284. 10.1093/annonc/mdu441 25210016 Mellman, I., and R.M. Steinman. 2001. Dendritic cells: specialized and regulated antigen processing machines. Cell. 106 :255–258. 10.1016/S0092-8674(01)00449-4 11509172 Moore, M.W., F.R. Carbone, and M.J. Bevan. 1988. Introduction of soluble protein into the class I pathway of antigen processing and presentation. Cell. 54 :777–785. 10.1016/S0092-8674(88)91043-4 3261634 Nair-Gupta, P., A. Baccarini, N. Tung, F. Seyffer, O. Florey, Y. Huang, M. Banerjee, M. Overholtzer, P.A. Roche, R. Tampé, 2014. TLR signals induce phagosomal MHC-I delivery from the endosomal recycling compartment to allow cross-presentation. Cell. 158 :506–521. 10.1016/j.cell.2014.04.054 25083866 Neijssen, J., C. Herberts, J.W. Drijfhout, E. Reits, L. Janssen, and J. Neefjes. 2005. Cross-presentation by intercellular peptide transfer through gap junctions. Nature. 434 :83–88. 10.1038/nature03290 15744304 Pettitt, S.J., Q. Liang, X.Y. Rairdan, J.L. Moran, H.M. Prosser, D.R. Beier, K.C. Lloyd, A. Bradley, and W.C. Skarnes. 2009. Agouti C57BL/6N embryonic stem cells for mouse genetic resources. Nat. Methods. 6 :493–495. 10.1038/nmeth.1342 19525957 Sánchez-Paulete, A.R., F.J. Cueto, M. Martínez-López, S. Labiano, A. Morales-Kastresana, M.E. Rodríguez-Ruiz, M. Jure-Kunkel, A. Azpilikueta, M.A. Aznar, J.I. Quetglas, 2016. Cancer Immunotherapy with immunomodulatory anti-CD137 and anti-PD-1 monoclonal antibodies requires BATF3-dependent dendritic cells. Cancer Discov. 6 :71–79. 10.1158/2159-8290.CD-15-0510 26493961 Savina, A., C. Jancic, S. Hugues, P. Guermonprez, P. Vargas, I.C. Moura, A.-M. Lennon-Duménil, M.C. Seabra, G. Raposo, and S. Amigorena. 2006. NOX2 controls phagosomal pH to regulate antigen processing during crosspresentation by dendritic cells. Cell. 126 :205–218. 10.1016/j.cell.2006.05.035 16839887 Savina, A., A. Peres, I. Cebrian, N. Carmo, C. Moita, N. Hacohen, L.F. Moita, and S. Amigorena. 2009. The small GTPase Rac2 controls phagosomal alkalinization and antigen crosspresentation selectively in CD8+ dendritic cells. Immunity. 30 :544–555. 10.1016/j.immuni.2009.01.013 19328020 Sedlik, C., J. Vigneron, L. Torrieri-Dramard, F. Pitoiset, J. Denizeau, C. Chesneau, P. de la Rochere, O. Lantz, C. Thery, and B. Bellier. 2014. Different immunogenicity but similar antitumor efficacy of two DNA vaccines coding for an antigen secreted in different membrane vesicleassociated forms. J. Extracell. Vesicles. 3 :24646. 10.3402/jev.v3.24646 Topalian, S.L., F.S. Hodi, J.R. Brahmer, S.N. Gettinger, D.C. Smith, D.F. McDermott, J.D. Powderly, R.D. Carvajal, J.A. Sosman, M.B. Atkins, 2012. Safety, activity, and immune correlates of anti-PD-1 antibody in cancer. N. Engl. J. Med. 366 :2443–2454. 10.1056/NEJMoa1200690 22658127 Wakim, L.M., and M.J. Bevan. 2011. Cross-dressed dendritic cells drive memory CD8+ T-cell activation after viral infection. Nature. 471 :629–632. 10.1038/nature09863 21455179 Winzler, C., P. Rovere, M. Rescigno, F. Granucci, G. Penna, L. Adorini, V.S. Zimmermann, J. Davoust, and P. Ricciardi-Castagnoli. 1997. Maturation stages of mouse dendritic cells in growth factor-dependent long-term cultures. J. Exp. Med. 185 :317–328. 10.1084/jem.185.2.317 9016880 Yatim, N., H. Jusforgues-Saklani, S. Orozco, O. Schulz, R. Barreira da Silva, C. Reis e Sousa, D.R. Green, A. Oberst, and M.L. Albert. 2015. RIPK1 and NF-κB signaling in dying cells determines cross-priming of CD8+ T cells. Science. 350 :328–334. 10.1126/science.aad0395 26405229 Yewdell, J.W., and B.P. Dolan. 2011. Immunology: cross-dressers turn on T cells. Nature. 471 :581–582. 10.1038/471581a 21455165 Zeelenberg, I.S., M. Ostrowski, S. Krumeich, A. Bobrie, C. Jancic, A. Boissonnas, A. Delcayre, J.-B. Le Pecq, B. Combadière, S. Amigorena, and C. Théry. 2008. Targeting tumor antigens to secreted membrane vesicles in vivo induces efficient antitumor immune responses. Cancer Res. 68 :1228–1235. 10.1158/0008-5472.CAN-07-3163 18281500

PMC005xxxxxx/PMC5584142.txt

==== Front J Cell Biol J Cell Biol jcb jcb The Journal of Cell Biology 0021-9525 1540-8140 The Rockefeller University Press 28716842 201606043 10.1083/jcb.201606043 Research Articles Article 10 28 38 Outer nuclear membrane protein Kuduk modulates the LINC complex and nuclear envelope architecture Kud modulates the LINC complex and NE architecture Ding Zhao-Ying 1* http://orcid.org/0000-0003-4764-7592 Wang Ying-Hsuan 1* Huang Yu-Cheng 1 Lee Myong-Chol 1 Tseng Min-Jen 1 Chi Ya-Hui 2 http://orcid.org/0000-0002-4307-1584 Huang Min-Lang 1 1 Department of Life Science, National Chung-Cheng University, Chiayi, Taiwan 2 Institute of Biotechnology and Pharmaceutical Research, National Health Research Institutes, Miaoli, Taiwan Correspondence to Min-Lang Huang: biomlh@ccu.edu.tw * Z.-Y. Ding and Y.-H. Wang contributed equally to this paper. 04 9 2017 216 9 28272841 08 6 2016 25 5 2017 15 6 2017 © 2017 Ding et al. 2017 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). LINC complexes connect the inner and outer nuclear membrane (ONM) to transduce nucleocytoskeletal force. Ding et al. identify an ONM protein, Kuduk/TMEM258, which modulates the quality of LINC complexes and regulates the nuclear envelope architecture, nuclear positioning, and the development of ovarian follicles. Linker of nucleoskeleton and cytoskeleton (LINC) complexes spanning the nuclear envelope (NE) contribute to nucleocytoskeletal force transduction. A few NE proteins have been found to regulate the LINC complex. In this study, we identify one, Kuduk (Kud), which can reside at the outer nuclear membrane and is required for the development of Drosophila melanogaster ovarian follicles and NE morphology of myonuclei. Kud associates with LINC complex components in an evolutionarily conserved manner. Loss of Kud increases the level but impairs functioning of the LINC complex. Overexpression of Kud suppresses NE targeting of cytoskeleton-free LINC complexes. Thus, Kud acts as a quality control mechanism for LINC-mediated nucleocytoskeletal connections. Genetic data indicate that Kud also functions independently of the LINC complex. Overexpression of the human orthologue TMEM258 in Drosophila proved functional conservation. These findings expand our understanding of the regulation of LINC complexes and NE architecture. Ministry of Science and Technology, Taiwan http://dx.doi.org/10.13039/501100004663 103-2311-B-194-001-MY3 ==== Body pmcIntroduction The nuclear envelope (NE) consists of two lipid bilayer membranes that separate the nucleoplasm from the cytoplasm. The outer nuclear membrane (ONM) is continuous with the ER. The inner nuclear membrane (INM) has an underlying filamentous meshwork of lamin proteins called the nuclear lamina. The nuclear membranes and the lamina contain NE proteins that have important functions in regulating NE rigidity, gene expression, and chromosome organization. Dysfunctions in NE proteins impair NE architecture and cause human diseases such as rapid aging and cancers (Burke and Stewart, 2014). The linker of nucleoskeleton and cytoskeleton (LINC) complexes, which are highly conserved throughout evolution, consist of Klarsicht (Klar)/ANC-1/SYNE homology (KASH) and Sad1/UNC-84 (SUN) domain proteins (hereafter referred to as KASH and SUN proteins; Chang et al., 2015). KASH proteins span the ONM by the KASH domain, which bears a carboxyl tail that binds to the SUN domain of INM-resident SUN proteins in the perinuclear space (PNS). This KASH–SUN interaction forms a stable structure bridging the ONM and INM (Sosa et al., 2012). Cytoplasmic extensions of KASH proteins bind to cytoskeletal filaments, and SUN proteins interact with INM proteins and with the nuclear lamina. Therefore, the LINC complex controls nucleocytoskeletal force transduction and thereby contributes to nuclear migration and cytoskeletal organization (Chang et al., 2015). Mutations of the genes encoding LINC complexes lead to nuclear dysmorphology and defective nuclear positioning in mouse skeletal muscle (Zhang et al., 2007; Lüke et al., 2008; Lei et al., 2009; Puckelwartz et al., 2009). Mutations of the human LINC complex genes cause human genetic disorders such as arthrogryposis, cerebellar ataxia, deafness, and Emery–Dreifuss muscular dystrophy (Gros-Louis et al., 2007; Attali et al., 2009; Puckelwartz et al., 2009; Horn et al., 2013; Wang et al., 2015). Aberrant expressions of KASH and SUN proteins are causative in lung and breast cancers (Lv et al., 2015; Matsumoto et al., 2015). KASH and SUN proteins anchor at the NE through the “diffusion retention” model (Boni et al., 2015; Ungricht et al., 2015). SUN proteins retain KASH proteins through a physical interaction. Upon depletion of the Drosophila melanogaster SUN protein Klaroid (Koi), the two KASH proteins Klar and muscle-specific protein 300 (Msp300) are no longer localized at the NE (Kracklauer et al., 2007; Technau and Roth, 2008). Moreover, anchoring of SUN proteins at the INM can depend on the nuclear lamina proteins’ lamins and their associated proteins at the INM (Chang et al., 2015). In Drosophila and Caenorhabditis elegans carrying mutations of the lamin genes, SUN proteins are not localized at the NE (Lee et al., 2002; Kracklauer et al., 2007). It remains unknown whether proteins at the ONM regulate the LINC complex. In this study, we identified the Drosophila protein Kuduk (Kud) at the ONM, where it associates with LINC components. Kud regulates NE architecture, nuclear positioning, and the development of ovarian follicles through LINC-dependent and -independent mechanisms. Overexpression of the human orthologue TMEM258 in Drosophila proved functional conservation. These findings improve our knowledge about the regulation of the LINC complex and the NE and might contribute to a better understanding of the pathology and treatment of the human diseases related to this complex and TMEM258. Results The conserved protein Kud is required for the survival and growth of ovarian follicle cells Proteins in the UPF0197 family are short proteins with 79 aa on average and are evolutionarily conserved in metazoans (Fig. 1 A). In this study, we studied the Drosophila member Kud encoded by the CG9669 gene. To determine its function, we generated gene knockout flies by homologous recombination (Fig. S1, A–C). The homozygous mutants displayed growth retardation (Fig. S1 D) and died as larvae, indicating that kud is a gene that is essential for development. We observed homozygous mutant cells in heterozygous flies by the Flippase (FLP)/FLP recombination target (FRT) technique (Xu and Rubin, 1993) and found defects in ovarian follicle cells, which enwrap the germ cell clusters in ovarioles. At 114 h after clone induction, apoptotic cells were present in more than half of the mutant clones, which were GFP negative, but not in the controls (Fig. 1, B and C). To distinguish the mutant clones more easily, we overexpressed GFP in the mutant clones using mosaic analysis with a repressible cell marker (MARCM; Lee and Luo, 1999). We found that loss of Kud reduced the cell size to 50% of the controls (Fig. 1, D and E). To examine whether the defects resulted from the loss of Kud, we overexpressed Kud C-terminally tagged with CFP (Kud-CFP) in the mutant cells. The production of Kud-CFP significantly rescued the phenotypes of kud mutant cells (Fig. 1, C and E). The human orthologous protein, transmembrane protein 258 (TMEM258), is 65% identical to Kud (Fig. 1 A). Overexpression of TMEM258 in kud mutant cells inhibited apoptosis (Fig. 1 C). These data indicate that kud expression is required for cellular growth and survival and that its human orthologue is conserved functionally. Figure 1. The evolutionarily conserved protein Kud is required for ovarian follicle development. (A) UPF0197 proteins have been conserved in evolution; TM1 and TM2 are putative TMs. The NCBI accession numbers of the UPD0197 proteins from top to bottom are Q9VVA8, Q6PBS6, P61165, P61166, Q76LT9, and Q6DDB3. Asterisks indicate conserved residues, colons indicate residues with strongly similar properties, and periods indicate residues with weakly similar properties. (B and B’) Apoptotic cells were marked with activated caspase 3 (Caspase3*), and the kud mutant cells lacking GFP expression are outlined by dashed lines. (C) Quantification of the clones containing apoptotic cells: n = 78, 153, and 78 from left to right. (D and D′) The boundaries of follicle cells were visualized by the plasma membrane protein Disc large (Dlg in red), and the kud mutant cells expressing GFP are outlined with dashed lines. Bars, 10 µm. (E) Quantification of normalized cell areas: n (number of clones) = 9, 13, and 15 from left to right. ***, P < 0.001. Error bars indicate means ± SD. Conserved subcellular localizations at the NE and in the cytoplasm Proteins in the UPF0197 family have two putative transmembrane domains (TMs; Fig. 1 A), suggestive of membrane localization. Indeed, cell fractionation analyses indicated that the overexpressed TMEM258-Myc was present in the membrane fraction but not in the soluble fraction of human 293T cells (Fig. 2 A), indicating that TMEM258 is membrane associated. Overexpressed TMEM258-Myc was found in the cytoplasm and at the NE, where it colocalized with the nuclear lamina marker, Lamin B1 (Fig. 2 B). Cytoplasmic but not NE localization of TMEM258 has been previously reported (Adachi et al., 2002). This disparity in protein distribution could be caused by differences in expression levels in cells or in the antibodies used. Figure 2. Conserved subcellular localizations at the NE and in the cytoplasm. (A) TMEM258 is membrane associated. Western blot (WB) of soluble (Sol) and membrane (Mem) fractions of TMEM258-Myc–expressing 293T cells probed with anti-GAPDH (cytoplasm), anti-Calnexin (ER and ONM), anti-Emerin (INM), and anti-Myc antibodies. (B–C′′) The overexpressed TMEM258-Myc in 293T cells (B) and the endogenous Kud in ovarian follicle cells (C) locate at the cytoplasm and colocalize with NE marker lamins (B′ and C′, arrows). Merges are shown in B′′ and C′′. Bars, 10 µm. To visualize endogenous Kud in Drosophila, we raised a polyclonal antibody against its aa 1–18. We initially tested the antibody specificity by immunoblotting. Western blotting using larval lysates revealed that the antibody recognized a band ∼12 kD in the WT larva. The band was absent in kud homozygous mutants and was reduced in kud knockdown flies (Fig. S1 E). In addition, Western blotting using S2 cells overexpressing Kud-GFP revealed that the antibody recognized both the endogenous Kud and Kud-GFP (Fig. S1 F). When we used S2 cells overexpressing KudNGA-GFP, where the N terminus 18 aa was replaced with glycine and alanine, the antibody recognized only the endogenous Kud but not KudNGA-GFP (Fig. S1 F). These experiments showed that the antibody is specific for Kud. The antibody is also suitable for detecting native proteins because immunostaining revealed that Kud signals were highly reduced in kud mutant cells (Fig. S1 G). Kud in WT follicle cells were present in the cytoplasm and colocalized with LaminDm0 (LamDm0), the B-type lamin (Fig. 2 C). Double staining using the antibody and organelle markers revealed that the cytoplasmic Kud partially colocalized with the ER membrane protein Calnexin 99A (Cnx99A) but not markers of ER lumen, Golgi, or mitochondria (Fig. S2). These data indicate that the cytoplasmic Kud can localize at the ER membrane. Collectively, our data indicate that the subcellular localizations of Kud are conserved at the NE and in the cytoplasm. Kud spans the ONM To determine the topology of TMEM258 at the NE, we performed protease protection assays in 293T cell. However, proteinase K treatments did not alter the molecular weights of the bands recognized by antibodies of the N terminus and C terminus of TMEM258 (Fig. S3, A and B). This resistance to proteinase K digestion might result from the structure of TMEM258, which was not determined in this study. Thus, biochemical subcellular fractionation experiments might not be useful to determine the topology of TMEM258 at the NE. However, to determine the topology of Kud at the NE, we compared the signals of NE proteins in larval body muscles permeabilized using either Triton X-100 or digitonin. Digitonin permeabilizes membranes by binding to cholesterol, but the NE contains less cholesterol than the plasma membrane. Thus, low levels of digitonin can permeabilize only the plasma membrane but not the NE, and higher levels of digitonin can permeabilize the ONM or even both nuclear membranes. We found that in body muscles treated with digitonin, the permeability of nuclear membranes of individual nuclei were unequal, possibly because of the different depths of nuclei in the tissues. Therefore, we used LamDm0 as an indicator of a permeable INM. Upon treatment with Triton X-100, the N and C termini of Kud colocalized with LamDm0 in WT tissues and in muscle overexpressing C-terminally HA-tagged Kud (Kud-HA), respectively (Fig. 3, A and C). However, upon treatment with digitonin, both termini could be detected when the INM was intact, indicated by lack of immunoactivity of LamDm0 (Fig. 3, B and D), suggesting that they are not exposed to the nucleoplasm. Interestingly, we found some nuclei with the N terminus signal but without the C terminus HA signal after treatment with digitonin (Fig. 3, E and F). These data suggest that the two termini do not reside at the same side of a given membrane and that the N and C termini are exposed to the cytoplasm and to the PNS, respectively. Figure 3. Kud spans the ONM. (A–F′) Larval muscles of indicated genotypes were permeabilized with Triton X-100 or digitonin as indicated. The muscles were stained with antibodies against LamDm0 (A, B, C, and D), Kud (A′, B′, E′, and F′), or HA (C′, D′, E, and F). Bars, 10 µm. (G–H) Proteinase K protection assays of intact nuclei were performed. Kud-HA–expressing larval muscles were permeabilized with digitonin and were treated in the absence (G) or presence of proteinase K (PK; H). The tissues were then immunostained with anti-HA and anti-Kud antibodies in the presence of Triton X-100. The N terminus anti-Kud signals were highly reduced, whereas the HA signal of the C terminus remained (white arrows). Arrows indicate nuclei. (I) Helical projection of aa 19–39 of Kud. Color code for residues: gray, alanine and glycine; purple, threonine; yellow, hydrophobic. The polar residues defined by HeliQuest (http://heliquest.ipmc.cnrs.fr) include threonine and glycine. (J) A schematic diagram of the possible topology of Kud. Next, we combined the digitonin permeabilization experiment with protease protection assay in larval muscles to examine the predicted topology. If the NE is intact when tissues are treated with digitonin, proteinase K should digest Kud’s cytoplasmic N terminus but not the PNS-localized C terminus. As expected, after proteinase K treatment in muscle overexpressing Kud-HA, signals of the N terminus were highly reduced, whereas the HA signal of the C terminus remained (Fig. 3, G and H). Furthermore, we analyzed the sequences of TMs using HeliQuest software (Gautier et al., 2008) and found that the hydrophobic and polar residues of predicted TM1 (aa 19–39) potentially segregate to two opposite faces (Fig. 3 I). This implies that the segment is amphipathic and does not cross the ONM. Thus, the segment is considered to be an intramembrane (IM) domain. Collectively, these data suggest that the N terminus of Kud is exposed to the cytoplasm, the IM resides at the cytoplasmic surface and orients in a parallel way to the ONM, the TM (aa 54–74) spans the ONM, and the C terminus resides in the PNS (Fig. 3 J). Kud associates with LINC complex components in an evolutionarily conserved manner In a two hybrid–based protein interaction mapping experiment, Kud was found to interact physically with the ONM protein Klar (Giot et al., 2003). Consistently, we found colocalization of Kud and Klar at the NE (Fig. 4 A). We then performed coimmunoprecipitation (coIP) experiments to examine the association of Kud and Klar. Kud and its orthologues are short; thus, they are unable to extend far from the ONM. We reasoned that if the interaction between Kud orthologues and KASH proteins had been conserved evolutionarily, the interacting parts of KASH proteins should be conserved and near to the NE. The best candidates are the conserved KASH domains, which consist of a TM spanning the ONM and a ∼30-aa C terminus in the PNS (Sosa et al., 2013). Indeed, we found that the KASH domains of Klar and of the other KASH protein, Msp300, respectively associated with Kud-His in S2 cells (Fig. 4 B). These results suggest that the KASH domain is sufficient for the interaction with Kud. Figure 4. Kud associates with LINC complex components in an evolutionarily conserved manner. (A) Kud colocalizes with Klar (A′ and A′′) at the NE (arrow) of larval muscles. Bar, 10 µm. (B–D) CoIP experiments were performed with the transfected S2 cell lysates (B) and with the transfected 293T cell lysates (C and D). (B) Kud was associated with the KASH domains of Msp300 and Klar, respectively. The interactions were specific because the coIP experiments did not detect the NPC protein Nup62 or the ER marker Cnx99A. (C) Nesprin 1 and Nesprin 2 associated with TMEM258, but deletion of the KASH domain abolished this interaction. (D) Association between TMEM258, SUN1, and the KASH domain. The association is specific because the coIP experiments did not detect the INM marker Emerin, the ER marker STIM1, or the chromatin-binding protein barrier to autointegration factor (BAF). IgG was used as negative control. Each coIP experiment was repeated three times. IP, immunoprecipitation; WB, Western blot. To examine whether the human orthologous protein TMEM258 would associate with human KASH proteins, we overexpressed proteins in 293T cells and found that the KASH proteins Nesprin 1 and Nesprin 2 both coimmunoprecipitated with TMEM258. The associations were lost after deletion of the KASH domain (Fig. 4 C). Moreover, the KASH domain alone also interacted with TMEM258 (Fig. 4 D), suggesting that the KASH domain is sufficient for this interaction. We also found an association between TMEM258 and the SUN protein SUN1 (Fig. 4 D), suggesting that TMEM258 associates with the LINC complex. These data indicate that Kud interacts with the LINC complex in an evolutionarily conserved manner. Kud regulates myonuclear positioning through the LINC complex The association between Kud and the LINC complex prompted us to examine whether Kud is involved in muscle development, where Klar and Msp300 cooperatively regulate myonuclear positioning (Elhanany-Tamir et al., 2012). First, we used loss-of-function phenotypes of the LINC complex as positive controls. Complete loss of Klar or Koi caused severe nuclear clustering (Fig. 5, A, B, and G). Overexpression of the KASH domain, which lacks the cytoplasmic residues of the KASH proteins and is cytoskeleton free, caused dominant-negative effects (Starr and Han, 2002; Wilhelmsen et al., 2006; Starr, 2009). Consistent with the effect of the KASH domain, overexpression of the KASH domain of Klar caused significant defects in myonuclear spacing (Fig. 5, C and G). Next, we found that reduction of Kud, either from kud heterozygosity or RNAi expression in the heterozygote mutant, resulted in myonuclear clustering (Fig. 5, D and G), indicating that Kud is similar to the LINC complex in regulating myonuclear positioning. Figure 5. Kud regulates myonuclear positioning through the LINC complex. (A–F) Positions of myonuclei were marked with LamDm0, and muscle fibers were marked with phalloidin. Bars, 50 µm. (G) Quantification of nuclear clustering: n = 19, 17, 18, 20, 20, 31, 9, 26, 29, 18, 26, 29, 38, and 29 from left to right. Corresponding numbers of myonuclei in any cluster are indicated by different colors (right). *, P < 0.05; **, P < 0.01; ***, P < 0.001. Error bars indicate means ± SD. Overexpression of Kud in muscle caused minor or no defects in nuclear spacing, depending on the different transgenic lines (Fig. 5 G). However, the defects were enhanced synergistically by heterozygosity of klar and koi, respectively, although the heterozygosity alone caused no defect (Fig. 5, E and G). Overexpression of TMEM258 in the klar heterozygote caused similar effects, suggestive of functional conservation (Fig. 5, F and G). These data indicate that increases in Kud levels affect LINC-dependent nuclear positioning. Together with the physical interaction between Kud and the LINC complex, these genetic interactions strongly suggest that Kud regulates myonuclear positioning through the LINC complex. Depletion of Kud increases LINC complex components to impair the cells Because of the physical and genetic interactions between Kud and the LINC complex, we examined whether they would affect each other. We found that the LINC complex was not required to retain Kud at the NE because Kud was normally present at the NE in klar mutants and in koi mutants as in the control follicle cells and muscles (Fig. S4, A–F). Interestingly, we found that Kud affected the LINC complex. Msp300 was observed by using the Msp300-GFP-trap, in which the endogenous Msp300 is fused with GFP (Nagarkar-Jaiswal et al., 2015). GFP signals in kud mutant cells (62%, n = 112) were stronger than those in adjacent control cells (2%, n = 87; Fig. 6 A). Furthermore, 70% of kud mutant cells exhibited an increase in Klar (n = 101) when compared with the control cells (4%, n = 86; Fig. 6 B). Additionally, we observed the expression of Koi because it determines the NE localization of KASH proteins (Kracklauer et al., 2007; Technau and Roth, 2008). We found that Koi was also increased in the kud mutant cells (80%, n = 84) when compared with the control cells (5%, n = 62; Fig. 6 B). These data indicate that Kud depletion increases levels of LINC complex components. Figure 6. Depletion of Kud increases LINC complexes to impair the cells. (A and A′) kud mutant cells lacking the nuclear marker His-RFP (A) are outlined by dashed lines. The expression level of Msp300 was visualized by msp300–GFP-trap (A′). (A′′) Merged image. (B–B′′) The expression levels of Klar (B′) and Koi (B′′) were increased in kud mutant cells, which are outlined by dashed lines (GFP positive). (C) Quantification of the clones containing apoptotic cells: n = 53, 72, 86, 104, and 117 from left to right. (D) Quantification of normalized cell areas: n (number of clones) = 9, 13, 16, and 10 from left to right. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Error bars indicate means ± SD. (E–F′) The kud mutant cells coexpressing Myc-tagged Klar-N and GFP are outlined by dashed lines. The Klar antibody recognizes a C-terminal region of Klar that Klar-N lacks. The overexpressed Klar-N (E’) and endogenous Klar (F′) localized at plasma membrane are indicated (arrows). The NE distributions of endogenous Klar are indistinguishable in control (white arrowheads) and in kud mutant cells (yellow arrowheads; F and F′). Bars, 10 µm. The increase in LINC complexes in the kud mutant cells raised the possibility that this might be a cause for the kud loss-of-function phenotypes. If so, reduction of LINC components in kud mutants should rescue the phenotypes. Confirming this, heterozygosity of msp300 reduced the apoptosis level (Fig. 6 C). In the double mutant clones of kud klar, in which Klar was removed completely from the kud mutant cells, apoptosis and cell size reductions were rescued significantly (Fig. 6, C and D). Interestingly, in the kud mutant clones, overexpression of the cytoplasmic domain of Klar (Klar-N; Fischer et al., 2004) significantly rescued apoptosis (Fig. 6 C). Because there was an accumulation of endogenous Klar at the plasma membrane where Klar-N was localized (Fig. 6, E and F) and mutant cells with an increase of Klar at the NE were reduced to 25.4% (n = 111), Klar-N seemed to rescue these abnormalities by trapping the endogenous Klar and thereby reducing its NE targeting. Furthermore, removing one copy of koi—thus theoretically halving the KASH protein content at the NE—was sufficient to rescue apoptosis in the kud mutant follicle cells (Fig. 6 C), suggesting that the increased LINC complex components at the NE were responsible for the phenotype. Kud might suppress anchorage of the cytoskeleton-free KASH isoforms at the NE to maintain LINC functions We wondered how LINC complexes could increase but become harmful for kud mutant cells because overexpression of Klar or Koi in the WT did not result in any defects as seen in the kud mutant follicles (not depicted). Endogenous KASH proteins exhibit various isoforms including those bearing the KASH domain but lacking the cytoskeleton-binding domain (Rajgor and Shanahan, 2013). Theoretically, these cytoskeleton-free KASH isoforms can compete with the KASH-binding sites at the NE to disrupt the LINC complex and thereby reduce nucleocytoskeletal connections. Consistently, in C. elegans and mice, overexpression of the KASH domain causes dominant-negative effects because it disrupts LINC complexes and displaces the endogenous KASH proteins from the NE (Starr and Han, 2002; Wilhelmsen et al., 2006; Starr, 2009). Similarly, we found that overexpression of the KASH domain disrupted myonuclear spacing (Fig. 5 G). Furthermore, we generated clones overexpressing KASH domains in WT follicle cells using a flip-out system (Struhl and Basler, 1993). This overexpression also displaced the endogenous Klar in 38% of follicle cells (n = 105; Fig. 7 A), although such overexpression did not cause any obvious defects (not depicted). This confirmed that increases in cytoskeleton-free KASH isoforms impair LINC function. We found that overexpression of the KASH domain of Klar significantly enhanced apoptosis in the kud mutant cells (Fig. 7 B). A previously described construct of this KASH domain (Klar-C; Guo et al., 2005) showed similar results (Fig. 7 B). These data indicate that defective LINC function leads to apoptosis upon loss of Kud, and that Kud prevents the apoptotic effect of the dominant-negative KASH domain. These results prompted us to examine whether Kud would affect the anchorage of the KASH domain at the NE. When clonally overexpressed in follicles, the KASH domain of Klar was localized at the NE of 82% of the cells (n = 233; Fig. 7 C). Interestingly, overexpression of Kud significantly reduced the NE distribution of the KASH domain: only 8% of the cells (n = 336) retained the NE localization (Fig. 7 D). This treatment did not affect anchorage of endogenous Klar at the NE (Fig. 7 E). These data indicate that Kud can suppress the anchorage of the cytoskeleton-free KASH domain at the NE to maintain LINC functions. Figure 7. Kud suppresses the anchorage of cytoskeleton-free KASH proteins at the NE. (A and A′) HA-KASH–expressing cells are outlined by dashed lines. The NE distribution of Klar was lower in these cells (yellow arrows) than that in control cells (white arrows). (B) Quantification of the clones containing apoptotic cells: n = 78, 43, and 84 from left to right. *, P < 0.05. Error bars indicate means ± SD. (C–D′) HA-KASH was coexpressed with mCD8::GFP (C) and Kud-CFP (D), respectively. The KASH domain was found at the NE in control cells (C′) but was dispersed from the NE in Kud-CFP–overexpressing cells (D′). Arrows indicate the NE. (E and E′) The NE distribution of Klar (E′) is indistinguishable in Kud-CFP–overexpressing cells (outlined by dashed lines; yellow arrows) and in adjacent control cells (white arrows). Bars, 10 µm. Kud might down-regulate the level of LINC complex components through autophagy How does Kud regulate the level of LINC complex? Because the B-type lamin LamDm0 determines the NE targeting of the LINC complex (Kracklauer et al., 2007), we tested whether the level of LamDm0 would be affected by depletion of Kud. However, the level of LamDm0 in kud mutant cells was similar to the control (Fig. S4 G), indicating that Kud does not affect the LINC complex via LamDm0. The protein but not the mRNA level of Klar was significantly increased in kud mutant larvae (Fig. 8, A and B), indicating that Kud negatively regulates the level of KASH proteins and that this regulation occurs in tissues other than follicles. We further examined whether LINC complexes would be regulated by autophagy or proteasome-mediated degradation. Proteasomes mediate the degradation of SUN proteins (Chen et al., 2012). However, blocking such proteasome degradation by knockdown of Rpn6, a subunit of 26S proteasome, did not increase the level of LINC complex components (cells with increased Klar: 2.9%, n = 104; increased Koi: 1%, n = 104). These data suggest that proteasome-dependent degradation might not participate. In contrast, we found that autophagy is involved. Activation of autophagy by overexpression of Atg1 reduced the level of Klar in 63% of cells (n = 104; Fig. 8 C), indicating that Klar can be degraded via autophagy. This treatment also reduced the level of Koi in 9% of cells (n = 122; Fig. 8 D), suggesting that autophagy might also play a role for degrading SUN proteins. In addition, autophagy is activated in late-stage follicle cells (Barth et al., 2011), which could be observed using LysoTracker, a lysosome and autolysosome marker. Interestingly, LysoTracker signals were highly reduced in kud mutant follicle cells (Fig. 8, E and F), and the level of the autophagosome marker Atg8a was increased in kud mutant larvae (Fig. 8 G). These data suggest that loss of Kud might block the autophagic flux. Altogether, our data suggest that Kud might down-regulate LINC complex components through positively regulating autophagy. Figure 8. Kud might down-regulate the level of LINC complex components. (A, B, and G) RT-PCR (A) and Western blotting (WB; B and G) were performed with larval muscle or whole larval lysates. Numbers shown are means ± SD. The normalized mRNA level of klar in third and second instar larvae of WT and in kud mutants was equivalent (A). The normalized protein levels of Klar in WT, kud, and transheterozygote mutants over the deficient condition are shown in B. IP, immunoprecipitation. (C–D′) Atg1 and GFP were coexpressed in follicles (outlined by dashed lines; C and D). In control cells (white arrows), Klar (C′) and Koi (D′) were present at the NE, and the expression was reduced by Atg1 overexpression (yellow arrows). (E and E′) The kud mutant cells (GFP positive) are outlined with dashed lines. The lysosomes and autolysosomes in cytoplasm were marked using LysoTracker (Ltr). (F) Quantification of LysoTracker-positive dots per cell. n = 29 and 41. Error bars indicate means ± SD. (G) The normalized protein levels of Atg8a are shown. Bars, 10 µm. *, P < 0.05; ***, P < 0.001. Domain studies for Kud–KASH association and the function of Kud To ascertain which part of Kud is important for associating with KASH proteins, we generated Kud variants with alterations in each domain (Fig. 9 A). CoIP analyses were performed by coexpressing the Kud variants and the KASH domain of Klar in S2 cells. Substitutions of the N terminus and the loop region with peptides containing glycine or alanine (KudNGA or KudLGA) did not affect the Kud–KASH association (Fig. 9, B and C). Furthermore, to evaluate the importance of Kud IM and TM domains, we replaced the domains with TMs of the ER protein jagunal to generate KudIM-J and KudTM-J as well as TMs of the plasma membrane proteins Neuroglian (Nrg) and Unzipped (Uzip) to generate KudIM-N and KudTM-U. Only KudTM-U exhibited reduced Kud–KASH association, but other TM substitution variants retained a similar strength of association compared with the WT form (Fig. 9, B and C). Interestingly, the strength of the association seemed to correlate with NE targeting, because except for KudTM-U, all Kud variants could target to the NE in muscles (Fig. 9, A, D, and E). Because KudTM-J can localize at the NE, we reason that the reduction in NE targeting of KudTM-U might result from replacement of the TM rather than from its loss and that the TM of Uzip in KudTM-U might promote localization in downstream compartments in the secretory pathway. Together with our findings that the depletion of klar or koi did not affect the NE targeting of Kud (Fig. S4, A–F), these data suggest that NE targeting of Kud is required for Kud–KASH association. In addition, each domain could not work singly but via a compensatory relationship to contribute to NE targeting and the association, because altering single domains did not affect NE targeting of Kud and the Kud–KASH association. Figure 9. Domain studies for the Kud–KASH association and the function of Kud. (A) Schematic diagram showing Kud variants. GA, glycine and alanine; NrgTM, the TM domain of Nrg; UzipTM, the TM domain of Uzip. (B and C) CoIP of Klar’s KASH domain and Kud variants expressed in S2 cells. The interactions were specific because the coIP did not detect the NPC protein Nup62 (B). The efficiency with which Kud variants were coimmunoprecipitated with the KASH domain was quantified by three independent experiments (C). Data shown are means ± SEM. IP, immunoprecipitation; WB, Western blot. (D–E′) Subcellular localization of Kud and KudTM-U. Ring-shaped structures, suggestive of the NE, appeared in larval muscle expressing Kud (D) but not in KudTM-U-expressing muscle (E). LamDm0 marked the NE (E′). Bars, 10 µm. (F) Percentages of cells with the increased Klar. (G) Percentages of clones containing apoptotic cells. The numbers in the bars are cells (F) and clones (G) scored. *, P < 0.05; ***, P < 0.001. To examine the importance of domains for the functions of Kud, we overexpressed Kud variants in kud mutant follicle cells. We found that overexpression of KudTM-U could not rescue the level of Klar as Kud did and could not suppress the apoptosis of mutant cells (Fig. 9, F and G). To avoid the problem of losing NE localization of KudTM-U, we used KudΔN, KudIM-J, and KudTM-J, which can localize at the NE. The variants partially rescued the level of Klar but did not achieve the rescue efficiency of Kud (Fig. 9, F and G). These data suggest that each domain of Kud participates in regulating the level of Klar. Furthermore, Kud and KudΔN, but not KudIM-J or KudTM-J, significantly suppressed the apoptosis (Fig. 9, F and G), suggesting that IM and TM domains of Kud are critical for Kud functions. Because KudIM-J and KudTM-J can associate with the KASH protein, these data suggest that in addition to the association with KASH protein, other mechanisms dependent on IM and TM domains are required for functions of Kud. Kud regulates myonuclear morphology and has LINC-independent functions Knockdown of genes encoding Nesprins alters nuclear shape (Zhang et al., 2007; Lüke et al., 2008). We examined whether Kud and the LINC complex would affect myonuclear morphology in Drosophila. Myonuclei of larval body walls were marked using LamDm0. Nuclear roundness was determined by calculating the nuclear contour ratio (CR), which for a circle is 1, and this number decreases in misshapen nuclei (Goldman et al., 2004). Complete loss of Koi or Klar altered nuclear roundness (Fig. 10, A, B, and E) and caused ectopic nuclear LamDm0 foci (Fig. 10, F–H and L). The LamDm0 foci were attached to the NE and were positive for the INM marker but negative for the nuclear pore complex (NPC) marker Mab414 (Fig. 10, H and I). These observations suggest that the foci are of type I nucleoplasmic reticulum (NR), where the INM but not the ONM invaginates into the nucleoplasm, identical to those found upon knockdown of Nesprin-encoding genes (Zhang et al., 2007). These data confirm that the LINC complex affects myonuclear morphology in Drosophila. Figure 10. Kud regulates myonuclear morphology. (A–D, F–K′, and M–N′) Nuclei were marked with LamDm0. (A–D) The indicated myonuclei are magnified below each image. (E) Quantification of the nuclear roundness, represented by the CR (see the Quantification section in Materials and methods). n > 200 nuclei for each genotype. (F–K′ and M–N′) Confocal microscopy cross sections of the regions indicated by dashed arrows are shown below each image. The bubble-shaped nuclear foci were attached to the NE (arrows). (H–I′) In klar mutant myonuclei, the INM marker Koi (H′) but not the NPC proteins (I′) were found at LamDm0 foci. (J–K′) In Kud-depleted myonuclei, Koi (J′) and the ONM marker KASH domain (K′) were found at the LamDm0 foci. (L) Quantification of nuclear foci. (M–N′) In Kud-overexpressing myonuclei, Koi (M′) and the ONM marker Kud-CFP (N′) were found at the LamDm0 foci. Bars, 10 µm. (O) Quantification of nuclear foci size. Red lines in the scatter dot plots show the mean values. *, P < 0.05; ***, P < 0.001. As Kud regulates LINC complexes, its dysfunctions might phenocopy the mutants of the LINC components in myonuclear morphology. In support of this, misshapen nuclei were observed upon reduction and overexpression of Kud, although only the latter case caused significant reduction in the CR (Fig. 10, C–E). These alterations in Kud levels also produced ectopic LamDm0 foci (Fig. 10, J–N). Overexpression of the human Kud orthologous protein TMEM258 resulted in misshapen nuclei and ectopic LamDm0 foci, confirming functional conservation of these proteins (Fig. 10, E and L). These data suggest that both Kud and the LINC complex are required for maintaining myonuclear morphology in Drosophila. Based on the physical and genetic interactions found in our study, Kud might regulate myonuclear morphology through a LINC-dependent mechanism. However, upon reduction or overexpression of Kud, the LamDm0 foci were positive for INM and ONM markers (Fig. 10, J, K, M, and N). These data indicate that the foci are invaginations of both nuclear membranes, a characteristic of the type II NR, and thus distinct from those seen in mutant forms of the LINC complex. Additionally, the number of LamDm0 foci upon reduction or overexpression of Kud was more than that in koi mutants (Fig. 10 L), and the foci in Kud-overexpressing cells were larger than those in koi mutants (Fig. 10 O). These differences imply that Kud might act through a LINC-independent mechanism. Consistent with this, homozygosity for the kud but not koi mutation impairs follicle development and causes death. Collectively, these data demonstrate that Kud modulates NE architecture and also functions in a LINC-independent manner. Discussion Kud is a newly identified regulator for the LINC complex and NE architecture In J.R.R. Tolkien’s novel, The Lord of the Rings, the Hobbits, who call themselves “Kuduk,” are short and protect the One Ring of Power. Thus, we named the protein Kud because of its short length and importance for the ring-shaped NE. To our knowledge, Kud is a newly described ONM protein and the first LINC regulator found at the ONM. Our data support the idea that Kud modulates NE architecture via LINC-dependent and -independent mechanisms. The findings add to our understanding of regulation in the LINC complex and the NE architecture. Kud-mediated quality control of the LINC complex Our data suggest that Kud is a KASH-interacting protein regulating the LINC complex. As to the lack of evidence of direct interactions between Kud and KASH proteins, we do not exclude the possibility that the interaction is indirect. In addition, the association between TMEM258 and SUN1 suggests that Kud might interact with the KASH proteins of LINC complexes. As KASH proteins can interconnect via the cytoplasmic domains (Lu et al., 2012; Taranum et al., 2012), Kud might interact with the KASH proteins of LINC complexes through associating with other KASH proteins that do not participate in LINC complexes. Further studies should examine how Kud interacts with KASH proteins. Based on our finding, we speculate that Kud might regulate the LINC complex in several ways. First, partners of KASH proteins have been suggested to keep the three KASH proteins interacting with a SUN trimer apart (Sosa et al., 2012). Such Kud–KASH association might allow Kud to regulate the organization of KASH proteins. Second, because of the alternative initiation of transcription and splicing, mammalian Nesprins have KASH isoforms, which contain the KASH domain, without or with cytoskeleton-interacting domains of different lengths (Rajgor and Shanahan, 2013). Switching KASH isoforms might alter cytoskeletal connections of the LINC complex to regulate muscle development and embryonic stem cell differentiation (Randles et al., 2010; Smith et al., 2011). Our data support the idea that Kud suppresses the NE anchorage of cytoskeleton-free KASH proteins to ensure that the LINC complexes are linked to the cytoskeleton. Therefore, Kud might act as a quality control for the cytoskeletal connections of LINC complexes. Third, LINC complexes can align linearly along dorsal actin cables to form higher-ordered assemblies, the transmembrane actin-associated nuclear (TAN) lines. Formation of the TAN lines depends on the cytoskeletal connections of KASH proteins (Luxton et al., 2010). Thus, Kud might also benefit the formation of the higher-ordered LINC assemblies. Collectively, it appears that Kud regulates the quality of the LINC complex, thus modulating nucleocytoskeletal connections during development. Kud might promote autophagic degradation of the LINC complex Components of the NE can be degraded by nuclear autophagy, which is evolutionarily conserved (Luo et al., 2016). However, the substrates and regulatory mechanisms are largely unknown. Our data support the idea that LINC complexes might be substrates of nuclear autophagy and that Kud could suppress the level of KASH proteins by positively regulating autophagy. We speculate that the cytoskeleton-free KASH proteins are prone to be excluded from the TAN line and that the association with Kud might allow Kud-mediated autophagy to eliminate cytoskeleton-free KASH proteins selectively. Besides this, our data indicate that Kud-mediated autophagy might play a minor role for regulating the level of Koi. Further studies should identify which mechanisms downstream of Kud are responsible for regulating the level of Koi. Kud might regulate nuclear morphology and autophagy to affect global cellular functions Given that Kud regulates NE architecture and the LINC complex and that NE targeting is required for the Kud–KASH association, we suggest that NE-localized Kud plays a crucial role. It is still worthy to examine the subcellular localization and functions of cytoplasmic Kud. Besides, our data suggest that in addition to the Kud–KASH association, Kud might participate in other mechanisms dependent on IM and TM domains. TMs can dimerize through the (small)xxx(small) motif, in which “small” means small amino acids such as glycine and serine (Russ and Engelman, 2000; Schneider and Engelman, 2004). Interestingly, the TM of Kud has a conserved tandem repeat of the dimerization motif SLCASIFLG (Fig. 1 A). This observation along with the dimerization of bacterially expressed Kud (Fig. S3 C) suggest a homophilic interaction. Oligomerization of membrane proteins and insertion of wedgelike amphipathic helices can regulate membrane curvature (McMahon and Gallop, 2005; Drin and Antonny, 2010). It would be interesting to examine whether the homophilic interaction of Kud and the presumptively amphipathic IM might play a role in regulating nuclear morphology. Finally, nuclear morphology can control gene expression and DNA repair (Walters et al., 2012), and nuclear autophagy can degrade NE, NE proteins, and chromatin (Luo et al., 2016). Therefore, we reason that Kud controls the homeostasis of the LINC complex and global factors through regulation of nuclear morphology and autophagy, thus including LINC-independent cellular functions. TMEM258 is associated with human diseases We show in this study that Kud is required for the development of Drosophila ovarian follicles and muscles. However, depletion of Kud did not impair the nuclear migration of photoreceptors (Fig. S5), which depends on klar and msp300 (Mosley-Bishop et al., 1999; Patterson et al., 2004; Technau and Roth, 2008). These data indicate that Kud functions in a tissue-specific manner, as with the LINC complex and other NE proteins (Worman and Schirmer, 2015). Our data show that the human orthologous protein TMEM258 exhibits conserved functions, suggesting that dysregulation of this protein might affect LINC functions and the NE to cause defects in specific tissues. In addition, we have demonstrated that overexpression of Kud and TMEM258 can result in cellular defects, suggesting that an increase in TMEM258 might cause diseases. Interestingly, in the dominantly inherited disorder spinocerebellar ataxia type 20, the chromosomal region containing TMEM258 is duplicated, and the expression level of this gene is increased (Knight et al., 2008). In addition, misshapen nuclei and NR are characteristics of cancer cells (Malhas and Vaux, 2014). The large noncoding RNA ANRIL (antisense noncoding RNA in the INK4 locus), which is associated with cancers and metastasis, positively regulates the expression of TMEM258 (Bochenek et al., 2013; Qiu et al., 2015). Graham et al. (2016) recently reported that increased TMEM258 expression is associated with inflammatory bowel disease. Further studies should examine whether the pathology of TMEM258-associated diseases involves alterations of the LINC complex and NE architecture. Materials and methods Fly stocks and incubation The stocks we used were as follows: klarΔ1–18 (Elhanany-Tamir et al., 2012), msp300compl (Technau and Roth, 2008), koiHRKO80.w (Kracklauer et al., 2007), 24B-GAL4 (Brand and Perrimon, 1993), UAS-Atg1 (Chen et al., 2008), UAS-6×myc–Klar-N (Fischer et al., 2004), UAS-6×myc–Klar-C (Fischer et al., 2004), UAS-mCD8-GFP (Gao et al., 1999), and UAS-Atg1 (Chen et al., 2008). The stock w1118 was used as a WT control. The following stocks were obtained from the Bloomington Drosophila Stock Center: Mef2-GAL4 (BL50742), 6934-Hid (BL6934), GAL4477[w–]; TM2/TM6 (BL26258), Df(3L)Exel9002 (BL7935), Msp300GFP (BL59757), and sqh-EYFP-Mito (BL7194). UAS-Rpn6 Ri (VDRC18021) was obtained from the Vienna Drosophila Resource Center. The following stocks were used to generate flies bearing mutant or overexpression clones: hs-FLP (BL26902), ubi-GFP.nls FRT2A (BL5825), tub-GAL80 FRT2A (BL5190), act-GAL4, UAS-GFP, Act>CD2>Gal4 (BL4779), and Act>CD2>Gal4; UAS-GFP (BL39760). The transgenic flies we generated were: UAS-Kud-CFP, UAS-Kud-GFP, UAS-Kud-HA, UAS-KudΔN-HA, UAS-KudNGA-GFP, UAS-KudLGA-GFP, UAS-KudIM-J-GFP, UAS-KudTM-J-GFP, UAS-KudIM-N-CFP, UAS-KudTM-U-CFP, UAS-kud Ri, UAS-TMEM258-CFP, and UAS-HA-KASH. All flies were incubated at 25°C for observing follicle cells and at 29°C for higher-level expressions of transgenes in larval muscles. Generation of mutant and overexpression clones in ovarian follicle cells Female adults were heat shocked at 37°C for 1 h and then were mated and cultured in vials containing fresh yeast-supplemented food. Ovaries were dissected at 96–114 h after clone inductions for immunostaining. Female genotypes and the abbreviations for generating mutant clones are listed as follows: (a) kud–/– (using FRT/FLP technique): hs-FLP/+; kud FRT2A/ubi-GFP.nls FRT2A; (b) kud–/– (using the mosaic analysis with a repressible cell marker technique): hs-FLP/+; act>GFP/+; kud FRT2A/tub-GAL80 FRT2A; (c) kud–/– klar–/–: hs-FLP/+; klar kud FRT2A/ubi-GFP.nls FRT2A; (d) kud–/– msp300–/+: hs-FLP/+; msp300/+; kud FRT2A/ubi-GFP.nls FRT2A; (e) kud–/– koi–/+: hs-FLP/+; koi/+; kud FRT2A/ubi-GFP.nls FRT2A; (f) kud–/– msp300GFP/+: hs-FLP/+; msp300GFP/+; kud FRT2A/ubi-GFP.nls FRT2A; (g) kud–/– + HA-KASH: hs-FLP/+; act> HA-KASH /+; kud FRT2A/tub-GAL80 FRT2A; and (h) kud–/– + Myc-Klar-N (or Myc-Klar-C): hs-FLP/+; act> Myc-Klar-N (or Myc-Klar-C)/+; kud FRT2A/tub-GAL80 FRT2A. Female genotypes for generating overexpression clones are listed as follows: (a) Act>CD2>Gal4−/+; hs-FLP/UAS-X (and UAS-Y); The X and Y mean one of the overexpressed proteins: HA-KASH, Kud-CFP, or mCD8GFP; and (b) Atg1 + GFP: Act>CD2>Gal4; UAS-GFP/UAS-Atg1; hs-FLP/+. Preparation of ovarian follicles and larval muscles and immunostaining Adult female flies were put on ice for 10 min. Ovaries were dissected and teased apart from the abdomen in PBS on Sylgard plates and then shaken in a fixation solution (volume ratio of 4% formaldehyde and 1% NP-40 in PBS: heptane = 1:3) for 20 min. The egg chambers were dispersed by pipetting the ovaries vigorously. Stage 10–12 egg chambers were collected and washed in PBS with 0.3% Triton X-100 (PBST). Stage 12 egg chambers were used for observing apoptosis, cell areas, and the subcellular localization of Kud and LysoTracker. Stage 10 egg chambers were used for observing other proteins. Third instar larvae were put on ice for 10 min and pinned on Sylgard plates. Larvae were slit open, cleaned of fat bodies and organs, and fixed in 4% formaldehyde in PBS for 20 min, and then they were washed with PBST and blocked in 5% normal donkey serum in PBST for immunostaining. For digitonin experiments, the larval muscles were treated with 20 µg/ml digitonin on ice for 5 min for permeabilization, and PBS was substituted for PBST for subsequent immunostaining procedures. For protease protection assays, the digitonin-permeabilized larval muscles were washed with PBS and then treated with or without 50 µg/ml proteinase K on ice for 5 min. The muscles were washed with PBS and fixed in 4% formaldehyde for 20 min and then were blocked in 5% normal donkey serum in PBST for immunostaining. The first antibodies used in immunostaining included mouse anti-Dlg (4F3; 1:50), mouse anti-GFP (GFP-G1; 1:100), mouse anti-LamDm0 (ADL84.12; 1:100), mouse anti–Klar-C (9C10; 1:30), mouse anti-Cnx99A (6-2-1; 1:100), and rat anti-Elav (7E8A10; 1:200) were obtained from the Developmental Studies Hybridoma Bank. The antibody for Drosophila Kud was generated in rabbits against the peptide sequence of aa 1–18 (1:600; MDVMQRYVSPVNPAVFPH; LTK BioLaboratories). The antibody for Drosophila Koi was generated in rabbits against the peptide sequence of amino acids 79–96 (1:2,000; DYSSDDMTPDAKRKQNSI; LTK BioLaboratories). The other commercial antibodies used were rabbit anti-GFP–conjugated Alexa Fluor 488 (1:500; Thermo Fisher Scientific), rabbit anti–activated caspase 3 (1:250; Cell Signaling Technology), mouse anti-Myc (1:200; Santa Cruz Biotechnology, Inc.), mouse anti-HA (1:50; Santa Cruz Biotechnology, Inc.), mouse anti–NPC proteins (Mab414; 1:200; Abcam), mouse anti-KDEL (10C3; 1:50; Santa Cruz Biotechnology, Inc.), mouse anti–Golgi p120 (7H6D7C2; 1:100; EMD Millipore), rabbit anti-HA (1:400; Cell Signaling Technology), goat anti–human lamin B1 (1:100; Santa Cruz Biotechnology, Inc.), and rhodamine-phalloidin (1:200; Invitrogen). Secondary antibodies including goat anti–mouse Alexa Fluor 488, anti–rabbit Alexa Fluor 488, and donkey anti–goat Alexa Fluor 546 were obtained from Invitrogen. Goat anti–mouse Cy3, anti–rabbit Cy3, and anti–rabbit Cy5 were obtained from Jackson ImmunoResearch Laboratories, Inc. For labeling with LysoTracker DND-99 (Thermo Fisher Scientific), ovaries were dissected in Schneider’s Drosophila Medium (Thermo Fisher Scientific), washed briefly, and incubated for 5 min in 25 µM LysoTracker. Then, they were washed three times in PBS and fixed in 4% paraformaldehyde for 20 min, washed three times in PBST, and then mounted and imaged immediately. Confocal microscopy After immunostaining, muscles and egg chambers were mounted in PBST for imaging using Plan-Apochromat 20× 0.6 NA or 63× 1.4 NA oil immersion objectives and 100× 1.4 NA oil immersion lenses with a confocal microscope (LSM 510; ZEISS). The fluorescence images were processed using Photoshop CS5 (Adobe). Cell culture, fractionation, immunoprecipitation, and Western blotting Drosophila S2 cells were cultured as described in the Drosophila Expression System (Invitrogen) and were transfected using commercial Calcium Phosphate Transfection kits (Invitrogen) or TransIT-Insect Transfection Reagent (Mirus). Human 293T cells were cultured in DMEM supplemented with 10% FBS and transfected as described in the procedure manual for the TurboFect Transfection Reagent (Thermo Fisher Scientific). For subcellular fractionation, 293T cells were processed following the manufacturer’s instructions of the Cell Membrane Protein Extraction kit (Biokit Biotechnology). For immunoprecipitation, lysates of equal numbers of cells were incubated with protein A beads (Protein A Sepharose CL-4B; GE Healthcare) coupled, respectively, with mouse antipolyhistidine (His; 1:100; Santa Cruz Biotechnology, Inc.), mouse anti-HA (1:40; Santa Cruz Biotechnology, Inc.), mouse anti-Myc (1:40; Santa Cruz Biotechnology, Inc.), mouse anti-GFP[9F9.F9] (1:120; Abcam), mouse IgG (1:250; Santa Cruz Biotechnology, Inc.), rabbit anti-FLAG (1:100; Sigma-Aldrich), and rabbit IgG (1:250; Abcam) antibodies overnight at 4°C. After incubation, beads were washed once with a low-stringency wash buffer (50 mM Tris, pH 7.5, 150 mM NaCl, and 0.1% NP-40) and twice with a high-stringency buffer (50 mM Tris, pH 7.5, and 0.1% NP-40), and then they were boiled with 1× SDS loading buffer to elute the proteins. Proteins were resolved by 8% or 12% SDS–PAGE and subjected to standard Western blotting techniques. The primary and secondary antibodies for Western blotting were rabbit anti-Kud (1:1,000), mouse anti-HA (1:200; Santa Cruz Biotechnology, Inc.), mouse anti-His (1:2,000; Santa Cruz Biotechnology, Inc.), mouse anti-Myc (1:300; Santa Cruz Biotechnology, Inc.), mouse anti-Nup62 [Mab414] (1:500; Abcam), mouse anti-Cnx99A (6-2-1; 1:500; Developmental Studies Hybridoma Bank), mouse anti-GAPDH (1:5,000; Novus Biologicals), mouse anti-Emerin[8A1] (1:600; Developmental Studies Hybridoma Bank), rabbit anti-Calnexin (1:600; Novus Biologicals), mouse anti–Klar-C (9C10; 1:30, Developmental Studies Hybridoma Bank), rabbit anti-HA (1:1,500; Cell Signaling Technology), rabbit anti-Atg8a (1:2,500; provided by G.-C. Chen, Academia Sinica, Taipei, Taiwan), rabbit anti-FLAG (1:2,000; Sigma-Aldrich), rabbit anti-GFP (1:500; Santa Cruz Biotechnology, Inc.), goat anti-TMEM258 (1:300; Santa Cruz Biotechnology, Inc.), goat anti–mouse HRP (1:5,000; Jackson ImmunoResearch Laboratories, Inc.), goat anti–rabbit HRP (1:5,000; Jackson ImmunoResearch Laboratories, Inc.), and donkey anti–goat HRP (1:5,000; Jackson ImmunoResearch Laboratories, Inc.). RT-PCR Total RNA was purified from WT and kud mutant larvae with TRIzol and then the RNA was reverse-transcribed into cDNA with SuperScript cDNA Synthesis kit (Invitrogen). 300 ng cDNA was amplified by primers specific for Klar and GAPDH with Tag polymerase (Promega). The mRNA level of Klar was normalized to the level of GAPDH. Quantification The normalized cell area of follicles was calculated from the areas of five mutant cells based on that of control cells in one egg chamber. Images of myonuclear foci (>5 µm2) were acquired from larval muscles 6 and 7 of abdominal segments 2 and 3. The numbers of foci per myonucleus were the means of the total numbers of myonuclear foci in one segment. For CR calculations, the area and perimeter of each nucleus were measured using ImageJ (National Institutes of Health). The CR was calculated for each nucleus using the formula 4π area/perimeter2. For foci size calculations, the area of each was measured. Myonuclear positions were observed in larval muscles 6 and 7 of abdominal segments 5 and 6. In quantification of myonuclei clusters, a group of myonuclei was defined as having a distance between any two of less than half of their diameter. Statistical significance was calculated from the percentages of clusters containing more than two myonuclei. RT-PCR and Western blotting results were quantified using ImageJ, and the intensity was normalized to WT. Statistical analysis was performed with Prism 5.0 (GraphPad Software). Statistical analysis was performed using Student’s t test unless otherwise noted. Asterisks in graphs indicate the significance of P values comparing indicated group with controls unless specifically indicated (*, P < 0.05; **, P < 0.01; ***, P < 0.001). Generation of kud knockout flies The kud knockout lines were generated by homologous recombination as described previously (Huang et al., 2008). In brief, the crossing strategies were divided into three steps: targeting, screening, and mapping crosses. In targeting crosses, virgin females of a transgenic line bearing the donor DNA (“P {kud}”) were crossed with 6934-hid males (yw/Y, hs-hid; hs-FLP, hs-I-SceI/CyO, hs-hid) for egg laying over 24 h. The vials were then heat-shocked twice at 38°C for 1.5 h during 2 d. The linear donor DNA fragments (P[kud]Rpr+) were generated extrachromosomally by FLP and I-SceI enzymes and inserted into the targeted chromosome. In screening crosses, virgin females (yw; hs-FLP,hs-I-SceI/P[kud]Rpr+) from the targeting crosses were mated with balancer males expressing GAL4477 (yw/Y; GAL4477w-]; TM2/TM6B) to eliminate any failed positive candidates. In mapping crosses, the candidates of kud knockout lines with red eyes (yw/Y; hs-FLP, hs-I-SceI/GAL4477w-]; kudKO/TM6B or TM2) were selected and crossed with balancer flies (Dr/TM6B) individually to maintain them. Generation of constructs and transgenic flies The whole coding region of Kud was amplified from the S2 cell cDNA library and cloned into pGEM–T-Easy vector to generate pGEM–T-Kud (Promega). 3HA DNA was subcloned into pGEM–T-Kud, and C-terminally tagged Kud-3HA was subcloned into pUAST. KudΔN, KudIM-N, and KudTM-U was amplified by multiple-step PCR from the templates possessing Kud-3HA, Nrg TM, and Uzip TM DNA, respectively (Ding et al., 2011) and cloned into pGEM–T-Easy vector. The DNA segments were subcloned into pUAST-CFP vector to generate the C-terminally tagged Kud variants (Drosophila Genomics Resource Center), and KudΔN-3HA was subcloned into pUAST. KudLGA, KudNGA, KudIM-J, and KudTM-J DNA were synthesized by Quantum Biotechnology. The synthesized DNA segments were cloned into pMX vectors. To generate the C-terminally tagged Kud variants, these DNA segments were subcloned into pUAST-GFP vectors whose GFP was amplified from genomic DNA of UAS-Syt1-GFP fly. sym-pUAST-kud was generated by subcloning the full-length Kud into sym-pUAST (Giordano et al., 2002), which can produce double-stranded RNA in vivo. C-terminally tagged KASH(Klar)-3HA and KASH(Msp300)-3HA were amplified by three-step PCR from the S2 cell cDNA library and pGEMT-Kud-3HA and were cloned into pGEM–T-Easy vector. N-terminally tagged 3HA-KASH(Klar) was amplified by three-step PCR from pMT/V5-KASH(Klar)-3HA. The amplified cDNAs were cloned into pGEM–T-Easy vector and was subcloned into pMT/V5 or pUAST. pMT/V5-Kud-His was generated by replacing the full-length Kud into pMT/V5-Uzip-His (Ding et al., 2011). TMEM258 was amplified by PCR from cDNA of 293T cells. KASH(Syne1)-3HA was amplified by three-step PCR from the cDNA library of 293T cell and pGEMT-Kud-3HA. The amplified cDNAs were cloned into pGEM–T-Easy vector and were subcloned into pcDNA3.1/myc-His vector (Thermo Fisher Scientific). pcDNA3-hSUN1-dHA, pcDNA3-dFLAG-hSYNE1, pcDNA3-dFLAG-hSYNE1 (1–922; KASH deletion), pcDNA3-dFLAG-hSYNE2, pcDNA3-dFLAG-hSYNE2 (1–498; KASH deletion), pcDNA3-dFLAG-BAF (Chi et al., 2007), and pcDNA3-dFLAG-STIM1 were provided by Y.-H. Chi (National Health Research Institutes, Zhunan, Taiwan). Full-length Syne1 v3 (983 aa) was cloned from cDNA of KIAA0796, and full-length Syne2 v4 (556 aa) was cloned from RT-PCR. pUAST clones were microinjected into w1118 early stage embryos, which carried transposase Δ2–3 for generating UAS transgenic flies. Constructs of pMT/V5, pUAST, and pWA-GAL4 were used to express proteins in S2 cells. pcDNA3 constructs were used to express proteins in human 293T cells. Online supplemental material Fig. S1 illustrates the generation of kud knockout mutants. Fig. S2 indicates that cytoplasmic Kud can localize at the ER membrane but not the ER lumen, Golgi, or mitochondria. Fig. S3 illustrates a proteinase K protection assay of the topology of TMEM258 and dimerization of Kud. Fig. S4 shows that loss of LINC components does not affect the NE targeting of Kud, and loss of Kud does not affect the level of LamDm0. Fig. S5 shows that mutation in kud does not impair the nuclear migration of photoreceptors in Drosophila eyes. Acknowledgments We thank H. Elhanany-Tamir for klar mutant, M. Technau for msp300 mutant, M. P. Kracklauer for koi mutant, M.-D. Lin for KDEL antibody, G.-C. Chen for UAS-Atg1 fly and Atg8a antibody, the Bloomington Drosophila Stock Center for public fly strains, the Developmental Studies Hybridoma Bank for antibodies, FlyCore in Taiwan for assistance, H.-H. Lee of National Taiwan University for the fly for homologous recombination, L.-C. Chi and Y.-C. Liu for assistance in experiments, and C.-T. Chien of Academia Sinica for discussion and comments. This work was supported by the Ministry of Science and Technology, Taiwan grant 103-2311-B-194-001-MY3. The authors declare no competing financial interests. Author contributions: Z.-Y. Ding and Y.-H. Wang performed most of the experiments. Y.-C. Huang generated some plasmids and initially observed phenotypes. M.-C. Lee generated kud knockout flies. M.-J. Tseng helped with human cell culture system. Y.-H. Chi provided several plasmids used in human cell experiments and commented on the manuscript. Z.-Y. Ding, Y.-H. Wang, and M.-L. Huang designed and interpreted the experiments and wrote the paper. M.-L. Huang supervised the project. Abbreviations used: coIP coimmunoprecipitation CR contour ratio FRT flippase recombination target IM intramembrane INM inner nuclear membrane KASH Klarsicht/ANC-1/SYNE homology LINC linker of nucleoskeleton and cytoskeleton NE nuclear envelope NPC nuclear pore complex NR nucleoplasmic reticulum ONM outer nuclear membrane PNS perinuclear space SUN Sad1/UNC-84 TAN transmembrane actin-associated nuclear TM transmembrane domain ==== Refs Adachi, N., Z.E. Karanjawala, Y. Matsuzaki, H. Koyama, and M.R. Lieber. 2002. Two overlapping divergent transcription units in the human genome: The FEN1/C11orf10 locus. OMICS. 6 :273–279. 10.1089/15362310260256927 12427278 Attali, R., N. Warwar, A. Israel, I. Gurt, E. McNally, M. Puckelwartz, B. Glick, Y. Nevo, Z. Ben-Neriah, and J. Melki. 2009. Mutation of SYNE-1, encoding an essential component of the nuclear lamina, is responsible for autosomal recessive arthrogryposis. Hum. Mol. Genet. 18 :3462–3469. 10.1093/hmg/ddp290 19542096 Barth, J.M., J. Szabad, E. Hafen, and K. Köhler. 2011. Autophagy in Drosophila ovaries is induced by starvation and is required for oogenesis. Cell Death Differ. 18 :915–924. 10.1038/cdd.2010.157 21151027 Bochenek, G., R. Häsler, N.E. El Mokhtari, I.R. König, B.G. Loos, S. Jepsen, P. Rosenstiel, S. Schreiber, and A.S. Schaefer. 2013. The large non-coding RNA ANRIL, which is associated with atherosclerosis, periodontitis and several forms of cancer, regulates ADIPOR1, VAMP3 and C11ORF10. Hum. Mol. Genet. 22 :4516–4527. 10.1093/hmg/ddt299 23813974 Boni, A., A.Z. Politi, P. Strnad, W. Xiang, M.J. Hossain, and J. Ellenberg. 2015. Live imaging and modeling of inner nuclear membrane targeting reveals its molecular requirements in mammalian cells. J. Cell Biol. 209 :705–720. 10.1083/jcb.201409133 26056140 Brand, A.H., and N. Perrimon. 1993. Targeted gene expression as a means of altering cell fates and generating dominant phenotypes. Development. 118 :401–415.8223268 Burke, B., and C.L. Stewart. 2014. Functional architecture of the cell’s nucleus in development, aging, and disease. Curr. Top. Dev. Biol. 109 :1–52.24947235 Chang, W., H.J. Worman, and G.G. Gundersen. 2015. Accessorizing and anchoring the LINC complex for multifunctionality. J. Cell Biol. 208 :11–22. 10.1083/jcb.201409047 25559183 Chen, C.Y., Y.H. Chi, R.A. Mutalif, M.F. Starost, T.G. Myers, S.A. Anderson, C.L. Stewart, and K.T. Jeang. 2012. Accumulation of the inner nuclear envelope protein Sun1 is pathogenic in progeric and dystrophic laminopathies. Cell. 149 :565–577. 10.1016/j.cell.2012.01.059 22541428 Chen, G.C., J.Y. Lee, H.W. Tang, J. Debnath, S.M. Thomas, and J. Settleman. 2008. Genetic interactions between Drosophila melanogaster Atg1 and paxillin reveal a role for paxillin in autophagosome formation. Autophagy. 4 :37–45. 10.4161/auto.5141 17952025 Chi, Y.H., K. Haller, J.M. Peloponese Jr., and K.T. Jeang. 2007. Histone acetyltransferase hALP and nuclear membrane protein hsSUN1 function in de-condensation of mitotic chromosomes. J. Biol. Chem. 282 :27447–27458. 10.1074/jbc.M703098200 17631499 Ding, Z.Y., Y.H. Wang, Z.K. Luo, H.F. Lee, J. Hwang, C.T. Chien, and M.L. Huang. 2011. Glial cell adhesive molecule unzipped mediates axon guidance in Drosophila. Dev. Dyn. 240 :122–134. 10.1002/dvdy.22508 21117153 Drin, G., and B. Antonny. 2010. Amphipathic helices and membrane curvature. FEBS Lett. 584 :1840–1847. 10.1016/j.febslet.2009.10.022 19837069 Elhanany-Tamir, H., Y.V. Yu, M. Shnayder, A. Jain, M. Welte, and T. Volk. 2012. Organelle positioning in muscles requires cooperation between two KASH proteins and microtubules. J. Cell Biol. 198 :833–846. 10.1083/jcb.201204102 22927463 Fischer, J.A., S. Acosta, A. Kenny, C. Cater, C. Robinson, and J. Hook. 2004. Drosophila klarsicht has distinct subcellular localization domains for nuclear envelope and microtubule localization in the eye. Genetics. 168 :1385–1393. 10.1534/genetics.104.028662 15579692 Gao, F.B., J.E. Brenman, L.Y. Jan, and Y.N. Jan. 1999. Genes regulating dendritic outgrowth, branching, and routing in Drosophila. Genes Dev. 13 :2549–2561. 10.1101/gad.13.19.2549 10521399 Gautier, R., D. Douguet, B. Antonny, and G. Drin. 2008. HELIQUEST: a web server to screen sequences with specific α-helical properties. Bioinformatics. 24 :2101–2102. 10.1093/bioinformatics/btn392 18662927 Giordano, E., R. Rendina, I. Peluso, and M. Furia. 2002. RNAi triggered by symmetrically transcribed transgenes in Drosophila melanogaster. Genetics. 160 :637–648.11861567 Giot, L., J.S. Bader, C. Brouwer, A. Chaudhuri, B. Kuang, Y. Li, Y.L. Hao, C.E. Ooi, B. Godwin, E. Vitols, 2003. A protein interaction map of Drosophila melanogaster. Science. 302 :1727–1736. 10.1126/science.1090289 14605208 Goldman, R.D., D.K. Shumaker, M.R. Erdos, M. Eriksson, A.E. Goldman, L.B. Gordon, Y. Gruenbaum, S. Khuon, M. Mendez, R. Varga, and F.S. Collins. 2004. Accumulation of mutant lamin A causes progressive changes in nuclear architecture in Hutchinson–Gilford progeria syndrome. Proc. Natl. Acad. Sci. USA. 101 :8963–8968. 10.1073/pnas.0402943101 15184648 Graham, D.B., A. Lefkovith, P. Deelen, N. de Klein, M. Varma, A. Boroughs, A.N. Desch, A.C. Ng, G. Guzman, M. Schenone, 2016. TMEM258 is a component of the oligosaccharyltransferase complex controlling ER stress and intestinal inflammation. Cell Reports. 17 :2955–2965. 10.1016/j.celrep.2016.11.042 27974209 Gros-Louis, F., N. Dupré, P. Dion, M.A. Fox, S. Laurent, S. Verreault, J.R. Sanes, J.P. Bouchard, and G.A. Rouleau. 2007. Mutations in SYNE1 lead to a newly discovered form of autosomal recessive cerebellar ataxia. Nat. Genet. 39 :80–85. 10.1038/ng1927 17159980 Guo, Y., S. Jangi, and M.A. Welte. 2005. Organelle-specific control of intracellular transport: Distinctly targeted isoforms of the regulator Klar. Mol. Biol. Cell. 16 :1406–1416. 10.1091/mbc.E04-10-0920 15647372 Horn, H.F., Z. Brownstein, D.R. Lenz, S. Shivatzki, A.A. Dror, O. Dagan-Rosenfeld, L.M. Friedman, K.J. Roux, S. Kozlov, K.T. Jeang, 2013. The LINC complex is essential for hearing. J. Clin. Invest. 123 :740–750.23348741 Huang, J., W. Zhou, A.M. Watson, Y.N. Jan, and Y. Hong. 2008. Efficient ends-out gene targeting in Drosophila. Genetics. 180 :703–707. 10.1534/genetics.108.090563 18757917 Knight, M.A., D. Hernandez, S.J. Diede, H.G. Dauwerse, I. Rafferty, J. van de Leemput, S.M. Forrest, R.J. Gardner, E. Storey, G.J. van Ommen, 2008. A duplication at chromosome 11q12.2–11q12.3 is associated with spinocerebellar ataxia type 20. Hum. Mol. Genet. 17 :3847–3853. 10.1093/hmg/ddn283 18801880 Kracklauer, M.P., S.M. Banks, X. Xie, Y. Wu, and J.A. Fischer. 2007. Drosophila klaroid encodes a SUN domain protein required for Klarsicht localization to the nuclear envelope and nuclear migration in the eye. Fly (Austin). 1 :75–85. 10.4161/fly.4254 18820457 Lee, T., and L. Luo. 1999. Mosaic analysis with a repressible cell marker for studies of gene function in neuronal morphogenesis. Neuron. 22 :451–461. 10.1016/S0896-6273(00)80701-1 10197526 Lee, K.K., D. Starr, M. Cohen, J. Liu, M. Han, K.L. Wilson, and Y. Gruenbaum. 2002. Lamin-dependent localization of UNC-84, a protein required for nuclear migration in Caenorhabditis elegans. Mol. Biol. Cell. 13 :892–901. 10.1091/mbc.01-06-0294 11907270 Lei, K., X. Zhang, X. Ding, X. Guo, M. Chen, B. Zhu, T. Xu, Y. Zhuang, R. Xu, and M. Han. 2009. SUN1 and SUN2 play critical but partially redundant roles in anchoring nuclei in skeletal muscle cells in mice. Proc. Natl. Acad. Sci. USA. 106 :10207–10212. 10.1073/pnas.0812037106 19509342 Lu, W., M. Schneider, S. Neumann, V.M. Jaeger, S. Taranum, M. Munck, S. Cartwright, C. Richardson, J. Carthew, K. Noh, 2012. Nesprin interchain associations control nuclear size. Cell. Mol. Life Sci. 69 :3493–3509. 10.1007/s00018-012-1034-1 22653047 Lüke, Y., H. Zaim, I. Karakesisoglou, V.M. Jaeger, L. Sellin, W. Lu, M. Schneider, S. Neumann, A. Beijer, M. Munck, 2008. Nesprin-2 Giant (NUANCE) maintains nuclear envelope architecture and composition in skin. J. Cell Sci. 121 :1887–1898. 10.1242/jcs.019075 18477613 Luo, M., X. Zhao, Y. Song, H. Cheng, and R. Zhou. 2016. Nuclear autophagy: An evolutionarily conserved mechanism of nuclear degradation in the cytoplasm. Autophagy. 12 :1973–1983. 10.1080/15548627.2016.1217381 27541589 Luxton, G.W., E.R. Gomes, E.S. Folker, E. Vintinner, and G.G. Gundersen. 2010. Linear arrays of nuclear envelope proteins harness retrograde actin flow for nuclear movement. Science. 329 :956–959. 10.1126/science.1189072 20724637 Lv, X.B., L. Liu, C. Cheng, B. Yu, L. Xiong, K. Hu, J. Tang, L. Zeng, and Y. Sang. 2015. SUN2 exerts tumor suppressor functions by suppressing the Warburg effect in lung cancer. Sci. Rep. 5 :17940. 10.1038/srep17940 26658802 Malhas, A.N., and D.J. Vaux. 2014. Nuclear envelope invaginations and cancer. Adv. Exp. Med. Biol. 773 :523–535. 10.1007/978-1-4899-8032-8_24 24563364 Matsumoto, A., M. Hieda, Y. Yokoyama, Y. Nishioka, K. Yoshidome, M. Tsujimoto, and N. Matsuura. 2015. Global loss of a nuclear lamina component, lamin A/C, and LINC complex components SUN1, SUN2, and nesprin-2 in breast cancer. Cancer Med. 4 :1547–1557. 10.1002/cam4.495 26175118 McMahon, H.T., and J.L. Gallop. 2005. Membrane curvature and mechanisms of dynamic cell membrane remodelling. Nature. 438 :590–596. 10.1038/nature04396 16319878 Mosley-Bishop, K.L., Q. Li, L. Patterson, and J.A. Fischer. 1999. Molecular analysis of the klarsicht gene and its role in nuclear migration within differentiating cells of the Drosophila eye. Curr. Biol. 9 :1211–1220. 10.1016/S0960-9822(99)80501-6 10556085 Nagarkar-Jaiswal, S., P.T. Lee, M.E. Campbell, K. Chen, S. Anguiano-Zarate, M.C. Gutierrez, T. Busby, W.W. Lin, Y. He, K.L. Schulze, 2015. A library of MiMICs allows tagging of genes and reversible, spatial and temporal knockdown of proteins in Drosophila. eLife. 4 :e05338. 10.7554/eLife.05338 25824290 Patterson, K., A.B. Molofsky, C. Robinson, S. Acosta, C. Cater, and J.A. Fischer. 2004. The functions of Klarsicht and nuclear lamin in developmentally regulated nuclear migrations of photoreceptor cells in the Drosophila eye. Mol. Biol. Cell. 15 :600–610. 10.1091/mbc.E03-06-0374 14617811 Puckelwartz, M.J., E. Kessler, Y. Zhang, D. Hodzic, K.N. Randles, G. Morris, J.U. Earley, M. Hadhazy, J.M. Holaska, S.K. Mewborn, 2009. Disruption of nesprin-1 produces an Emery Dreifuss muscular dystrophy-like phenotype in mice. Hum. Mol. Genet. 18 :607–620. 10.1093/hmg/ddn386 19008300 Qiu, J.J., Y.Y. Lin, J.X. Ding, W.W. Feng, H.Y. Jin, and K.Q. Hua. 2015. Long non-coding RNA ANRIL predicts poor prognosis and promotes invasion/metastasis in serous ovarian cancer. Int. J. Oncol. 46 :2497–2505.25845387 Rajgor, D., and C.M. Shanahan. 2013. Nesprins: from the nuclear envelope and beyond. Expert Rev. Mol. Med. 15 :e5. 10.1017/erm.2013.6 23830188 Randles, K.N., T. Lam, C.A. Sewry, M. Puckelwartz, D. Furling, M. Wehnert, E.M. McNally, and G.E. Morris. 2010. Nesprins, but not sun proteins, switch isoforms at the nuclear envelope during muscle development. Dev. Dyn. 239 :998–1009. 10.1002/dvdy.22229 20108321 Russ, W.P., and D.M. Engelman. 2000. The GxxxG motif: A framework for transmembrane helix-helix association. J. Mol. Biol. 296 :911–919. 10.1006/jmbi.1999.3489 10677291 Schneider, D., and D.M. Engelman. 2004. Motifs of two small residues can assist but are not sufficient to mediate transmembrane helix interactions. J. Mol. Biol. 343 :799–804. 10.1016/j.jmb.2004.08.083 15476801 Smith, E.R., X.Y. Zhang, C.D. Capo-Chichi, X. Chen, and X.X. Xu. 2011. Increased expression of Syne1/nesprin-1 facilitates nuclear envelope structure changes in embryonic stem cell differentiation. Dev. Dyn. 240 :2245–2255. 10.1002/dvdy.22717 21932307 Sosa, B.A., A. Rothballer, U. Kutay, and T.U. Schwartz. 2012. LINC complexes form by binding of three KASH peptides to domain interfaces of trimeric SUN proteins. Cell. 149 :1035–1047. 10.1016/j.cell.2012.03.046 22632968 Sosa, B.A., U. Kutay, and T.U. Schwartz. 2013. Structural insights into LINC complexes. Curr. Opin. Struct. Biol. 23 :285–291. 10.1016/j.sbi.2013.03.005 23597672 Starr, D.A. 2009. A nuclear-envelope bridge positions nuclei and moves chromosomes. J. Cell Sci. 122 :577–586. 10.1242/jcs.037622 19225124 Starr, D.A., and M. Han. 2002. Role of ANC-1 in tethering nuclei to the actin cytoskeleton. Science. 298 :406–409. 10.1126/science.1075119 12169658 Struhl, G., and K. Basler. 1993. Organizing activity of wingless protein in Drosophila. Cell. 72 :527–540. 10.1016/0092-8674(93)90072-X 8440019 Taranum, S., I. Sur, R. Müller, W. Lu, R.N. Rashmi, M. Munck, S. Neumann, I. Karakesisoglou, and A.A. Noegel. 2012. Cytoskeletal interactions at the nuclear envelope mediated by nesprins. Int. J. Cell Biol. 2012 :736524. 10.1155/2012/736524 22518138 Technau, M., and S. Roth. 2008. The Drosophila KASH domain proteins Msp-300 and Klarsicht and the SUN domain protein Klaroid have no essential function during oogenesis. Fly (Austin). 2 :82–91. 10.4161/fly.6288 18820478 Ungricht, R., M. Klann, P. Horvath, and U. Kutay. 2015. Diffusion and retention are major determinants of protein targeting to the inner nuclear membrane. J. Cell Biol. 209 :687–703. 10.1083/jcb.201409127 26056139 Walters, A.D., A. Bommakanti, and O. Cohen-Fix. 2012. Shaping the nucleus: Factors and forces. J. Cell. Biochem. 113 :2813–2821. 10.1002/jcb.24178 22566057 Wang, J.Y., I.S. Yu, C.C. Huang, C.Y. Chen, W.P. Wang, S.W. Lin, K.T. Jeang, and Y.H. Chi. 2015. Sun1 deficiency leads to cerebellar ataxia in mice. Dis. Model. Mech. 8 :957–967. 10.1242/dmm.019240 26035387 Wilhelmsen, K., M. Ketema, H. Truong, and A. Sonnenberg. 2006. KASH-domain proteins in nuclear migration, anchorage and other processes. J. Cell Sci. 119 :5021–5029. 10.1242/jcs.03295 17158909 Worman, H.J., and E.C. Schirmer. 2015. Nuclear membrane diversity: underlying tissue-specific pathologies in disease? Curr. Opin. Cell Biol. 34 :101–112. 10.1016/j.ceb.2015.06.003 26115475 Xu, T., and G.M. Rubin. 1993. Analysis of genetic mosaics in developing and adult Drosophila tissues. Development. 117 :1223–1237.8404527 Zhang, Q., C. Bethmann, N.F. Worth, J.D. Davies, C. Wasner, A. Feuer, C.D. Ragnauth, Q. Yi, J.A. Mellad, D.T. Warren, 2007. Nesprin-1 and -2 are involved in the pathogenesis of Emery–Dreifuss muscular dystrophy and are critical for nuclear envelope integrity. Hum. Mol. Genet. 16 :2816–2833. 10.1093/hmg/ddm238 17761684

PMC005xxxxxx/PMC5716290.txt

==== Front J Cell Biol J Cell Biol jcb jcb The Journal of Cell Biology 0021-9525 1540-8140 The Rockefeller University Press 28978644 201707172 10.1083/jcb.201707172 Research Articles Article 19 35 34 Segregation in the Golgi complex precedes export of endolysosomal proteins in distinct transport carriers Sorting of endolysosomal proteins in the Golgi Chen Yu * http://orcid.org/0000-0002-0602-210X Gershlick David C. * Park Sang Yoon * http://orcid.org/0000-0002-5673-6370 Bonifacino Juan S. Cell Biology and Neurobiology Branch, Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, MD Correspondence to Juan S. Bonifacino: juan.bonifacino@nih.gov * Y. Chen, D.C. Gershlick, and S.Y. Park contributed equally to this paper. 4 12 2017 216 12 41414151 31 7 2017 18 8 2017 24 8 2017 This is a work of the U.S. Government and is not subject to copyright protection in the United States. Foreign copyrights may apply. 2017 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). Chen et al. present evidence that two sets of newly synthesized endolysosomal proteins segregate in the Golgi complex before their export in two distinct populations of transport carriers, by mechanisms that are respectively dependent or independent of sorting signal–adaptor interactions. Biosynthetic sorting of newly synthesized transmembrane cargos to endosomes and lysosomes is thought to occur at the TGN through recognition of sorting signals in the cytosolic tails of the cargos by adaptor proteins, leading to cargo packaging into coated vesicles destined for the endolysosomal system. Here we present evidence for a different mechanism in which two sets of endolysosomal proteins undergo early segregation to distinct domains of the Golgi complex by virtue of the proteins’ luminal and transmembrane domains. Proteins in one Golgi domain exit into predominantly vesicular carriers by interaction of sorting signals with adaptor proteins, but proteins in the other domain exit into predominantly tubular carriers shared with plasma membrane proteins, independently of signal–adaptor interactions. These findings demonstrate that sorting of endolysosomal proteins begins at an earlier stage and involves mechanisms that partly differ from those described by classical models. National Institute of Child Health and Human Development https://doi.org/10.13039/100000071 National Institutes of Health https://doi.org/10.13039/100000002 ZIA HD001607 ==== Body pmcIntroduction The endolysosomal system of eukaryotic cells comprises an array of membrane-enclosed organelles, including early, late, and recycling endosomes, as well as lysosomes. Transmembrane proteins that reside in these compartments (hereafter referred to as endolysosomal proteins) are synthesized in the ER and subsequently transported through the cis, medial, and trans cisternae of the Golgi stack and the TGN (collectively referred to as the “Golgi complex”; Braulke and Bonifacino, 2009). The proteins are eventually delivered to the endolysosomal system either directly from the Golgi complex (Harter and Mellman, 1992; Waguri et al., 2003; Ang et al., 2004) or indirectly after transport to the plasma membrane and endocytosis (Lippincott-Schwartz and Fambrough, 1986; Braun et al., 1989; Janvier and Bonifacino, 2005). Key determinants of sorting to the endolysosomal system are signals present in the cytosolic domains of the proteins (Bonifacino and Traub, 2003), which are recognized by adaptor proteins (APs) that are components of protein coats (Robinson, 2004). Signal–adaptor interactions promote incorporation of the proteins into coated transport carriers that participate in the delivery of proteins to the endolysosomal system. Despite progress in the characterization of these molecular mechanisms, however, many aspects of endolysosomal protein sorting in the context of the whole cell remain poorly understood. These aspects include the step in the biosynthetic pathway when endolysosomal proteins diverge from plasma membrane proteins, the extent to which specific endolysosomal proteins follow the direct or indirect pathways, the nature of the transport carriers involved in either pathway, the particular signal–adaptor interactions that mediate protein incorporation into these carriers, and the possible existence of other sorting determinants. Addressing these issues in intact, living cells has proved difficult because of limitations in the ability to visualize the transport of newly synthesized endolysosomal proteins with sufficient temporal and spatial resolution and without temperature or drug manipulations that perturb the structure and function of the Golgi complex. In this study, we have taken advantage of recent methodological developments that allow synchronization of protein transport through the biosynthetic pathway, as well as live-cell and superresolution imaging, to examine how endolysosomal proteins are sorted in the Golgi complex. Specifically, we have used the “retention using selective hooks” (RUSH) system (Boncompain et al., 2012) to track the biosynthetic transport of three transmembrane proteins with different steady-state distributions in the endolysosomal system: the cation-dependent mannose-6-phosphate receptor (CD-MPR; localized to the TGN and early/late endosomes), the transferrin receptor (TfR; plasma membrane, early endosomes, and recycling endosomes), and lysosomal-associated membrane protein (LAMP) 1 (late endosomes and lysosomes). Our analyses reveal an unexpected level of complexity in the mechanisms of endolysosomal protein sorting at the Golgi complex. We find that CD-MPR undergoes an early segregation from TfR and LAMP1 in the Golgi complex, well before their export in transport carriers. This segregation is independent of signals in the cytosolic tails but dependent on the transmembrane and luminal domains of the proteins. The CD-MPR subsequently leaves the Golgi in a population of predominantly vesicular transport carriers in a manner dependent on a cytosolic dileucine-based signal that interacts with clathrin-associated GGA adaptors. These carriers do not translocate toward the plasma membrane but directly deliver the CD-MPR to endosomes. The TfR and LAMP1, on the other hand, are exported in a population of predominantly tubular carriers destined for the plasma membrane, independently of cytosolic sorting signals and their cognate adaptors. The sorting signals in TfR and LAMP1 and the clathrin-associated AP-2 complex are, however, required for endocytosis of TfR and LAMP1 as a requisite for their eventual delivery to endosomes and lysosomes, respectively. These findings demonstrate that early segregation of different sets of endolysosomal proteins in the Golgi complex precedes their export in two distinct populations of transport carriers involved in the direct and indirect pathways. Our study also highlights distinct requirements for signal–adaptor interactions in the exit of different endolysosomal proteins from the Golgi complex. Results Newly synthesized endolysosomal proteins exit the Golgi complex in two distinct sets of transport carriers Transport of newly synthesized TfR, LAMP1, and CD-MPR through the biosynthetic pathway was analyzed using the RUSH system (Boncompain et al., 2012). The endolysosomal proteins (i.e., “reporter” proteins) were genetically fused to a streptavidin-binding peptide (SBP) and a fluorescent protein (GFP or mCherry) and coexpressed with streptavidin fused to the ER-retrieval signal KDEL (Munro and Pelham, 1987; i.e., “hook” protein; Fig. 1 a). For simplicity, the reporter proteins will hereafter be referred to as TfR, LAMP1, and CD-MPR, with the implicit understanding that they are modified for use in the RUSH system. As expected, coexpression of the reporter proteins with the hook proteins resulted in their accumulation in the ER (Fig. 1 b, 0 min). Addition of the vitamin biotin competed off the SBP–streptavidin interaction, resulting in synchronous release of the proteins from the ER (Fig. 1 b and Video 1), and their eventual transport to their corresponding locations in the endolysosomal system (Fig. S1 a). Coexpression of combinations of the reporter proteins showed that they all exited the ER in the same transport carriers (Fig. S1 b) and arrived simultaneously in the Golgi complex at 15–25 min after the addition of biotin (Fig. 1, b and c). At 20–35 min, the proteins began to exit the Golgi complex in pleiomorphic transport carriers similar to those previously shown to mediate various post-Golgi transport events (Hirschberg et al., 1998; Polishchuk et al., 2000, 2003, 2006; Puertollano et al., 2003). Interestingly, we noticed the existence of two distinct populations of carriers: predominantly tubular carriers containing both TfR and LAMP1 but not CD-MPR (Fig. 2, a, b, d, and e; and Video 2), and predominantly vesicular carriers containing CD-MPR but not TfR (Fig. 2 c and Video 3; best observed with higher time resolution in Video 4). For simplicity, we refer to these carriers as “tubular” and “vesicular,” respectively, notwithstanding that they display substantial variation in shape and size. Further analyses of the tubular carriers showed that they were enriched in the vesicular stomatitis virus glycoprotein (VSV-G; Fig. 2 f), a marker of transport carriers destined for the plasma membrane (Hirschberg et al., 1998; Polishchuk et al., 2006; Micaroni et al., 2013). In contrast, they were devoid of internalized transferrin (Tf; Fig. S2, a and b), indicating that they were not derived from recycling endosomes. In addition, total internal reflection fluorescence (TIRF) microscopy showed direct fusion of carriers containing TfR (Fig. S3 a and Video 5) but not internalized Tf (Fig. S3 b and Video 6), with the plasma membrane. Together, these observations indicated that TfR/LAMP1 tubular carriers bud from the Golgi and directly fuse to the plasma membrane without passing through recycling endosomes. Figure 1. Structure, localization, and ER exit of RUSH reporter proteins. (a) Schematic representation of streptavidin–KDEL “hook” and TfR, LAMP1, and CD-MPR “reporter” proteins used in the RUSH experiments. FP, fluorescent protein (GFP or mCherry). In all figures, green and red lettering corresponds to constructs tagged with GFP and mCherry, respectively. (b) RUSH imaging series of three reporter cargos, TfR, LAMP1, and CD-MPR, expressed in HeLa cells, from Video 1. Before the addition of biotin (time 0), the three cargos exhibit a typical ER localization. At 21 min after biotin addition, the cargos localize to the Golgi. At later times, they exit the Golgi, reaching their final destination after 60 min. Bars, 5 µm. (c) Kinetics of trafficking of RUSH cargos through the Golgi complex. The normalized intensity of the masked perinuclear region indicated in b was measured across the whole time course and plotted as a function of time. Values are mean ± SEM; n = 12 cells for each cargo. Notice that the three reporter proteins released from the ER are transported into the Golgi complex at about the same time. Figure 2. Endolysosomal proteins are exported from the Golgi complex in two distinct populations of transport carriers. (a–c) HeLa cells coexpressing streptavidin–KDEL with each of the indicated reporter proteins were treated with biotin and imaged live by spinning-disk confocal microscopy. The left panel shows single frames captured at the indicated times after addition of biotin. The right panel shows magnified images of the boxed areas. (d–f) HeLa cells coexpressing streptavidin–KDEL with combinations of the indicated reporter constructs were analyzed as in a–c. The left columns show single frames captured at the indicated times after addition of biotin (from Video 2 in d and Video 3 in e). The right columns show magnified images of the boxed areas. Bars: (low magnification) 5 µm; (high magnification) 1 µm. Early segregation of endolysosomal proteins in the Golgi complex Airyscan superresolution imaging of cells fixed 30 min after the addition of biotin confirmed the presence of TfR and LAMP1 in tubules budding from the Golgi complex (Fig. 3, a and b) and CD-MPR in a distinct population of vesicles (Fig. 3, d and e). Surprisingly, whereas TfR and LAMP1 colocalized throughout the entire Golgi structure (Fig. 3, a and c; Pearson’s coefficient = 0.95, similar to that of the same reporter protein tagged with different fluorescent proteins; Fig. S4, a and b), TfR and CD-MPR were largely segregated to different Golgi domains (Pearson’s coefficient = 0.37; Fig. 3, d and f). Live-cell Airyscan imaging after the addition of biotin showed that TfR and LAMP1 continuously colocalized from their entry into the Golgi complex to their exit in tubular carriers (Fig. 3, g and i; and Video 7). In contrast, TfR and CD-MPR started to segregate shortly after their entry into the Golgi complex, and their segregation increased over time (Fig. 3, h and i; and Video 8). Similar Golgi segregation and budding patterns were observed in a different cell line, U2OS (Fig. S4, c and d). These observations thus revealed that segregation of the CD-MPR from other endolysosomal proteins begins in the Golgi complex, before their incorporation into distinct transport carriers. The sorting receptor sortilin behaved similarly to the CD-MPR (Fig. S4, e and f), demonstrating that it belongs to the same subset of endolysosomal cargos as the CD-MPR. Figure 3. Segregation of endolysosomal proteins in the Golgi complex. (a–f) HeLa cells coexpressing streptavidin–KDEL with combinations of the indicated reporter proteins were fixed 30 min after the addition of biotin and imaged by Airyscan microscopy. (a and d) Golgi complexes of representative cells. (b and e) Magnified views of box 1. (c and f) Magnified views of box 2 and plots of fluorescence intensity along the white dashed lines. (g and h) HeLa cells coexpressing streptavidin–KDEL with combinations of the indicated reporter proteins were imaged live by Airyscan microscopy. The top rows show Golgi complexes from representative cells. The middle rows show magnifications of the boxed region. The bottom row shows plots of fluorescence intensity along the white dashed lines. Bars, 1 µm. (i) Pearson’s coefficients (r) of data sets of which g and h are representative. Values were normalized to 1.0 at the first time point and are represented as mean ± SEM (n = 7 cells for each pair of cargos). ns, not significant; *, P < 0.1; **, P < 0.01; ***, P < 0.001. Times indicated in g–i are normalized to observable initiation of tubule budding, allowing comparative statistics. Exit of LAMP1 and TfR in tubular carriers is independent of signal–adaptor interactions Sorting of TfR and LAMP1 to endosomes and lysosomes, respectively, is dependent on sorting signals fitting the YXXØ motif (where X is any amino acid and Ø a bulky hydrophobic amino acid) in the cytosolic tail of the proteins (Traub and Bonifacino, 2013; Fig. 4 a). In the case of TfR, an additional signal comprising the sequence GDNS (Fig. 4 a) contributes to sorting to the basolateral plasma membrane in polarized epithelial cells (Odorizzi and Trowbridge, 1997). Strikingly, mutation of the YXXØ and GDNS signals in TfR (residues Y20, G31, D32, N33, and S34 to alanines) and the YXXØ signal in LAMP1 (residue Y404 to alanine; Fig. 4 a) did not prevent incorporation of these proteins into the Golgi-derived tubular carriers, although it resulted in subsequent accumulation of the proteins at the plasma membrane (Fig. 4, b and c; and Video 9). YXXØ motifs are recognized by the AP complexes AP-1, AP-2, AP-3, and AP-4 (Fig. 5 a; Traub and Bonifacino, 2013) and the GDNS motif by AP-1 (Gravotta et al., 2012). CRISPR/Cas9 knockout (KO) of subunits of these complexes (Fig. 5 b) had no effect on exit of TfR and LAMP1 from the Golgi complex in tubular carriers (Fig. 5 c). KO of the AP-2 μ2 subunit, however, caused accumulation of TfR and LAMP1 at the plasma membrane (Fig. 5 d and Fig. S5, a–c), as previously shown by siRNA knockdown (Motley et al., 2003; Janvier and Bonifacino, 2005). KO of subunits of the other complexes did not prevent transport of LAMP1 to lysosomes (Fig. 5 d). These experiments indicated that interactions of cytosolic sorting signals with AP complexes are dispensable for export of TfR and LAMP1 from the Golgi complex in tubular carriers, but an interaction with AP-2 is subsequently required for endocytic delivery of the proteins to endosomes and lysosomes. Figure 4. Role of adaptor-binding motifs in export and segregation of endolysosomal proteins at the Golgi complex. (a) Sequences from the cytosolic domains of TfR, LAMP1, and CD-MPR. Motifs that bind to AP complexes in each protein are highlighted in red. Mutations are indicated with blue letters. (b and c) HeLa cells coexpressing streptavidin–KDEL together with TfR and LAMP1 reporter constructs having mutations in AP-binding motifs, namely TfR-Y20A/GDNS31-34AAAA (b) or LAMP1-Y404A (c), were imaged live by spinning-disk confocal microscopy. Images are single frames from Video 9. The times after addition of biotin are indicated. (Right) Magnifications of the boxed regions. Bars: (low magnification) 5 µm; (high magnification) 1 µm. (d and e) HeLa cells coexpressing streptavidin–KDEL together with GGA1–GFP and CD-MPR (d) or CD-MPR-L274A/L275A (e) reporter proteins were fixed 30 min after the addition of biotin and imaged by Airyscan microscopy. Bar, 2 µm. Arrows indicate carriers containing GGA1–GFP. (f) Airyscan microscopy of HeLa cells coexpressing streptavidin–KDEL together with CD-MPR-L274A/L275A and TfR-Y20A/GDNS31-34AAAA reporter proteins 30 min after the addition of biotin. The inset shows magnified images of the boxed regions. Bars, 1 µm. Figure 5. AP complexes are dispensable for cargo sorting into Golgi-derived tubular carriers. (a) Schematic representation of AP-1, AP-2, AP-3, and AP-4. (b) Confirmation of KO by immunoblot analysis of endogenous targets. Notice that AP-2 μ2 KO is not complete. Cells with complete KO of AP-2 μ2 were not found in the screening. WB, Western blotting. (c) Images from spinning-disk, live-cell microscopy of LAMP1 or TfR reporter proteins in AP-KO cell lines at the indicated times after biotin addition. Tubular carriers containing LAMP1 or TfR reporters were found in all of the AP-KO cells. Bars: 10 µm; (insets) 1 μm. (d) The LAMP1 reporter protein was expressed in each AP-KO cell line. Cells were fixed 60 min after the addition of biotin and stained for an endogenous lysosomal marker (LAMTOR4) to assess the requirement of AP complexes for transport to lysosomes. Bars: 5 µm; (insets) 1 µm. Exit of CD-MPR in vesicular carriers depends on interaction of sorting signals with GGA proteins In contrast to TfR and LAMP1, CD-MPR has been shown to exit the Golgi by virtue of the interaction of a DXXLL motif in the cytosolic tail of the receptor (Fig. 4 a) with the monomeric clathrin adaptors GGA1, GGA2, and GGA3 (Puertollano et al., 2001, 2003; Zhu et al., 2001). Indeed, we observed that CD-MPR-containing vesicles budding from the Golgi were decorated with GGA1 (Fig. 4 d). Moreover, mutation of the DXXLL signal (residues L274 and L275 to alanine; Fig. 4 a) prevented exit of CD-MPR in the GGA1-coated carriers (Fig. 4 e). Hence, unlike Golgi export of TfR and LAMP1 in tubular carriers, export of CD-MPR in vesicular carriers depends on a specific signal–adaptor interaction. It is worth noting that the DXXLL mutant of the CD-MPR was not diverted to tubules by default but was retained in the Golgi complex. We also observed that mutation of the cytosolic sorting signals in both the TfR and CD-MPR did not prevent their segregation into different Golgi subdomains (Fig. 4 f), indicating that this phenomenon is independent of interactions with APs. Transmembrane and luminal domains determine segregation of endolysosomal proteins in the Golgi complex What then are the determinants of segregation in the Golgi complex and exit into tubules? To address this question, we constructed chimeric proteins having different combinations of the luminal, transmembrane, and cytosolic domains of LAMP1 and CD-MPR (both type I transmembrane proteins; Fig. 6 a) and compared their transport with that of the TfR (a type II transmembrane protein) using the RUSH system (Fig. 6, b–h). We observed that LAMP1 chimeras having the transmembrane and/or luminal domains of the CD-MPR (termed MML, LML, and MLL) were not incorporated into Golgi-derived tubules (Fig. 6 b) and were segregated from the TfR in the Golgi complex (Fig. 6, e, g, and h). In contrast, a chimera having the transmembrane and luminal domains of LAMP1 and the cytosolic tail of CD-MPR (LLM) was exported into tubules (Fig. 6 b) and colocalized with TfR in the Golgi complex (Fig. 6 f). These results indicated that transmembrane and/or luminal domains determine protein segregation within the Golgi complex that precedes exit into distinct transport carriers. Figure 6. The luminal and transmembrane domains of endolysosomal proteins determine their intra-Golgi segregation. (a) Schematic representation of chimeric proteins generated by swapping luminal, transmembrane, and cytosolic domains from LAMP1 (L) and CD-MPR (M). The chimeras were fused to a fluorescent protein and SBP for use as reporter proteins in the RUSH system (Fig. 1 a). (b) HeLa cells coexpressing streptavidin–KDEL together with the indicated chimeras and TfR as reporter proteins were fixed 30 min after the addition of biotin and analyzed by Airyscan microscopy. Bars, 1 µm. Images show the presence or absence of the chimeras in tubular carriers emanating from the Golgi complex. (c–h) Cells in b were similarly imaged for the distribution of the chimeras in the Golgi complex. r, Pearson’s coefficient. Magnified images and plots of fluorescence intensity along the whited dashed lines are shown at right. Bars, 1 µm. Discussion Cargo segregation in the early Golgi complex The TGN has been classically regarded as the Golgi subcompartment where newly synthesized proteins destined for secretory vesicles, the plasma membrane, endosomes, and lysosomes are sorted into distinct populations of transport carriers (Griffiths and Simons, 1986). In this classical view, all cargo proteins remain mixed throughout the Golgi complex until they are packaged into their corresponding transport carriers. Several findings, however, challenge the notion that all cargo sorting in the Golgi complex occurs at the TGN. First, transmembrane cargos were shown to partition from Golgi enzymes during their transport through the Golgi cisternae (Patterson et al., 2008). In addition, the contents of different types of secretory granule were found to segregate from one another as early as in the cis-Golgi cisternae (Clermont et al., 1992). Other studies showed that proteoglycans aimed for the apical or basolateral surfaces of polarized epithelial cells acquire different carbohydrate modifications in the Golgi complex (Tveit et al., 2005; Vuong et al., 2006), and high-mannose forms of basolateral and apical proteins exhibit different detergent solubility, all suggestive of segregation in the early Golgi complex (Alfalah et al., 2005). Using superresolution and live-cell imaging, we now provide direct evidence that two sets of transmembrane proteins destined for the endolysosomal system undergo progressive segregation into distinct Golgi domains before their export into distinct transport carriers. These findings indicate that cargo sorting can occur early in the Golgi complex, even for proteins that are targeted to the same organellar system. The Golgi domains to which proteins are segregated could correspond to the center or the rims of the same cisternae or to different cisternae. They could also be discrete ministacks laterally connected as part of a larger Golgi ribbon (Yano et al., 2005; Puthenveedu et al., 2006). Ultrastructural methods in combination with the RUSH system will be required to determine the exact identity of these domains. Early segregation determines export from different Golgi sites Early segregation likely determines the sites of cargo export from the Golgi complex. For proteins that require interaction with clathrin adaptors (e.g., CD-MPR, sortilin; Nielsen et al., 2001; Puertollano et al., 2001; Zhu et al., 2001), the early segregation domain must be connected to the TGN, which is where clathrin coats are located in the Golgi complex (Klumperman et al., 1993). Proteins that exit independently of clathrin adaptors (e.g., TfR, LAMP1, and plasma membrane proteins; Pols et al., 2013), on the other hand, could be exported from a nonclathrin TGN domain or an earlier cisterna. In support of this latter possibility, electron tomography studies showed the presence of a still unidentified nonclathrin, “lace-like” coat at the rims of Golgi cisternae proximal to, but distinct from, the TGN (Ladinsky et al., 1994). These sites were proposed to mediate protein export to the plasma membrane (Ladinsky et al., 1994), although evidence for this function remains to be obtained. Furthermore, fluorescence microscopy of nocodazole-fragmented Golgi ministacks suggested that plasma membrane–directed cargos such as VSV-G exit from the Golgi stack before reaching the TGN (Tie et al., 2016). Thus, early cargo segregation in the Golgi stack likely determines export from different Golgi subcompartments. Export of endolysosomal proteins in distinct transport carriers Our studies also show that endolysosomal proteins leave the Golgi complex in two types of transport carrier. Both types have variable sizes and shapes, although they tend to be predominantly tubular or vesicular in appearance. The characteristics of these carriers likely reflect the Golgi compartments from which they arise. The carriers containing TfR and LAMP1 probably correspond to the VSV-G Golgi–to–plasma membrane carriers previously characterized by correlative light-electron microscopy (Polishchuk et al., 2000). These carriers were found to be large (up to 1.7 µm in diameter), tubular-saccular, and devoid of protein coats (Polishchuk et al., 2000), suggestive of their origin in noncoated areas of the Golgi complex. As in our study, they were shown to fuse with the plasma membrane en bloc, without intersecting any other compartment along the way (Hirschberg et al., 1998; Polishchuk et al., 2000). Other immunoelectron studies also demonstrated export of LAMP1 from the TGN in noncoated carriers devoid of cation-independent mannose-6-phosphate receptor and AP-1 but containing VPS41 and VAMP7 (Pols et al., 2013). The latter carriers may correspond to a population that follows the direct pathway and that are less than the level of detection in our assays. Correlative light-electron microscopy of CD-MPR- and GGA1-containing carriers also showed them to be large, convoluted tubular-vesicular structures. However, they have associated clathrin-coated profiles, indicative of their origin at the TGN (Polishchuk et al., 2006). Previous studies showed that, unlike the VSV-G/TfR/LAMP-1 carriers, the CD-MPR/GGA1 carriers did not translocate toward the plasma membrane but merged with endosomes (Puertollano et al., 2003; Waguri et al., 2003; Polishchuk et al., 2006). These properties of the TfR/LAMP1 and CD-MPR carriers are consistent with their being the mediators of transport in the indirect and direct pathways, respectively, to the endolysosomal system. The use of the direct pathway by the CD-MPR fits in with its role as an intracellular sorting receptor for lysosomal hydrolase precursors. Molecular determinants of early Golgi segregation and Golgi export Molecular dissection of the endolysosomal proteins used in our study revealed that they have different types of sorting determinant. The initial segregation in the Golgi stack is independent of sorting signals in the cytosolic tails but dependent on the transmembrane and/or luminal domains of the proteins. Transmembrane domains could mediate partitioning into specific lipid domains or interactions with other transmembrane proteins, as shown for other sorting events (Nishikawa and Nakano, 1993; Emery et al., 2003; Alfalah et al., 2005; Patterson et al., 2008; Kaiser et al., 2011). Luminal domains could segregate proteins by promoting oligomerization or aggregation in the special environment of the Golgi complex, as also shown in other settings (Compton et al., 1989; Dintzis et al., 1994; Colomer et al., 1996; Wolins et al., 1997; Paladino et al., 2004). In this regard, it is worth noting that some constitutive secretory cargos, such as the cartilage oligomeric matrix protein and lysozyme C, bind in a Ca2+-dependent manner to the Golgi protein Cab45, which facilitates their export into a specific population of secretory carriers (Crevenna et al., 2016). It is conceivable that a similar mechanism could operate for segregation of transmembrane proteins through their luminal domains. The subsequent packaging of proteins into Golgi export carriers has long been thought to depend on interactions of cytosolic sorting signals with TGN-associated adaptors such as the GGAs, AP-1, and AP-4. However, for the proteins examined in our study, this appears to be true only for the CD-MPR and sortilin, which require a GGA-binding signal for exit into vesicular carriers. In contrast, exit of TfR and LAMP1 in tubular carriers is independent of sorting signals and of the AP-1, AP-2, AP-3, and AP-4 adaptors. This is in line with the carriers’ being the same that transport plasma membrane proteins. AP-2 is subsequently required for endocytosis of TfR and LAMP1 from the plasma membrane, as previously shown by RNAi studies (Motley et al., 2003; Janvier and Bonifacino, 2005). The fact that AP-2 is the only YXXØ-interacting adaptor required for sorting of TfR and LAMP1 to endosomes and lysosomes demonstrates the critical role of endocytosis in this process. Collectively, these observations lend further support to the notion that CD-MPR follows mainly the direct pathway, and TfR and LAMP1 the indirect pathway, for transport to endosomes and lysosomes. Caveats in the interpretation of our findings Although the use of the RUSH system has allowed us to track the biosynthetic transport of endolysosomal transmembrane proteins in unprecedented detail, there are several caveats in the interpretation of our experiments. The most important one is that the expression levels of the reporter proteins are likely higher than those of their endogenous counterparts. The mechanics of RUSH could generate a wave of newly synthesized reporter proteins moving through the secretory pathway, potentially creating abnormal structures or altering the properties of the organelles along the way. Overexpression of the reporter proteins could also saturate sorting dependent on signals and adaptors (Marks et al., 1996). To avoid these problems, in our study we imaged cells expressing moderate levels of the reporter proteins: high enough for detection of transport intermediates but not so high that they changed the appearance of the organelles. In this regard, electron microscopy of cells expressing reporter proteins 25 min after their release from the ER showed normal appearance of the Golgi complex (unpublished data). In addition, saturation of sorting mechanisms would have been expected to homogenize the distribution of different reporters among Golgi domains, transport carriers, and destination organelles, but this was clearly not the case in our studies. These considerations notwithstanding, we cannot rule out the existence of alternative processes, such as populations of CD-MPR following the indirect pathway and TfR and LAMP1 following the direct pathway to some extent. Hypothetical model for sorting of endolysosomal proteins in the Golgi complex Our results suggest a two-step process for the sorting of endolysosomal proteins in the Golgi complex (Fig. 7). In the first step, two sets of proteins become segregated to different domains of the Golgi stack by virtue of transmembrane and/or luminal domains. In the second step, proteins segregated to one domain (e.g., CD-MPR) exit the Golgi complex in vesicular carriers bound for the endolysosomal system, in a process that is dependent on recognition of cytosolic sorting signals by clathrin adaptors (i.e., the GGAs). Proteins in the other domain, in contrast, leave the Golgi complex in tubular carriers directed to the plasma membrane, independently of sorting signals and clathrin adaptors. This model differs from the classical model in that different sets of endolysosomal proteins are presorted in the early Golgi and that some of those proteins leave the Golgi complex independently of signal–adaptor interactions. Figure 7. Model depicting the sorting of endolysosomal proteins in the Golgi complex. Endolysosomal proteins are delivered from the ER to the Golgi complex in the same transport carriers. Once in the Golgi complex, sets of endolysosomal proteins segregate to distinct domains. One domain gives rise to tubular carriers in which endolysosomal and plasma membrane proteins leave the Golgi independently of cytosolic sorting signals and AP complexes. The other domain is the source of vesicular carriers into which endolysosomal proteins are sorted through interaction of cytosolic sorting signals with GGA proteins. Materials and methods Recombinant DNAs Plasmid constructs to synchronize the traffic of the TfR, LAMP1, CD-MPR, VSV-G, sortilin, and LAMP1–CD-MPR chimeric reporter proteins (Figs. 1 a, 4 a, and 6 a) through the secretory pathway were generated by replacing sequences encoding the reporter proteins in the original bicistronic RUSH constructs (gift of F. Perez and G. Boncompain, Curie Institute, Paris, France; Boncompain et al., 2012) in which streptavidin KDEL was used as the hook. The type I transmembrane proteins LAMP1 (UniProt accession no. P14562), CD-MPR (UniProt accession no. P20645), sortilin (UniProt accession no. Q99523), and VSV-G (UniProt accession no. P04882) were modified by insertion of the SBP and a fluorescent protein (EGFP or mCherry) at their luminal N termini, immediately after the signal peptide. The type II transmembrane protein TfR (UniProt accession no. P02786) was tagged with SBP and a fluorescent protein at its luminal C terminus. Tagging in the luminal domain avoided potential interference with cytosolic sorting signals. TfR-Y20A/GDNS31-34AAAA, LAMP1-Y404A, and CD-MPR-L274A/L275A mutants were generated using the QuickChange Site-Directed Mutagenesis kit (Agilent Technologies). Chimeras combining LAMP1 and CD-MPR domains (Fig. 6 a) were generated in the bicistronic RUSH construct using Gibson assembly. LAMP1: luminal, amino acids 22–371; transmembrane, amino acids 372–395; cytosolic, amino acids 396–407. CD-MPR: luminal, amino acids 27–185; transmembrane, amino acids 186–210; cytosolic, amino acids 211–277. EGFP-tagged Rab4a and GGA1 were described previously (Puertollano et al., 2001; Chen et al., 2012). Antibodies Mouse monoclonal antibodies to AP-1 γ (1:5,000 for immunoblotting, anti–Adaptin γ; catalog no. 610385), AP-2 μ2 (1:5,000 for immunoblotting, anti-AP50; catalog no. 611351), AP-3 δ (1:5,000 for immunoblotting, anti–Adaptin δ; catalog no. 611329), AP-4 ε (1:5,000 for immunoblotting, anti–Adaptin ε; catalog no. 612018), and actin (1:5,000 for Western blotting, anti-actin; catalog no. 612657) were purchased from BD Biosciences. HRP-conjugated goat anti–mouse antibody (1:5,000 for immunoblotting; catalog no. sc-2004) was purchased from Santa Cruz Biotechnology. Mouse monoclonal antibody to cation-independent mannose-6-phosphate receptor (1:100 for immunofluorescence microscopy, MEM-238; ab8093) was purchased from Abcam. Rabbit monoclonal antibody to LAMTOR4 (1:1,000 for immunofluorescence microscopy; catalog no. 13140) was purchased from Cell Signaling Technology. Alexa-conjugated secondary antibodies including Alexa Fluor 488–conjugated donkey anti–rabbit antibody (1:1,000 for immunofluorescence microscopy; catalog no. A21206) and Alexa Fluor 555–conjugated donkey anti–mouse antibody (1:1,000 for immunofluorescence microscopy; catalog no. A31570) antibodies were purchased from Invitrogen. Transfection and RUSH For fixed-cell imaging experiments, 40,000 HeLa cells per well were seeded on a 12-well plate containing 18-mm cover glasses (Marienfeld) coated with fibronectin 1 d before transfection. For live-cell imaging experiments, 40,000 HeLa cells per well were seeded on 2-well Nunc Lab-Tek chambers coated with fibronectin 1 d before transfection. Cells were transfected using FuGENE 6 (E2691; Promega). 6 μl FuGENE transfection reagent was diluted into 80 µl Opti-MEM (31985-070; GIBCO BRL), and, separately, 2 µg DNA was diluted into 20 µl Opti-MEM. For cotransfections, DNA plasmids were combined in an equimolar ratio. After 5 min, the DNA and transfection solutions were mixed and incubated for 20 min before being added to the cells. 20 h after transfection with the indicated plasmids, cells were imaged in 37°C prewarmed phenol red–free medium (20163-029; GIBCO BRL), supplemented with 25 mM Hepes. D-biotin (Sigma-Aldrich) at a final concentration of 40 µM was added to the chamber at time 0. Fluorescence microscopy Immunofluorescence microscopy was performed as previously described (Schindler et al., 2015). In brief, HeLa cells were fixed for 30 min at RT in 4% PFA, 4% sucrose, 0.1 mM CaCl2, and 1 mM MgCl2 in PBS. Cells were then permeabilized for 10 min at RT with 0.2% saponin in PBS, followed by incubation with the indicated antibodies. Live-cell imaging was conducted with an Eclipse Ti Microscope System (Nikon) equipped with an environmental chamber (temperature controlled at 37°C and CO2 at 5%) and NIS-Elements AR microscope imaging software. Spinning-disk confocal images were taken with a Plan Apo VC 60× objective (NA 1.40) and a high-speed electron-multiplying charge-coupled device camera (Evolve 512; Photometrics) mounted on the left portal. TIRF microscopic images were taken with an Apo TIRF 100× Oil DIC N2 objective (NA 1.49) and an electron-multiplying charge-coupled device camera (DU-897; Andor) mounted on the right portal. TIRF position was calibrated for each imaging experiment, and the focus was maintained using a Perfect Focus system. Dual-color imaging was done by fast switching of the excitation lasers, and images from green and red channels were aligned automatically. For TIRF imaging of Alexa Fluor 488–conjugated Tf, transfected cells were incubated in DMEM without FBS, with 1% BSA, for 30 min before being incubated for 1 h in Alexa 488–Tf (working concentration 25 µg/ml; T13342; Thermo Fisher Scientific) at 37°C. RUSH was performed under the constant presence of Alexa Fluor 488–Tf after addition of biotin (which doubled the volume of media, effectively halving the concentration of Tf for the duration of the imaging). Superresolution microscopic images were taken using an LSM 880 microscope with Airyscan (Zeiss) and a Plan Apochromat 63× objective (NA 1.40) with the settings recommended by the manufacturer. CRISPR/Cas9 KO cells HeLa-KO cell lines were generated using the CRISPR/Cas9 system (Ran et al., 2013). Target gRNA sequences (AP-1 γ1: 5′-TACATACCGATGTCGGAATG-3′; AP-2 μ2: 5′-CGATGTCATCTCGGTAGACT-3′; AP-3 δ: 5′-CCTTGTGGTTACGGATGCCG-3′; AP-4 ε: 5′-GCAATCAAGTTAGCCCAACA-3′) were cloned into the px330 CRISPR/Cas9 vector using the restriction enzyme BbsI. CRISPR/Cas9 constructs were transfected into HeLa cells, and cell lines derived from single colonies were validated by immunoblotting to confirm the loss of the target proteins. For AP-2 μ2, small amounts of the target protein were found in the validation screening, perhaps because of the lethality of complete AP-2 μ2 KO. Rapid accumulation of TfR–SBP–mCherry and SBP–mCherry–LAMP1 on the surface of AP-2 μ2 KO cells was considered confirmation of effective abrogation of AP-2 function (Fig. 5 d and Fig. S5 a). Flow cytometry HeLa cells were plated onto six-well plates and transfected with plasmids encoding SBP–GFP–CD-MPR or SBP–GFP–LAMP1 using FuGENE 6. 20 h after transfection, cells were incubated with biotin for 0 and 60 min. All further manipulations were on ice or at 4°C. Cells were detached from the plate by incubating with 10 mM EDTA in PBS for 20 min, pipetting up and down every 5 min. Cells were transferred to 1.5-ml microcentrifuge tubes followed by fixation in PBS containing 4% PFA for 20 min. Cells were washed four times by repeated centrifugation (4°C, 500 g, 5 min) in PBS to remove residual PFA. Fixed cells were stained with anti-GFP conjugated with Alexa Fluor 647 (565197; BD Biosciences) at a concentration of 2 µg/ml in PBS containing 3% BSA. Cells were filtered using Cell-Strainer-capped 5-ml round-bottom tubes (352235; Corning). A minimum of 50,000 cells per sample was analyzed using an LSRFortessa cell analyzer (BD Biosciences), gating for GFP-positive cells indicative of expression of the transgene. Data were analyzed using FlowJo software. Surface expression of protein was deduced by relative intensity of Alexa Fluor 647. Quantitative and statistical analyses All numerical results are reported as the mean ± SEM and represent data from a minimum of three independent experiments. Line plots were performed in ImageJ. For Fig. 3 (g–i), the first observable robust tubulation from the Golgi was considered time 0 to normalize time points. For live-cell imaging, Imaris was used to calculate the Pearson’s correlation of the voxels from the whole image in a z-stack. Images were thresholded at 0.05% of total intensity to reduce background. For each sample, at least seven cells were analyzed per sample. Data were normalized to time point minus 15 min in each data set. A two-tailed Student t test for unpaired data were used to evaluate single comparisons between different experimental groups using Microsoft Excel. For the kinetic analysis of RUSH, SBP–mCherry–LAMP1, SBP–GFP–CD-MPR, or TfR–SBP–mCherry was transfected into HeLa cells 1 d before the experiment. The images were taken by spinning-disk microscopy for 60 min with 3-min time intervals. Only cells with a total intensity at time 0 between 0.5 × 107 and 107 arbitrary units were selected to be neither overexpressing nor too affected by bleaching across the 60 min. Because of random lateral movement of the microscopy stage, some time course data sets were stabilized with the image stabilizer plugin for ImageJ (http://www.cs.cmu.edu/~kangli/code/Image_Stabilizer.html). The Golgi was masked and intensity measured across the whole time course in that region. The data set from each cell was normalized so the highest value was equal to 1. Online supplemental material Fig. S1 and Video 1 show synchronized transport and eventual destinations of RUSH cargos. Videos 2, 3, 4, 7, and 8 show distinct carriers derived from the Golgi complex for RUSH cargos, and Video 9 shows those for TfR mutant. Figs. S2 and S3 and Videos 5 and 6 show direct fusion of TfR-containing carriers with the plasma membrane. Fig. S4 shows control experiments for cargo segregation in the Golgi complex. Fig. S5 shows different routes for cargos to reach endosomes and lysosomes. Supplementary Material Supplemental Materials (PDF) Video 1 Video 2 Video 3 Video 4 Video 5 Video 6 Video 7 Video 8 Video 9 Acknowledgments We thank Frank Perez and Gaelle Boncompain for kindly providing the original RUSH constructs and Michal Jarnik for help with electron microscopy. This work was funded by the Intramural Program of the National Institute of Child Health and Human Development, National Institutes of Health (grant ZIA HD001607). The authors declare no competing financial interests. Author contributions: Y. Chen and J.S. Bonifacino conceived the project. Y. Chen, D.C. Gershlick, and S.Y. Park performed experiments and analyzed the data. All authors participated in the preparation of the figures and the writing of the manuscript. Abbreviations used: AP adaptor protein CD-MPR cation-dependent mannose-6-phosphate receptor KO knockout LAMP lysosomal-associated membrane protein RUSH retention using selective hooks SBP streptavidin-binding peptide Tf transferrin TfR Tf receptor TIRF total internal reflection fluorescence VSV-G vesicular stomatitis virus glycoprotein ==== Refs Alfalah, M., G. Wetzel, I. Fischer, R. Busche, E.E. Sterchi, K.P. Zimmer, H.P. Sallmann, and H.Y. Naim. 2005. A novel type of detergent-resistant membranes may contribute to an early protein sorting event in epithelial cells. J. Biol. Chem. 280 :42636–42643. 10.1074/jbc.M505924200 16230359 Ang, A.L., T. Taguchi, S. Francis, H. Fölsch, L.J. Murrells, M. Pypaert, G. Warren, and I. Mellman. 2004. Recycling endosomes can serve as intermediates during transport from the Golgi to the plasma membrane of MDCK cells. J. Cell Biol. 167 :531–543. 10.1083/jcb.200408165 15534004 Boncompain, G., S. Divoux, N. Gareil, H. de Forges, A. Lescure, L. Latreche, V. Mercanti, F. Jollivet, G. Raposo, and F. Perez. 2012. Synchronization of secretory protein traffic in populations of cells. Nat. Methods. 9 :493–498. 10.1038/nmeth.1928 22406856 Bonifacino, J.S., and L.M. Traub. 2003. Signals for sorting of transmembrane proteins to endosomes and lysosomes. Annu. Rev. Biochem. 72 :395–447. 10.1146/annurev.biochem.72.121801.161800 12651740 Braulke, T., and J.S. Bonifacino. 2009. Sorting of lysosomal proteins. Biochim. Biophys. Acta. 1793 :605–614. 10.1016/j.bbamcr.2008.10.016 19046998 Braun, M., A. Waheed, and K. von Figura. 1989. Lysosomal acid phosphatase is transported to lysosomes via the cell surface. EMBO J. 8 :3633–3640.2583113 Chen, Y., Y. Wang, J. Zhang, Y. Deng, L. Jiang, E. Song, X.S. Wu, J.A. Hammer, T. Xu, and J. Lippincott-Schwartz. 2012. Rab10 and myosin-Va mediate insulin-stimulated GLUT4 storage vesicle translocation in adipocytes. J. Cell Biol. 198 :545–560. 10.1083/jcb.201111091 22908308 Clermont, Y., A. Rambourg, and L. Hermo. 1992. Segregation of secretory material in all elements of the Golgi apparatus in principal epithelial cells of the rat seminal vesicle. Anat. Rec. 232 :349–358. 10.1002/ar.1092320304 1543259 Colomer, V., G.A. Kicska, and M.J. Rindler. 1996. Secretory granule content proteins and the luminal domains of granule membrane proteins aggregate in vitro at mildly acidic pH. J. Biol. Chem. 271 :48–55. 10.1074/jbc.271.1.48 8550606 Compton, T., I.E. Ivanov, T. Gottlieb, M. Rindler, M. Adesnik, and D.D. Sabatini. 1989. A sorting signal for the basolateral delivery of the vesicular stomatitis virus (VSV) G protein lies in its luminal domain: analysis of the targeting of VSV G-influenza hemagglutinin chimeras. Proc. Natl. Acad. Sci. USA. 86 :4112–4116. 10.1073/pnas.86.11.4112 2542964 Crevenna, A.H., B. Blank, A. Maiser, D. Emin, J. Prescher, G. Beck, C. Kienzle, K. Bartnik, B. Habermann, M. Pakdel, 2016. Secretory cargo sorting by Ca2+-dependent Cab45 oligomerization at the trans-Golgi network. J. Cell Biol. 213 :305–314. 10.1083/jcb.201601089 27138253 Dintzis, S.M., V.E. Velculescu, and S.R. Pfeffer. 1994. Receptor extracellular domains may contain trafficking information. Studies of the 300-kDa mannose 6-phosphate receptor. J. Biol. Chem. 269 :12159–12166.8163521 Emery, G., R.G. Parton, M. Rojo, and J. Gruenberg. 2003. The trans-membrane protein p25 forms highly specialized domains that regulate membrane composition and dynamics. J. Cell Sci. 116 :4821–4832. 10.1242/jcs.00802 14600267 Gravotta, D., J.M. Carvajal-Gonzalez, R. Mattera, S. Deborde, J.R. Banfelder, J.S. Bonifacino, and E. Rodriguez-Boulan. 2012. The clathrin adaptor AP-1A mediates basolateral polarity. Dev. Cell. 22 :811–823. 10.1016/j.devcel.2012.02.004 22516199 Griffiths, G., and K. Simons. 1986. The trans Golgi network: sorting at the exit site of the Golgi complex. Science. 234 :438–443. 10.1126/science.2945253 2945253 Harter, C., and I. Mellman. 1992. Transport of the lysosomal membrane glycoprotein lgp120 (lgp-A) to lysosomes does not require appearance on the plasma membrane. J. Cell Biol. 117 :311–325. 10.1083/jcb.117.2.311 1560028 Hirschberg, K., C.M. Miller, J. Ellenberg, J.F. Presley, E.D. Siggia, R.D. Phair, and J. Lippincott-Schwartz. 1998. Kinetic analysis of secretory protein traffic and characterization of Golgi to plasma membrane transport intermediates in living cells. J. Cell Biol. 143 :1485–1503. 10.1083/jcb.143.6.1485 9852146 Janvier, K., and J.S. Bonifacino. 2005. Role of the endocytic machinery in the sorting of lysosome-associated membrane proteins. Mol. Biol. Cell. 16 :4231–4242. 10.1091/mbc.E05-03-0213 15987739 Kaiser, H.J., A. Orłowski, T. Róg, T.K. Nyholm, W. Chai, T. Feizi, D. Lingwood, I. Vattulainen, and K. Simons. 2011. Lateral sorting in model membranes by cholesterol-mediated hydrophobic matching. Proc. Natl. Acad. Sci. USA. 108 :16628–16633. 10.1073/pnas.1103742108 21930944 Klumperman, J., A. Hille, T. Veenendaal, V. Oorschot, W. Stoorvogel, K. von Figura, and H.J. Geuze. 1993. Differences in the endosomal distributions of the two mannose 6-phosphate receptors. J. Cell Biol. 121 :997–1010. 10.1083/jcb.121.5.997 8099077 Ladinsky, M.S., J.R. Kremer, P.S. Furcinitti, J.R. McIntosh, and K.E. Howell. 1994. HVEM tomography of the trans-Golgi network: structural insights and identification of a lace-like vesicle coat. J. Cell Biol. 127 :29–38. 10.1083/jcb.127.1.29 7929568 Lippincott-Schwartz, J., and D.M. Fambrough. 1986. Lysosomal membrane dynamics: Structure and interorganellar movement of a major lysosomal membrane glycoprotein. J. Cell Biol. 102 :1593–1605. 10.1083/jcb.102.5.1593 2871029 Marks, M.S., L. Woodruff, H. Ohno, and J.S. Bonifacino. 1996. Protein targeting by tyrosine- and di-leucine-based signals: Evidence for distinct saturable components. J. Cell Biol. 135 :341–354. 10.1083/jcb.135.2.341 8896593 Micaroni, M., A.C. Stanley, T. Khromykh, J. Venturato, C.X. Wong, J.P. Lim, B.J. Marsh, B. Storrie, P.A. Gleeson, and J.L. Stow. 2013. Rab6a/a′ are important Golgi regulators of pro-inflammatory TNF secretion in macrophages. PLoS One. 8 :e57034. 10.1371/journal.pone.0057034 23437303 Motley, A., N.A. Bright, M.N. Seaman, and M.S. Robinson. 2003. Clathrin-mediated endocytosis in AP-2-depleted cells. J. Cell Biol. 162 :909–918. 10.1083/jcb.200305145 12952941 Munro, S., and H.R. Pelham. 1987. A C-terminal signal prevents secretion of luminal ER proteins. Cell. 48 :899–907. 10.1016/0092-8674(87)90086-9 3545499 Nielsen, M.S., P. Madsen, E.I. Christensen, A. Nykjaer, J. Gliemann, D. Kasper, R. Pohlmann, and C.M. Petersen. 2001. The sortilin cytoplasmic tail conveys Golgi-endosome transport and binds the VHS domain of the GGA2 sorting protein. EMBO J. 20 :2180–2190. 10.1093/emboj/20.9.2180 11331584 Nishikawa, S., and A. Nakano. 1993. Identification of a gene required for membrane protein retention in the early secretory pathway. Proc. Natl. Acad. Sci. USA. 90 :8179–8183. 10.1073/pnas.90.17.8179 8367481 Odorizzi, G., and I.S. Trowbridge. 1997. Structural requirements for basolateral sorting of the human transferrin receptor in the biosynthetic and endocytic pathways of Madin-Darby canine kidney cells. J. Cell Biol. 137 :1255–1264. 10.1083/jcb.137.6.1255 9182660 Paladino, S., D. Sarnataro, R. Pillich, S. Tivodar, L. Nitsch, and C. Zurzolo. 2004. Protein oligomerization modulates raft partitioning and apical sorting of GPI-anchored proteins. J. Cell Biol. 167 :699–709. 10.1083/jcb.200407094 15557121 Patterson, G.H., K. Hirschberg, R.S. Polishchuk, D. Gerlich, R.D. Phair, and J. Lippincott-Schwartz. 2008. Transport through the Golgi apparatus by rapid partitioning within a two-phase membrane system. Cell. 133 :1055–1067. 10.1016/j.cell.2008.04.044 18555781 Polishchuk, E.V., A. Di Pentima, A. Luini, and R.S. Polishchuk. 2003. Mechanism of constitutive export from the golgi: bulk flow via the formation, protrusion, and en bloc cleavage of large trans-Golgi network tubular domains. Mol. Biol. Cell. 14 :4470–4485. 10.1091/mbc.E03-01-0033 12937271 Polishchuk, R.S., E.V. Polishchuk, P. Marra, S. Alberti, R. Buccione, A. Luini, and A.A. Mironov. 2000. Correlative light-electron microscopy reveals the tubular-saccular ultrastructure of carriers operating between Golgi apparatus and plasma membrane. J. Cell Biol. 148 :45–58. 10.1083/jcb.148.1.45 10629217 Polishchuk, R.S., E. San Pietro, A. Di Pentima, S. Teté, and J.S. Bonifacino. 2006. Ultrastructure of long-range transport carriers moving from the trans Golgi network to peripheral endosomes. Traffic. 7 :1092–1103. 10.1111/j.1600-0854.2006.00453.x 16787435 Pols, M.S., E. van Meel, V. Oorschot, C. ten Brink, M. Fukuda, M.G. Swetha, S. Mayor, and J. Klumperman. 2013. hVps41 and VAMP7 function in direct TGN to late endosome transport of lysosomal membrane proteins. Nat. Commun. 4 :1361. 10.1038/ncomms2360 23322049 Puertollano, R., R.C. Aguilar, I. Gorshkova, R.J. Crouch, and J.S. Bonifacino. 2001. Sorting of mannose 6-phosphate receptors mediated by the GGAs. Science. 292 :1712–1716. 10.1126/science.1060750 11387475 Puertollano, R., N.N. van der Wel, L.E. Greene, E. Eisenberg, P.J. Peters, and J.S. Bonifacino. 2003. Morphology and dynamics of clathrin/GGA1-coated carriers budding from the trans-Golgi network. Mol. Biol. Cell. 14 :1545–1557. 10.1091/mbc.02-07-0109 12686608 Puthenveedu, M.A., C. Bachert, S. Puri, F. Lanni, and A.D. Linstedt. 2006. GM130 and GRASP65-dependent lateral cisternal fusion allows uniform Golgi-enzyme distribution. Nat. Cell Biol. 8 :238–248. 10.1038/ncb1366 16489344 Ran, F.A., P.D. Hsu, J. Wright, V. Agarwala, D.A. Scott, and F. Zhang. 2013. Genome engineering using the CRISPR-Cas9 system. Nat. Protoc. 8 :2281–2308. 10.1038/nprot.2013.143 24157548 Robinson, M.S. 2004. Adaptable adaptors for coated vesicles. Trends Cell Biol. 14 :167–174. 10.1016/j.tcb.2004.02.002 15066634 Schindler, C., Y. Chen, J. Pu, X. Guo, and J.S. Bonifacino. 2015. EARP is a multisubunit tethering complex involved in endocytic recycling. Nat. Cell Biol. 17 :639–650. 10.1038/ncb3129 25799061 Tie, H.C., D. Mahajan, B. Chen, L. Cheng, A.M. VanDongen, and L. Lu. 2016. A novel imaging method for quantitative Golgi localization reveals differential intra-Golgi trafficking of secretory cargoes. Mol. Biol. Cell. 27 :848–861. 10.1091/mbc.E15-09-0664 26764092 Traub, L.M., and J.S. Bonifacino. 2013. Cargo recognition in clathrin-mediated endocytosis. Cold Spring Harb. Perspect. Biol. 5 :a016790. 10.1101/cshperspect.a016790 24186068 Tveit, H., G. Dick, V. Skibeli, and K. Prydz. 2005. A proteoglycan undergoes different modifications en route to the apical and basolateral surfaces of Madin-Darby canine kidney cells. J. Biol. Chem. 280 :29596–29603. 10.1074/jbc.M503691200 15980070 Vuong, T.T., K. Prydz, and H. Tveit. 2006. Differences in the apical and basolateral pathways for glycosaminoglycan biosynthesis in Madin-Darby canine kidney cells. Glycobiology. 16 :326–332. 10.1093/glycob/cwj075 16394120 Waguri, S., F. Dewitte, R. Le Borgne, Y. Rouillé, Y. Uchiyama, J.F. Dubremetz, and B. Hoflack. 2003. Visualization of TGN to endosome trafficking through fluorescently labeled MPR and AP-1 in living cells. Mol. Biol. Cell. 14 :142–155. 10.1091/mbc.E02-06-0338 12529433 Wolins, N., H. Bosshart, H. Küster, and J.S. Bonifacino. 1997. Aggregation as a determinant of protein fate in post-Golgi compartments: Role of the luminal domain of furin in lysosomal targeting. J. Cell Biol. 139 :1735–1745. 10.1083/jcb.139.7.1735 9412468 Yano, H., M. Yamamoto-Hino, M. Abe, R. Kuwahara, S. Haraguchi, I. Kusaka, W. Awano, A. Kinoshita-Toyoda, H. Toyoda, and S. Goto. 2005. Distinct functional units of the Golgi complex in Drosophila cells. Proc. Natl. Acad. Sci. USA. 102 :13467–13472. 10.1073/pnas.0506681102 16174741 Zhu, Y., B. Doray, A. Poussu, V.P. Lehto, and S. Kornfeld. 2001. Binding of GGA2 to the lysosomal enzyme sorting motif of the mannose 6-phosphate receptor. Science. 292 :1716–1718. 10.1126/science.1060896 11387476

PMC005xxxxxx/PMC5839718.txt

==== Front J Gen Physiol J Gen Physiol jgp jgp The Journal of General Physiology 0022-1295 1540-7748 Rockefeller University Press 29382698 201711846 10.1085/jgp.201711846 Research Articles Research Article 514 Concatenated nicotinic acetylcholine receptors: A gift or a curse? Optimizing Cys-loop receptor concatenation http://orcid.org/0000-0003-1807-3331 Ahring Philip Kiær * Liao Vivian Wan Yu * http://orcid.org/0000-0002-0233-8350 Balle Thomas Faculty of Pharmacy, The University of Sydney, Sydney, Australia Correspondence to Philip Kiær Ahring: philip.ahring@sydney.edu.au * P.K. Ahring and V.W.Y. Liao contributed equally to this paper. 05 3 2018 150 3 453473 11 7 2017 15 11 2017 22 12 2017 © 2018 Ahring et al. 2018 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). Nicotine acetylcholine receptors can form countless heteromeric stoichiometries from a common set of subunits. Ahring et al. present the limitations of subunit concatenation and establish a refinement that achieves substantiated expression of uniform receptor pools from complex stoichiometric origins. Nicotinic acetylcholine receptors (nAChRs) belong to the Cys-loop receptor family and are vital for normal mammalian brain function. Cys-loop receptors are pentameric ligand-gated ion channels formed from five identical or homologous subunits oriented around a central ion-conducting pore, which result in homomeric or heteromeric receptors, respectively. Within a given Cys-loop receptor family, many different heteromeric receptors can assemble from a common set of subunits, and understanding the properties of these heteromeric receptors is crucial for the continuing quest to generate novel treatments for human diseases. Yet this complexity also presents a hindrance for studying Cys-loop receptors in heterologous expression systems, where full control of the receptor stoichiometry and assembly is required. Therefore, subunit concatenation technology is commonly used to control receptor assembly. In theory, this methodology should facilitate full control of the stoichiometry. In reality, however, we find that commonly used constructs do not yield the expected receptor stoichiometries. With ternary or more complex receptors, concatenated subunits must assemble uniformly in only one orientation; otherwise, the resulting receptor pool will consist of receptors with mixed stoichiometries. We find that typically used constructs of α4β2 nAChR dimers, tetramers, and pentamers assemble readily in both the clockwise and the counterclockwise orientations. Consequently, we investigate the possibility of successfully directing the receptor assembly process using concatenation. We begin by investigating the three-dimensional structures of the α4β2 nAChR. Based on this, we hypothesize that the minimum linker length required to bridge the C terminus of one subunit to the N terminus of the next is shortest in the counterclockwise orientation. We then successfully express receptors with a uniform stoichiometry by systematically shortening linker lengths, proving the hypothesis correct. Our results will significantly aid future studies of heteromeric Cys-loop receptors and enable clarification of the current contradictions in the literature. Bionomics Limited Australian Research Council https://doi.org/10.13039/501100000923 LP140100781 Australian National Health and Medical Research Council https://doi.org/10.13039/501100000925 APP1069417 ==== Body pmcIntroduction There could be several reasons for using concatenated subunits to expres heteromeric Cys-loop receptors. However, this technique is mostly used to direct receptor assembly to ensure expression of specific receptor pools. For instance, binary receptor combinations such as the nicotinic acetylcholine (ACh) receptor (nAChR) α4β2 can express as either (α4)3(β2)2 or (α4)2(β2)3. By using concatenated subunits, the cell surface receptor pool can be enriched with either combination (Zhou et al., 2003; Harpsøe et al., 2011). Because concatenated constructs could consist of anywhere between two to five subunits (dimers to pentamers), this technique promises unique control of the assembly process at the single-receptor level. As Cys-loop subunits have their C and N termini in the extracellular space, it is a straightforward process to concatenate them. This consists of manipulating the respective cDNAs such that all coding regions are located in the same expression cassette as one open reading frame. This is attained using a synthetic linker sequence bridging the C terminus of the first subunit to the N terminus of the next subunit. Typical linker sequences code for glutamine repeats or alanine–glycine–serine (AGS) repeats, as these amino acids are believed to have relatively minimal impact on normal subunit folding. Baumann et al. (2001) demonstrated that functional α1β2γ2 γ-amino-butyric-acid type A receptors (GABAARs) formed readily in Xenopus laevis oocytes upon the injection of concatenated dimer constructs of α1 and β2 in combination with a monomeric γ2 subunit. Later, Zhou et al. (2003) demonstrated successful expression of α4β2 nAChRs using a similar methodology. The most simple and efficient construct strategy used six repeats of an AGS sequence to link the β2 to the α4 subunit (β2-(AGS)6-α4). This was paired with the expression of either a monomeric α4 or β2 subunit. The specific methodologies developed in these studies have since been used extensively to answer basic scientific questions for both GABAARs and nAChRs (Baumann et al., 2002; Kuryatov and Lindstrom, 2011; Mazzaferro et al., 2011; Shu et al., 2012). For ease, the β2-(AGS)6-α4 construct is termed the β-6-α construct in this paper. Although the use of concatenated constructs is a powerful technique to study Cys-loop receptors, there are potential caveats that could affect experimental outcomes. First, the artificial linker sequence and accompanying structural constraints could change the receptor properties. Although this has not been commonly observed, results should be verified using other methodologies whenever possible. Second, the linker sequences could be subjected to proteolysis, thereby freeing subunits to assemble in an unintended manner. Although proteolysis never can be fully excluded, this has fortunately not been found to occur to any significant extent (Groot-Kormelink et al., 2006; Carbone et al., 2009). Third, linked constructs could form unexpected and unwanted receptors by themselves, thereby “polluting” the receptor pool. Indeed, such unwanted receptors are regularly observed; hence, linked dimer, trimer, or tetramer constructs should ideally be constructed such that they do not form functional receptors by themselves (Zhou et al., 2003; Groot-Kormelink et al., 2004; Kaur et al., 2009). Fourth, the assembly direction may not be fully controlled. This means several different receptor stoichiometries can arise in ternary receptor scenarios. Of the caveats listed above, we find the fourth to be the most disconcerting, as it can potentially lead to erroneous conclusions. Therefore, in the present study, we explore the degree to which the commonly used nAChR β2-6-α4 construct directs receptor subunit orientation. Unexpectedly, β2-6-α4 did not direct the orientation of linked subunits, nor did derived tetrameric or pentameric constructs. Because the β2-6-α4 construct is the “mother” of most used nAChR constructs, the implications of this are substantial. Furthermore, our data trigger the question of whether it is at all possible to direct subunit orientation using linked subunits. To address this, we studied 3-D models of Cys-loop receptor subunits and hypothesized that short linkers would direct receptor assembly in the counterclockwise orientation. We then designed a range of concatenated constructs with shorter linkers. Importantly, we found that although it is possible to direct subunit orientation, this first requires an optimization process to identify the “optimal” linker length for the specific subunits in question. This optimization, however, is a crucial step for the appropriate use of concatenation technology. Materials and methods Materials 3-(3-(pyridine-3-yl)-1,2,4-oxadiazol-5-yl)benzonitrile (NS9283) was synthesized at Saniona A/S as described by Timmermann et al. (2012). The structure was confirmed using mass spectrometric analysis and proton nuclear magnetic resonance spectroscopy and was of >98% purity. ACh and all salts or other chemicals not specifically mentioned were purchased from Sigma-Aldrich and were of analytical grade. Oligonucleotides were purchased from Sigma-Aldrich, and sequencing services were obtained from the Australian Genome Research Facility. Restriction enzymes, Q5 polymerase, T4 DNA ligase, and competent Escherichia coli 10-β bacteria were from New England Biolabs. Molecular biology Human cDNA for monomeric α4, β2, and α4VFL nAChR subunits and concatenated constructs β2-6-α4 and β2-6-α4-9-β2-6-α4 were kind gifts from Saniona A/S. New β2-xa-α4 concatenated constructs, where x = 9, 6, 3, 0, and −3 amino acid (a) linkers, were built from wild-type β2 and α4 subunits using PCR. In brief, AGS linker sequences were designed to contain a unique central restriction site; antisense β2 and sense α4 oligonucleotide sequences were then fabricated to traverse this site. The antisense β2 oligonucleotides caused deletion of the β2 stop codon and in-frame fusion to the AGS linker sequence. The sense α4 oligonucleotides caused omission of the α4 signal peptide and in-frame fusion to the AGS linker sequence. The remaining β2 sense and α4 antisense oligonucleotides were designed to match the respective wild-type sequences and include suitable restriction sites. Standard PCR reactions with β2 or α4 as a template were performed using Q5 polymerase, and PCR products were cloned into in-house vectors using restriction digestion and ligation. Correct introduction of linker sequences and fidelity of all coding sequences were verified by double-stranded sequencing. Thereafter, concatenated constructs were created by restriction digestion and ligation using the unique AGS linker and vector sites. Pentameric constructs were built in a similar manner, with each linker sequence containing its own unique restriction site. E. coli 10-β bacteria were used as hosts for plasmid expansion, and plasmid purifications were performed with standard kits (Qiagen). cRNA was produced from linearized cDNA using the mMESSAGE mMACHINE T7 Transcription kit (Ambion) according to the manufacturer’s description and stored at −80°C until use. Modeling The x-ray structure of the human α4β2 nAChR (Protein Data Bank accession no. 5KXI; Morales-Perez et al., 2016) was downloaded from the database (Berman et al., 2002) and prepared according to the protocol for protein preparation implemented in Maestro 10.4 (Schrödinger Release 2015–4; Schrödinger). In the published structure, 20 amino acids in the C-terminal end of the β2 subunit are disordered and are hence unresolved. Thus, relative to the wild-type protein, the β2 subunit chain ends with FLQPL373, and 374FQNYTTTTFLHSDHSAPSSK393 is missing. No attempts were made to model these missing residues. Instead, the shortest possible AGS-repeat linkers were inserted to create a clockwise construct linking the last visible residue in the C-terminal tail of the β2 subunit (chain B) to the N-terminal tail of the α4 subunit (chain A) or a counterclockwise construct connecting the last visible C-terminal residue of β2 (chain E) to the N terminus of α4 (chain A). In both cases, “shortest possible” was defined as the number of AGS repeats required to link the two terminals without causing significant distortion to either terminal after geometry optimization in a Macromodel (OPLS3 Force Field, GB/SA solvation model). Electrophysiology X. laevis oocytes were prepared as previously described by Mirza et al. (2008). In brief, to obtain isolated oocytes, lobes from the ovaries of adult female X. laevis frogs were removed as approved by the Animal Ethics Committee of The University of Sydney (reference number 2013/5915) and defolliculated using collagenase. Oocytes were injected with ∼50 nl of a 0.5-µg/µl cRNA mixture encoding the desired nAChR subunits and incubated for 2–3 d (unless otherwise noted) at 18°C in modified Barth’s solution (96 mM NaCl, 2.0 mM KCl, 1 mM MgCl2, 1.8 mM CaCl2, 5 mM HEPES, 2.5 mM sodium pyruvate, 0.5 mM theophylline, and 100 µg/ml gentamycin, pH 7.4). Electrophysiological recordings using the two-electrode voltage-clamp technique were performed with oocytes placed in a custom-built recording chamber and continuously perfused with a Ca2+-free saline solution termed CF buffer (115 mM NaCl, 2.5 mM KCl, 1.8 mM BaCl2, and 10 mM HEPES, pH 7.4). Pipettes were backfilled with 3 M KCl, and open pipette resistances ranged from 0.4 to 2 MΩ when submerged in CF buffer. Cells were voltage clamped at a holding potential of −60 mV using an Axon Geneclamp 500B amplifier (Molecular Devices). Oocytes with initial leak currents exceeding 200 nA when clamped were discarded. NS9283 was dissolved as a 100-mM stock solution in DMSO, which upon final dilution gave a maximal concentration of 0.1%. This DMSO concentration did not evoke any measurable currents from wild-type α4β2 receptors. Fresh ACh and NS9283 dilutions were prepared in CF buffer on the day of the experiment, and solutions were applied to the oocytes with a flow rate of 2.0 ml/min via a glass capillary. Each application lasted ∼30 s, and the application system ensured rapid solution exchange (in the order of a few seconds). Amplified signals were low-pass filtered at 20 Hz, digitized at 200 Hz by an Axon Digidata 1440A (Molecular Devices), and recorded using Clampex 10.2 (Molecular Devices). Experimental protocols A complete concentration–response relationship (CRR) consisting of six to eight individual concentrations of NS9283 or ACh was obtained from each oocyte. To ensure the reproducibility of evoked current amplitudes, a set of control applications was performed before the actual concentration–response applications. These control applications were three AChcontrol (10 µM) applications, one AChmax (3,160 µM) application, another three AChcontrol (10 µM) applications, and finally a buffer (no ACh) application. Thereafter followed six to eight applications of NS9283 coapplied with AChcontrol (10 µM) or ACh alone in increasing concentrations. In a few instances, an NS9283 CRR was obtained after an ACh CRR on the same oocyte. In these cases, the maximal ACh-evoked current amplitude for the given oocyte was considered as belonging only to the ACh CRR. Final datasets for NS9283 and ACh were assembled from experiments conducted on a minimum of two batches of oocytes. Data analysis Raw traces were analyzed using pClamp 10.2 (Molecular Devices). Traces were baseline subtracted during analysis, and responses to individual applications were quantified as peak current amplitudes. For experiments with ACh, peak current amplitudes (I) of full CRRs were fitted to the Hill equation and normalized to the maximal fitted response (Imax_fit_ACh) for each individual oocyte (i.e., I/Imax_fit_ACh). For experiments with NS9283, the compound was coapplied with AChcontrol (10 µM). Differences between AChcontrol-evoked current amplitudes in the absence or presence of NS9283 (I) were calculated as the percent change from the AChcontrol-evoked current (i.e., ([I−IACh_control] × 100)/IACh_control). All CRRs were fitted by nonlinear regression in Prism 7 (GraphPad) to a mono- or biphasic equation with a constrained Hill slope of 1 and efficacy at infinitely low compound concentrations set to 0, unless otherwise specified. Comparison of best approximation (monophasic vs. biphasic) was performed using the F test in Prism 7. Further statistical analysis was performed using Prism 7. Experimental strategy 1: ACh and NS9283 sensitivity of α4β2 nAChRs Depending on the stoichiometry, wild-type α4β2 nAChRs display either a monophasic (α4)2(β2)3 or a biphasic (α4)3(β2)2 ACh CRR, meaning that the data are best approximated by a first- or second-order equation as revealed by, for example, an F test (Fig. 1, A and B; Harpsøe et al., 2011; Mazzaferro et al., 2011). For the biphasic (α4)3(β2)2 receptor concentration–response curve, the first component reflects ACh binding and activation via two high-affinity α4–β2 interface–binding sites, and the second component reflects the additional activation of the same receptors via ACh binding to the low-affinity α4–α4 interface site (Indurthi et al., 2016). Such biphasic CRRs with interdependent variables are, however, inherently difficult to resolve, and the calculated fractions of the first component, as well as EC50 values, can vary considerably from relatively minor data variations. Hence, it can be virtually impossible to distinguish a pure receptor pool from one containing pollutant receptors whenever biphasic CRRs are involved. Figure 1. α4β2 nAChR stoichiometry and functional effects of ACh and NS9283. X. laevis oocytes were injected with cRNA mixtures of α4 and β2 or α4VFL and β2 subunits in 10:1 ratios and subjected to two-electrode voltage-clamp electrophysiology as described in Materials and methods. The 10:1 cRNA ratios were used to ensure uniform populations of (α4)3(β2)2 and (α4VFL)3(β2)2 receptors. Data for α4 and β2 injected in a 1:4 ratio ((α4)2(β2)3 receptor) are from Harpsøe et al. (2011). (A) Functional α4β2 nAChRs can express in 2α:3β or 3α:2β stoichiometries (left and middle, respectively). The stoichiometry affects the total number of ACh-binding sites, as the 3α:2β stoichiometry contains an additional site in the α4–α4 interface. Furthermore, NS9283 binds with high selectivity in the α4–α4 site, where it behaves as an agonist. Upon mutating three amino acids in the complementary face of the α4 subunit to give α4VFL, ACh sensitivity is increased in the α4VFL–α4VFL site, and NS9283 binding is lost (right). (B) ACh CRRs. Baseline-subtracted, ACh-evoked peak current amplitudes (I) for the indicated receptors were fitted to the Hill equation by nonlinear regression and normalized to the maximal fitted values (Imax fit ACh). Normalized responses are depicted as means ± SEM as a function of the ACh concentrations, and they are fitted to biphasic equations with a fixed bottom of 0 and a Hill slope of 1. Data were obtained from n = 9–14 experiments, and regression results are presented in Table 1. Data for the (α4)2(β2)3 receptor are from Harpsøe et al. (2011). (C) NS9283 CRRs. NS9283 enhancement of ACh-evoked currents was evaluated for (α4)3(β2)2 and (α4VFL)3(β2)2 receptors by coapplication with a submaximal control concentration of ACh (10 µM). Baseline-subtracted peak current amplitudes (I) were expressed as percent change from IACh_control and are depicted as means ± SEM as a function of the NS9283 concentration. Data points were fitted by nonlinear regression to the Hill equation with a fixed bottom of 0 and a Hill slope of 1. Data were obtained from n = 13–16 experiments, and regression results are presented in Table 1. Data for the (α4)2(β2)3 receptor are from Timmermann et al. (2012). (D) Hypothetically, injection of a cRNA mixture of α4, α4VFL, and β2 into oocytes could yield eight different receptors in the 3α:2β stoichiometry. Using NS9283 as a marker, these can be divided into those that are sensitive and those that are insensitive, depending on whether the α4VFL subunit is participating in the complementary position of the α4–α4 interface. To increase the assay sensitivity in the present work, we used the unique actions and binding property of the compound NS9283. Although originally identified as an allosteric modulator of α4β2 receptors, NS9283 has been found to have site-selective agonistic actions (Timmermann et al., 2012; Olsen et al., 2013, 2014). Essentially, NS9283 binds in the wild-type α4–α4 ACh-binding pocket, where it largely interacts with the same amino acids as ACh (Fig. 1 A). Thus, at nonsaturating ACh concentrations, NS9283 binding to the α4–α4 interface causes increased receptor activation of (α4)3(β2)2 receptors (Fig. 1 C and Table 1). However, NS9283 displays no actions on receptors that lack an α4–α4 interface, and because (α4)2(β2)3 receptors have a β2–β2 interface instead, no activity is observed at this stoichiometry. Table 1. Maximal fitted response and potency of ACh and NS9283 from wild-type and concatenated α4β2 nAChRs Construct Subunit ACh NS9283 Both Emax pEC50_1 pEC50_2 Frac n E31.6 µM Emax pEC50 n AChmax current % % % nA α4 (10:1) β2 102 ± 1 5.7 ± 0.3 3.9 ± 0.04 0.11 ± 0.03 14 760 ± 70 860 ± 60 5.3 ± 0.1 13 8,100 ± 900 (27) α4VFL (10:1) β2 104 ± 2 5.0 ± 0.1 3.9 ± 0.3 0.71 ± 0.12 9 −9.9 ± 1.8 No pos. pot. N/A 16 1,700 ± 300 (25) β-6-α 102 ± 2 5.8 ± 0.3 3.6 ± 0.1 0.11 ± 0.02 9 640 ± 90 900 ± 140 4.9 ± 0.2 9 1,000 ± 300 (18) β-6-α α4 101 ± 2 5.7 ± 0.6 3.8 ± 0.1 0.073 ± 0.038 10 860 ± 100 1,100 ± 100 5.1 ± 0.2 12 5,200 ± 800 (22) β-6-α α4VFL 106 ± 2 5.2 ± 0.2 3.8 ± 0.1 0.40 ± 0.08 12 210 ± 20 270 ± 30 5.0 ± 0.1 7 2,200 ± 400 (19) β-6-α (1:25) α4VFL 106 ± 1 5.3 ± 0.1 4.0 ± 0.1 0.39 ± 0.05 8 240 ± 20 300 ± 30 5.1 ± 0.1 6 1,200 ± 200 (12) β-9a-α 105 ± 3 5.7 ± 0.2 3.9 ± 0.1 0.28 ± 0.05 11 100 ± 20 130 ± 30 5.0 ± 0.3 5 220 ± 60 (16) β-9a-α α4 103 ± 3 5.6 ± 0.4 3.7 ± 0.1 0.13 ± 0.05 8 680 ± 40 950 ± 80 4.9 ± 0.1 6 3,000 ± 600 (14) β-9a-α α4VFL 109 ± 5 5.3 ± 0.3 3.7 ± 0.2 0.32 ± 0.09 5 200 ± 10 260 ± 20 5.0 ± 0.1 8 2,600 ± 1,000 (13) β-6a-α 103 ± 2 5.8 ± 0.1 4.5 ± 0.4 0.70 ± 0.12 6 45 ± 9 51 ± 7 5.4 ± 0.2 6 94 ± 18 (12)a β-6a-α α4 100 ± 1 6.1 ± 0.4 3.9 ± 0.04 0.059 ± 0.018 10 710 ± 50 870 ± 60 5.1 ± 0.1 6 1,800 ± 300 (16) β-6a-α α4VFL 105 ± 3 5.6 ± 0.4 4.2 ± 0.2 0.31 ± 0.14 7 140 ± 10 180 ± 10 5.0 ± 0.1 8 170 ± 40 (15) β-3a-α 103 ± 1 5.7 ± 0.1 4.5 ± 0.2 0.64 ± 0.11 14 7.5 ± 8.2 Inconclusive N/A 11 120 ± 30 (25)b β-3a-α α4 102 ± 1 5.7 ± 0.2 4.0 ± 0.03 0.11 ± 0.02 17 610 ± 30 720 ± 30 5.2 ± 0.1 18 1,500 ± 200 (35) β-3a-α α4VFL 104 ± 1 5.7 ± 0.1 4.4 ± 0.1 0.41 ± 0.06 17 22 ± 3 27 ± 4 5.2 ± 0.2 24 890 ± 130 (41) β-0a-α N/A N/A N/A N/A N/A N/A N/A N/A N/A 15 ± 4 (24)c β-0a-α α4 103 ± 2 5.8 ± 0.7 3.7 ± 0.05 0.053 ± 0.028 11 840 ± 90 1,100 ± 100 5.0 ± 0.1 9 130 ± 40 (20)d β-0a-α α4VFL 101 ± 1 5.4 ± 0.5 4.6 ± 0.2 0.28 ± 0.34 14 −6.0 ± 2.4 No pos. pot. N/A 15 95 ± 18 (29)d β-(-3a)-α N/A N/A N/A N/A N/A N/A N/A N/A N/A No current (18) β-(-3a)-α α4 N/A N/A N/A N/A N/A N/A N/A N/A N/A 36 ± 6 (27) β-(-3a)-α α4VFL N/A N/A N/A N/A N/A N/A N/A N/A N/A   15 ± 2 (21) X. laevis oocytes were injected with the indicated cRNA mixtures in a 1:1 ratio (unless otherwise indicated) and subjected to two-electrode voltage-clamp electrophysiology after 2–3 d of incubation time as described in Materials and methods; also see Fig. 1. Data points were fitted to either a monophasic or a biphasic equation with the bottom set to 0 and a Hill slope set to 1 by nonlinear regression. For ACh, biphasic fitting represented the preferred model for all datasets as determined by an F test. Fitted maximal responses of ACh and NS9283 are presented as Emax ± SEM in percentages, with associated potencies presented as pEC50 ± SEM in M for the indicated number of individual oocytes. Observed NS9283 responses at the 31.6-μM concentration are presented as E31.6 μM ± SEM in percentages. The mean maximal current obtained with applications of 3.16 mM ACh is presented as AChmax current ± SEM in nanoamperes for all tested oocytes for each construct. N/A, not applicable; No pos. pot., no positive potentiation; Inconclusive, meaningful fitting not possible, as only ∼50% of the oocytes displayed a positive NS9283 response. a Selected oocytes due to low AChmax-evoked current amplitudes. b No AChmax-evoked currents in approximately three out of four oocytes. c Only two oocytes show >25 nA AChmax-evoked current amplitude. d Mixture of oocytes from 3 and 5 d of incubation. Experimental strategy 2: Using the combination of NS9283 and α4VFL Potent binding and efficacy of NS9283 are dependent on the presence of three E-loop amino acids (H142, Q150, and T152) in the complementary face of the α4 subunit (Olsen et al., 2014). Therefore, by point mutating these to the corresponding amino acids in β2 (V136, F144, and L146 [the VFL motif]), no NS9283 efficacy is observed in receptors with an α4VFL–α4VFL interface (Fig. 1, A and C). Yet, these same three mutations cause the agonist-binding pocket in the α4VFL–α4VFL interface to resemble that of an α4–β2 interface, which leads to increased sensitivity of the second component of the ACh concentration–response curve (Fig. 1 B). Although NS9283 shows no activity at (α4VFL)3(β2)2 receptors, the situation is more complex when only one of the subunits is mutated in the α4–α4 interface. In the case of an α4VFL–α4 interface, the VFL mutations do not face the NS9283-binding site, and therefore the compound still shows full response (Fig. 1 D). Conversely, in the case of an α4–α4VFL interface, the three VFL mutations face the NS9283-binding pocket, and NS9283 no longer binds with sufficient potency to allow any activity (Fig. 1 D). Therefore, NS9283 can be used to pinpoint the position of the α4VFL subunit in an α4–α4 interface; if NS9283 displays efficacy, the mutant a4VFL subunit is primary; if not, the subunit is complementary. Experimental strategy 3: Concentrations at which NS9283 shows selectivity Like most other compounds, NS9283 only remains selective in a certain concentration range. At concentrations <10 µM, NS9283 appears fully selective for the α4–α4 interface. This is evidenced by the increasing current amplitudes at the (α4)3(β2)2 receptor in the presence of AChcontrol (10 µM) and the lack of actions at (α4)2(β2)3 or (α4VFL)3(β2)2 receptors (Fig. 1 C). At concentrations of 31.6 µM or higher, NS9283 still shows increased activity at (α4)3(β2)2 receptors; however, this is accompanied by inhibition of the AChcontrol-evoked current amplitudes at (α4)2(β2)3 and (α4VFL)3(β2)2 receptors (Fig. 1 C). For (α4VFL)3(β2)2 receptors, an inhibition of −9.9 ± 1.8% (n = 16) and −27 ± 2% (n = 9) is observed at 31.6 and 100 µM, respectively (note, the 100-µM data point is not included in the illustration). Given that NS9283 binds to Ls-AChBP with a Ki value of 67 µM (Olsen et al., 2014), the observed inhibition is consistent with binding to the ACh-binding pocket in α4–β2 interfaces at high concentrations. As NS9283 is not an agonist at the wild-type α4–β2 interfaces (Olsen et al., 2014), such binding would lead to competitive antagonism instead. Therefore, in the experiments performed in this study, we used a maximal concentration of 31.6 µM NS9283. This represents a compromise between the desire to observe the highest possible activity in receptor pools with a high proportion of sensitive receptors and the desire to avoid inhibition in pools with a low proportion of sensitive receptors. Experimental strategy 4: Evaluating NS9283 efficacy at a fixed AChcontrol concentration When comparing efficacy of a modulatory compound at different receptor types in the presence of an agonist, it is best practice to measure at the same open-channel probability (Ahring et al., 2016). In cases where the maximal open-channel probability is not known, it is usually approximated by using the same degree of agonist response (e.g., EC10). However, this strategy is only valid when comparing efficacies between receptor populations that can be assumed uniform. In this study, the response of NS9283 was measured at receptors arising from a range of constructs. One of the key findings was that many of these constructs did not lead to uniform receptor expression but to mixed receptor pools. These receptor pools contained both NS9283-sensitive and NS9283-insensitive receptors. To further complicate the situation, the receptors had variant ACh sensitivity. In such scenarios, a difference in measured NS9283 response is (a) a reflection of the change in the mean ACh CRR such that a given ACh concentration leads to altered percent activation (e.g., EC10 to EC30) and (b) a reflection of a change in percentage of NS9283-sensitive receptors among nonsensitive ones. Therefore, an attempt to adjust the AChcontrol concentrations according to ACh CRRs in cases with mixed receptor pools is counterproductive. Finally, it is important to note that a main goal in this study was to develop a technology that allows expression of pure receptor populations. Consequently, the specific experiments were designed such that this is achieved when NS9283 displays no efficacy. Results In the present study, we relied on the unique selectivity of NS9283 in combination with an NS9283-resistant mutant α4VFL subunit to decipher the absolute stoichiometry of expressed α4β2 nAChRs. NS9283 selectively binds in the α4–α4 interface of wild-type (α4)3(β2)2 receptors to increase receptor gating at submaximal ACh concentrations. The three mutations in the α4VFL subunit make the ACh-binding pocket in an α4VFL–α4VFL interface resemble that of an α4–β2 interface, increasing ACh sensitivity but abolishing NS9283 sensitivity (Fig. 1 A). Further detail regarding the selectivity of NS9283, its sensitivity to the VFL mutations, and the reason why NS9283 responses are compared for the 31.6-µM concentration in these results is presented in Materials and methods (Experimental strategies 1–4). Binary and ternary receptors Heterologously expressed Cys-loop receptors often assemble in multiple stoichiometries. It is therefore imperative to consider what to expect in a situation when a receptor pool contains two or more receptor subpopulations. Expressing binary α4β2 nAChRs in oocytes by injection of wild-type α4 and β2 cRNA is simple because there are only two stoichiometries, 2α:3β and 3α:2β, that form functional receptors (Fig. 1 A; Harpsøe et al., 2011; Mazzaferro et al., 2011). Although injection of equimolar amounts of cRNA generally leads to a mixed receptor pool, biasing cRNA ratios allows for uniform receptor pool formation (Zwart and Vijverberg, 1998; Harpsøe et al., 2011). Using this methodology, α4β2 receptors were expressed in 2α:3β and 3α:2β stoichiometries by coinjection of α4 and β2 cRNA in 1:4 and 10:1 ratios, respectively. These two receptor pools are easily distinguished by their sensitivity to ACh and NS9283 (Fig. 1, B and C; and Table 1). Although the ACh CRR for the 2α:3β stoichiometry was well approximated by the Hill equation (i.e., a monophasic, or first-order, equation), the ACh CRR for the 3α:2β stoichiometry was best approximated by a biphasic, or second-order, equation. Furthermore, although 31.6 µM NS9283 displayed no positive response at the 2α:3β stoichiometry, the response at the 3α:2β stoichiometry was 760%. The binary scenario with a mutant α4VFL subunit coinjected with wild-type β2 cRNA in a 10:1 ratio displayed ACh sensitivity intermediate to that of the two wild-type stoichiometries, although no positive efficacy was observed with NS9283 (Fig. 1, B and C; and Table 1). These data are in good agreement with previous studies (Harpsøe et al., 2011; Timmermann et al., 2012; Olsen et al., 2013). Expression of ternary scenarios is inherently more complex. Even when biasing toward the 3α:2β stoichiometry, injection of a cRNA mixture consisting of wild-type α4, mutant α4VFL, and wild-type β2 can theoretically lead to assembly of eight different receptors that contain zero to three α4VFL subunits (Fig. 1 D). It is rarely possible to distinguish these receptors from one another. By using NS9283 sensitivity as an assessment, they can be separated into two groups depending on the characteristics of the α4–α4 interface, as described in Materials and methods (Experimental strategy 2). Consequently, for ternary receptor scenarios, a concatenation methodology represents the only avenue to ensure receptor expression of specific stoichiometries. The linker in the β2-6-α4 concatenated construct does not direct the orientation of linked subunits In our initial studies with linked subunits, we employed the widely used concatenated β2-6-α4 dimer construct (Zhou et al., 2003). As previously described, this particular construct contains a linker of six AGS repeats connecting the C terminus of the β2 to the mature N terminus of the α4 subunit. β-6-α Robust ACh-evoked currents with amplitudes in the microampere range were observed from oocytes injected with β-6-α cRNA alone (Table 1). For ACh, the CRR was best approximated with a biphasic equation, revealing two EC50 values of ∼2 and 250 µM and a first component fraction of 0.11 (Fig. 2 A and Table 1). Furthermore, 31.6 µM NS9283 increased the AChcontrol (10 µM) current by 640% (Fig. 2, B and C; and Table 1). These data are similar to observations at the wild-type 3α:2β receptors obtained from α4 and β2 in the biased 10:1 cRNA ratio described in the previous section (t test NS9283 response: P = 0.30). This indicates that injection of the dimer construct alone leads to receptors that are predominantly of the 3α:2β stoichiometry, which is corroborated by the findings of Jin and Steinbach (2011) using a similar β-6-α construct. Disregarding the possibility of linker proteolysis, the simplest explanation for the observations are pentameric receptors composed of three sets of linked dimers with a “dangling” β2 subunit (Fig. 2 D). As it is unknown whether or how the linker dictates assembly, there are three possible ways in which such receptors can assemble with clockwise and counterclockwise orientations of the dimers (Fig. 2 D). Although this represents the simplest explanation, more complex assemblies including dangling α4 subunits or di-pentamers could also exist. Figure 2. ACh and NS9283 sensitivity and potential stoichiometry of receptors from the concatenated β-6-α construct. X. laevis oocytes were subjected to two-electrode voltage-clamp electrophysiology as described in Materials and methods. (A and B) ACh (A) and NS9283 (B) CRRs were obtained from oocytes injected with the β-6-α dimer construct alone or coinjected with monomeric α4 or α4VFL subunits in a 1:1 ratio. The linker sequence is shown in Table 3. Electrophysiological data were evaluated as described in Materials and methods; also see Fig. 1. Data from n = 6–12 experiments are depicted as means ± SEM as a function of the ACh or NS9283 concentration, and regression results are presented in Table 1. Data for wild-type receptors from monomeric subunits in Fig. 1 are indicated as dashed lines. (C) Representative traces illustrating NS9283 responses at oocytes injected with β-6-α, β-6-α and α4, or β-6-α and α4VFL. Bars above the traces indicate the 30-s application time and concentrations of applied compounds. (D) The simplest way in which a dimeric β-6-α construct could lead to functional 3α:2β stoichiometry receptors is three sets of linked dimers assembling with a dangling β2 subunit. This, again, could be envisioned to lead to three different assemblies because the two dimers in each receptor can be oriented in the clockwise, the counterclockwise, or both orientations when viewed from the synaptic cleft. Note that other, more complex assemblies cannot be excluded. (E) When coinjecting β-6-α and α4VFL, four different possible assemblies involving two dimer constructs and one monomeric subunit could arise. Of the four possibilities, one can likely be considered nonfunctional, given that all three α subunits are placed consecutively (right). If one or both dimers assemble in the clockwise orientation, the receptor will mimic wild-type 3α:2β receptors with respect to NS9283 sensitivity (middle). However, if both dimer constructs assemble in the counterclockwise orientation, the receptor will mimic wild-type 2α:3β receptors (left). β-6-α and α4 Coinjection of β-6-α and α4 cRNA resulted in receptors of the 3α:2β stoichiometry. Based on visual inspection of the ACh CRR and an NS9283 response of 860%, the receptors appear identical to the ones obtained from α4 and β2 cRNA in a biased ratio, but also to those from the β-6-α dimer alone (ANOVA NS9283 response: F = 1.45, P = 0.25; Fig. 2, A–C; and Table 1). Given a known high propensity of monomeric subunits to integrate with linked subunits (Groot-Kormelink et al., 2004) and a 2:1 molar ratio of α4 to β-6-α in the cRNA mixture, it is reasonable to assume that the receptor pool in this case predominantly consists of “true” pentameric receptors (i.e., two sets of dimers with one monomeric α4 subunit). β-6-α and α4VFL Robust ACh-evoked currents with amplitudes in the microampere range were observed upon coinjection of cRNA for β-6-α with monomeric α4VFL; however, the resulting receptor pool did not appear related to any that were hitherto observed (Fig. 2, A–C; and Table 1). Judged from visual inspection of the ACh CRR, some resemblance was noted to receptors obtained with α4VFL and β2 cRNA in a biased ratio (Fig. 2 A). Yet, although both displayed biphasic ACh CRRs with similar EC50 values, different first-component fractions led to altered curve progressions. Furthermore, an NS9283 response of 210% was observed for β-6-α and α4VFL in comparison to no positive efficacy with α4VFL and β2 (Fig. 2 B). Overall, these “intermittent” data point toward a mixed receptor pool containing subpopulations where some receptors are NS9283 sensitive and others are not. To rule out the presence of major receptor subpopulations with dangling subunits, we performed similar experiments with a 1:25 cRNA ratio of β-6-α and α4VFL. In comparison with the 1:1 ratio, this did not significantly alter results for either ACh or NS9283 (Fig. 2, A and B). NS9283 gave rise to a 230% increase in AChcontrol-evoked currents for the 1:25 ratio versus a 210% increase for the 1:1 ratio (t test NS9283 response: P = 0.49). This implies that receptors with dangling subunits do not constitute substantial subpopulations in the receptor pools. Therefore, the receptor pool from β-6-α and α4VFL injection appears to consist primarily of pentamers from two linked dimers and one monomeric α4VFL subunit. Again, this is consistent with a known high propensity of monomeric subunits to integrate with linked subunits (Groot-Kormelink et al., 2004). Given that the data showed a mixed receptor pool with intermediate NS9283 response, these results could have only arisen from an ability of the linked dimers to orient themselves in both the clockwise and counterclockwise orientations. In total, there are four assembly possibilities in which the α4VFL subunit takes different positions between two dimers (Fig. 2 E). Of these four possibilities, one can likely be excluded, as it places all three α subunits consecutively (Fig. 2 E, right). Out of the remaining three, the scenario in which both dimers are oriented in the counterclockwise direction will lead to an α4–α4VFL interface that is insensitive to NS9283 (Fig. 2 E, left). In the remaining two scenarios, the VFL mutations are not facing the binding pocket of the α4–α4 interface, and the receptors formed are therefore still responsive to NS9283. Hence, the receptor pool arising from β-6-α and α4VFL injection contains at least two different receptor subpopulations, one sensitive and one insensitive to NS9283, and these can form only if the subunit assembly orientation is not directed by the construct. Increasing the number of linked subunits does not direct the orientation of linked subunits Several studies with concatenated Cys-loop receptors have relied on linking more than two subunits (Baumann et al., 2002; Baur et al., 2006; Groot-Kormelink et al., 2006; Carbone et al., 2009; Kuryatov and Lindstrom, 2011; Shu et al., 2012; Jin et al., 2014; Mazzaferro et al., 2014). Having observed that the β-6-α construct does not direct the orientation of subunit assembly, we decided to investigate whether the addition of further linked subunits would affect this. We therefore used a previously published concatenated tetrameric construct, β2-6-α4-9-β2-6-α4, which was made by fusing two dimer β-6-α constructs to an (AGS)9 linker (Carbone et al., 2009; Harpsøe et al., 2011). The three “extra” repeats used in comparison with the dimer (AGS)6 linker were included to compensate for the shorter C-terminal tail of α4 versus β2 to give similar linker lengths (Carbone et al., 2009). In the following results, the β2-6-α4-9-β2-6-α4 construct is termed β-6-α-9-β-6-α. β-6-α-9-β-6-α Relatively small AChmax-evoked currents with amplitudes of ∼50 nA were observed from oocytes injected with cRNA for β-6-α-9-β-6-α alone (Table 2). The receptor pool appeared to be a mixture containing 2α:3β and 3α:2β receptors with a fraction of 0.45 for the first ACh CRR component and an NS9283 response of 170% (Fig. 3, A and B; and Table 2). Hence, it is possible to form functional pentameric receptors from two linked tetramers, leaving three subunits dangling in at least two ways. Table 2. Maximal fitted response and potency of ACh and NS9283 from concatenated α4β2 nAChRs Construct Subunit ACh NS9283 Both Emax pEC50_1 pEC50_2 Frac n E31.6 µM Emax pEC50 n AChmax current % % % nA α4 (10:1) β2 102 ± 1 5.7 ± 0.3 3.9 ± 0.04 0.11 ± 0.03 14 760 ± 70 860 ± 60 5.3 ± 0.1 13 8,100 ± 900 (27) α4VFL (10:1) β2 104 ± 2 5.0 ± 0.1 3.9 ± 0.3 0.71 ± 0.12 9 −9.9 ± 1.8 No pos. pot. N/A 16 1,700 ± 300 (25) Tetrameric β-6-α-9-β-6-α 109 ± 5 5.7 ± 0.2 3.8 ± 0.2 0.45 ± 0.07 6 170 ± 20 220 ± 30 5.0 ± 0.2 6 49 ± 9 (12) β-6-α-9-β-6-α α4 100 ± 1 5.9 ± 0.8 3.8 ± 0.04 0.035 ± 0.020 8 1,100 ± 100 1,300 ± 100 5.2 ± 0.1 6 1,500 ± 300 (14) β-6-α-9-β-6-α α4VFL 109 ± 4 5.0 ± 0.2 3.6 ± 0.2 0.41 ± 0.09 9 130 ± 10 170 ± 20 5.0 ± 0.2 5 720 ± 130 (14) β-6-α-9-β-6-α (1:5) α4VFL 102 ± 2 5.5 ± 0.3 4.2 ± 0.2 0.32 ± 0.13 13 100 ± 10 130 ± 20 5.1 ± 0.2 6 850 ± 170 (19) β-6-α-9-β-6-α (1:25) α4VFL 106 ± 2 5.4 ± 0.1 4.0 ± 0.2 0.40 ± 0.12 9 130 ± 10 140 ± 10 5.4 ± 0.1 8 310 ± 40 (15) Pentameric β-21a-α-β-α-α 101 ± 1 5.9 ± 0.3 3.7 ± 0.02 0.060 ± 0.011 12 930 ± 30 1,100 ± 40 5.1 ± 0.1 14 2,300 ± 400 (15) β-21a-α-β-α-αVFL 105 ± 1 5.2 ± 0.1 4.0 ± 0.2 0.59 ± 0.07 14 41 ± 4 48 ± 5 5.3 ± 0.2 19 2,400 ± 400 (22) β-21a-α-α-β-α 100 ± 2 6.1 ± 0.6 4.0 ± 0.05 0.060 ± 0.025 14 780 ± 90 910 ± 70 5.3 ± 0.1 9 5,300 ± 700 (23) β-21a-α-αVFL-β-α 102 ± 2 5.3 ± 0.2 4.3 ± 0.2 0.46 ± 0.18 19 55 ± 6 75 ± 9 5.0 ± 0.2 23 2,000 ± 200 (39) β-21a-αVFL-α-β-α 104 ± 1 5.2 ± 0.1 4.1 ± 0.1 0.44 ± 0.1 12 85 ± 10 102 ± 10 5.2 ± 0.1 12 600 ± 60 (12) β-3a-α-β-α-α 102 ± 1 6.0 ± 0.3 4.0 ± 0.04 0.11 ± 0.02 9 640 ± 90 800 ± 90 5.1 ± 0.2 5 700 ± 120 (14) β-3a-α-β-α-αVFL 104 ± 1 5.4 ± 0.1 4.3 ± 0.1 0.53 ± 0.08 14 11 ± 2 13 ± 1a 5.8 ± 0.2 15 1,400 ± 400 (29) β-3a-αVFL-β-α-α 105 ± 1 5.6 ± 0.1 4.1 ± 0.1 0.39 ± 0.05 15 200 ± 20 240 ± 20 5.2 ± 0.1 9 440 ± 80 (24) β-3a-αVFL-α-β-α 105 ± 4 5.1 ± 0.3 3.6 ± 0.2 0.28 ± 0.10 9 270 ± 10 340 ± 20 5.1 ± 0.1 12 730 ± 120 (17) Data for ACh and NS9283 obtained from tetrameric or pentameric concatenated constructs are handled and presented as described in Table 1. Longer incubation times of 3–7 d were generally utilized to increase expression levels. Data for α4 and β2 or α4VFL and β2 are from Table 1 for reference purposes. No pos. pot., no positive potentiation. a No NS9283 efficacy observed in ∼50% of oocytes, resulting in poor fitting, as evidenced by the apparent increase in potency. Figure 3. ACh and NS9283 sensitivity and potential stoichiometry of receptors from concatenated tetrameric or pentameric constructs. X. laevis oocytes were subjected to two-electrode voltage-clamp electrophysiology. Electrophysiological data were evaluated as described in Materials and methods; also see Fig. 1. (A and B) ACh (A) and NS9283 (B) CRRs were obtained from oocytes injected with the tetrameric β-6-α-9-β-6-α (T) construct alone or coinjected with monomeric α4 or α4VFL subunits in a 1:1 ratio. Data from n = 5–13 experiments are depicted as means ± SEM as a function of the ACh or NS9283 concentration, and regression results are presented in Table 2. Data for wild-type receptors from monomeric subunits in Fig. 1 are indicated as dashed lines. (C) Functional 3α:2β stoichiometry receptors arising from coinjections of β-6-α-9-β-6-α and α4VFL could originate from assembly of the concatenated construct in either a clockwise or a counterclockwise orientation when viewed from the synaptic cleft. With respect to NS9283 sensitivity, receptors with the tetramer in the clockwise orientation will resemble wild-type 3α:2β receptors, whereas receptors with the tetramer in the counterclockwise orientation will resemble wild-type 2α:3β receptors. (D and E) ACh (D) and NS9283 (E) CRRs were obtained from oocytes injected with the indicated pentameric constructs, in which x indicates either the α4 or the α4VFL subunit in the fifth construct position. Data from n = 12–19 experiments are depicted as means ± SEM as a function of the ACh or NS9283 concentration, and regression results are presented in Table 2. (F) When injecting the pentameric construct including the α4VFL subunit, clockwise and counterclockwise assemblies lead to different receptors, and NS9283 behaves differentially as described in C. Note that in the receptor illustrations, the first construct linkers are indicated with bold purple font, and specific linker sequences are shown in Table 3. β-6-α-9-β-6-α and α4 As expected, coinjecting cRNA of β-6-α-9-β-6-α and monomeric α4 gave rise to receptors with ACh and NS9283 characteristics matching the wild-type 3α:2β stoichiometry receptors (Fig. 3, A and B; and Table 2). Mean AChmax-evoked currents were in the 1–2-μA range, demonstrating that a tetrameric construct is an efficient way of generating standard binary 3α:2β receptors. β-6-α-9-β-6-α and α4VFL Finally, when β-6-α-9-β-6-α and monomeric α4VFL cRNA were coinjected, the receptor pool appeared similar to that observed with the β-6-α and α4VFL combination, both with respect to their ACh CRR and NS9283 response (Fig. 3, A and B; and Tables 1 and 2). Although the difference in observed NS9283 response with 130% for β-6-α-9-β-6-α and α4VFL versus 210% for β-6-α and α4VFL was significant (t test NS9283 response: P = 0.010), both datasets reveal receptor pools mixed from NS9283-sensitive and NS9283-insensitive receptors. To evaluate whether subpopulations with dangling subunits constitute major contaminants, experiments were conducted using β-6-α-9-β-6-α and α4VFL cRNA ratios of 1:5 and 1:25. This represents a 20–100-fold molar surplus of the α4VFL subunit in the cRNA mixtures. Yet, the data for the three tested ratios were virtually identical to the NS9283 response, ranging from 100 to 130% (Fig. 3 B and Table 2). Thus, receptors formed from only linked tetramers with dangling subunits do not appear to be substantial subpopulations in the β-6-α-9-β-6-α and α4VFL receptor pools. The simplest explanation of the observed result is that linked subunits assemble readily in both clockwise and counterclockwise orientations, resulting in receptors with both NS9283-sensitive α4VFL–α4 and NS9283-insensitive α4–α4VFL interfaces (Fig. 3 C). Because linked tetramers assemble into functional receptors in both the clockwise and counterclockwise orientations, it appears unlikely that adding a fifth subunit would alter this flexibility. To verify this, we made two new pentameric constructs with a wild-type α4 or a mutant α4VFL subunit as the fifth and final subunit. These were termed β-21a-α-β-α-α and β-21a-α-β-α-αVFL, and although not of identical sequence, the linkers largely resemble those in the previously published tetrameric β-6-α-9-β-6-α construct above (Table 3). The chosen nomenclature indicates the number of inserted amino acids (a) between the first two subunits. Table 3. Linker sequences utilized for creating concatenated α4β2 nAChR constructs Linked set Full name C-term. Linker sequence Mature N term. C tail Linker N term. (to L anchor) Total Dimeric constructs β-6-α β2-6-α4 APSSK EG(AGS)6R AHAEE 23 21 6 50 β-9a-α β2-9a-α4 APSSK (AGS)3 AHAEE 23 9 6 38 β-6a-α β2-6a-α4 APSSK (AGS)2 AHAEE 23 6 6 35 β-3a-α β2-3a-α4 APSSK AGS AHAEE 23 3 6 32 β-0a-α β2-0a-α4 HSAPS GS AHAEE 21 2 6 29 β-(-3a)-α β2-(-3a)-α4 HSDHS GS AHAEE 18 2 6 26 Tetrameric and pentameric constructs β1-21a-α2- β21-21a-α42- APSSK (AGS)7 AHAEE 23 21 6 50 β1-3a-α2- β21-3a-α42- APSSK AGS AHAEE 23 3 6 32 -α2-β3- -α42-33a-β23- LAGMI (AGS)5LGS(AGS)5 TDTEE 8 33 6 47 -α2-α3- -α42-33a-α43- LAGMI (AGS)5LGS(AGS)5 AHAEE 8 33 6 47 -β3-α4- -β23-18a-α44- APSSK (AGS)2AGT(AGS)3 AHAEE 23 18 6 47 -α3-β4- -α43-33a-β24- LAGMI (AGS)7ATG(AGS)3 TDTEE 8 33 6 47 -α4-α5 -α44-33a-α45 LAGMI (AGS)4ATG(AGS)6 AHAEE 8 33 6 47 -β4-α5 -β24-20a-α45 APSSK TG(AGS)6 AHAEE 23 20 6 49 The five most proximal C-terminal (C-term.) amino acids and the five first amino acids from the predicted mature N terminus (N term.) are presented for α4 and β2 subunits along with the specific linker sequences introduced during concatenation. C-terminal tail lengths were obtained from NCBI (SwissProt accession nos. P43681 for α4 and P17787 for β2), whereas N-terminal lengths were derived from the definition of an α-helical hydrophobic leucine anchor as described in the Results. For tetrameric and pentameric constructs, numbers indicate the construct position of each subunit (e.g., α2: α4 subunit in the second construct position). β-21a-α-β-α-α With only wild-type subunits in the pentameric construct, the ACh CRR and NS9283 response were, as expected, similar to those observed with α4 and β2 cRNA in a biased ratio (Fig. 3, D and E; and Table 2). The mean AChmax-evoked currents were ∼2 μA. This demonstrates that pentameric constructs can lead to expression levels equal to shorter linked constructs, but this requires increased incubation time after cRNA injection. β-21a-α-β-α-αVFL With the mutated α4VFL as the final subunit in the pentameric complex, the data qualitatively resembled those from both the β-6-α and α4VFL and β-6-α-9-β-6-α and α4VFL combinations above. NS9283 modulation of expected potency was observed in all 19 tested oocytes for β-21a-α-β-α-αVFL, albeit a lower mean NS9283 response of 41% was observed for the pentameric construct compared with 100–130% for the tetrameric construct above. Nevertheless, the presence of NS9283-sensitive receptors suggests that linked pentamers assemble in both clockwise and counterclockwise orientations (Fig. 3 F). The position of the α4–α4 interface within a construct does not influence expression orientation Because injection of a tetrameric construct alone can lead to functional receptors, it remains a possibility that the receptor pools in the β-6-α-9-β-6-α and α4VFL and β-21a-α-β-α-αVFL scenarios above contained small populations of receptors with dangling subunits. With the α4VFL subunit either being a separate entity in the cRNA mixture or the final fifth subunit in the linked construct, it could be envisioned that minor receptor populations do not have this subunit included within their pentameric complex. Minor populations of receptors of only wild-type subunits might lead to an NS9283 response in the observed 41–130% range; this could therefore alter the conclusions. To examine this, we created three new, additional pentameric constructs in which the α4–α4 interface was placed in the construct center: β-21a-α-α-β-α, β-21a-α-αVFL-β-α, and β-21a-αVFL-α-β-α (Table 3). Interestingly, similar constructs were previously found to express very poorly (Carbone et al., 2009); however, no plausible reason for this finding was provided. β-21a-α-α-β-α When the pentameric construct containing only wild-type subunits was expressed in oocytes, the ACh and NS9283 CRRs revealed a receptor pool of archetypical 3α:2β stoichiometry receptors (Fig. 4, A–C; and Table 2). The NS9283 response of 780% for β-21a-α-α-β-α was not significantly different from the 760% observed with receptors obtained from the expression of monomeric subunits using a biased ratio (t test NS9283 response: P = 0.86). The mean maximal current amplitude of 5.3 µA for β-21a-α-α-β-α matched and even surpassed that of previously tested pentameric constructs, contrasting with the observations of Carbone et al. (2009). This indicates that the specific position of the α4–α4 motif within a pentameric construct has no particular impact on surface expression levels. Figure 4. ACh and NS9283 sensitivity and potential stoichiometry of receptors from concatenated pentameric constructs with the α4–α4 site in the second to third construct positions. X. laevis oocytes were subjected to two-electrode voltage-clamp electrophysiology. Electrophysiological data were evaluated as described in Materials and methods; also see Fig. 1. (A and B) ACh (A) and NS9283 (B) CRRs were obtained from oocytes injected with the indicated pentameric constructs, where x and y denote an α4 or an α4VFL subunit in the second and third construct positions. Data from n = 9–23 experiments are depicted as means ± SEM as a function of the ACh or NS9283 concentration, and regression results are presented in Table 2. Data for wild-type receptors from monomeric subunits in Fig. 1 are indicated as dashed lines. (C) Representative traces illustrating NS9283 responses at oocytes injected with β-21a-α-α-β-α, β-21a-α-αVFL-β-α, or β-21a-αVFL-α-β-α. Bars above the traces indicate the 30-s application time and concentrations of applied compounds. (D) For the β-21a-α-αVFL-β-α receptor (α4VFL subunit in the third construct position), NS9283-sensitive receptors originate from assembly of the linkers in a clockwise orientation (top right). However, for the β-21a-αVFL-α-β-α receptor (α4VFL subunit in the second construct position), NS9283-sensitive receptors are assembled with the linkers in a counterclockwise orientation (bottom left). Note that in the receptor illustrations, the first construct linkers are indicated with bold purple font, and specific linker sequences are shown in Table 3. β-21a-α-αVFL-β-α The ACh CRR showed greater resemblance to receptors from monomeric α4VFL and β2 when the pentameric construct contained an α4VFL subunit in the third construct position (Fig. 4 A). Nevertheless, with an NS9283 response of 55%, the receptor pool clearly contained both sensitive and insensitive receptors (Fig. 4, B and C; and Table 2). Given the construct position of the α4VFL subunit, sensitive receptors must have their linkers in a clockwise orientation in this scenario (Fig. 4 D, top). β-21a-αVFL-α-β-α Data for the pentameric construct with the α4VFL subunit in the second position were virtually identical to those observed for α4VFL in the third position, as described in the previous paragraph. The ACh CRR resembled that of monomeric α4VFL and β2 (Fig. 4 A), and an NS9283 response of 85% revealed a mixed receptor pool (Fig. 4, B and C; and Table 2). Because of the construct position of the α4VFL subunit, NS9283-sensitive receptors must have their linkers in the counterclockwise orientation in this scenario (Fig. 4 D, bottom). It is relatively easy to imagine how a pentameric construct could “loop out” the first or the last construct subunit and assemble into functional receptors from two or more pentamers with dangling subunits. However, it is difficult to envision how pentameric constructs with an α4VFL subunit placed in the center could assemble into functional receptors without the inclusion of the α4VFL subunit. All three tested pentameric constructs showed similar NS9283 responses in the range of 41–85%; hence, the collective data demonstrate that linked pentamers readily assemble in both orientations. Is it possible for linkers to direct the assembly orientation of Cys-loop receptors? Given the aforementioned data, it is essential to theoretically assess whether it is at all possible to direct the assembly orientation using subunit concatenation. The linker connects the C-terminal tail extruding from transmembrane helix four (TM4) of the first subunit to the mature N terminus of the second subunit. There appears to be no inherent proficiency of the first subunit to significantly influence the linker direction. Despite having a long C-terminal tail in comparison with other nAChR subunits, a linked β2 tail easily orients itself in both directions. The flexibility of this part of the protein is supported by a lack of electron densities associated with the last 20 amino acids in the recently published α4β2 nAChR x-ray structure (Morales-Perez et al., 2016). Hence, for linking purposes, the C-terminal tail could simply be considered as part of the linker, and any potential control of the assembly orientation would rely on the structural features of the second subunit. The extracellular domain of all Cys-loop subunits has an α-helical segment a few amino acids downstream of the mature N terminus, which points in the clockwise direction when viewed from the extracellular side (Fig. 5 A). The exact length and predicted number of α-helical turns vary between subunits, but the orientation of the helix is well defined. Based on the α4β2 x-ray structure, this helix is firmly attached to the remainder of the subunit through a series of conserved hydrophobic segments. These are termed LLxxLF in α4 and LVxxLL in β2, and both anchor into hydrophobic pockets on the subunits (Fig. 5 B). The hydrophobic patches ensure that the upstream N terminus remains pointed toward the complementary face of the subunit. Based on the 5KXI structure, the direct distance from the last amino acid in transmembrane segment four of β2 (Leu479) to the N-terminal tip of mature α4 (Ala34) in counterclockwise and clockwise directions is 61 and 81 Å, respectively. In reality, the distance on the surface of the protein is longer, so to give a more realistic estimate, we used molecular modeling to estimate the lowest number of AGS repeats required to connect β2 (Leu479) to α4 (Ala34) without causing significant distortions after energy minimization. This modeling approach used the linker as a flexible ruler and estimated that 8 and 10 AGS repeats were sufficient to connect the two terminals in the counterclockwise and clockwise directions (Fig. 5 C). This translates into a proposed difference in distance between the two orientations corresponding to approximately six amino acids or 20 Å. We therefore hypothesized that a clockwise orientation of linked subunits requires a longer minimal linker sequence than the counterclockwise one. This suggests that it is possible to obtain a uniform population of receptors in the counterclockwise orientation once the linker is sufficiently short. Figure 5. 3-D structure of the human α4β2 nAChR (from Protein Data Bank accession no. 5KXI; Morales-Perez et al., 2016) showing direction of N-terminal α-helix and modeled linkers. The α4 subunit is colored green, and β2 is blue. (A) Top view of the (α4)2(β2)3 receptor with the N-terminal helixes shown in red. (B) Top view of the α4β2 dimer with the N-terminal LLxxLF motif shown as red spheres. The surface of the protein constituting hydrophobic residues on the top of the α4 and β2 subunits is colored yellow. The LLxxLF motif interacts with the hydrophobic surface to form a hydrophobic patch. (C) α4β2 dimer with modeled long clockwise and short counterclockwise linkers. The shortest possible number of AGS repeats required to connect the first residue after TM4 of the β2 subunit (Leu479) to the mature N terminus of α4 (Ala34) is shown and had lengths of 10 and 8 repeats, respectively. (D) Illustration of total linker length calculations for new concatenated β-xa-α dimer constructs. Note that only parts of the cDNA sequences close to the linkers are shown. Strategy and design of new concatenated dimer constructs When designing concatenated constructs with defined linker lengths, it is necessary to devise a strategy for calculating existing N- and C-terminal amino acids, as many of these end up as de facto parts of the linker. Calculating the length of the C-terminal tail is a straightforward process, as TM4 is well defined. For the N terminus, however, there is no obvious fix point from which to count “protruding” amino acids. Based on an alignment of all Cys-loop GABAAR and nAChR subunits, we define an N-terminal fix point as the first hydrophobic anchor in the α-helical segment mentioned in the previous paragraph; the conserved leucine corresponds to the first L in L32VEHLL for β2 and L40LKKLF for α4 (Fig. 5 B). The number of amino acids preceding this leucine in the proposed mature nAChR peptides varies from subunit to subunit. It can even vary for a given subunit depending on the predicted signal peptide cleavage. Out of 16 mammalian nAChR subunits, 8 (α1, α3, α7, β1, β2, ε, γ, and δ) have six amino acids preceding the leucine anchor; therefore, it appears reasonable to set six amino acids as a consensus for this receptor class. With this definition, mature β2 starts with T26DTEE and α4 with A34HAEE, which closely matches what was used by Zhou et al. (2003). By strict definition, the linker in the original β-6-α construct is not only six AGS repeats but EG(AGS)6R, as extra amino acids were added for restriction site purposes. Using our calculation method, the total linker length in β-6-α is 50 amino acids, and the linkers used for the new pentameric β-21a-α-β-α-α construct above had a length of 47–50 amino acids (Fig. 5 D and Table 3). To avoid the introduction of charged amino acids when designing new concatenated dimer constructs, we used primarily AGS repeats. Five new constructs were designed with total linker lengths of 38, 35, 32, 29, and 26 amino acids corresponding to the introduction of 9, 6, 3, 0, and −3 amino acids (Fig. 5 D and Table 3). These linker lengths span the predicted distances from the modeling. For clarity, a new nomenclature of β-xa-α was used, where x denotes the number of added amino acids (a). Shortened β-xa-α linkers can lead to a fixed counterclockwise orientation of linked subunits β-9a-α Injection of only β-9a-α cRNA into oocytes led to robust ACh-evoked currents, albeit with a fivefold loss of mean peak current amplitudes from 1,000 to 220 nA when compared with the β-6-α construct above (Table 1). Although the ACh CRR was best approximated by a biphasic equation, the fraction of the first component increased from 0.11 to 0.28, and the NS9283 response decreased from 640 to 100% (Fig. 6, A–C; and Table 1). Collectively, these changes suggest that the receptor pool contained a higher proportion of receptors in the 2α:3β stoichiometry when the dimer linker was shortened. Data from coinjections with α4 or α4VFL cRNA were very similar to the corresponding β-6-α data above, with an NS9283 response at 200% for β-9a-α and α4VFL versus 210% for β-6-α and α4VFL. Hence, the β-9a-α linker was still long enough to allow assembly in both orientations (Fig. 6 D). Figure 6. ACh and NS9283 sensitivity and potential stoichiometry of receptors from concatenated β-xa-α dimer constructs. X. laevis oocytes were subjected to two-electrode voltage-clamp electrophysiology. Electrophysiological data were evaluated as described in Materials and methods; also see Fig. 1. (A and B) ACh (A) and NS9283 (B) CRRs were obtained from oocytes coinjected with β-xa-α and the monomeric α4VFL subunit in a 1:1 ratio. X represents the number of amino acids in the linker, and specific linker sequences are shown in Table 3. Data from n = 5–24 experiments are depicted as means ± SEM as a function of the ACh or NS9283 concentration, and regression results are presented in Table 1. Data for the receptor obtained from monomeric α4VFL and β2 subunits in Fig. 1 are indicated with dashed lines. (C) Representative traces illustrating NS9283 responses at oocytes injected with β-9a-α and α4VFL, β-6a-α and α4VFL, β-3a-α and α4VFL, or β-0a-α and α4VFL. Bars above the traces indicate the 30-s application time and concentrations of applied compounds. (D) Based on the lack of NS9283 efficacy observed in B, receptors originating from coinjection of β-0a-α with the monomeric α4VFL subunit have the concatenated construct assembled exclusively in the counterclockwise orientation when viewed from the synaptic cleft. β-6a-α The most noticeable impact observed from shortening the linker by one AGS repeat was fewer oocytes responding robustly to ACh applications when only cRNA for β-6a-α was injected (Table 1). Although the ACh CRR for β-6a-α and α4VFL did display closer visual resemblance to that of α4VFL and β2 from monomeric subunits, the NS9283 response was still substantial at 140% (Fig. 6, A–C; and Table 1), demonstrating assembly in both orientations (Fig. 6 D). β-3a-α In contrast, further shortening by one AGS repeat to create the β-3a-α construct yielded significantly different data. When cRNA for β-3a-α was injected alone, only one out of four oocytes responded to ACh applications with low current amplitudes. Although coinjections with α4VFL yielded receptors with robust ACh-evoked currents, these displayed low responsiveness to NS9283 with a response of 22% (Fig. 6, A–C; and Table 1). This indicates that the receptor pool contained linked dimers primarily, but not exclusively, in the counterclockwise orientation (Fig. 6 D). Importantly, this preference was accomplished without a detrimental loss of ACh-evoked current amplitudes or changes in the NS9283 response of the 3α:2β receptors obtained by coinjection of β-3a-α and α4 (Table 1). β-0a-α Additional shortening of the linker to create the β-0a-α construct resulted in a substantial loss of maximal ACh-evoked current amplitudes for all three tested cRNA mixtures. Therefore, additional incubation of the oocytes for 2–3 d was necessary to obtain current amplitudes >100 nA for an AChmax application. Nevertheless, the ACh and NS9283 responsiveness of oocytes coinjected with β-0a-α and α4 appeared identical to that of wild-type 3α:2β stoichiometry receptors, indicating that the linked dimer retains normal functionality. Remarkably, coinjection of β-0a-α and α4VFL yielded receptors that appeared identical to those from monomeric α4VFL and β2 subunits in a biased ratio (Fig. 6, A–C; and Table 1). No hint of NS9283 enhancement was noted in individual oocytes (n = 15), which demonstrated an exclusive counterclockwise dimer orientation (Fig. 6 D). β-(-3a)-α As expected based on the data for the β-0a-α construct, removing an additional three amino acids from the linker to create the β-(−3a)-α construct aggravated issues with current amplitudes. Although ACh-evoked currents were observed (Table 1), they rarely approached the amplitudes necessary to perform full CRRs for ACh and NS9283 (set at >100 nA for an initial AChmax application). Thus, when the total linker length in a dimer construct is shortened to 32 amino acids or less (Table 3), the resulting receptor pool predominantly contains receptors assembled from dimers in the counterclockwise orientation. This proves the hypothesis that a counterclockwise orientation represents the shortest distance for a linked construct. However, the data also indicate that obtainment of exclusive counterclockwise expression comes at a cost of low current amplitudes. For practical purposes, it may be advantageous to accept the risk of a small polluting receptor population. Pentameric constructs with one short linker express primarily in the counterclockwise orientation Given that a shortened dimer construct expresses predominantly in the counterclockwise orientation, the same should theoretically also be applicable for pentameric constructs. Two points are worth noting in this context. First, when creating concatenated constructs of more than two subunits, it is logical to assume that if the first linker directs the orientation, subsequent linked subunits can only extend in the same direction. The used linkers are simply not long enough to bridge across several subunits, as would be required for altering direction. Second, if a linker is optimized to ensure only one assembly direction, it is likely to be tightly packed to the bridging subunits. This implies that a linker could pack tightly across the C loop of an agonist-binding interface, which intuitively should be avoided, as it may influence normal binding and function of the receptor. Therefore, when designing new constructs, we chose to use a total length of 32 amino acids for the first linker and keep the length of consecutive linkers at 47 amino acids (Table 3). Although a 32–amino acid linker did not lead to exclusive counterclockwise expression with the β-3a-α construct, it was deemed the best compromise, as pentameric constructs can be expected to give overall lower current amplitudes. The two new constructs were β-3a-α-β-α-α and β-3a-α-β-α-αVFL. β-3a-α-β-α-α When cRNA for β-3a-α-β-α-α was injected into oocytes, the resulting receptors appeared identical to those of wild-type 3α:2β stoichiometry from monomeric α4 and β2 in a biased ratio. With an NS9283 response of 640%, no significant difference (t test NS9283 response: P = 0.36) was noted, compared with the 760% observed at 3α:2β stoichiometry receptors (Fig. 7, A and B; and Table 2). Figure 7. ACh and NS9283 sensitivity and potential stoichiometry of receptors from concatenated pentameric constructs with short first linkers. X. laevis oocytes were subjected to two-electrode voltage-clamp electrophysiology. Electrophysiological data were evaluated as described in Materials and methods; also see Fig. 1. (A and B) ACh (A) and NS9283 (B) CRRs were obtained from oocytes injected with the indicated pentameric constructs in which x denotes an α4 or an α4VFL subunit in the fifth construct position. Data from n = 5–15 experiments are depicted as means ± SEM as a function of the ACh or NS9283 concentration, and regression results are presented in Table 2. Data for wild-type receptors from monomeric subunits in Fig. 1 are indicated as dashed lines. (C) Representative traces illustrating NS9283 responses at oocytes injected with β-3a-α-β-α-α (top) or β-3a-α-β-α-αVFL (bottom). Bars above the traces indicate the 30-s application time and concentrations of applied compounds. (D) Based on the NS9283 responses observed with x = α4VFL in B, the receptor pool consists mainly of pentamers with the linkers assembled in the counterclockwise orientation when viewed from the synaptic cleft. Note that in the receptor illustrations, the first construct linkers are indicated with bold purple font, and specific linker sequences are shown in Table 3. β-3a-α-β-α-αVFL In contrast, receptors in oocytes injected with the β-3a-α-β-α-αVFL construct resembled those from monomeric α4VFL and β2. The mean NS9283 response was 11%; however, it is noteworthy that no effect was observed in ∼25% of the oocytes. This indicates that the receptor pool for β-3a-α-β-α-αVFL is dominated by receptors expressed in the counterclockwise orientation. Substituting an α4 with an α4VFL subunit in each position in an (α4)3(β2)2 receptor With a methodology for expressing pentamers in a preferred orientation, it becomes possible to evaluate the effects of introducing an α4VFL subunit in each of the three possible positions in a 3α:2β stoichiometry receptor. Data for α4VFL in the complementary side position of the α4–α4VFL interface are addressed by the β-3a-α-β-α-αVFL construct in the previous paragraph. To investigate the other two positions, two new pentameric constructs were made: β-3a-αVFL-β-α-α (wild-type α4–α4 interface) and β-3a-αVFL-α-β-α (α4VFL–α4 interface). β-3a-αVFL-β-α-α When the α4VFL subunit is placed between two β2 subunits, the three mutations are located in the complementary side of a β2–α4VFL interface, and because of the construct symmetry, this is not dependent on expression orientation. As β2–α4 interfaces are not believed to bind ACh, it would be expected that the mutations have no effect on an ACh CRR. If this was the case, the ACh CRR should look similar to that of wild-type 3α:2β stoichiometry receptors; however, as first noted by Lucero et al. (2016), this is in fact not the case. Instead, the ACh CRR visually appeared more similar to that of α4VFL and β2, albeit the fitted parameters for EC50 values and fractions were not identical (Fig. 8 A and Table 2). How mutations in a nonbinding interface can exert such an effect is presently unclear. Nevertheless, the β-3a-αVFL-β-α-α receptors remained sensitive to NS9283 binding in the α4–α4 interface (Fig. 8 B and Table 2). The observed response of 250% could seem low compared with wild-type 3α:2β stoichiometry receptors, but this is to be expected because a fixed AChcontrol concentration of 10 µM was used in all experiments (~EC40 for β-3a-αVFL-β-α-α receptors vs. ~EC15 for α4 and β2). Figure 8. ACh and NS9283 sensitivity of receptors from α4VFL subunit containing concatenated pentameric constructs with short first linkers. X. laevis oocytes were subjected to two-electrode voltage-clamp electrophysiology. Electrophysiological data were evaluated as described in Materials and methods; also see Fig. 1. (A and B) ACh (A) and NS9283 (B) CRRs were obtained from oocytes injected with the indicated pentameric constructs. Data from n = 9–15 experiments are depicted as means ± SEM as a function of the ACh or NS9283 concentration, and regression results are presented in Table 2. Data for the β-3a-α-β-α-αVFL construct are from Fig. 7, and data for wild-type receptors from monomeric subunits in Fig. 1 are indicated as dashed lines. The preferred expression orientation of each pentamer is indicated for the NS9283 data. (C) Representative traces illustrating NS9283 responses at oocytes injected with β-3a-αVFL-β-α-α or β-3a-αVFL-α-β-α. Bars above the traces indicate the 30-s application time and concentrations of applied compounds. β-3a-αVFL-α-β-α Finally, with the construct in which the preferred counterclockwise orientation leads to α4VFL–α4 interfaces, the ACh CRR visually most resembled that of wild-type 3α:2β stoichiometry receptors (Fig. 8 A). That said, the fitted parameters were not identical, and in particular, the fraction of the first component of the curve was almost threefold higher (Table 2). Presumably, this was because of the presence of a small population of receptors expressed in the clockwise orientation, which showed higher sensitivity to low ACh concentrations. With an observed NS9283 response of 270%, the receptors responded less than wild-type 3α:2β stoichiometry receptors (Fig. 8 B and Table 2). There could be two reasons for this: (1) the presence of a population of receptors that did not respond to NS9283, or (2) the α4VFL–α4 interface might not have responded to NS9283 fully like a wild-type α4–α4 interface does. The higher first component fraction of the ACh CRR and the fact that NS9283 displayed normal potency in the α4VFL–α4 interface suggest that the first reason is more likely. Discussion For heteromeric Cys-loop receptors, the method of subunit concatenation holds great promise. Theoretically, this technique allows for detailed experimental control at the single-subunit level that would not otherwise be possible. In reality, when working with α4β2 nAChR–concatenated constructs based on published designs, we observed that the resulting receptor pools were not always uniform. Although the technique worked superbly for wild-type binary receptors, this was not the case for ternary receptors (including binary receptors with mutations in one of the subunits, which technically make them ternary receptors). In this study, we therefore investigated to what degree a concatenation strategy can be used to direct subunit assembly and what is required for this to work reliably. The β-6-α construct first published by Zhou et al. (2003) exemplifies a case in which ternary receptors can convolute the data. Coexpression with a monomeric α4VFL subunit clearly resulted in a mixed receptor pool, as evidenced from the response to NS9283. The most plausible explanation is that receptor subpopulations arose from dimer assembly in clockwise, counterclockwise, or both directions. Assuming no linker direction bias, this could lead to a receptor pool containing four different ternary receptors. The exact percentage of these different receptor types is not easily established; however, the data support the notion of at least two abundant receptor subpopulations. Although it might be speculated that increasing the number of linked subunits in the cDNA construct would alleviate the problem, this was not the case. NS9283 data for the previously published (Carbone et al., 2009; Harpsøe et al., 2011) tetrameric β-6-α-9-β-6-α construct coexpressed with a monomeric α4VFL subunit or a similarly designed fully pentameric β-21a-α-β-α-αVFL construct still revealed mixed receptor pools. Furthermore, new pentameric constructs with the α4VFL subunit positioned in the construct center (β-21a-α-αVFL-β-α and β-21a-αVFL-α-β-α) also led to mixed receptor pools. Hence, it can be concluded that typically used concatenated α4β2 constructs do not ensure expression of pure ternary receptor pools, as the linked subunits can assemble readily in both orientations. When evaluating models of Cys-loop receptors, it is clear that the C-terminal tail is a flexible entity with no specific features that are likely to significantly direct subunit assembly orientation. This is in good agreement with data for the β-6-α construct, and for linking purposes, it is therefore sensible to consider the C-terminal tail merely as part of the linker sequence. Although it is easy to determine the length of a C-terminal tail based on TM4, it is less straightforward to predict the length of a corresponding N-terminal “protrusion.” We defined the first amino acid of mature nAChR subunits and total linker lengths based on a hydrophobic anchor in the α-helical segment located at the top of all Cys-loop receptor subunits. In this context, perhaps the most important feature of this α-helical segment was the fact that it was pointing in a clockwise direction when viewed from the synaptic cleft. This brought the N terminus toward the subunit complementary face and caused a difference in length between a linker bridging the TM4 of one subunit to the N terminus of the next subunit in the clockwise versus the counterclockwise direction. Our calculations suggested that the clockwise linker should be approximately six amino acids longer than the counterclockwise one. Hence, we hypothesized that a uniform receptor pool with concatenated constructs expressing only in the counterclockwise orientation is attainable when the linkers are short. To evaluate whether a shorter linker would direct subunit assembly, we decreased the total calculated linker length of dimer constructs progressively from 38 down to 26 amino acids, 3 amino acids at a time. Based on this, it appeared that a linker of 35 amino acids was too long (substantial NS9283 response), whereas a linker of 26 amino acids was too short (peak current amplitudes that were too low). The “sweet spot” appeared to be a linker length of ∼32 amino acids. The β-3a-α dimer showed a low NS9283 response when coexpressed with the α4VFL subunit, but showed wild-type 3α:2β stoichiometry behavior when coexpressed with the α4 subunit. This demonstrates that the receptor pool has dimers primarily oriented in the counterclockwise direction. Although a small subpopulation with dimers in the clockwise orientation was present, this appeared sufficiently negligible for most practical purposes. Hence, our hypothesis that an optimized short linker can direct dimer subunit assembly was proven experimentally. Having discovered that 32 amino acids represent a good compromise for a dimeric construct linker length, we next evaluated whether this would also be applicable for pentameric constructs. Three constructs were made in which a single α4VFL subunit replaced a wild-type α4 subunit in each of the three possible locations in the 3α:2β stoichiometry receptor. Importantly, a low NS9283 response (11%) was observed at the receptor containing an α4–α4 VFL interface versus a substantial response (270%) at the receptor with an α4VFL–α4 interface. This demonstrates that pentameric constructs, like their dimeric counterparts, express in a preferred counterclockwise orientation when the first construct linker is 32 amino acids. Again, it is important to note that although this linker length might represent a good compromise, it did not ensure a 100% uniform receptor population. When considering the experimental data for the shortened dimeric constructs, it appears that the actual difference between a linker traversing the clockwise versus the counterclockwise orientation was less than what had been predicted from the 3-D x-ray structures. Exclusive counterclockwise orientation was only observed with finely optimized short linkers, and the margin between exclusivity and loss of function appeared narrow in the order of a few amino acids. Hence, the models likely overestimated the real distance difference, most likely because of the flexibility of TM4 and potential unwinding of the helices, which had not been observed in the experimentally determined structure. The consequences of these findings are substantial. The β-6-α construct contains a total linker length of 50 amino acids (using our calculation method), and similar linkers have been used to design various concatenated constructs in laboratories across the world. This raises fundamental questions as to the validity of published data whenever resulting receptors are of a ternary nature. There are many examples in which this was the case: (a) concatenated dimeric or pentameric constructs were used to show that the α5 subunit can occupy two positions in the (α4β2)2α5 receptor, either as a nonbinding subunit or as a replacement of one β2 subunit (Jin et al., 2014); (b) by introducing point mutations and expressing a dimer construct with monomeric subunits, Jain et al. (2016) concluded that both α5 and β3 can participate in what was termed “unorthodox ACh-binding sites”; (c) pentameric constructs were used to show that the role or equality of subunits differ, depending on their specific positions within the α4β2 receptor complex (Mazzaferro et al., 2014); and (d) by introducing point mutations in the β2 subunit, Lucero et al. (2016) concluded that the two canonical α4–β2 ACh agonist sites make significantly different contributions to receptor activation for α4β2 receptors in the 2α:3β (but not the 3α:2β) stoichiometry. However, all of these studies correspond to ternary receptor scenarios, and in the absence of strong evidence regarding linker orientation, the respective conclusions may be erroneous. Given the extensive use of the concatenation technology, it is puzzling that the complexity of this technique is not widely recognized. One explanation for this lies in the data resolution with which many constructs have been studied. Fundamentally, ACh CRRs, biphasic ones in particular, rarely yield data of sufficient resolution to determine whether a receptor pool is truly uniform. Another explanation relates to the prior use of ambiguous data to draw conclusions. Zhou et al. (2003) attempted to determine the assembly orientation of their constructs by coexpression with the β4 subunit and evaluating for a cytisine response. They concluded that linked β-6-α dimers express in a clockwise orientation; however, this relied on speculation regarding expected cytisine responses from receptors with mixed α4–β2 and α4–β4 agonist interfaces. Therefore, to ensure sufficient resolution and unambiguous data, we relied on the site-selective agonism of NS9283. At the highest tested concentrations of NS9283 (31.6 µM), an EC15 ACh-evoked current is increased by ∼700%. This allows for robust detection of even a low percentage of responding receptors in pools of primarily nonresponding receptors. Prior observations with GABAARs support that the findings observed in this study are likely to be universal to all Cys-loop receptors. Baumann et al. (2001) elegantly applied concatenated receptors to answer fundamental questions regarding the stoichiometry of an α1β2γ2 receptor. Although their linker lengths were shorter than typical nAChR linkers (37–41 vs. 50 amino acids using our calculation method), occurrence of GABA-evoked currents with some construct combinations (e.g., αβ/γ) can only be explained by dimers expressing in both orientations. Hence, for GABAARs, similar considerations will have to be considered when evaluating data. Conclusion In the present study, we demonstrate that typically used linker lengths in concatenated nAChR constructs do not ensure uniform receptor expression of ternary receptors. This is because the linked subunits can orient themselves readily in both the clockwise and the counterclockwise directions. For ternary or more complex scenarios, this leads to receptor pools containing mixed stoichiometries. By shortening construct linkers, it is possible to use concatenation to ensure expression of receptor pools with a preference toward subunits in a counterclockwise orientation. However, substantial optimization is required to obtain exclusive expression for a given set of subunits. This is because linker length changes of only a few angstroms can make a substantial difference. Consequently, it is of the utmost importance to carefully assess the used concatenated constructs when evaluating the validity of historical data or when designing new experiments. Acknowledgments We wish to thank Troels E. Sørensen for insightful discussions. This work was conducted as part of an Australian Research Council Linkage project in collaboration with Bionomics Limited. This work was supported by Bionomics Limited, the Australian Research Council (grant LP140100781), and the Australian National Health and Medical Research Council (grant APP1069417). The authors declare no competing financial interests. Author contributions: P.K. Ahring designed the research. V.W.Y. Liao, P.K. Ahring, and T. Balle performed the research. P.K. Ahring analyzed the data and drafted the manuscript. All authors approved the final version of the manuscript. Richard W. Aldrich served as editor. ==== Refs Ahring, P.K., L.H. Bang, M.L. Jensen, D. Strøbæk, L.Y. Hartiadi, M. Chebib, and N. Absalom. 2016. A pharmacological assessment of agonists and modulators at α4β2γ2 and α4β2δ GABAA receptors: The challenge in comparing apples with oranges. Pharmacol. Res. 111 :563–576. 10.1016/j.phrs.2016.05.014 27178730 Baumann, S.W., R. Baur, and E. Sigel. 2001. Subunit arrangement of gamma-aminobutyric acid type A receptors. J. Biol. Chem. 276 :36275–36280. 10.1074/jbc.M105240200 11466317 Baumann, S.W., R. Baur, and E. Sigel. 2002. Forced subunit assembly in alpha1beta2gamma2 GABAA receptors. Insight into the absolute arrangement. J. Biol. Chem. 277 :46020–46025. 10.1074/jbc.M207663200 12324466 Baur, R., F. Minier, and E. Sigel. 2006. A GABA(A) receptor of defined subunit composition and positioning: concatenation of five subunits. FEBS Lett. 580 :1616–1620. 10.1016/j.febslet.2006.02.002 16494876 Berman, H.M., T. Battistuz, T.N. Bhat, W.F. Bluhm, P.E. Bourne, K. Burkhardt, Z. Feng, G.L. Gilliland, L. Iype, S. Jain, 2002. The Protein Data Bank. Acta Crystallogr. D Biol. Crystallogr. 58 :899–907. 10.1107/S0907444902003451 12037327 Carbone, A.L., M. Moroni, P.J. Groot-Kormelink, and I. Bermudez. 2009. Pentameric concatenated (alpha4)(2)(beta2)(3) and (alpha4)(3)(beta2)(2) nicotinic acetylcholine receptors: subunit arrangement determines functional expression. Br. J. Pharmacol. 156 :970–981. 10.1111/j.1476-5381.2008.00104.x 19366353 Groot-Kormelink, P.J., S.D. Broadbent, J.P. Boorman, and L.G. Sivilotti. 2004. Incomplete incorporation of tandem subunits in recombinant neuronal nicotinic receptors. J. Gen. Physiol. 123 :697–708. 10.1085/jgp.200409042 15148328 Groot-Kormelink, P.J., S. Broadbent, M. Beato, and L.G. Sivilotti. 2006. Constraining the expression of nicotinic acetylcholine receptors by using pentameric constructs. Mol. Pharmacol. 69 :558–563. 10.1124/mol.105.019356 16269534 Harpsøe, K., P.K. Ahring, J.K. Christensen, M.L. Jensen, D. Peters, and T. Balle. 2011. Unraveling the high- and low-sensitivity agonist responses of nicotinic acetylcholine receptors. J. Neurosci. 31 :10759–10766. 10.1523/JNEUROSCI.1509-11.2011 21795528 Indurthi, D.C., T.M. Lewis, P.K. Ahring, T. Balle, M. Chebib, and N.L. Absalom. 2016. Ligand Binding at the 4-4 Agonist-Binding Site of the 42 nAChR Triggers Receptor Activation through a Pre-Activated Conformational State. PLoS One. 11 :e0161154. 10.1371/journal.pone.0161154 27552221 Jain, A., A. Kuryatov, J. Wang, T.M. Kamenecka, and J. Lindstrom. 2016. Unorthodox Acetylcholine Binding Sites Formed by α5 and β3 Accessory Subunits in α4β2* Nicotinic Acetylcholine Receptors. J. Biol. Chem. 291 :23452–23463. 10.1074/jbc.M116.749150 27645992 Jin, X., and J.H. Steinbach. 2011. A portable site: a binding element for 17β-estradiol can be placed on any subunit of a nicotinic α4β2 receptor. J. Neurosci. 31 :5045–5054. 10.1523/JNEUROSCI.4802-10.2011 21451042 Jin, X., I. Bermudez, and J.H. Steinbach. 2014. The nicotinic α5 subunit can replace either an acetylcholine-binding or nonbinding subunit in the α4β2* neuronal nicotinic receptor. Mol. Pharmacol. 85 :11–17. 10.1124/mol.113.089979 24184962 Kaur, K.H., R. Baur, and E. Sigel. 2009. Unanticipated structural and functional properties of delta-subunit-containing GABAA receptors. J. Biol. Chem. 284 :7889–7896. 10.1074/jbc.M806484200 19141615 Kuryatov, A., and J. Lindstrom. 2011. Expression of functional human α6β2β3* acetylcholine receptors in Xenopus laevis oocytes achieved through subunit chimeras and concatamers. Mol. Pharmacol. 79 :126–140. 10.1124/mol.110.066159 20923852 Lucero, L.M., M.M. Weltzin, J.B. Eaton, J.F. Cooper, J.M. Lindstrom, R.J. Lukas, and P. Whiteaker. 2016. Differential alpha4(+)/(-)beta2 Agonist-binding Site Contributions to alpha4beta2 Nicotinic Acetylcholine Receptor Function within and between Isoforms. J. Biol. Chem. 291 :2444–2459. 10.1074/jbc.M115.684373 26644472 Mazzaferro, S., N. Benallegue, A. Carbone, F. Gasparri, R. Vijayan, P.C. Biggin, M. Moroni, and I. Bermudez. 2011. Additional acetylcholine (ACh) binding site at alpha4/alpha4 interface of (alpha4beta2)2alpha4 nicotinic receptor influences agonist sensitivity. J. Biol. Chem. 286 :31043–31054. 10.1074/jbc.M111.262014 21757735 Mazzaferro, S., F. Gasparri, K. New, C. Alcaino, M. Faundez, P. Iturriaga Vasquez, R. Vijayan, P.C. Biggin, and I. Bermudez. 2014. Non-equivalent ligand selectivity of agonist sites in (α4β2)2α4 nicotinic acetylcholine receptors: a key determinant of agonist efficacy. J. Biol. Chem. 289 :21795–21806. 10.1074/jbc.M114.555136 24936069 Mirza, N.R., J.S. Larsen, C. Mathiasen, T.A. Jacobsen, G. Munro, H.K. Erichsen, A.N. Nielsen, K.B. Troelsen, E.O. Nielsen, and P.K. Ahring. 2008. NS11394 [3′-[5-(1-hydroxy-1-methyl-ethyl)-benzoimidazol-1-yl]-biphenyl-2-carbonitrile], a unique subtype-selective GABAA receptor positive allosteric modulator: in vitro actions, pharmacokinetic properties and in vivo anxiolytic efficacy. J. Pharmacol. Exp. Ther. 327 :954–968. 10.1124/jpet.108.138859 18791063 Morales-Perez, C.L., C.M. Noviello, and R.E. Hibbs. 2016. X-ray structure of the human α4β2 nicotinic receptor. Nature. 538 :411–415. 10.1038/nature19785 27698419 Olsen, J.A., J.S. Kastrup, D. Peters, M. Gajhede, T. Balle, and P.K. Ahring. 2013. Two distinct allosteric binding sites at α4β2 nicotinic acetylcholine receptors revealed by NS206 and NS9283 give unique insights to binding activity-associated linkage at Cys-loop receptors. J. Biol. Chem. 288 :35997–36006. 10.1074/jbc.M113.498618 24169695 Olsen, J.A., P.K. Ahring, J.S. Kastrup, M. Gajhede, and T. Balle. 2014. Structural and functional studies of the modulator NS9283 reveal agonist-like mechanism of action at α4β2 nicotinic acetylcholine receptors. J. Biol. Chem. 289 :24911–24921. 10.1074/jbc.M114.568097 24982426 Shu, H.J., J. Bracamontes, A. Taylor, K. Wu, M.M. Eaton, G. Akk, B. Manion, A.S. Evers, K. Krishnan, D.F. Covey, 2012. Characteristics of concatemeric GABA(A) receptors containing α4/δ subunits expressed in Xenopus oocytes. Br. J. Pharmacol. 165 :2228–2243. 10.1111/j.1476-5381.2011.01690.x 21950777 Timmermann, D.B., K. Sandager-Nielsen, T. Dyhring, M. Smith, A.M. Jacobsen, E.O. Nielsen, M. Grunnet, J.K. Christensen, D. Peters, K. Kohlhaas, 2012. Augmentation of cognitive function by NS9283, a stoichiometry-dependent positive allosteric modulator of α2- and α4-containing nicotinic acetylcholine receptors. Br. J. Pharmacol. 167 :164–182. 10.1111/j.1476-5381.2012.01989.x 22506660 Zhou, Y., M.E. Nelson, A. Kuryatov, C. Choi, J. Cooper, and J. Lindstrom. 2003. Human alpha4beta2 acetylcholine receptors formed from linked subunits. J. Neurosci. 23 :9004–9015 (PubMed).14534234 Zwart, R., and H.P. Vijverberg. 1998. Four pharmacologically distinct subtypes of alpha4beta2 nicotinic acetylcholine receptor expressed in Xenopus laevis oocytes. Mol. Pharmacol. 54 :1124–1131. 10.1124/mol.54.6.1124 9855643

PMC005xxxxxx/PMC5839719.txt

==== Front J Gen Physiol J Gen Physiol jgp jgp The Journal of General Physiology 0022-1295 1540-7748 Rockefeller University Press 29453293 201711880 10.1085/jgp.201711880 Research Articles Communication 511 503 501 Protein phosphorylation maintains the normal function of cloned human Cav2.3 channels Phosphorylation prevents Cav2.3 channel run-down http://orcid.org/0000-0002-6376-6391 Neumaier Felix Alpdogan Serdar Hescheler Jürgen http://orcid.org/0000-0003-2816-2696 Schneider Toni Institute for Neurophysiology, University of Cologne, Cologne, Germany Correspondence to Felix Neumaier: felix@neumaier-net.de 05 3 2018 150 3 491510 14 8 2017 22 12 2017 24 1 2018 © 2018 Neumaier et al. 2018 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). Cav2.3 Ca2+ channels are subject to cytosolic regulation, which has been difficult to characterize in native cells. Neumaier et al. demonstrate the role of phosphorylation in the function of these channels and suggest a close relationship between voltage dependence and the phosphorylation state. R-type currents mediated by native and recombinant Cav2.3 voltage-gated Ca2+ channels (VGCCs) exhibit facilitation (run-up) and subsequent decline (run-down) in whole-cell patch-clamp recordings. A better understanding of the two processes could provide insight into constitutive modulation of the channels in intact cells, but low expression levels and the need for pharmacological isolation have prevented investigations in native systems. Here, to circumvent these limitations, we use conventional and perforated-patch-clamp recordings in a recombinant expression system, which allows us to study the effects of cell dialysis in a reproducible manner. We show that the decline of currents carried by human Cav2.3+β3 channel subunits during run-down is related to adenosine triphosphate (ATP) depletion, which reduces the number of functional channels and leads to a progressive shift of voltage-dependent gating to more negative potentials. Both effects can be counteracted by hydrolysable ATP, whose protective action is almost completely prevented by inhibition of serine/threonine but not tyrosine or lipid kinases. Protein kinase inhibition also mimics the effects of run-down in intact cells, reduces the peak current density, and hyperpolarizes the voltage dependence of gating. Together, our findings indicate that ATP promotes phosphorylation of either the channel or an associated protein, whereas dephosphorylation during cell dialysis results in run-down. These data also distinguish the effects of ATP on Cav2.3 channels from those on other VGCCs because neither direct nucleotide binding nor PIP2 synthesis is required for protection from run-down. We conclude that protein phosphorylation is required for Cav2.3 channel function and could directly influence the normal features of current carried by these channels. Curiously, some of our findings also point to a role for leupeptin-sensitive proteases in run-up and possibly ATP protection from run-down. As such, the present study provides a reliable baseline for further studies on Cav2.3 channel regulation by protein kinases, phosphatases, and possibly proteases. Deutsche Forschungsgemeinschaft https://doi.org/10.13039/501100001659 SCHN 387/21-1 ==== Body pmcIntroduction Electrophysiological recordings from excised cell patches or dialyzed cells are almost invariably hampered by time-dependent changes in voltage-gated ion channel function. The most common form of these phenomena, termed run-down or washout, is a progressive decline of ionic currents and is thought to reflect changes in intracellular signaling cascades, which occur secondary to the loss or dilution of cytosolic factors (Becq, 1996). It can be preceded by a transient current facilitation (run-up), which may reflect voltage- and time-dependent repriming (i.e., recovery from inactivation) or modification of signaling cascades that tonically inhibit these currents (Tiaho et al., 1993; Elhamdani et al., 1994, 1995). Although run-down remains a major obstacle for studies on voltage-gated Ca2+ channel (VGCC) function, it has also provided insight into the manifold regulation of these channels in intact cells. For example, the decline of L-type Ca2+ currents has been linked to several interrelated processes, which may include loss of ATP and other cytoplasmic factors, progressive protein dephosphorylation, decoupling of guanosine-5′-triphosphate (GTP)–binding proteins, and possibly increased proteolysis of the channels (Chad et al., 1987; McDonald et al., 1994; Kepplinger and Romanin, 2005; Xu et al., 2016; Yu et al., 2016). In P/Q-, N-, and certain neuronal L-type Ca2+ channels on the other hand, run-down appears to involve depletion of membrane PIP2, a mechanism also thought to mediate M1 muscarinic receptor-dependent inhibition of these channels (Wu et al., 2002; Suh et al., 2010). Much less is known about the run-down of “pharmaco-resistant” R-type currents, which are mainly mediated by Cav2.3-type VGCCs. R-type and R-type–like currents have been shown to exhibit both run-up and run-down (Cota, 1986; Hilaire et al., 1997; Benquet et al., 1999; Almog and Korngreen, 2009), but low expression levels and the need for pharmacological isolation have generally prevented further characterization of the two processes in native cells. The human embryonic kidney (HEK-293) cell line is widely used for heterologous expression of recombinant ion channels and receptors because it contains few endogenous channels, whereas most signaling pathways for regulation and posttranslational processing are operational (Toth et al., 1996; Thomas and Smart, 2005; Clare, 2006). Apart from circumventing the need for R-type current isolation, HEK-293 cells have a simple and uniform shape, which facilitates reproducible manipulation of their intracellular milieu. We therefore used conventional and perforated-patch-clamp recordings together with different inhibitors and cytosolic factors to study the effects of cell dialysis in a stably transfected HEK-293 cell line expressing human Cav2.3+β3 channel subunits. Our findings show that the decline of macroscopic currents during run-down can partly be accounted for by changes in channel voltage dependence and that it can be prevented or slowed down by provision of intracellular ATP and in perforated-patch recordings. Protection from run-down depended on ATP-hydrolysis and was not related to lipid kinase-mediated PIP2 synthesis or phosphorylation of tyrosine residues but was sensitive to inhibition of serine/threonine kinases. Protein kinase inhibition in intact cells also reduced peak current densities and reproduced the effects of run-down on channel voltage-dependence. Together, these findings indicate that run-down involves constitutive dephosphorylation of sites on the channels themselves or an associated protein and that ATP promotes phosphorylation of these sites by one or more endogenous kinases. Interestingly, our findings also indicate that the current facilitation during run-up could involve activation of leupeptin (Leu)-sensitive proteases, which may also influence the protective action of ATP. Materials and methods Cell culture Human embryonic kidney (HEK-293) cells stably transfected with human Cav2.3d and β3 Ca2+ channel subunits (Nakashima et al., 1998) were cultured under normal growth conditions (37°C and 5% CO2) in Dulbecco’s modified Eagle medium (DMEM; Sigma-Aldrich) supplemented with 10% FCS and antibiotics (1% penicillin–streptomycin and selection markers: 1 mg/ml geneticin [G-418] and 200 µg/ml hygromycin B). Cells were routinely passaged twice a week by using 0.05% trypsin/0.02% EDTA. For electrophysiological recordings, cells were seeded on nitric acid–washed glass coverslips and used within 24–48 h after plating. Electrophysiological recordings Cells were voltage-clamped by using the whole-cell configuration of the patch-clamp technique (Hamill et al., 1981). Pipettes were prepared from thick-walled borosilicate glass capillaries (1.5/0.84-mm-outside/inside diameter; World Precision Instruments) by using a P97 Micropipette puller (Sutter Instruments). Resistance of the resulting electrodes was between 1.5 and 6.5 MΩ (mean = 3.8 ± 0.1 MΩ in 339 recordings) when filled with standard internal solution. The bath was connected to ground via 140 mM sodium chloride agar bridges. Currents were sampled at 20 or 50 kHz and filtered at 10 kHz by using an EPC9 amplifier (HEKA) controlled with HEKA’s Pulse software. Leak and capacitive currents were subtracted online by use of a −P/5 protocol. Recordings obtained with voltage ramps were leak-corrected by a linear fit to the current recorded between −80 and −50 mV, which was extrapolated to +60 mV and subtracted from the whole recording. Series resistance (Rs) was compensated electronically by up to 90% (mean Rs after compensation = 2.7 ± 0.1 MΩ) and continuously monitored throughout the measurements. The charging time-constant after compensation was always ≤100 µs (mean = 38 ± 1 µs), and the maximum uncompensated Rs-error at the time of peak current was ≤2.5 mV (mean = 0.82 ± 0.04 mV). All experiments were performed at room temperature, from a holding potential of −80 mV and, unless noted otherwise, in cells with a whole-cell capacitance (Cs) between 5 and 25 pF (mean = 15 ± 1 pF), as estimated from the slow capacitance compensation of the amplifier. Perforated patch recordings were performed by using β-escin, which was dissolved in type I ultrapure water to prepare a 25 mM stock solution, protected from light and stored at −20°C for up to 2 wk. Before the recordings, aliquots of the stock solution were diluted in standard internal solution and vortex-mixed for 1 min to give a final escin concentration of 40 µM. After patch formation, gradual perforation was observed as a progressive increase in the speed and amplitude of capacitive transients during 5-ms voltage-steps to −75 mV. Cells showing signs of spontaneous patch rupture (i.e., sudden increase in the amplitude and speed of capacitive transients) were omitted or used as control recordings to exclude that β-escin itself affected the channel properties or the time course of changes in IBa. Recording solutions and drugs All solutions for electrophysiological experiments were prepared by using type 1 ultrapure water (Milli-Q by Millipore Corporation or Purelab Flex 2 by ELGA Labwater) and, unless noted otherwise, reagents purchased from Sigma-Aldrich. During the recordings, cells were constantly perfused with external solution containing (mM) 120 NaCl, 5 BaCl2 or CaCl2, 5 KCl, 1 MgCl2, 20 TEA chloride, 10 glucose, 10 HEPES, and 0.1 BaEDTA with the pH adjusted to 7.4 by using NaOH and osmolarity of 300–320 mOsm. The solution was filtered through 0.2-µm polyethersulfone membranes and applied to cells at a rate of ∼2–4 ml/min by using a gravity-driven perfusion system controlled by manual precision flow regulators (Sarstedt). The standard intracellular solution was composed of (mM) 130 CsCl, 5 oxaloacetic acid, 5 creatine, 5 pyruvic acid, 10 EGTA, and 10 HEPES with the pH adjusted to 7.3 by using CsOH and osmolarity of 275–295 mOsm. It was stored at −20°C, thawed on the day of the experiments, filtered through 0.2-µm surfactant-free cellulose acetate membranes (Corning), and kept on ice between the recordings. The liquid junction potential between internal and external solution (calculated by using the JPcalc algorithm in pClamp 10; Molecular Devices) was ∼5.5 mV. Because no correction for the liquid junction potential was done, all voltages shown were actually 5.5 mV more negative. Nucleotide triphosphates (NaATP, MgATP, and MgGTP) were dissolved in 10 mM HEPES to prepare 100 mM stock solutions (pH = 7.3), stored at −20°C and diluted in standard internal solution immediately before the recordings. Calmodulin (CaM) was dissolved in type I ultrapure water to prepare a 100 µM stock solution and stored at −20°C. AMP-PNP was used as the lithium salt and dissolved directly in the standard internal solution. Deltamethrin, wortmannin, U73122, genistein, and staurosporine (stauro) were dissolved in DMSO to prepare 2–50 mM stock solutions, protected from light and stored at −20°C. Before the experiments, they were diluted in standard internal solution, and the same amount of DMSO (0.1–0.25% vol/vol) without drug was added to the internal solution used for parallel control recordings. To account for the short half-life in aqueous media, all solutions containing wortmannin were discarded after a maximum of 20 min after its dilution into standard internal or external solution, respectively. Cytosolic extract was prepared on the day of the experiments by freeze/thaw lysis of ∼2.1 × 107 cells in 1.5 ml ice-cold pipette solution. After centrifugation at 0°C for 5 min, the supernatant was filtered through a 0.2-µm surfactant-free cellulose membrane (Corning), an aliquot was removed for quantification of the protein concentration, and the rest was immediately used for the recordings. Whole-cell protocols Unless noted otherwise, time-course recordings were performed by repetitive application at 0.03 Hz of a 30-ms test pulse to +10 mV followed by 10-ms repolarization at −50 mV to record well resolved tail-currents. In some experiments, a 50-ms voltage ramp from −80 mV to +60 mV (2.8 V/s) was used instead of the voltage steps to monitor changes in the quasi steady-state voltage dependence. Because recordings with drugs added to the internal solution could not be used as their own control, we always performed parallel control recordings with standard internal solution. To study changes in current immediately after establishing the whole-cell configuration, cells were clamped at the holding potential of −80 mV before disruption of the patch. After patch rupture, capacitance was quickly compensated by using the auto function of the amplifier, and stimulation started after a delay of no more than 10 s. Because initial current amplitudes were small and several seconds were often required for Rs to stabilize completely, the first response was always recorded without Rs compensation. We then stopped stimulation, adjusted the Rs compensation of the amplifier to achieve maximal compensation without ringing, and restarted the stimulation protocol. To reduce variation introduced by differences in the time required for stabilization of Rs, initial values (Istart) for statistical comparison were always calculated as the average of the first two responses, which also provided a compromise between the lack of Rs compensation during acquisition of the first response and the time that had already evolved when the second response was recorded. To construct steady-state current-voltage (IV) relationships, peak currents recorded with a protocol consisting of 25-ms test pulses to potentials between −80 mV and +60 mV (10-mV increments at 0.1 Hz) were normalized by the maximum current amplitude at time 0 and plotted as a function of the test-pulse potential. Instantaneous IV (IIV) relationships were obtained with a protocol consisting of a fixed 10-ms prepulse to +60 mV followed by 40-ms test pulses to potentials between −80 mV and +60 mV (10-mV increments at 0.1 Hz). Instantaneous (tail) current amplitudes recorded during the test pulses were normalized by the maximum tail current amplitude at time 0 and plotted as a function of the prepulse potential. The voltage dependence of activation was assessed by a protocol consisting of 25-ms prepulses to potentials between −80 mV and +60 mV (10-mV increments at 0.1 Hz) followed by a fixed 10-ms test pulse to −50 mV. Tail-current amplitudes recorded during the test pulse were normalized to the maximum tail-current amplitude and plotted as a function of the prepulse potential to construct isochronous activation curves. The fraction of channels available for activation from different holding potentials was assessed by a protocol consisting of 3-s conditioning prepulses at potentials between −120 mV and +10 mV (10-mV increments at 0.1 Hz) followed by a fixed 35-ms test pulse to +10 mV. To construct prepulse inactivation curves, peak current amplitudes recorded during the test pulse were normalized by the maximum amplitude and plotted as a function of the prepulse potential. Data analysis and statistics Leak-subtracted current traces were directly analyzed with PulseFit (HEKA) or exported for further processing with Microsoft Excel 2010 and OriginLab Pro (version 9; OriginLab). Time-course recordings were quantified in terms of the current increase during run-up, the duration of run-up, and the residual current after 7 min of run-down, as detailed in the results section. To compare the effect of various test solutions, each of these parameters was also normalized by the mean value observed in parallel control recordings according to Eq. 1:Δpt=pt−pc¯,(1) where pt is the value of parameter p measured in a cell with test solution t, pc¯ is the mean value of p in parallel control recordings, and Δpt is the normalized value of p, which should provide an estimate for the amount of change over control. IV relationships were fitted with a combined Ohm-Boltzmann equation (Eq. 2):I=(Vm−Vrev)*gIV/(1+exp(−(Vm−V0.5)/k)),(2) where I is the (normalized) peak current density measured at the test potential Vm, Vrev is the apparent reversal potential, gIV is the maximum slope conductance, V0.5 is the voltage eliciting half-maximal inward currents, and k is the slope factor. Isochronous activation and prepulse inactivation curves were fitted with single Boltzmann equations (Eq. 3):I/Imax=A2+(A1−A2)/(1+exp((Vm−V0.5)/k)),(3) where I/Imax is the normalized current at the prepulse potential Vm, V0.5 is the voltage of half-maximal activation (V0.5act) or inactivation (V0.5inact), k is the activation (kact) or inactivation (kinact) slope factor, and A1 and A2 are the initial and final values, respectively. All fits were performed by using the Levenberg-Marquardt least-squares algorithm, and the goodness of fit was judged based on residual plots and adjusted χ2 values. Smooth curves in the figures represent fits to average data whereas values given in the text are average data from fits to individual measurements. Values in the text and figures are expressed as mean ± SEM based on the number of independent experiments. Statistical significance was assessed with OriginLab Pro 9 by using a repeated-measures ANOVA followed by Bonferroni's post hoc analysis when comparing mean values from the same cells or a one-way ANOVA followed by Bonferroni's post hoc analysis when comparing multiple independent mean values. Homogeneity of variances between groups was tested by using Levene’s test for equality of variances on the squared deviations. In the case of heteroscedastic data (P < 0.05 in Levene’s test and ratio of largest-to-smallest variance ≥4), statistical significance was assessed with Minitab (version 17; Minitab Inc.) by using Welch’s ANOVA and the Games-Howell multiple-comparison method. The secondary Cav2.3 channel structure in Fig. 14 was visualized by using the web-based tool Protter for interactive protein feature visualization (available at http://wlab.ethz.ch/protter/start/; Omasits et al., 2014). Phosphorylation sites in the same figure were predicted at the highest threshold with the Group-based Prediction Software for Prediction of Kinase-specific Phosphorylation Sites 3.0 (available at http://gps.biocuckoo.org; Xue et al., 2008). Results Evolution and run-down of cloned Cav2.3 channel currents Fig. 1 summarizes changes over time in macroscopic Cav2.3 channel currents carried by 5 mM Ba2+ and evoked by voltage-steps to 10 mV in dialyzed HEK-293 cells. Peak current amplitudes recorded immediately after establishing the whole-cell configuration (Istart) were small and tended to increase for several minutes (run-up) until they reached a maximum value (Imax). This was invariably followed by a progressive but somewhat slower current decline (run-down), so that the response that remained 7 min after complete run-up (I7min) amounted to approximately half of the maximum current amplitude (Fig. 1, A and B). Figure 1. Evolution and run-down of Cav2.3 channel currents carried by 5 mM Ba2+. (A) Time course of changes in peak whole-cell currents evoked by repetitive step depolarization to 10 mV every 30 s (n = 10). For consistency with the later figures, time 0 is defined as the time of maximum inward current (Imax). Gray lines in the background indicate time course of changes in individual recordings. Also shown are mean current traces corresponding to the time points indicated next to the traces and by filled symbols in the graph. (B) Same data as in A but normalized to the maximum peak current amplitude (Imax). Gray lines in the background indicate the time course of changes in individual recordings. Also shown above is the lack of spontaneous changes in Rs over the time course of the recordings. (C) Same current traces as in A but scaled to their maximum amplitude to visualize changes in their shape. Values are expressed as mean ± SEM. We quantified the time course of changes by estimating (1) the time required after patch rupture for currents to reach Imax (= time to Imax), (2) the ratio between Imax and Istart (= increase), and (3) the ratio between I7min and Imax (= remaining IBa). Under standard recording conditions (Fig. 1), with mean Rs and Cs values of 6 ± 1 MΩ and 14 ± 1 pF, respectively (n = 10), run-up took 4 ± 1 min and was associated with a 2.7 ± 0.2-fold increase of peak IBa at 10 mV, whereas the residual current amplitude after partial run-down amounted to 42 ± 4% of its maximum value. As illustrated by inspection of scaled current traces (Fig. 1 C), the initial increase of currents was paralleled by a decrease of fractional inactivation during the 30-ms test-pulses from 53 ± 4% immediately after establishing the whole-cell configuration to 40 ± 4% (P < 0.001) after complete run-up. On the other hand, run-down tended to increase inactivation, although quantification of this effect was often confounded by the reduced signal-to-noise ratio due to current decline (but see also Fig. 13 E). The general time course of changes was very similar among cells with different peak current densities, not attributable to voltage loss or Rs variation (Fig. 1 B) and unaffected by inclusion of two 3-min resting periods without stimulation (not depicted). Factors involved in run-up and run-down Increased mechanical tension associated with fluid flow has been shown to cause run-up and functional alterations in some other VGCCs (Peng et al., 2005; Park et al., 2007), but the time-dependent changes in Cav2.3 channel current were still observed when recordings were performed under static (i.e., without perfusion) conditions (n = 3; not depicted). As would be expected for a process that involves diffusion-controlled depletion of metabolic substrates or other cytosolic factors on the other hand, the time course but not magnitude of changes in IBa depended on the absolute values of Rs (i.e., cell–pipette coupling) and Cs (i.e., cell size), so that combined use of smaller pipette tip diameters (Rs = 12 ± 1 MΩ) and larger cells (Cs = 29 ± 6 pF) resulted in a significant slowing of run-up and run-down compared with standard recording conditions (Fig. 2, A–D). Unless noted otherwise, the following experiments were all performed under standard recording conditions to facilitate fast and reproducible cell dialysis. Figure 2. Dependence of run-up and run-down on cell dialysis, Ca2+, and voltage. (A) Comparison of time-course recordings performed under standard recording conditions (Ctrl, with Rs = 6 ± 1 MΩ and Cs = 14 ± 1 pF; same data as in Fig. 1) or with high-resistance pipettes (Rs = 12 ± 1 MΩ) in large cells (Cs = 29 ± 6 pF) to retard cytosolic dilution (high Cs, n = 6). (B) Comparison of the run-up duration, determined as the time required for currents to reach their maximum amplitude (same cells as in A). (C) Comparison of the residual current after partial run-down, determined as the ratio between current amplitudes after 7 min of run-down and at time 0 (same cells as in A). (D) Comparison of the current increase during run-up, determined as the ratio between maximum and initial current amplitude (same cells as in A). (E) Comparison of time-course recordings performed with 5 mM Ba2+ (n = 5) or Ca2+ (n = 5) as the charge carrier. Dotted lines are extrapolated single exponential fits to the evolution of currents during run-up, which yielded the time-constant indicated above. (F) Comparison of the run-up duration observed with Ba2+ or Ca2+ as the charge carrier (same cells as in E). (G) Mean current traces evoked by a 50-ms voltage ramp from −80 mV to 60 mV immediately after establishing the whole-cell configuration (Istart, blue solid line), at the time of maximum IBa (Imax, black solid line) and after 7 min of run-down (I7min, orange broken line; n = 7). Inset: Same current traces but scaled to their maximum value to illustrate the lack of changes in position and shape during run-up and subsequent left shift of the curve during run-down. **, P < 0.01; ***, P < 0.001 vs. Ctrl in B and C or vs. Ba2+ in F (one-way ANOVA with Bonferroni post-hoc correction). Values are expressed as mean ± SEM. As shown in Fig. 2 E, the characteristic sequence of changes was also observed when recordings were performed with 5 mM Ca2+ as the charge carrier (Fig. 2 E), although run-up was significantly faster (Fig. 2 F) and appeared to be more complete when compared with parallel recordings with 5 mM Ba2+. Moreover, Istart (but not Imax or I7min) was significantly larger in recordings performed with Ca2+, suggesting that run-up but not run-down may be initiated or facilitated by a Ca2+-dependent mechanism. Because changes in the ionic conditions during cell dialysis could affect currents by altering the driving force, we also examined the quasi-steady-state voltage dependence of IBa with a 50-ms voltage-ramp from −80 to 60 mV (Fig. 2 G). Run-up was similar over the whole voltage range and not associated with changes in the apparent reversal potential (Vrev), so that initial (Istart, blue solid line) and run-up (Imax, black solid line) currents overlapped almost completely when scaled to their maximum amplitude (Fig. 2 G, inset). Currents recorded after partial run-down (I7min, orange dotted line) still had the same Vrev, but were shifted to more negative stimulus voltages, suggesting that their decline is not the result of a change in driving force but influenced by the test potential. Run-down involves separable changes in gating and maximum conductance The macroscopic conductance of a uniform population of voltage-gated channels can be approximated by the product of (1) the voltage- and time-dependent channel open probability (PO), (2) the holding potential-dependent availability for activation (PF), and (3) the voltage-independent maximum macroscopic conductance (Gmax), which depends on single-channel conductance (γ), the total number of channels (N), and the maximum values of PO (PO,max) and PF (PF,max). In this framework, run-down could involve reduced activation, increased inactivation, and/or an intrinsically voltage-dependent decrease of Gmax. In a first attempt to distinguish between these possibilities, we compared steady-state (IV) and instantaneous (IIV) current-voltage relationships, determined at the onset of run-down and in 5-min intervals thereafter (Fig. 3, A–C). Because IIV currents were recorded immediately after a brief prepulse to open most channels available for activation, they should be little affected by changes in the voltage-dependence of Po. Fig. 3 D plots the fraction of IV and IIV currents that remained after partial run-down for 5 min as a function of the test potential. The decline of steady-state currents increased with voltage but reached voltage-independent values at positive test potentials (where PO approaches PO,max), which is consistent with the shift of ramp-evoked currents (Fig. 2 G). When measured with the IIV protocol on the other hand, run-down was about the same at different test potentials and similar to IV-current decline at depolarized voltages (Fig. 3 D). The same pattern was observed at the later time points examined, so that the kinetics of IIV-current run-down were well represented by the exponential decline in slope conductance between −80 and 0 mV (τ = 7 ± 2 min, gIIV in Fig. 3 E). The time course of IV-current decline was markedly delayed during weak depolarization but similar when it was measured at sufficiently positive test potentials (τ = 9 ± 1 min) or in terms of the maximum slope conductance (τ = 10 ± 2 min, gIV in Fig. 3 F). Based on these findings, the decrease of macroscopic conductance by itself was independent of the test-pulse potential and had an approximately exponential time course, so the apparent voltage-dependence must have derived from changes in the voltage dependence of PO, which (partly) counteracted IV-current decline and even caused some transient stimulation at negative test potentials. Figure 3. Voltage dependence of steady-state and instantaneous current decline. (A) Mean steady-state (IV) currents (n = 6) recorded at the indicated test potentials before (black lines) and after 5 min of run-down (orange lines). To allow for comparison of the relative change at different test potentials, they have been normalized to the peak current amplitude observed at each voltage before run-down. (B) Mean instantaneous (IIV) currents recorded from the same cells as in A before (black lines) and after 5 min of run-down (orange lines). To allow for comparison of the relative change at different test potentials, they have been normalized as in A. (C) Comparison of steady-state IV (squares) and IIV (circles) relationships, determined immediately after complete run-up (open symbols) and after 5 min of run-down (orange symbols; same cells as in A and B). (D) Voltage dependence of the fraction of IV (squares) and IIV (circles) current remaining after 5 min of run-down, determined as the ratio between current amplitudes after 5 min of run-down and current amplitudes at time 0 (same cells as in A–C). (E) Mean changes over time in the fraction of IIV current recorded at −20 mV (open circles with dotted line) and +20 mV (black circles with solid line) and in the slope conductance (gIIV) between −80 and 0 mV (same cells as in A–D). (F) Mean changes over time in the fraction of IV current recorded at −20 mV (open squares with dotted line) and +20 mV (black squares with solid line) and in the maximum slope conductance (gIV) determined from IV relationships (same cells as in A–E). Values are expressed as mean ± SEM. Run-down is paralleled by changes in channel voltage dependence To assess how changes in the voltage dependence of PO and PF are involved in current decline, we compared IV, activation, and prepulse inactivation curves (for details on the voltage protocols see Materials and methods), determined at different time points after the onset of run-down. Fig. 4 summarizes the results obtained in a total of 49 cells (Cslow = 29 ± 2 pF), from which meaningful currents could be recorded for at least 21 min. The shape of IV current lost to run-down (Fig. 4 A, inset), was essentially identical to the shape at time 0 and contained a prominent ON-gating current component, suggesting that run-down reduced the total number of active channels (i.e., PF at −80 mV, PF,max, and/or N). It was paralleled by progressive but unequal hyperpolarizing shifts of activation and prepulse inactivation, which are illustrated in Fig. 4 (C and D). The half-activation voltage (V0.5act) significantly decreased from 0 ± 1 mV at time 0 to −6 ± 1 mV (P < 0.001) after 21 min of run-down, corresponding to a shift by roughly −6 mV. In the same time, the half-inactivation voltage (V0.5inact) significantly decreased from −52 ± 1 mV to −69 ± 1 mV (P < 0.001), corresponding to a shift by roughly −17 mV. In both cases, the rate of shift in individual recordings could be well described by linear fits to the changes in half-point as a function of time, with slopes of −0.33 ± 0.02 mV/min (adjusted R2 = 0.90 ± 0.02) for activation and −0.82 ± 0.03 mV/min (adjusted R2 = 0.95 ± 0.01) for inactivation. Based on these findings, PF at −80 mV was reduced by ∼5, 12, and 19% after 7, 14, and 21 min of run-down, respectively (Fig. 4 C), which corresponds to a constant rate of roughly 1%/min. In the same time, gIV declined by ∼25, 47, and 61%, so that reduced availability for activation alone could only explain part of the IV-current decline at positive test potentials (Fig. 4 B). Together with a complete loss of gating currents during run-down, these findings suggested that the decline in conductance involves additional (holding potential–independent) changes in the number of active channels (i.e., N and/or PF,max). This was also evident in time-course recordings where the holding potential after partial run-down was made more negative, which led to considerable but incomplete recovery of the current at 10 mV (not depicted). Figure 4. Changes in channel voltage dependence during run-down. (A) Families of mean IV currents (n = 8) evoked by 25 ms voltage steps to test potentials between −80 mV and +60 mV at the onset of run-down (0 min) and in 7-min intervals thereafter. They have been normalized to the maximum peak IBa at time 0. Inset: Current lost to run-down at the indicated test potentials, determined by subtracting from the normalized current traces recorded at time 0 the corresponding traces recorded after 7 (orange lines), 14 (purple lines), and 21 min (green lines) of run-down. (B) Mean IV relationships measured at the same time points as in A and normalized to the maximum peak current amplitude recorded at time 0 (n = 49). (C) Voltage dependence of activation (right) and prepulse inactivation (left), determined at the same time points as in A (same cells as in B). (D) Time course of changes in the half-activation (V0.5act) and half-inactivation (V0.5inact) voltages, determined from single Boltzmann fits to the data in C. (E) Time course of changes in the activation (kact) and inactivation (kinact) slope factors, determined from single Boltzmann fits to the data in C. Values are expressed as mean ± SEM. In addition to the shift, run-down reduced the steepness of both curves in Fig. 4 C, which was reflected in a gradual increase of the activation and inactivation slope factors (kact and kinact) from 9.0 ± 0.2 and 9.2 ± 0.1 mV/e-fold change at time 0 to 10.5 ± 0.3 (P < 0.001) and 11.1 ± 0.3 (P < 0.001) mV/e-fold change after 21 min of run-down, respectively (Fig. 4 E). Hydrolysable ATP provides protection from run-down Because one of the most common reasons for “washout” of ionic currents is a decrease in the level of intracellular high-energy compounds or other cytosolic factors, we next tested if different forms and concentrations of nucleotide triphosphates, a cytosolic extract, and/or perforated-patch recordings are effective in altering run-down. To this end, the time course of changes in peak IBa under various experimental conditions was quantified as described above and compared with parallel control recordings performed with standard internal solution (Figs. 5 and 6). To facilitate comparison of effects between the different experiments, Fig. 7 summarizes mean changes expressed relative to the corresponding control recordings (i.e., Δ Increase, Δ Time to Imax, and Δ Remaining IBa), which were determined by subtracting from each individual value with the indicated test solutions the mean value determined in parallel control recordings (for details see Materials and methods). As can be seen in Fig. 5, addition of 1–5 mM MgATP to the pipette solution significantly increased the current fraction that remained after 7 min of run-down compared with parallel control recordings (Fig. 5, A and B) and produced a concentration-dependent slowing of run-up (Fig. 5, A and C). These effects were not attributable to changes in intracellular free Mg2+ (∼0.4 mM with 5 mM MgATP) because addition of 1 mM MgCl2 alone (∼0.5 mM free Mg2+) to the standard internal solution affected neither run-up nor run-down (Fig. 5 E). Moreover, with a pipette solution containing 5 mM NaATP and no added free Mg2+, there was still a significant slowing of run-up and a significant increase of the current fraction remaining after 7 min of run-down (Fig. 7, A and C). Figure 5. Hydrolysable ATP provides protection from Cav2.3 channel run-down. Comparison between time-course recordings performed with the indicated test substances and parallel control recordings with standard internal solution. (A) Mean data from 5 recordings with 1 mM MgATP (orange squares), 8 recordings with 5 mM MgATP (orange circles), and 13 parallel control recordings (open squares). (B) Comparison of the residual IBa after 7 min of run-down under control conditions and with 1 or 5 mM MgATP in the pipette solution (same cells as in A). (C) Comparison of the run-up duration under control conditions and with 1 or 5 mM MgATP in the pipette solution (same cells as in A). (D) Comparison of the current increase under control conditions and with 1–5 mM MgATP in the pipette solution (same cells as in A). (E) Mean data from five recordings with 1 mM MgCl2 (orange squares) and seven parallel control recordings (open squares). (F) Mean data from six recordings with 0.3 mM MgGTP (orange squares) and five parallel control recordings (open squares). (G) Mean data from eight recordings with 5 mM AMP-PNP (orange squares) and seven parallel control recordings (open squares). **, P < 0.01; ***, P < 0.001 vs. Ctrl (one-way ANOVA with Bonferroni post hoc correction). Values are expressed as mean ± SEM. Figure 6. ATP protection from run-down is reproduced by perforated-patch recordings. Comparison of time-course recordings with 40 µM β-escin in the absence and presence of MgATP. (A) Mean data from 11 perforated-patch recordings with standard internal solution (orange diamonds), 6 perforated-patch recordings with 5 mM MgATP (orange circles), and 3 perforated-patch recordings with standard internal solution in which the patch ruptured during perforation (open squares). All recordings were performed with 40 µM β-escin as the perforating agent. (B) Comparison of the residual IBa after 7 min of run-down in recordings with patch rupture and in perforated recordings without or with 5 mM MgATP (same cells as in A). (C) Comparison of the run-up duration in recordings with patch rupture and in perforated recordings without or with 5 mM MgATP (same cells as in A). (D) Comparison of the current increase in recordings with patch rupture and in perforated recordings without or with 5 mM MgATP (same cells as in A). *, P < 0.05; **, P < 0.01 vs. Ctrl (one-way ANOVA with Bonferroni post hoc correction). Values are expressed as mean ± SEM. Figure 7. Overview of changes in run-up and run-down observed under various experimental conditions. Data are from the same cells as in Figs. 5 and 6 with additional data from 7 recordings with cytosolic extract (Cytosol) and 6 parallel control recordings, 9 recordings with 5 mM NaATP and 7 parallel control recordings, and 13 recordings performed with Ca2+ as the charge carrier and 5 mM MgATP in the presence or absence of 3 µM CaM plus 5 parallel control recordings. For comparison between the different experimental conditions, all values obtained with a given test solution were normalized by subtracting the mean value observed in parallel control recordings and then averaged, so the derived parameters provide a measure for the mean change relative to control (for details see Materials and methods). Symbols for statistical significance refer to the comparison of normalized values between different experimental conditions. (A) Comparison of mean changes relative to parallel control recordings in the duration of run-up (Δ Time to Imax). (B) Comparison of mean changes relative to parallel control recordings in the magnitude of run-up (Δ Increase). (C) Comparison of mean changes relative to parallel control recordings in the residual current after 7 min of run-down (Δ Remaining IBa). *, P < 0.05; **, P < 0.01; ***, P < 0.001 (one-way ANOVA with Bonferroni post hoc correction in A and C or Welch’s ANOVA with the Games-Howell multiple-comparison method in B). Values are expressed as mean ± SEM. ATP protection from run-down was about the same with different charge carriers (i.e., Ba2+ vs. Ca2+) and not altered when the nucleotide was combined with 3 µM CaM (Fig. 7 C). Supplementation of the pipette solution with 0.3 mM MgGTP (Fig. 5 F) or 5 mM of the nonhydrolyzable ATP-analogue AMP-PNP (Fig. 5 G) on the other hand affected neither run-up nor run-down (Fig. 7). Together, these findings argue against a role of direct nucleotide binding and indicate a requirement for ATP-hydrolysis and transfer of the phosphate group. Note also that (with Ba2+ as the charge carrier) none of the manipulations tested significantly altered the magnitude of run-up (Figs. 5 B and 7 B) or the absolute values of Imax (not depicted), suggesting that ATP can slow but not prevent the processes underlying run-up. The apparent increase in the magnitude of run-up by ATP observed with Ca2+ as the charge carrier (Fig. 7 B) was also not related to a more pronounced up-regulation of ICa per se because absolute values of Imax were the same as in parallel control recordings (not depicted). Although there was a tendency for Istart to be smaller in recordings with ATP, this effect could hardly be ascribed to an action of the nucleotide. Because the faster run-up kinetics in Ca2+ hampered accurate determination of Istart, it might instead have resulted from small differences between groups in the time required for Rs to stabilize after patch-rupture. ATP protection is reproduced by perforated-patch recordings In principle, exogenous ATP might counteract run-down by preventing a decrease of cytosolic nucleotide levels or by increasing them above the normal value in intact cells, thereby stimulating the channels through an unrelated mechanism. To distinguish between these two possibilities, we performed perforated-patch recordings with β-escin, diluted into the same internal recording solution as in ruptured-patch experiments. ATP-diffusion through β-escin pores has been demonstrated, but perforating efficiency is concentration and time dependent, so washout should be considerably slower than in ruptured-patch recordings (Arnould et al., 1996; Fan and Palade, 1998; Fu et al., 2003). As illustrated in Fig. 6 A and summarized in Fig. 7, the time course of changes during perforated recordings was very similar to that observed in ruptured recordings with ATP. Importantly, and consistent with a role of ATP-depletion for the current decline, perforated recordings were almost as effective in reducing run-down as provision of ATP in ruptured recordings. Moreover, the protective effects of perforated recordings and ATP were not additive, so that provision of 5 mM MgATP in perforated recordings produced the same effects as in ruptured recordings (Figs. 6 B and 7 C). Collectively, these findings indicate that both, perforated recording and/or provision of ATP reduced run-down by preventing depletion of the nucleotide during the recordings. This is in contrast to the effects on run-up duration, which were much more pronounced in perforated recordings and further increased in the presence of ATP (Figs. 6 C and 7 A). Moreover, because our measurements were started only after Rs had reached values ≤15 MΩ, we most likely missed part of the run-up process, so these results may still underestimate the true slowing of run-up in perforated recordings. Thus, the initial but not maximum peak current density was significantly larger in perforated recordings (not depicted), and this was reflected in a reduced magnitude of run-up (Figs. 6 D and 7 B). Together, these observations point to the involvement of additional cytosolic factors during run-up, which are more effectively retained in perforated-patch recordings. However, a cytosolic extract prepared in standard internal solution (4.32 mg protein/ml) was ineffective in altering run-up or run-down (Fig. 7), possibly because these factors depend on the additional presence of ATP. ATP stabilizes channel gating and maximum conductance All the above findings supported the assumption that Cav2.3 channel run-down and, to some extent, run-up in dialyzed cells are related to a depletion of cytosolic ATP. Because it was of interest how ATP affects the different alterations associated with run-down, we reexamined the changes in channel voltage dependence during current decline with a pipette solution supplemented with 5 mM MgATP. As illustrated in Fig. 8 A, inclusion of ATP completely abolished the early phase of current decline, so that on average, gIV after 7 min of run-down amounted to 97 ± 7% of its initial value (n = 6). In addition, the stabilizing action of ATP was associated with an almost complete prevention of time-dependent changes in channel voltage dependence and sensitivity (Fig. 8, B–E), suggesting that basal ATP-dependent modulation alters channel-gating behavior and is required for maintaining it in a functional state. Figure 8. Hydrolysable ATP stabilizes Cav2.3 channel gating and function. (A) Mean IV relationships measured at the onset of run-down and in 7-min intervals thereafter with a pipette solution containing 5 mM MgATP (n = 6). (B) Voltage dependence of activation (right) and prepulse inactivation (left), determined at the same time points as in A (same cells as in A). (C) Time course of changes in the half-activation voltage during run-down observed in the absence (black squares, same cells as in Fig. 4) and presence (orange circles, same cells as in A) of 5 mM MgATP. (D) Time course of changes in the half-inactivation voltage during run-down observed in the absence (black squares, same cells as in Fig. 4) and presence (orange circles, same cells as in A) of 5 mM MgATP. (E) Time course of changes in the inactivation slope factor during run-down observed in the absence (black squares, same cells as in Fig. 4) and presence (orange circles, same cells as in A) of 5 mM MgATP. Values are expressed as mean ± SEM. The effects of ATP are not related to lipid kinase–mediated PIP2 synthesis Because run-down of several neuronal VGCCs and its reversal by MgATP have been linked to depletion and resynthesis of membrane PIP2, respectively (Wu et al., 2002; Suh et al., 2010), we next examined the effects of wortmannin (WM), a potent and irreversible lipid kinase inhibitor (Wipf and Halter, 2005). When MgATP in the pipette solution was combined with 10 µM WM (ATP+WM in Fig. 9, A–D), it still significantly increased the current fraction remaining after 7 min of run-down (Fig. 9 B) and markedly slowed run-up (Fig. 9 C). Relative to recordings with MgATP alone, WM actually tended to enhance the protective effects (Fig. 10 C), suggesting that it increased ATP availability because of reduced consumption by lipid kinases. Likewise, pretreatment of cells by incubation in extracellular solution containing 10 µM WM for 15 min (ATP+WM PT in Fig. 9, A–D) did not significantly impair the stabilizing action of MgATP (Fig. 9 B) nor did it alter the time course of changes observed in parallel control recordings. Unlike acute treatment (i.e., ATP+WM) however, WM pretreatment markedly enhanced the ATP-induced slowing of run-up in a subset of cells (Fig. 9 C), so that on average, the latter effect was stronger but also more variable (Fig. 10 A). A similar modification of ATP-induced changes in the duration of run-up was observed in only one of seven recordings with ATP+WM (Fig. 9 C), suggesting that PIP2 depletion rather than acute inhibition of its resynthesis may be required to modify ATP-effects on run-up. Brief application of WM inhibits PIP2 replenishment without affecting PIP2 hydrolysis (Zhang et al., 2003), so the variable effectiveness of WM pretreatment might have been related to differences in lipid turnover and degree of actual PIP2 depletion between cells. Regardless of the exact effects on run-up however, the protection by ATP against run-down was clearly unaffected by WM, suggesting that it was not related to lipid kinase–mediated mechanisms and arguing against a role of PIP2 depletion for Cav2.3 channel run-down in our system. Figure 9. ATP protection from run-down is not related to increased PIP2 synthesis. (A) Mean data from seven recordings with 5 mM MgATP and 10 µM WM (ATP + WM), nine recordings with 5 mM MgATP after WM pretreatment (ATP + WM PT), and 13 parallel control recordings. (B) Comparison of the residual IBa after 7 min of run-down under control conditions, with ATP and WM and with ATP after WM pretreatment (same cells as in A). (C) Comparison of the run-up duration under control conditions, with ATP and WM and with ATP after WM pretreatment (same cells as in A). (D) Comparison of the current increase under control conditions, with ATP and WM and with ATP after WM pretreatment (same cells as in A). (E) Mean data from five recordings with 1 mg/ml BSA and nine parallel control recordings. (F) Mean data from six recordings with 5 µM U73122 and eight parallel control recordings. **, P < 0.01; ***, P < 0.001 vs. Ctrl (one-way ANOVA with Bonferroni post hoc correction). Values are expressed as mean ± SEM. Figure 10. Overview of changes in run-up and run-down observed under various experimental conditions. Data are from the same cells as in Figs. 9, 11, and 13 with additional data from six recordings with 5 mM MgATP + 100 µM GS (+GS) and four parallel control recordings and three recordings after a 60-min pretreatment with 10 µM DM and five parallel control recordings. For comparison between the different experimental conditions, all values obtained with a given test solution were normalized as described in Fig. 7 and Materials and methods. Symbols for statistical significance refer to the comparison of normalized values between different experimental conditions. (A) Comparison of mean changes relative to parallel control recordings in the duration of run-up (Δ Time to Imax). (B) Comparison of mean changes relative to parallel control recordings in the magnitude of run-up (Δ Increase). (C) Comparison of mean changes relative to parallel control recordings in the residual current after 7 min of run-down (Δ Remaining IBa). *, P < 0.05; **, P < 0.01; ***, P < 0.001 (one-way ANOVA with Bonferroni post hoc correction in A and C or Welch’s ANOVA with the Games-Howell multiple-comparison method in B). Values are expressed as mean ± SEM. Another line of reasoning has been that not PIP2 depletion itself, but rather accumulation of one of its fatty acid cleavage products, arachidonic acid (AA), is responsible for run-down of some neuronal VGCCs (Liu et al., 2001, 2006; Liu and Rittenhouse, 2003). However, neither 1 mg/ml of the AA-scavenger BSA (Fig. 9 E) nor 5 µM of the phospholipase C inhibitor U73122 (Fig. 9 F) affected the time course of run-up or run-down when they were added to the (ATP-free) internal solution (Fig. 10, A–C). The effects of ATP depend on protein kinase–mediated phosphorylation We next examined the role of changes in protein phosphorylation and dephosphorylation, which may result from ATP-depletion and have been implicated in the run-down of L-type high-voltage-activated Ca2+ channels (Armstrong and Eckert, 1987; Hilgemann, 1997). As summarized in Fig. 10 (A–C), the effects of MgATP were well preserved and again even somewhat enhanced when it was combined with 100 µM of the protein tyrosine kinase inhibitor genistein (GS). Combination with 10 µM of the broad-spectrum serine/threonine kinase inhibitor stauro on the other hand completely abolished MgATP effects on the duration of run-up, so the time to Imax with MgATP+stauro was not significantly different from the time course in parallel control recordings (Fig. 11, A–C; and Fig. 10 A). Moreover, although MgATP in the presence of stauro still significantly increased the fraction of current remaining after 7 min of run-down relative to parallel control recordings (Fig. 11, A and B), the net effect was significantly reduced when compared with that observed in other recordings with MgATP. Thus, expressed relative to the corresponding control recordings, MgATP+stauro was significantly less effective in increasing the current fraction after 7 min of run-down than MgATP alone, MgATP+WM, or MgATP+GS (Fig. 10 C). Figure 11. ATP protection from run-down involves serine/threonine phosphorylation. (A) Mean data from 10 recordings with 5 mM MgATP and 10 µM stauro (ATP+Stauro), 4 recordings with 10 µM stauro only (Stauro), and 12 parallel control recordings (Ctrl). (B) Comparison of the residual IBa after 7 min of run-down in recordings with Stauro or ATP+Stauro and in parallel control recordings (same cells as in A). (C) Comparison of the run-up duration in recordings with Stauro or ATP+Stauro and in parallel control recordings (same cells as in A). (D) Mean data from seven recordings after pretreatment with 1 µM stauro (Stauro PT) and five parallel control recordings (Ctrl). Note that current amplitudes are expressed in pA/pF to illustrate the reduced peak current densities after stauro pretreatment. (E) Comparison of the initial peak current densities observed after stauro pretreatment and in parallel control recordings (same cells as in D). (F) Comparison of the maximum peak current densities observed after stauro pretreatment and in parallel control recordings (same cells as in D). (G) Same data as in D but normalized to the initial current amplitudes to illustrate changes in the magnitude of run-up. (H) Comparison of the current increase during run-up after stauro pretreatment and in parallel control recordings (same cells as in D). (I) Comparison of the run-up duration after stauro pretreatment and in parallel control recordings (same cells as in D). *, P < 0.05; **, P < 0.01 (one-way ANOVA with Bonferroni post hoc correction). Values are expressed as mean ± SEM. There was no evident effect of stauro in the absence of ATP (Fig. 11, A–C), indicating that it acted by reducing ATP-protection from run-down. From these findings it follows that the ATP-induced slowing of run-up and most but not all the protection from run-down must have involved protein kinase–mediated phosphorylation of serine/threonine residues. On the other hand, run-down in the absence of ATP might reflect a progressive decrease of channel phosphorylation because of changes in the balance between constitutive de- and rephosphorylation. To test this assumption, we examined the effects of a sustained (30–60-min) incubation of cells in extracellular solution containing 1 µM stauro, which should mimic run-down by reducing basal protein phosphorylation. Consistent with (partial) run-down of the channels before establishment of the recordings, stauro pretreatment increased the number of cells lacking macroscopic currents and significantly reduced both Istart and Imax in the remaining cells (Fig. 11, D–F). Interestingly, it also significantly increased the magnitude and duration of run-up (Fig. 11, G–I), although the latter effect may have been related to the appearance of a plateau phase with little change in current amplitudes for several minutes. Subsequently, run-down proceeded with a normal time course, so the decrease of currents after 7 min was essentially the same as in parallel control recordings. To assess how these effects were related to the ATP-sensitive gating changes, we also examined the effects of stauro pretreatment on channel voltage dependence (Fig. 12). As illustrated in Fig. 12 (B and C), activation and prepulse inactivation curves recorded in pretreated cells before the onset of run-down showed a significant shift to more negative test potentials when compared with the results obtained in untreated cells, although the voltage-sensitivity (i.e., kact and kinact) was similar (not depicted). Figure 12. Protein kinase inhibition in intact cells reproduces the gating changes during run-down in dialyzed cells. (A) Voltage dependence of activation (right) and prepulse inactivation (left), determined at the onset of run-down (time 0) and in 7-min intervals thereafter in nine cells pretreated with 1 µM stauro for 30–60 min. (B) Comparison of half-activation voltages at the onset of run-down in cells pretreated with 1 µM stauro (same cells as in A) and under control conditions (same cells as in Fig. 4). (C) Comparison of half-inactivation voltages at the onset of run-down in cells pretreated with 1 µM stauro and under control conditions (same cells as in B). (D) Time-course of changes in half-activation voltages during run-down observed in cells pretreated with 1 µM stauro and under control conditions (same cells as in B). (E) Time course of changes in half-inactivation voltages during run-down observed in cells pretreated with 1 µM stauro and under control conditions (same cells as in B). (F) Time course of changes in activation slope factors during run-down observed in cells pretreated with 1 µM stauro and under control conditions (same cells as in B). ***, P < 0.001 vs. Ctrl (one-way ANOVA with Bonferroni post hoc correction). Values are expressed as mean ± SEM. In addition, stauro pretreatment tended to accelerate the voltage shifts during run-down (Fig. 12, D and E), although comparison of the exact time course is confounded by differences in cell size between the experiments, so no firm conclusions can be drawn from this finding. Interestingly however, it also effectively prevented the decrease in activation voltage sensitivity during run-down (Fig. 12 F) without altering the changes in inactivation voltage sensitivity (not depicted), possibly pointing to the involvement of multiple mechanisms at distinct sites. Because the serine/threonine phosphatase calcineurin has been shown to interact with neuronal VGCCs (Chad and Eckert, 1986; Fomina and Levitan, 1997) and implicated in the run-down of certain K+ channels (Horváth et al., 2002), we also tested the effects of pretreating cells with a 60-min incubation in extracellular solution containing 10 µM of the membrane-permeable calcineurin-inhibitor deltamethrin (DM). However, the time course of changes in IBa observed in DM-pretreated cells (recorded with ATP-free pipette solution) was not significantly different from that in parallel control recordings (Fig. 10). Run-up may involve activation of Leu-sensitive proteases Finally, we performed experiments with Leu, a protease inhibitor that has been shown to prevent an irreversible ATP-resistant component of run-down in L-type VGCCs (Chad and Eckert, 1986; Elhamdani et al., 1994). Unexpectedly, and in contrast to all other manipulations tested, MgATP+100 µM Leu significantly decreased the magnitude of run-up (Fig. 13, A–C), so that Istart was similar but Imax was approximately half of the value observed in parallel control recordings. In addition, Leu not only abolished the MgATP-dependent slowing of run-up but actually produced a significant decrease of the time to Imax when it was combined with ATP (Fig. 13 D). Apart from altering the time course and net increase of currents, ATP+Leu reduced cell-to-cell variability in the magnitude of run-up (variance = 0.89 in control recordings vs. 0.16 with MgATP+Leu, P < 0.05) and diminished the initial slowing of inactivation (Fig. 13 C), indicating that both processes may be related to activation of Leu-sensitive proteases and influenced by variable endogenous protease and/or protease inhibitor levels. Figure 13. Run-up may involve activation of Leu-sensitive proteases. (A) Mean data from 10 recordings with 5 mM MgATP and 100 µM Leu (ATP+Leu), 9 recordings with 100 µM Leu only (Leu), and 17 parallel control recordings (Ctrl). (B) Same data as in A but normalized to the initial current amplitude to illustrate differences in the magnitude of run-up. (C) Comparison of the current increase during run-up with ATP+Leu, Leu only, and in parallel control recordings (same cells as in A). (D) Comparison of the run-up duration with ATP+Leu, Leu only, and in parallel control recordings (same cells as in A). (E) Time course of changes in inactivation during the 30-ms test pulses with ATP+Leu, Leu only, and in parallel control recordings (same cells as in A). **, P < 0.01 (one-way ANOVA with Bonferroni post hoc correction in D or Welch’s ANOVA with the Games-Howell multiple-comparison method in C). Values are expressed as mean ± SEM. Curiously, Leu also suppressed the protective effect of MgATP against run-down (Fig. 13 A), so that expressed relative to the corresponding control recordings, MgATP+Leu was significantly less effective against rundown than MgATP alone, MgATP+WM, MgATP after WM pretreatment, or MgATP+GS (Fig. 10 C). When the same concentration of Leu (100 µM) was used in the absence of MgATP, it had similar but quantitatively much less marked effects on run-up (Fig. 13, A–D) and no effect on the degree of run-down (Fig. 10 C) or the inactivation changes during run-up (Fig. 13 E). Discussion In the present study, we used conventional and perforated-patch-clamp recordings in a recombinant expression system to assess changes in Cav2.3 channel currents during cell dialysis. Our findings recapitulate studies about their native counterparts and show that dialysis with ATP-free internal solutions produces a characteristic sequence of run-up and run-down, the time-course of which depends on Rs, cell size, and recording configuration. The exponential decline in conductance during run-down by itself was voltage independent but paralleled by a progressive shift of channel voltage dependence to more negative test potentials. The voltage-shift reduced channel availability at the holding potential (i.e., PF) but proceeded at a constant rate and could only partly account for the complete loss of gating currents during run-down. Therefore, most of the decline in conductance was related to changes in the total number of functional channels (i.e., N and/or PF,max) and possibly to other factors (i.e., decrease of γ or PO,max). Run-down and all the associated biophysical changes could be slowed or prevented by provision of ATP, and this protective action was almost completely abolished by inhibition of serine/threonine kinases. Protein kinase inhibition also mimicked the effects of run-down in intact cells, so it reduced overall peak current densities and hyperpolarized the voltage dependence of gating relative to untreated cells. The effects of ATP could be replicated neither by a nonhydrolyzable ATP analogue nor by GTP, which argues against a role of direct nucleotide binding or G-protein interactions but is consistent with the reported selectivity of protein kinases for ATP (Becher et al., 2013). Finally, run-down was not influenced by dialysis with a PLC inhibitor or the AA scavenger BSA, and the effects of ATP were unaffected by inhibition of lipid kinases, suggesting that run-down was not related to PIP2 hydrolysis, or accumulation of its cleavage product AA. Based on these findings, we conclude that (1) ATP-protects from Cav2.3 channel run-down by maintaining phosphorylation of serine/threonine residues on the channels themselves or an associated protein, (2) constitutive dephosphorylation of these sites in dialyzed cells affects channel gating and reduces the total number of functional channels, and (3) (de-)phosphorylation of at least some of the sites can also regulate channel function in intact cells. In addition, some of our findings point to a role of Leu-sensitive proteases for Cav2.3 channel up-regulation during run-up and for the effects of protein phosphorylation on run-down. Comparison with run-down in other voltage-gated Ca2+ channels Since its first description, Ca2+ channel run-down has been consistently linked to diffusion-controlled dilution of ATP and other cytosolic components, but the underlying processes seem to differ among channels. ATP-induced protection from L-type Ca2+ channel run-down involves several interrelated processes, which may include changes in lipid turnover (Wu et al., 2002; Kaur et al., 2015), direct nucleotide binding (Feng et al., 2014), and phosphorylation of the channels by PKA, CaMKII and possibly other kinases (Wang et al., 2009; Xu et al., 2016). The latter counteracts constitutive dephosphorylation by opposing protein phosphatases and has been proposed to stabilize conformations that can be reprimed by voltage, CaM, or calpastatin and/or are more resistant to proteolytic degradation (Chad et al., 1987; Wang et al., 2009; Sun et al., 2014). This is in contrast to the situation in Cav2.1 and Cav2.2 channels, where run-down has been linked to constitutive hydrolysis of membrane PIP2, which can be prevented by ATP through lipid kinase–mediated PIP2 resynthesis (Wu et al., 2002; Suh et al., 2010). Although we have not directly tested the effects of PIP2, our present findings demonstrate that lipid kinases are not involved in ATP-induced protection from Cav2.3 channel run-down and that the current decline is not related to accumulation of AA. Together with a previous study, where PIP2-depletion had no effect on Cav2.3 channels (Suh et al., 2010), these findings argue against a major role of PIP2 hydrolysis for Cav2.3 channel run-down. Instead, ATP appears to maintain Cav2.3 channel function through increased phosphorylation of sites on the channels themselves or an associated protein, which may be required to counteract constitutive dephosphorylation. It has been shown that HEK-293 cells contain endogenous kinases and phosphatases, which can regulate the activity of transfected Ca2+ channels (Perez-Reyes et al., 1994; Johnson et al., 1997; Fuller et al., 2010; Aita et al., 2011; Blesneac et al., 2015). In addition, Cav2.3 channels are a well-known substrate for phosphorylation by various protein kinases (Fig. 14), at least some of which can be constitutively active in HEK-293 cells (Perez-Reyes et al., 1994; Crump et al., 2006). Although no firm conclusions can be drawn with regard to the exact sites or kinases involved in Cav2.3 channel maintenance, our findings provide a reliable baseline for further studies. They also raise the question how protein phosphorylation might be involved in the maintenance Cav2.1 and Cav2.2 channels, which still exhibit significant rundown in the presence of exogenous PIP2 (Gamper et al., 2004) or with mutations that reduce PIP2 sensitivity (Zhen et al., 2006). Because there can be cross talk with lipid signaling (Wu et al., 2002) and PIP2-hydrolysis partially inhibited Cav2.3 channels after full activation by PKC (Jeong et al., 2016), it will also be interesting to examine the exact relevance of lipid turnover for this form of regulation. Figure 14. Phosphorylation sites and potential PEST regions in human Cav2.3 channels. The secondary structure was visualized with Protter based on the human protein database entry Q15878, which includes exon 19 and exon 45 encoded stretches of the full-length Cav2.3d-splice variant. Predictions were performed at the highest threshold by using the Group-based System Software for Prediction of Kinase-specific Phosphorylation Sites 3.0 and at a threshold score of +5.0 by using epestfind for detection of potential proteolytic cleavage sites, respectively. CAMKI, Ca2+/CaM-dependent protein kinase I; CAMKII, Ca2+/CaM-dependent protein kinase II; CAMKL, Ca2+/CaM-dependent protein kinase-like kinases; MAPK, mitogen-activated protein kinase; SGK, serum and glucocorticoid-regulated kinase 1; ter, terminal. Molecular mechanisms and potential implications Very little is known about the molecular and structural mechanisms of Ca2+ channel run-down, but possibilities that have been considered include disruption of the linkage between voltage sensors and activation gate, entry into a permanent but existing inactivated state, and spontaneous drops in the total number of channels due to internalization or degradation. Our results do not resolve the exact processes underlying Cav2.3 channel run-down but do argue against a significant decoupling between voltage-sensors and activation gate because run-down abolished both ionic and gating currents. In addition, Leu was ineffective against run-down and actually reduced the protective effects of ATP, suggesting that the current decline was not directly related to increased proteolytic degradation (but see next section). Together, these findings contrast with studies that L-type–channel gating currents are not diminished by run-down (Hadley and Lederer, 1991; Costantin et al., 1999) and that Leu prevents an irreversible component of run-down in these channels (Chad et al., 1987). We can only speculate that dephosphorylation of Cav2.3 channels themselves or an interacting protein induces a conformational change that leads to terminal inactivation and/or facilitates their removal from the membrane. More importantly, our findings reveal that part of the current decline can be attributed to a gradual development of inactivation due to changes in channel voltage dependence. Based on the lack of saturation or concurrent changes in Vrev, these effects may have been related to changes in the fraction of applied membrane voltage sensed by the channels. Interestingly, and in contrast to the exponential decline in conductance, the gating changes occurred at a constant rate, which is difficult to reconcile in terms of a conformational change. An attractive hypothesis that remains to be substantiated is that progressive dephosphorylation could shift channels among their states in a simple manner, such as altering the internal surface charge. It has been quantitatively demonstrated that the bulk electrostatic effects of dephosphorylation are sufficient to produce hyperpolarizing voltage shifts during run-down of other voltage-gated ion channels (Perozo and Bezanilla, 1990, 1991). Regardless of the exact mechanism, our findings document a close relationship between protein phosphorylation and channel voltage dependence, which may influence the basal features of current carried by cloned Cav2.3 channels. Based on previous studies about their native counterparts, it seems reasonable to propose that this is also relevant under physiological conditions. Thus, Cav2.3 channels are the third most extensively phosphorylated ion channels in mouse brain (Cerda et al., 2011; Fig. 14), and depolarization of intact hippocampal slices has been shown to induce bulk changes in their phosphorylation state (Hell et al., 1995). Run-up and the role of calpain-like proteases Unexpectedly, our findings also indicate that Cav2.3 channel run-up may involve activation of calpain-like proteases (CLPs) and that it can be slowed but not prevented by ATP through increased protein phosphorylation. The process started immediately upon patch rupture, which can hardly be accounted for by ATP depletion but might reflect the loss of small cytosolic protease inhibitors because it was much slower in perforated recordings. In addition, run-up was observed with Ba2+ as the charge carrier but significantly faster and more complete with Ca2+, which is consistent with studies that show that Ba2+ can partially substitute for Ca2+ in activating CLPs (DeMartino and Croall, 1985; McDonald et al., 1994; Seydl et al., 1995). That phosphorylation suppresses most CLPs (Shiraha et al., 2002; Smith et al., 2003), while PIP2 is well known to be required for activation and may considerably lower their Ca2+ requirement (Tompa et al., 2001; Leloup et al., 2010), could explain why (1) ATP, but not AMP-PNP or ATP+stauro, slowed run-up and (2) pretreatment with WM (i.e., partial depletion of membrane PIP2) further increased the ATP-induced slowing in some cells. Dialysis or pretreatment with stauro alone had no effect or even increased run-up, suggesting that the process was not related to proteolytic activation of protein kinases. With this in mind and considering its fast onset, we can only speculate that run-up might involve partial proteolysis of the channels themselves and/or associated Cavβ3 subunits. Both proteins contain several PEST motifs or PEST-like regions in their C and N termini (Fig. 14), and there is convincing evidence for a functional relevance of these sites as potential cleavage sites. For example, deletion of PEST-like regions in the Cavβ3 subunit increased its half-life, stimulated currents mediated by coexpressed Cav2.2 channels, and reduced voltage-dependent inactivation in HEK-293 cells (Sandoval et al., 2006). Likewise, intracellular protease application has been shown to produce strong stimulation of VGCCs and a partial loss of fast voltage-dependent inactivation, which has been linked to C-terminal cleavage of a conserved autoinhibitory region in the pore-forming Cavα1 subunit (Wei et al., 1994; Klöckner et al., 1995; Gao et al., 2001; Mikala et al., 2003). Given that C-terminal truncation of L-, N- and P/Q-type VGCCs by endogenous proteases has been demonstrated before in HEK-293 cells (Kubodera et al., 2003; Gomez-Ospina et al., 2006) and in vivo (Gerhardstein et al., 2000; Abele and Yang, 2012), it seems not so far off that the structural determinants for Cav2.3 channel run-up could also be located in this region. With regard to the Leu effects on run-down, it may be important that L-type channels can also be cleaved at PEST regions within the core of the Cavα1 subunit, which has been shown to disrupt channel function (Groth et al., 2014; Michailidis et al., 2014). That Cav2.3 channels lack such regions might account for the inability of Leu to sustain ATP-induced protection. On the other hand, this cannot explain the puzzling finding that Leu actually impaired the protective effects of ATP on run-down. We can only speculate that the latter might be related to differential phosphorylation of long and short forms or changes in the accessibility of certain sites. Because Leu could also alter the effects of ATP in some other way, further studies will clearly be required to delineate the underlying processes. However, our finding that protein kinase inhibition alone replicated most of the gating changes during run-down in intact cells leads us to conclude that proteolysis is either not required for these effects or that it does also occur in intact cells. Conclusion In summary, our findings show that run-down of cloned Cav2.3 channels in dialyzed cells is associated with progressive changes in channel voltage-dependence and a decrease in the total number of functional channels, which can be prevented by ATP through maintained protein phosphorylation and replicated by protein kinase inhibition in intact cells. These findings distinguish the process from run-down in other Cav2 channels and suggest that one or more sites on the channels themselves or an associated protein must be phosphorylated to maintain them in a functional state. In addition, changes in the phosphorylation of a subset of sites may directly influence channel gating, possibly through bulk electrostatics. Although additional studies will be required to delineate the underlying mechanisms and their relevance in native cells, our results provide a reliable baseline that could stimulate further work on Cav2.3 channel regulation by protein kinases, phosphatases, and possibly proteases. Acknowledgments We thank Tobias Pook for technical help and supply of software for facilitated analysis and Mrs. Renate Clemens for her excellent and permanent assistance. This work was financially supported by the Deutsche Forschungsgemeinschaft (SCHN 387/21-1). The authors declare no competing financial interests. Author contributions: F. Neumaier contributed to the conception and design of the work, performed the experiments, analyzed and interpreted the data, prepared the figures, wrote the first draft of the manuscript, and coordinated its critical revision. T. Schneider contributed to the conception and design of the work, interpretation of the data, and critical revision of the manuscript. S. Alpdogan and J. Hescheler contributed to analysis and interpretation of the data and drafting of the manuscript. All authors have approved the final version of the manuscript and agree to be accountable for all aspects of the work. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed. Experiments were carried out in the Institute for Neurophysiology at the University of Cologne, Germany. Sharona E. Gordon served as editor. ==== Refs Abele, K., and J. Yang. 2012. Regulation of voltage-gated calcium channels by proteolysis. Sheng Li Xue Bao. 64 :504–514.23090491 Aita, Y., N. Kurebayashi, S. Hirose, and A.D. Maturana. 2011. Protein kinase D regulates the human cardiac L-type voltage-gated calcium channel through serine 1884. FEBS Lett. 585 :3903–3906. 10.1016/j.febslet.2011.11.011 22100296 Almog, M., and A. Korngreen. 2009. Characterization of voltage-gated Ca2+ conductances in layer 5 neocortical pyramidal neurons from rats. PLoS One. 4 :e4841. 10.1371/journal.pone.0004841 19337371 Armstrong, D., and R. Eckert. 1987. Voltage-activated calcium channels that must be phosphorylated to respond to membrane depolarization. Proc. Natl. Acad. Sci. USA. 84 :2518–2522. 10.1073/pnas.84.8.2518 2436233 Arnould, T., D. Janssens, C. Michiels, and J. Remacle. 1996. Effect of aescine on hypoxia-induced activation of human endothelial cells. Eur. J. Pharmacol. 315 :227–233. 10.1016/S0014-2999(96)00645-0 8960888 Becher, I., M.M. Savitski, M.F. Savitski, C. Hopf, M. Bantscheff, and G. Drewes. 2013. Affinity profiling of the cellular kinome for the nucleotide cofactors ATP, ADP, and GTP. ACS Chem. Biol. 8 :599–607. 10.1021/cb3005879 23215245 Becq, F. 1996. Ionic channel rundown in excised membrane patches. Biochim. Biophys. Acta. 1286 :53–63. 10.1016/0304-4157(96)00002-0 8634323 Benquet, P., J.L. Guen, G. Dayanithi, Y. Pichon, and F. Tiaho. 1999. omega-AgaIVA-sensitive (P/Q-type) and -resistant (R-type) high-voltage-activated Ba(2+) currents in embryonic cockroach brain neurons. J. Neurophysiol. 82 :2284–2293. 10.1152/jn.1999.82.5.2284 10561406 Blesneac, I., J. Chemin, I. Bidaud, S. Huc-Brandt, F. Vandermoere, and P. Lory. 2015. Phosphorylation of the Cav3.2 T-type calcium channel directly regulates its gating properties. Proc. Natl. Acad. Sci. USA. 112 :13705–13710. 10.1073/pnas.1511740112 26483470 Cerda, O., J.-H. Baek, and J.S. Trimmer. 2011. Mining recent brain proteomic databases for ion channel phosphosite nuggets. J. Gen. Physiol. 137 :3–16. 10.1085/jgp.201010555 21149544 Chad, J.E., and R. Eckert. 1986. An enzymatic mechanism for calcium current inactivation in dialysed Helix neurones. J. Physiol. 378 :31–51. 10.1113/jphysiol.1986.sp016206 2432251 Chad, J., D. Kalman, and D. Armstrong. 1987. The role of cyclic AMP-dependent phosphorylation in the maintenance and modulation of voltage-activated calcium channels. Soc. Gen. Physiol. Ser. 42 :167–186.2850609 Clare, J.J. 2006. Functional expression of ion channels in mammalian systems. In Expression and Analysis of Recombinant Ion Channels. J.J. Clare and D.J. Trezise, editors. Wiley-VCH, Weinheim, Germany. 79–109. Costantin, J.L., N. Qin, M.N. Waxham, L. Birnbaumer, and E. Stefani. 1999. Complete reversal of run-down in rabbit cardiac Ca2+ channels by patch-cramming in Xenopus oocytes; partial reversal by protein kinase A. Pflugers Arch. 437 :888–894. 10.1007/s004240050859 10370067 Cota, G. 1986. Calcium channel currents in pars intermedia cells of the rat pituitary gland. Kinetic properties and washout during intracellular dialysis. J. Gen. Physiol. 88 :83–105. 10.1085/jgp.88.1.83 2426390 Crump, S.M., R.N. Correll, E.A. Schroder, W.C. Lester, B.S. Finlin, D.A. Andres, and J. Satin. 2006. L-type calcium channel alpha-subunit and protein kinase inhibitors modulate Rem-mediated regulation of current. Am. J. Physiol. Heart Circ. Physiol. 291 :H1959–H1971. 10.1152/ajpheart.00956.2005 16648185 DeMartino, G.N., and D.E. Croall. 1985. Calcium-dependent proteases from liver and heart. Prog. Clin. Biol. Res. 180 :117–126.2994078 Elhamdani, A., J.-L. Bossu, and A. Feltz. 1994. Evolution of the Ca2+ current during dialysis of isolated bovine chromaffin cells: effect of internal calcium. Cell Calcium. 16 :357–366. 10.1016/0143-4160(94)90029-9 7859250 Elhamdani, A., J.-L. Bossu, and A. Feltz. 1995. ATP and G proteins affect the runup of the Ca2+ current in bovine chromaffin cells. Pflugers Arch. 430 :410–419. 10.1007/BF00373917 7491266 Fan, J.-S., and P. Palade. 1998. Perforated patch recording with beta-escin. Pflugers Arch. 436 :1021–1023. 10.1007/PL00008086 9799421 Feng, R., J. Xu, E. Minobe, A. Kameyama, L. Yang, L. Yu, L. Hao, and M. Kameyama. 2014. Adenosine triphosphate regulates the activity of guinea pig Cav1.2 channel by direct binding to the channel in a dose-dependent manner. Am. J. Physiol. Cell Physiol. 306 :C856–C863. 10.1152/ajpcell.00368.2013 24553186 Fomina, A.F., and E.S. Levitan. 1997. Control of Ca2+ channel current and exocytosis in rat lactotrophs by basally active protein kinase C and calcineurin. Neuroscience. 78 :523–531. 10.1016/S0306-4522(96)00571-4 9145807 Fu, L.-Y., F. Wang, X.-S. Chen, H.-Y. Zhou, W.-X. Yao, G.-J. Xia, and M.-X. Jiang. 2003. Perforated patch recording of L-type calcium current with beta-escin in guinea pig ventricular myocytes. Acta Pharmacol. Sin. 24 :1094–1098.14627491 Fuller, M.D., M.A. Emrick, M. Sadilek, T. Scheuer, and W.A. Catterall. 2010. Molecular mechanism of calcium channel regulation in the fight-or-flight response. Sci. Signal. 3 :ra70. 10.1126/scisignal.2001152 20876873 Gamper, N., V. Reznikov, Y. Yamada, J. Yang, and M.S. Shapiro. 2004. Phosphatidylinositol [correction] 4,5-bisphosphate signals underlie receptor-specific Gq/11-mediated modulation of N-type Ca2+ channels. J. Neurosci. 24 :10980–10992. 10.1523/JNEUROSCI.3869-04.2004 15574748 Gao, T., A.E. Cuadra, H. Ma, M. Bunemann, B.L. Gerhardstein, T. Cheng, R.T. Eick, and M.M. Hosey. 2001. C-terminal fragments of the alpha 1C (CaV1.2) subunit associate with and regulate L-type calcium channels containing C-terminal-truncated alpha 1C subunits. J. Biol. Chem. 276 :21089–21097. 10.1074/jbc.M008000200 11274161 Gerhardstein, B.L., T. Gao, M. Bünemann, T.S. Puri, A. Adair, H. Ma, and M.M. Hosey. 2000. Proteolytic processing of the C terminus of the alpha(1C) subunit of L-type calcium channels and the role of a proline-rich domain in membrane tethering of proteolytic fragments. J. Biol. Chem. 275 :8556–8563. 10.1074/jbc.275.12.8556 10722694 Gomez-Ospina, N., F. Tsuruta, O. Barreto-Chang, L. Hu, and R. Dolmetsch. 2006. The C terminus of the L-type voltage-gated calcium channel CaV1.2 encodes a transcription factor. Cell. 127 :591–606. 10.1016/j.cell.2006.10.017 17081980 Goswami, T., X. Li, A.M. Smith, E.M. Luderowski, J.J. Vincent, J. Rush, and B.A. Ballif. 2012. Comparative phosphoproteomic analysis of neonatal and adult murine brain. Proteomics. 12 :2185–2189. 10.1002/pmic.201200003 22807455 Groth, R.D., N.N. Tirko, and R.W. Tsien. 2014. CaV1.2 calcium channels: Just cut out to be regulated? Neuron. 82 :939–940. 10.1016/j.neuron.2014.05.030 24908477 Hadley, R.W., and W.J. Lederer. 1991. Properties of L-type calcium channel gating current in isolated guinea pig ventricular myocytes. J. Gen. Physiol. 98 :265–285. 10.1085/jgp.98.2.265 1658192 Hamill, O.P., A. Marty, E. Neher, B. Sakmann, and F.J. Sigworth. 1981. Improved patch-clamp techniques for high-resolution current recording from cells and cell-free membrane patches. Pflugers Arch. 391 :85–100. 10.1007/BF00656997 6270629 Hell, J.W., C.T. Yokoyama, L.J. Breeze, C. Chavkin, and W.A. Catterall. 1995. Phosphorylation of presynaptic and postsynaptic calcium channels by cAMP-dependent protein kinase in hippocampal neurons. EMBO J. 14 :3036–3044.7621818 Hilaire, C., S. Diochot, G. Desmadryl, S. Richard, and J. Valmier. 1997. Toxin-resistant calcium currents in embryonic mouse sensory neurons. Neuroscience. 80 :267–276. 10.1016/S0306-4522(97)00101-2 9252237 Hilgemann, D.W. 1997. Cytoplasmic ATP-dependent regulation of ion transporters and channels: Mechanisms and messengers. Annu. Rev. Physiol. 59 :193–220. 10.1146/annurev.physiol.59.1.193 9074761 Horváth, F., L. Erdei, B. Wodala, U. Homann, and G. Thiel. 2002. Deltamethrin rescues run down of K+ outward rectifying channels in Vicia faba guard cells. Acta Biol. Szeged. 46 :19–20. Huttlin, E.L., M.P. Jedrychowski, J.E. Elias, T. Goswami, R. Rad, S.A. Beausoleil, J. Villén, W. Haas, M.E. Sowa, and S.P. Gygi. 2010. A tissue-specific atlas of mouse protein phosphorylation and expression. Cell. 143 :1174–1189. 10.1016/j.cell.2010.12.001 21183079 Jeong, J.-Y., H.-J. Kweon, and B.-C. Suh. 2016. Dual regulation of R-type Cav2.3 channels by M1 muscarinic receptors. Mol. Cells. 39 :322–329. 10.14348/molcells.2016.2292 26923189 Johnson, B.D., J.P. Brousal, B.Z. Peterson, P.A. Gallombardo, G.H. Hockerman, Y. Lai, T. Scheuer, and W.A. Catterall. 1997. Modulation of the cloned skeletal muscle L-type Ca2+ channel by anchored cAMP-dependent protein kinase. J. Neurosci. 17 :1243–1255.9006969 Kaur, G., A. Pinggera, N.J. Ortner, A. Lieb, M.J. Sinnegger-Brauns, V. Yarov-Yarovoy, G.J. Obermair, B.E. Flucher, and J. Striessnig. 2015. A polybasic plasma membrane binding motif in the I-II linker stabilizes voltage-gated Cav1.2 calcium channel function. J. Biol. Chem. 290 :21086–21100. 10.1074/jbc.M115.645671 26100638 Kepplinger, K.J.F., and C. Romanin. 2005. The run-down phenomenon of Ca2+ channels. In Voltage-Gated Calcium Channels. Springer, Boston. 219–230. Klöckner, U., G. Mikala, M. Varadi, G. Varadi, and A. Schwartz. 1995. Involvement of the carboxyl-terminal region of the α1 subunit in voltage-dependent inactivation of cardiac calcium channels. J. Biol. Chem. 270 :17306–17310. 10.1074/jbc.270.29.17306 7615531 Kubodera, T., T. Yokota, K. Ohwada, K. Ishikawa, H. Miura, T. Matsuoka, and H. Mizusawa. 2003. Proteolytic cleavage and cellular toxicity of the human α1A calcium channel in spinocerebellar ataxia type 6. Neurosci. Lett. 341 :74–78. 10.1016/S0304-3940(03)00156-3 12676347 Leloup, L., H. Shao, Y.H. Bae, B. Deasy, D. Stolz, P. Roy, and A. Wells. 2010. m-Calpain activation is regulated by its membrane localization and by its binding to phosphatidylinositol 4,5-bisphosphate. J. Biol. Chem. 285 :33549–33566. 10.1074/jbc.M110.123604 20729206 Liu, L., and A.R. Rittenhouse. 2003. Arachidonic acid mediates muscarinic inhibition and enhancement of N-type Ca2+ current in sympathetic neurons. Proc. Natl. Acad. Sci. USA. 100 :295–300. 10.1073/pnas.0136826100 12496347 Liu, L., C.F. Barrett, and A.R. Rittenhouse. 2001. Arachidonic acid both inhibits and enhances whole cell calcium currents in rat sympathetic neurons. Am. J. Physiol. Cell Physiol. 280 :C1293–C1305. 10.1152/ajpcell.2001.280.5.C1293 11287343 Liu, L., R. Zhao, Y. Bai, L.F. Stanish, J.E. Evans, M.J. Sanderson, J.V. Bonventre, and A.R. Rittenhouse. 2006. M1 muscarinic receptors inhibit L-type Ca2+ current and M-current by divergent signal transduction cascades. J. Neurosci. 26 :11588–11598. 10.1523/JNEUROSCI.2102-06.2006 17093080 Lundby, A., A. Secher, K. Lage, N.B. Nordsborg, A. Dmytriyev, C. Lundby, and J.V. Olsen. 2012. Quantitative maps of protein phosphorylation sites across 14 different rat organs and tissues. Nat. Commun. 3 :876. 10.1038/ncomms1871 22673903 McDonald, T.F., S. Pelzer, W. Trautwein, and D.J. Pelzer. 1994. Regulation and modulation of calcium channels in cardiac, skeletal, and smooth muscle cells. Physiol. Rev. 74 :365–507. 10.1152/physrev.1994.74.2.365 8171118 Michailidis, I.E., K. Abele-Henckels, W.K. Zhang, B. Lin, Y. Yu, L.S. Geyman, M.D. Ehlers, E.A. Pnevmatikakis, and J. Yang. 2014. Age-related homeostatic midchannel proteolysis of neuronal L-type voltage-gated Ca2+ channels. Neuron. 82 :1045–1057. 10.1016/j.neuron.2014.04.017 24908485 Mikala, G., I. Bodi, U. Klöckner, M. Varadi, G. Varadi, S.E. Koch, and A. Schwartz. 2003. Characterization of auto-regulation of the human cardiac α1 subunit of the L-type calcium channel: importance of the C-terminus. Mol. Cell. Biochem. 250 :81–89. 10.1023/A:1024910605389 12962146 Munton, R.P., R. Tweedie-Cullen, M. Livingstone-Zatchej, F. Weinandy, M. Waidelich, D. Longo, P. Gehrig, F. Potthast, D. Rutishauser, B. Gerrits, 2007. Qualitative and quantitative analyses of protein phosphorylation in naive and stimulated mouse synaptosomal preparations. Mol. Cell. Proteomics. 6 :283–293. 10.1074/mcp.M600046-MCP200 17114649 Nakashima, Y.M., S.M. Todorovic, A. Pereverzev, J. Hescheler, T. Schneider, and C.J. Lingle. 1998. Properties of Ba2+ currents arising from human α1E and α1Ebeta3 constructs expressed in HEK293 cells: Physiology, pharmacology, and comparison to native T-type Ba2+ currents. Neuropharmacology. 37 :957–972. 10.1016/S0028-3908(98)00097-5 9833625 Omasits, U., C.H. Ahrens, S. Müller, and B. Wollscheid. 2014. Protter: Interactive protein feature visualization and integration with experimental proteomic data. Bioinformatics. 30 :884–886. 10.1093/bioinformatics/btt607 24162465 Park, S.W., D. Byun, Y.M. Bae, B.H. Choi, S.H. Park, B. Kim, and S.I. Cho. 2007. Effects of fluid flow on voltage-dependent calcium channels in rat vascular myocytes: fluid flow as a shear stress and a source of artifacts during patch-clamp studies. Biochem. Biophys. Res. Commun. 358 :1021–1027. 10.1016/j.bbrc.2007.05.024 17524365 Peng, S.Q., R.K. Hajela, and W.D. Atchison. 2005. Fluid flow-induced increase in inward Ba2+ current expressed in HEK293 cells transiently transfected with human neuronal L-type Ca2+ channels. Brain Res. 1045 :116–123. 10.1016/j.brainres.2005.03.039 15910769 Perez-Reyes, E., W. Yuan, X. Wei, and D.M. Bers. 1994. Regulation of the cloned L-type cardiac calcium channel by cyclic-AMP-dependent protein kinase. FEBS Lett. 342 :119–123. 10.1016/0014-5793(94)80484-2 8143862 Perozo, E., and F. Bezanilla. 1990. Phosphorylation affects voltage gating of the delayed rectifier K+ channel by electrostatic interactions. Neuron. 5 :685–690. 10.1016/0896-6273(90)90222-2 2223093 Perozo, E., and F. Bezanilla. 1991. Phosphorylation of K+ channels in the squid giant axon. A mechanistic analysis. J. Bioenerg. Biomembr. 23 :599–613. 10.1007/BF00785813 1917910 Rinschen, M.M., M.-J. Yu, G. Wang, E.S. Boja, J.D. Hoffert, T. Pisitkun, and M.A. Knepper. 2010. Quantitative phosphoproteomic analysis reveals vasopressin V2-receptor-dependent signaling pathways in renal collecting duct cells. Proc. Natl. Acad. Sci. USA. 107 :3882–3887. 10.1073/pnas.0910646107 20139300 Sandoval, A., N. Oviedo, A. Tadmouri, T. Avila, M. De Waard, and R. Felix. 2006. Two PEST-like motifs regulate Ca2+/calpain-mediated cleavage of the CaVbeta3 subunit and provide important determinants for neuronal Ca2+ channel activity. Eur. J. Neurosci. 23 :2311–2320. 10.1111/j.1460-9568.2006.04749.x 16706839 Seydl, K., J.-O. Karlsson, A. Dominik, H. Gruber, and C. Romanin. 1995. Action of calpastatin in prevention of cardiac L-type Ca2+ channel run-down cannot be mimicked by synthetic calpain inhibitors. Pflugers Arch. 429 :503–510. 10.1007/BF00704155 7617440 Shiraha, H., A. Glading, J. Chou, Z. Jia, and A. Wells. 2002. Activation of m-calpain (calpain II) by epidermal growth factor is limited by protein kinase A phosphorylation of m-calpain. Mol. Cell. Biol. 22 :2716–2727. 10.1128/MCB.22.8.2716-2727.2002 11909964 Smith, S.D., Z. Jia, K.K. Huynh, A. Wells, and J.S. Elce. 2003. Glutamate substitutions at a PKA consensus site are consistent with inactivation of calpain by phosphorylation. FEBS Lett. 542 :115–118. 10.1016/S0014-5793(03)00361-2 12729909 Suh, B.-C., K. Leal, and B. Hille. 2010. Modulation of high-voltage activated Ca2+ channels by membrane phosphatidylinositol 4,5-bisphosphate. Neuron. 67 :224–238. 10.1016/j.neuron.2010.07.001 20670831 Sun, W., R. Feng, H. Hu, F. Guo, Q. Gao, D. Shao, D. Yin, H. Wang, X. Sun, M. Zhao, 2014. The Ca2+-dependent interaction of calpastatin domain L with the C-terminal tail of the Cav1.2 channel. FEBS Lett. 588 :665–671. 10.1016/j.febslet.2014.01.019 24462690 Thomas, P., and T.G. Smart. 2005. HEK293 cell line: A vehicle for the expression of recombinant proteins. J. Pharmacol. Toxicol. Methods. 51 :187–200. 10.1016/j.vascn.2004.08.014 15862464 Tiaho, F., J. Nargeot, and S. Richard. 1993. Repriming of L-type calcium currents revealed during early whole-cell patch-clamp recordings in rat ventricular cells. J. Physiol. 463 :367–389. 10.1113/jphysiol.1993.sp019599 8246188 Tompa, P., R. Töth-Boconádi, and P. Friedrich. 2001. Frequency decoding of fast calcium oscillations by calpain. Cell Calcium. 29 :161–170. 10.1054/ceca.2000.0179 11162853 Toth, P.T., L.R. Shekter, G.H. Ma, L.H. Philipson, and R.J. Miller. 1996. Selective G-protein regulation of neuronal calcium channels. J. Neurosci. 16 :4617–4624.8764650 Trinidad, J.C., A. Thalhammer, C.G. Specht, A.J. Lynn, P.R. Baker, R. Schoepfer, and A.L. Burlingame. 2008. Quantitative analysis of synaptic phosphorylation and protein expression. Mol. Cell. Proteomics. 7 :684–696. 10.1074/mcp.M700170-MCP200 18056256 Trinidad, J.C., D.T. Barkan, B.F. Gulledge, A. Thalhammer, A. Sali, R. Schoepfer, and A.L. Burlingame. 2012. Global identification and characterization of both O-GlcNAcylation and phosphorylation at the murine synapse. Mol. Cell. Proteomics. 11 :215–229. 10.1074/mcp.O112.018366 22645316 Tweedie-Cullen, R.Y., J.M. Reck, and I.M. Mansuy. 2009. Comprehensive mapping of post-translational modifications on synaptic, nuclear, and histone proteins in the adult mouse brain. J. Proteome Res. 8 :4966–4982. 10.1021/pr9003739 19737024 Wang, W.-Y., L.-Y. Hao, E. Minobe, Z.A. Saud, D.-Y. Han, and M. Kameyama. 2009. CaMKII phosphorylates a threonine residue in the C-terminal tail of Cav1.2 Ca2+ channel and modulates the interaction of the channel with calmodulin. J. Physiol. Sci. 59 :283–290. 10.1007/s12576-009-0033-y 19340532 Wei, X., A. Neely, A.E. Lacerda, R. Olcese, E. Stefani, E. Perez-Reyes, and L. Birnbaumer. 1994. Modification of Ca2+ channel activity by deletions at the carboxyl terminus of the cardiac α1 subunit. J. Biol. Chem. 269 :1635–1640.7507480 Wipf, P., and R.J. Halter. 2005. Chemistry and biology of wortmannin. Org. Biomol. Chem. 3 :2053–2061. 10.1039/b504418a 15917886 Wiśniewski, J.R., N. Nagaraj, A. Zougman, F. Gnad, and M. Mann. 2010. Brain phosphoproteome obtained by a FASP-based method reveals plasma membrane protein topology. J. Proteome Res. 9 :3280–3289. 10.1021/pr1002214 20415495 Wu, L., C.S. Bauer, X.-G. Zhen, C. Xie, and J. Yang. 2002. Dual regulation of voltage-gated calcium channels by PtdIns(4,5)P2. Nature. 419 :947–952. 10.1038/nature01118 12410316 Xu, J., L. Yu, E. Minobe, L. Lu, M. Lei, and M. Kameyama. 2016. PKA and phosphatases attached to the CaV1.2 channel regulate channel activity in cell-free patches. Am. J. Physiol. Cell Physiol. 310 :C136–C141. 10.1152/ajpcell.00157.2015 26561637 Xue, Y., J. Ren, X. Gao, C. Jin, L. Wen, and X. Yao. 2008. GPS 2.0, a tool to predict kinase-specific phosphorylation sites in hierarchy. Mol. Cell. Proteomics. 7 :1598–1608. 10.1074/mcp.M700574-MCP200 18463090 Yu, L., J. Xu, E. Minobe, A. Kameyama, L. Yang, R. Feng, L. Hao, and M. Kameyama. 2016. Role of protein phosphatases in the run down of guinea pig cardiac Cav1.2 Ca2+ channels. Am. J. Physiol. Cell Physiol. 310 :C773–C779. 10.1152/ajpcell.00199.2015 26739491 Zhang, H., L.C. Craciun, T. Mirshahi, T. Rohács, C.M. Lopes, T. Jin, and D.E. Logothetis. 2003. PIP2 activates KCNQ channels, and its hydrolysis underlies receptor-mediated inhibition of M currents. Neuron. 37 :963–975. 10.1016/S0896-6273(03)00125-9 12670425 Zhen, X.-G., C. Xie, Y. Yamada, Y. Zhang, C. Doyle, and J. Yang. 2006. A single amino acid mutation attenuates rundown of voltage-gated calcium channels. FEBS Lett. 580 :5733–5738. 10.1016/j.febslet.2006.09.027 17010345

PMC005xxxxxx/PMC5839720.txt

==== Front J Gen Physiol J Gen Physiol jgp jgp The Journal of General Physiology 0022-1295 1540-7748 Rockefeller University Press 29437858 201711907 10.1085/jgp.201711907 Research Articles Research Article 501 507 509 Coarse-grained molecular dynamics simulations reveal lipid access pathways in P-glycoprotein Lipid access pathways in P-glycoprotein Barreto-Ojeda Estefania Corradi Valentina http://orcid.org/0000-0003-2587-0232 Gu Ruo-Xu http://orcid.org/0000-0001-5507-0688 Tieleman D. Peter Department of Biological Sciences and Centre for Molecular Simulation, University of Calgary, Calgary, Alberta, Canada Correspondence to D. Peter Tieleman: tieleman@ucalgary.ca 05 3 2018 150 3 417429 19 9 2017 17 1 2018 © 2018 Barreto-Ojeda et al. 2018 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). P-glycoprotein contributes to multidrug resistance by exporting a broad range of substrates across the cell membrane. Using molecular dynamics simulations, Barreto-Ojeda et al. identify key lipid-binding sites and reveal lipid access pathways toward the cavity of the transporter. P-glycoprotein (P-gp) exports a broad range of dissimilar compounds, including drugs, lipids, and lipid-like molecules. Because of its substrate promiscuity, P-gp is a key player in the development of cancer multidrug resistance. Although P-gp is one of the most studied ABC transporters, the mechanism by which its substrates access the cavity remains unclear. In this study, we perform coarse-grained molecular dynamics simulations to explore possible lipid access pathways in the inward-facing conformation of P-gp embedded in bilayers of different lipid compositions. In the inward-facing orientation, only lipids from the lower leaflet access the cavity of the transporter. We identify positively charged residues at the portals of P-gp that favor lipid entrance to the cavity, as well as lipid-binding sites at the portals and within the cavity, which is in good agreement with previous experimental studies. This work includes several examples of lipid pathways for phosphatidylcholine and phosphatidylethanolamine lipids that help elucidate the molecular mechanism of lipid binding in P-gp. Canadian Institutes of Health Research https://doi.org/10.13039/501100000024 Alberta Innovates - Health Solutions https://doi.org/10.13039/501100000145 Alberta Innovates - Technology Futures https://doi.org/10.13039/501100000146 Canadian Institutes for Health Research https://doi.org/10.13039/501100000024 MFE-140949 Canada Foundation for Innovation https://doi.org/10.13039/501100000196 Canada Research Chairs https://doi.org/10.13039/501100001804 ==== Body pmcIntroduction P-glycoprotein (P-gp), a member of the ATP-binding cassette (ABC) transporter superfamily, is responsible for the translocation of a wide range of compounds across the plasma membrane (Chen et al., 1986; Loo and Clarke, 1994). P-gp is found in several tissues facing an excretory compartment (e.g., in the apical surface of epithelial cells of the liver, pancreas, and small intestine; Thiebaut et al., 1987). It is also found in the brush border of kidneys and the endothelial cells of the blood–brain barrier (Eckford and Sharom, 2006; Sharom, 2014). Natural substrates of P-gp are toxins and xenobiotics, against which P-gp provides a mechanism of defense by exporting them outside the cell. Substrates binding to P-gp are structurally dissimilar compounds (Aller et al., 2009; Leong et al., 2012; Zhang et al., 2014), which include metabolites, drug molecules like analgesics and antibiotics, drug-like molecules such as diagnostic dyes, and lipids (van Helvoort et al., 1996; Bosch et al., 1997; Ambudkar et al., 2003; Marcoux et al., 2013; Ferreira et al., 2015). In addition, several studies have linked the overexpression of ABC transporters, most notably P-gp, with mechanisms of resistance in various cancer cells (Fletcher et al., 2010). Because of its broad substrate specificity, P-gp confers multidrug resistance (MDR) by pumping several pharmacological agents to the extracellular medium, preventing their accumulation inside the cell (Gros and Buschman, 1993; Sharom, 2014). The development of MDR is considered one of the major challenges in cancer treatment (Ambudkar et al., 2003; Szakács et al., 2006; Chen et al., 2016), and because P-gp acts as a key player, the study of its underlying mechanism is considerably important to improve current chemotherapeutic treatments (Gottesman, 1993; Sharom, 2014). In addition to having a drug transporter function, P-gp has also been proposed to act as a lipid flippase (Higgins and Gottesman, 1992), flipping lipids from the inner to the outer leaflet (Sharom, 2014). Direct interaction between lipid-based molecules and P-gp has been identified in previous studies (van Helvoort et al., 1996; Bosch et al., 1997; Marcoux et al., 2013; Sharom, 2014). Competition between transport and inhibition of ATPase activity suggests that lipids are not only active modulators of the transport mechanism of P-gp, but also of substrates (Sharom, 2011, 2014). The high affinity of the transporter for lipids and lipid-based molecules such as miltefosine and edelfosine, which are both phospholipid-based anticancer agents, has been linked with the flippase model (Sharom, 2014). Given the crucial role in cancer MDR, further understanding of the factors that determine lipid binding is essential for targeting P-gp inhibition as a means to increased drug delivery in chemotherapy treatments. The P-gp structure consists of two units linked by a flexible linker. Each unit includes an N-terminal transmembrane domain (TMD) followed by a C-terminal nucleotide-binding domain (NBD). The TMDs, with six transmembrane helices each (TMD1, H1–H6; TMD2, H7–H12), form the translocation pathway for substrates. TMDs and NBDs are coupled by intracellular loops. As for any other ABC transporter, the substrate translocation mechanism of P-gp is ATP driven (Loo and Clarke, 1995). ATP binding and hydrolysis at the NBDs induce NBD dimerization and dissociation, which in turn drive conformational changes in the TMDs (Loo and Clarke, 1994; Aller et al., 2009). To allow transport, P-gp switches between inward-facing and outward-facing states, which is relevant for substrate binding and release, respectively (Sharom, 2014; Chen et al., 2016). In the inward-facing conformation (Li et al., 2014), the two TMDs (TMD1 and TMD2) form a cavity open toward the cytoplasm that is enclosed by helices H1, H3, H5, H7, H11, and H12. In addition, two portals (referred to as P1 and P2) allow access of molecules directly from the bilayer. P1 is constituted by helices H4 and H6, and P2 is constituted by H10 and H12. Although several theoretical and experimental studies have targeted substrate binding (Leong et al., 2012; Marcoux et al., 2013; Ferreira et al., 2015), the mechanism by which P-gp can recognize substrates remains elusive. In this study, as a step toward a better understanding of how substrates access the translocation cavity, we use molecular dynamics (MD) simulations at a coarse-grained (CG) level of detail to investigate lipid pathways and lipid–protein interaction in P-gp using the Martini force field (Marrink et al., 2007). CG models are a powerful tool to investigate biomolecular processes of larger systems, or longer timescales, because of their simplified level of detail while maintaining chemical specificity (Marrink et al., 2007). CG models are particularly suitable to reach system size and timescales that are relevant to comprehensively study substrate pathways (Ingólfsson et al., 2014a). The Martini force field has been widely used to resolve lipid–protein interactions in diverse biomolecular systems (Marrink and Tieleman, 2013; Chavent et al., 2016). For example, Karo et al. (2012) explored the binding of mitochondrial creatine kinase to the membrane. Sengupta and Chattopadhyay (2012) analyzed the cholesterol-binding sites in the serotonin1A receptor. Schmidt et al. (2013) investigated phosphatidylinositol phosphate–binding events in the potassium channel Kir2.2. Given the importance of the role of lipids as possible substrates in P-gp, we investigated lipid–protein interactions using the crystallographic structure of mouse P-gp in the inward-facing conformation (Li et al., 2014), which was embedded in membranes of different ratios of POPC and POPE lipids. To investigate the P-gp lipid pathways, we first analyzed lipid access to the cavity. In the Martini force field (Marrink et al., 2007), even though the headgroups of POPC and POPE are zwitterionic, the interaction parameters differ based on hydrogen-bonding capability and solvation-free energy, among other properties (Marrink et al., 2004; de Jong et al., 2013). Because only the headgroup of POPE is a donor for hydrogen bonds, a different lipid–protein interaction might be possible, and given the membrane composition, it may berelevant for mechanistic details. For this reason, we first evaluated a possible selectivity between phosphatidylcholine (PC) and phosphatidylethanolamine (PE) lipids. Second, we studied the interactions between lipids and protein residues at the portals and inside the cavity. Finally, we analyzed the dynamics of the lipids inside the cavity of P-gp and identified key interactions between the lipid headgroups and the lipid tails with the residues in the cavity. In agreement with previous experimental studies (Aller et al., 2009; Marcoux et al., 2013), our results highlight the importance of positively charged residues at the portals, such as Arg355, which may facilitate lipid access to the cavity via electrostatic interactions. Materials and methods We used the refined structure of mouse P-gp as the input structure for MD simulations (Protein Data Bank accession no. 4M1M; Li et al., 2014). We conducted CG MD simulations of P-gp embedded in lipid bilayers of a different composition (Table 1). Because the most abundant phospholipids in the upper and lower leaflet for the mammalian plasma membrane (Ingólfsson et al., 2014b) are PC and PE, our simulations include pure POPC and POPE bilayers, a 1:1 symmetric POPC/POPE bilayer, and two asymmetric bilayers: one with the same PC/PE ratio as reported for the plasma membrane (Ingólfsson et al., 2014b), and, for comparison, one with a PC/PE ratio inverted between the two leaflets. All simulations were performed using the GROMACS 4.6 package (Hess et al., 2008) and the standard Martini force field 2.2 for protein and 2.0 for lipids (Marrink et al., 2007; Hess et al., 2008; Monticelli et al., 2008; Chavent et al., 2016). Table 1. Lipid composition of the bilayers System Composition S1 POPC S2 POPE S3 POPC/POPE 1:1 S4 Upper leaflet POPC/POPE 16:3, lower 11:16 S5 Upper leaflet POPC/POPE 11:16, lower 16:3 Simulation setup The atomistic structure of P-gp was converted to the corresponding CG model using the Martinize (de Jong et al., 2013) building tool. An elastic network was applied for atom pairs within a 1.0-nm cutoff distance (Periole et al., 2009). All protein bilayer systems were built using the INSANE (INSert membrANE) CG tool (Wassenaar et al., 2015). Membranes were solvated with standard Martini water. Ions were also added to neutralize the protein and to reach a final concentration of 150 mM NaCl. Each simulation box of 140 × 140 × 180 Å contained one copy of P-gp in CG representation, ∼600 lipids, ∼23,000 CG water, ∼250 Na+, and ∼260 Cl−. Energy minimization for all systems was performed using the steepest-descent algorithm with a force tolerance of 100 kJ mol−1nm−1 for convergence. For electrostatic interactions, we used a shift function to turn off the interactions between 0 and 1.2 nm. The relative dielectric constant was set to εr = 15, as is standard in nonpolarizable Martini. All subsequent equilibration simulations were performed for 8 ns in the NPT ensemble with a 10-fs time step. The Berendsen barostat (Berendsen et al., 1984) kept the systems at one bar with a relaxation time constant of τp = 1 ps, with semiisotropic pressure coupling. The reference temperature was set at 310K and controlled by a velocity-rescaling thermostat (Bussi et al., 2007) with a relaxation time constant of τT = 0.5 ps. Position restraints on the protein, with a force constant of 1,000 kJ mol−1nm−2, were applied gradually, first on all of the proteins, and then on the backbone beads only. Production runs were performed with a 20-fs integration time step. In the production simulations, the Berendsen barostat (Berendsen et al., 1984) was used to maintain the systems at one bar with a relaxation time constant of τp = 5 ps. The reference temperature of 310K was controlled by the velocity rescaling thermostat (Bussi et al., 2007) with a relaxation time constant of τT = 2 ps. The electrostatic interactions were treated as described above for energy minimization. For production run, each system was simulated for 20 µs. Lipid access and lipid–protein interaction analysis Lipid access The main cavity is defined here as formed by the residues on the helices lining the inner portion of the P-gp cavity (i.e., the side chain of the residues located along the helices H1, H3, and H5 and H7, H11, and H12). Residues on H2, H8, and H9 (external helices) and H4, H6, H10, and residues Tyr994 to Ser1002 of H12 (portals) were excluded. A lipid was considered as accessing the cavity if the corresponding PO4 bead was found for >200 consecutive nanoseconds within 8 Å of any of the side chain residues of the main cavity. Among different values, the 8 Å cutoff was the optimum distance to evaluate access events with no overestimation. The window of 200 ns guarantees that the access events are not caused by thermal fluctuations. Numerical analyses were performed using open-source Python NumPy and MDTraj libraries (van der Walt et al., 2011; McGibbon et al., 2015). Lipid occupancy inside the P-gp cavity A contact between a lipid molecule and the cavity was counted if the corresponding PO4 bead was located within 5 Å from any of the residues in the cavity of P-gp. We will refer to this as the contact condition. Initially, our analysis considered all possible residue–phosphate pairs interacting within 5, 8, and 10 Å. We found ∼450 residues of P-gp interacting with lipid headgroups within a cutoff of 10 Å. When the cutoff distance was reduced to 7 Å, the number of residues involved in lipid–protein interactions decreased by 60% compared with the 10-Å cutoff. Because our aim was to find the most representative lipid-binding sites, we reduced the cutoff to 5 Å, which registered ∼160 residues corresponding to the top 30% set of residues interacting with lipid headgroups. For each system, the number of lipids that satisfied the contact condition at the same time was calculated, representing the number of lipids that simultaneously remained inside the cavity of P-gp. For this, the restriction that a contact must be at least 200 ns was not applied. We also calculated the time during which a given number of lipid molecules occupied the cavity simultaneously. Lipid–protein interactions Portals To elucidate the lipid pathways in P-gp, we performed lipid–protein interaction analysis to identify the residues at the portals of P-gp following the contact condition. The portals of the protein are delimited by helices H4 and H6 (P1) and H10 and H12 (P2). We counted a binding event when the PO4 bead of a lipid molecule was located within 5 Å of the residues at the portals of P-gp. At the portals, the residues considered were Lys230 to His241 of H4, Ser345 to Gly356 of H6, Leu875 to Lys883 of H10, and Tyr994 to Ser1002 of H12. The analysis of binding events was performed on the lipid molecules selected from the previously described lipid access. Lipid-binding sites were identified as the top five residues for which the contact condition was satisfied during the last 15 µs of simulation time. The maximum contact time registered was normalized to 1.0. A color map distinguishes three levels of interaction: very high interactions are between 0.8 and 1(red); high interactions are between 0.4 and 0.7 (magenta); and moderate interactions are <0.4 (gray). The level of interaction was calculated for residues at the portals (Fig. 3) and in the cavity (Fig. 4). Cavity of P-gp Similar to the analysis of the lipid–protein interactions at the portals, a lipid molecule was considered bound if the corresponding PO4 bead was located within 5 Å of the residues at the cavity of P-gp for at least 200 ns. The lipid-binding sites and the levels of interaction follow the same definition as explained above for the top 10 residues, in addition to those interacting exclusively with PC (blue) and PE (cyan), as illustrated in Fig. 4. Sequence alignment The sequences were retrieved from the UniProt Knowledgebase (accession nos. P21447, P08183, P21440, and P21439; The UniProt Consortium, 2017) for mouse and human P-gp and mouse and human ABCB4, respectively. The sequence alignment was performed using MAFFT (Katoh et al., 2002; Katoh and Standley, 2013). Lipids inside the cavity To analyze the dynamics of the lipid headgroups and tails inside the P-gp cavity, we considered the PO4 bead of each lipid and the last bead of each tail (i.e., C4A and C4B beads). The lipid-binding sites for each tail were identified by the contact condition. For each frame, we defined the middle plane of the membrane z’ = 0 as the mean z coordinates between the upper and lower leaflet. Region I is defined as the highest region in the P-gp cavity, z’ > 0; region II corresponds to the region between the lower leaflet and the middle plane of the bilayer, where z’ < 0. Measurements of the z coordinate for the phosphate (PO4) bead and the last bead of each tail (oleoyl tail, C4A bead; palmitoyl tail, C4B bead) were performed. Occupancy maps Fractional lipid occupancy was calculated for POPE and POPC lipid types over the last 15 µs of each simulation system using the VolMap plugin in VMD 1.9.1 with a grid resolution of 1 Å (Humphrey et al., 1996). For the systems with pure POPC or POPE bilayers, maximum occupancy was found at 0.78 and 0.72, respectively. The maximum occupancy was found at 0.73 for the system with a 1:1 POPC/POPE ratio and at 0.78 and 0.74 for the asymmetric bilayer with the plasma ratio and the asymmetric system with the inverted plasma ratio, respectively. For visualization purposes, the isosurfaces in Fig. 6 are displayed at 20%, 50%, and 70% of the maximum value, set at 0.72 for all the systems, and the maps are superimposed with the protein structure extracted from the last frame of each simulation. Results Lipid access events were identified by the interaction between the lipid headgroup and residues of helices H1, H3, and H5 and H7, H11, and H12 following the contact condition (see Materials and methods, Lipid access). The number of lipid access events was determined by the number of lipid headgroups accessing the main cavity for both PC and PE lipid types. From our simulations, we found that these events involve only lipid molecules from the lower leaflet. The level of detail from the CG approach, together with the extensive sampling for lipid diffusion in our simulations, allowed us to observe reversible lipid binding in the cavity of P-gp as well as lipid confinement inside the cavity. Two examples are illustrated in Fig. 1. On the left, at 0 µs (red), the selected lipid is outside the cavity; during the first microsecond of the simulation, it entered the cavity through P1, where it remained for <1 µs. After ∼9 µs (white), the lipid again entered the cavity through P2, where it stayed for the remaining simulation time (blue). The right panel shows a lipid molecule that remained for ∼16 µs inside the cavity before diffusing outside. Despite the differences between POPE and POPC regarding the headgroup subtype previously mentioned, the results show no indication of selectivity toward any of the lipid types (Fig. 2 A). The differences in the number of access events for POPC and POPE lipids in the different membranes are not substantial. In what follows, we generally combine the results for both lipids from all the simulation systems. Figure 1. Two examples of PO4 bead trajectories found in the pure POPC bilayer. (Left) At 0 µs, the lipid molecule resides outside of the cavity. It accesses the cavity twice during the simulation time, first during the first microsecond of the simulation through P1 and then after ∼9 µs through P2. (Right) The selected lipid molecule remains inside the cavity for ∼16 µs before exiting through P1. P-gp is shown as a transparent surface, and the PO4 bead of the selected lipids is shown as a sphere, colored as a function of time. Figure 2. Number of lipid access events, lipid occupancy, and lipid-binding sites in the cavity and portals of P-gp. (A) Number of lipid access events. For each system (Table 1, S1–S5), the total number of lipid molecules PC (blue) and PE (cyan) in the lower leaflet is shown with the number of access events observed for each lipid type. (B) Lipid occupancy in the cavity. The number of PC or PE molecules occupying the cavity during the total simulation time is shown. Up to four lipid molecules can access the cavity simultaneously. (C) Lipid binding at the portals. Amount of simulation time for which binding events are detected at P1 and P2 for both PC and PE lipids. (D) Lipid binding in the cavity. Amount of simulation time for which binding events are detected for transmembrane helices H1, H5, H11, and H12. The interactions were detected along H1, H5, H11, and H12. Binding events at H3, H7, and H10 were not observed in our simulations. (E and F) Binding sites of lipid headgroups and tails in the cavity of P-gp normalized over the maximum number of interactions. (E) Mainly residues along H11 and H12 registered interactions with PO4 beads. (F) Binding sites for the oleoyl and palmitoyl tail were found mainly at H5. Lipid occupancy inside the P-gp cavity Using the contact definition described in Materials and methods, no PO4 beads of PC or PE lipids were found in the cavity for ∼45% and ∼60% of the simulation time, respectively (Fig. 2 B). Instead, water molecules were found inside. Overall, a single PC lipid remains in the cavity for ∼40% of the time, whereas a single PE remains during ∼30% of the simulation time. Transiently, up to a maximum of four lipid molecules can occupy the cavity simultaneously, although for short periods of time (<0.05%), as shown in Fig. 2 B. Lipid–protein interactions at the portals To identify the binding sites for POPC or POPE lipids, we looked at the residues located at the portals (helices H4 and H6 at P1; H10 and H12 at P2) that, during the simulation time, interacted with the headgroup of each lipid type. Every time a residue satisfied the contact condition, one binding event was counted (see Materials and methods). As shown in Fig. 2 C, overall, P1 registered more binding events than P2 for both lipid types. We then calculated the number of contacts between the PO4 beads to identify key binding residues at the portals involved in the lipid transit. Despite the chemical difference between POPC and POPE headgroups, the predominant binding sites found at the portals were the same for both lipid types: Arg355 located at H6 and Tyr994 at H12 (Fig. 3). Additional interactions were registered with the residues Lys230 and Lys235 at H4, Ser345 at H6, and Lys881 and Asp993 at H12. Figure 3. Binding residues for PC and PE lipids at the portals of P-gp. Binding residues are highlighted according to the color map. The maximum contact time registered was normalized to 1.0. A color map distinguishes three levels of interaction: very high interactions correspond to between 0.8 and 1 (red); high interactions correspond to between 0.4 and 0.7 (magenta); and moderate interactions correspond to <0.4 (gray). Lys230, Phe235, Ser345, and Arg355 at P1 and Asp993 and Tyr994 at P2 registered the top highest 10% of interactions. Lipid–protein interactions inside the cavity of P-gp Similar to the previous analysis for the residues at the portals, we identified residues inside the cavity of P-gp involved in interactions with lipid molecules. PC and PE binding events mostly involved residues of H12 (Fig. 2 D). PC registered over ∼50% of the simulation time. PE lipid binding was observed over ∼45% of the simulation time. Binding residues of H1, H5, and H11 registered interactions during <15% of the time, whereas residues at H3, H7, and H10 did not show binding events over the course of the simulation. Inside the cavity, the lipid–protein interactions mainly involved the residues Val984, Val987, and Ser988 at H12 and residues Phe938 and Thr941 located along H11 (Fig. 2 E and Fig. 4 B), all of them in the upper half of the P-gp cavity. Additional binding events were found for residues His60 of H1, Ser294 of H5, and Phe339 of H6 in TMD1 (Fig. 4 A). We observed a very high similarity between the PC and PE binding residues. Only two binding residues showed selectivity for lipid types: Ala338 for PC and Gly266 for PE, both with moderate interaction. The set of identified binding residues is conserved in mouse and human P-gp and ABCB4, as shown in the sequence alignment in Fig. 4 (C and D). In addition, a hot spot for lipid binding was observed at Arg828 on H9. However, given the orientation of the residue pointing toward the external side of the protein, the observed lipid–protein interaction does not occur in the cavity. This result will be addressed in the Discussion. Figure 4. Binding residues in the cavity of P-gp. (A and B) Residues Ser294 and Phe339 of TMD1 (A) and Val984, Val897, Ser988, and Ser989 of TMD2 (B) are highlighted in the atomistic representation of P-gp. Gly296 and Ala338 registered interactions only with headgroups of PC (blue) and PE (cyan), respectively. (C and D) Sequence alignment of TMD1 (C) and TMD2 (D) for human (h) and mouse (m) P-gp and ABCB4 highlighting the significant interacting residues, colored according to the same color scheme as in A and B and Fig. 3. Dynamics of the lipid molecules in the P-gp cavity To illustrate the dynamics of the PO4 bead and lipid tails inside the cavity of P-gp, we calculated the mean z coordinate over windows of 1 µs and plotted over the course of the simulation time (Fig. 5). In all our simulations, we observed that the PO4 bead remained mainly in the lower region of the membrane (Fig. 5, A and C) for both lipid types PC and PE. The tails of the lipids, however, occupy distinct regions and extend to the upper region of the cavity, where oscillatory and flipping events were detected (Fig. 5, B and D). We also identified the lipid tail–binding sites in the uppermost region of the cavity, as established by the contact condition (see Materials and methods). For the tails, we observed binding events mostly at H5 and, to a lesser extent, at H7. No binding residues were found on H1, H3, H11, or H12. Residues Phe299 and Leu301 showed very high interaction with the tails for both lipid types. Frequent interactions were also identified between the oleoyl tail and Tyr303 and Tyr306 and between the palmitoyl tail and Phe299 and Ala304 (Fig. 2 F). Figure 5. Dynamics of headgroups and tails inside the cavity of P-gp. (A–D) The mean z coordinate profile for the headgroup beads (PO4) and tail beads (C4A and C4B) plotted over 20 µs of simulation for a lipid that accesses the cavity two times (A and B) and a lipid that remains in the cavity (C and D). In A and C, P-gp is shown in the gray surface, and the position of the selected PO4 bead is shown as a colored sphere, from red to white to blue, as a function of time following the color map in Fig. 1. The z profiles of the PO4 bead (brown line) and the C4A and C4B beads (gray and black lines) were calculated over windows of 1 µs. The lipid occupancy maps highlight the ordering of the lipids in the immediate proximity of P-gp (see Materials and methods). We observed high occupancy values inside the cavity, independent of the lipid composition (Fig. 6). The presence of the lipids spanned the entire available volume due to the lipid tails reaching the upper region, reinforcing the results showed in Fig. 5. The lipid organization inside the cavity did not significantly differ between the PC/PE ratios considered in this work (Fig. 6, A and B). In addition, hot spots of lipid occupancy were also detected at the portals, where lipids compete for access to the cavity. Figure 6. Lipid occupancy maps for the POPC/POPE (1:1) and plasma ratio PC/PE systems. (A) POPC/POPE. (B) PC/PE. (Top) Lateral view of the bilayer. The protein is omitted for clarity, and the occupancy maps are clipped to highlight the occupancy signal in the inner region of the cavity. (Bottom) Top view from the extracellular side. From the last frame of the corresponding simulation system, residues with a high degree of interaction are shown as magenta (Phe339 and Phe938) and red spheres (Val987, Ser988, Val984, Tyr994, and Arg355), following the same color scheme from Figs. 3 and 4. PE and PC lipids are shown in cyan and blue lines, respectively. Lipid pathways Our simulations enabled us to identify trajectories of lipids that access the cavity of P-gp. Examples of trajectories of PC and PE lipids along 20 µs of simulation time are illustrated in Fig. 7. Simultaneous lipid uptake through the same portal can take place, as shown in Fig. 7 (A and C), during the last 8 µs of simulation (light blue color) in S4 (Table 1). No unique lipid pathway was detected because the approach to the portals is different for different lipids. The residence time of the lipids inside the cavity is different for every lipid. Figure 7. Lipid pathways found for PC and PE lipid types in plasma membrane ratio PC/PE. Different lipid headgroups (PO4 beads) access the cavity of P-gp through both portals as a function of time during the 20-µs simulation. (A and B) Two different views of PC lipid pathways. (C and D) Two different views of PE lipid pathways. P-gp is shown in a transparent surface representation. Each lipid shown is labeled in the order in which it accesses the cavity along the simulation by subindex i = [1,…,4]. Before accessing the cavity, the lipids are labeled as Li; after leaving the cavity, the lipids are labeled as Li’. Discussion P-gp is one of the best characterized ABC transporters (Ambudkar et al., 2003; Picchianti-Diamanti et al., 2014). Its broad substrate specificity allows for the transport of many chemically diverse molecules (Sharom, 2014; Chufan et al., 2015; Subramanian et al., 2016), and to date, several studies have suggested poly-specific drug-binding sites in the cavity of P-gp (Loo and Clarke, 1994; Borst et al., 2000; Chufan et al., 2015; Neumann et al., 2017). MD simulations have been extensively used to provide molecular details on the interactions between substrates and specific regions of the protein (Borst et al., 2000), most of them assuming specific binding sites. For example, using umbrella sampling methods, Subramanian et al. (2015) investigated spontaneous binding for morphine and nicardipine. More recently, Syed et al. (2017) used the drug-binding site in the transmembrane region from docking studies to test interactions with P-gp. For P-gp, it has been proposed that substrates access the cavity through the membrane (Eckford and Sharom, 2006; Subramanian et al., 2015), so lipid access pathways may also be relevant for elucidating drug transit and binding. The characterization of lipid-binding sites and lipid pathways in P-gp is relevant for a better understanding of drug binding to P-gp and other ABC transporters that are capable of translocating lipids (Tarling et al., 2013; Neumann et al., 2017). In addition, direct interaction between P-gp and lipid-based molecules such as miltefosine and edelfosine, which are lipid-based anticancer drug molecules, has been previously described (Eckford and Sharom, 2006; Syed et al., 2017). Toward this aim, we performed equilibrium CG MD simulations to explore how lipid molecules access the cavity of P-gp in the inward-facing conformation. With no previous assumption of specific binding sites, we identify key residues involved in the lipid binding at the portals and lipid occupancy in the cavity. Lipid access to the cavity of P-gp The structure of P-gp in the inward-facing conformation has two portals (P1 and P2) that provide access to the main cavity to molecules located in the lower leaflet of the membrane (Aller et al., 2009; Li et al., 2014). In all the analyzed simulations, we observed lipid diffusion toward the cavity of P-gp through the portals, but only for lipids of the inner leaflet, as previously suggested by Aller et al. (2009). No additional pathways for entrance to the cavity were detected. When comparing the two portals, a higher degree of lipid access was observed through P1 (Fig. 2 C). We link this result to the presence of the elbow helix of TMD2, which may screen the interactions at P2 between the lipid molecules and the residues located along H12, because it is partially covering H10. This feature also allows the direct exposure of Arg828 of H9, which faces the opening of P2, for lipid binding, although H9 is not included in the standard definition of portals followed in the literature (Aller et al., 2009; Li et al., 2014; Zhang et al., 2014). A limitation of the currently available structures of P-gp (Jin et al., 2012; Li et al., 2014; Szewczyk et al., 2015; Nicklisch et al., 2016; Esser et al., 2017) is the missing linker that connects NBD1 with TMD2. The structure of the linker has been reported for the human cystic fibrosis transmembrane conductance regulator (Liu et al., 2017), another member of the ABC transporter superfamily that acts as an anion channel where channel opening can be hampered by the presence of the linker (Liu et al., 2017). In P-gp, this flexible region has been proposed to be important for substrate specificity; additionally, it affects the conformational changes driven by ATP binding and hydrolysis at the NBD (Hrycyna et al., 1998; Sato et al., 2009; Ferreira et al., 2012; Esser et al., 2017). Although the presence of this linker may modify the substrate diffusion toward the cavity of P-gp (Ferreira et al., 2015) because of its location in the cytosolic side, oriented toward the opening between the TMD1 and TMD2, we believe it is unlikely that the linker will affect the lipid-binding sites reported in this study. Key lipid-binding residues in P-gp At the portals Our findings suggest that PC and PE lipids access the internal drug-binding pocket through the portals P1 and P2, which has also been reported for drug molecules in several studies (Loo and Clarke, 1994; Eckford and Sharom, 2006; Aller et al., 2009; Subramanian et al., 2015). It has also been suggested that a positively charged ring of Arg and Lys residues could drive substrates toward the cavity and promote lipid–protein interactions in ABC transporters in general via electrostatic interactions (Yu et al., 2015; Mehmood et al., 2016). From our simulations, two binding hot spots at P1 were identified: Lys230 of H4 and Arg355 of H6 (Fig. 3). This result is in perfect agreement with the experimental report from Marcoux et al. (2013), in which electrostatic interactions between the phosphate groups of lipid molecules and Lys230 were identified by mass spectrometry techniques. Lipid access events were also observed through P2, with key binding residues identified as Arg828 of H9, Lys881 of H10, and Asp993 and Tyr994 of H12 (Fig. 3). As mentioned above, Arg828 is not part of the canonical definition of portals (Li et al., 2014). However, because it is located at the bottom of H9 and its orientation is toward the exterior of the protein, facing H12, Arg828 is involved in the interactions for lipid access to the cavity through P2. Overall, the set of positively charged residues at the portals are key elements for lipid access because of the high affinity of Lys and Arg for phosphate groups via electrostatic interactions, promoting lipid entry into the cavity. In the cavity The funnel-shaped cavity of P-gp is lined by helices H1, H3, and H5 of TMD1 and H7, H11, and H12 of TMD2. Hydrophobic and aromatic residues in the inner cavity are important for substrate binding (Thiebaut et al., 1987; Aller et al., 2009; Martinez et al., 2014; Sharom, 2014; Mittra et al., 2017). Because of the amphipathic nature of lipids, we investigated the interactions between protein and lipid headgroups (PO4 beads) and lipid tails, respectively. The PO4 beads, once inside the cavity, established interactions mainly with residues of H11 and H12 (Fig. 2 D and Fig. 4 B). This result may reflect the structural architecture of the cavity of P-gp. The helical structure of H12 (residues 970–990) is interrupted by a second α helix (residues 991–994); therefore, the upper segment of H12 also belongs to the cavity. From our simulations, we observed interactions between the lipid headgroup and Phe938 on H11 (Fig. 2 E and Fig. 4 B), as reported by Aller et al. (2009) for the inhibitor verapamil. Additional interactions were also found for Ser989, Phe990, Pro992, Val984, Val987, and Ser988 on H12 and Thr941, Lys929, Lys 930, Met928, and Arg925 on H11 (Fig. 4). In the cavity of P-gp, our findings show the key role of Phe339 as a lipid-binding residue. This residue has also been shown to be involved in binding of the cyclic hexapeptide inhibitors cyclic-tris-(R)-valineselenazole (QZ59-RRR) and cyclic-tris-(S)-valineselenazole (QZ59-SSS; Li et al., 2014), while Zhang et al. (2015) corroborated the importance of Phe339 for doxorubicin access. If H12 drives many of the interactions between lipid headgroups and residues in the cavity, H5 appears to be the key player for the interactions between residues in the cavity of P-gp and the lipid tails (Fig. 2, E and F). Residues of H5 such as Leu300, Ile302, and Tyr303 were also reported as being binding sites for QZ59-RRR, QZ59-SSS, and verapamil (Aller et al., 2009). Overall, in the case of lipids, the tails extend toward the upper and more hydrophobic region of the cavity not explored by the PO4 beads (Fig. 5, B and D), with residues such as Phe299, Ile302, Tyr303, and Tyr306 mostly involved in the interactions with the oleoyl tail (Fig. 2 F). The location of substrates in the top of the cavity has been found in several experimental studies (Aller et al., 2009; Dolghih et al., 2011; Marcoux et al., 2013), in agreement with our simulations. Combined, these results reinforce the crucial role of H11, H12, and H5 in the interactions between protein and lipids as substrates. Many ABC transporters are involved in lipid transport (Borst et al., 2000; Eckford and Sharom, 2009; Fletcher et al., 2010; Quazi and Molday, 2011; Neumann et al., 2017). In particular, several studies have revealed lipid transport capabilities not only for P-gp (Sharom, 1997, 2014; Eckford and Sharom, 2009), but also for other members of the ABCB subfamily, such as ABCB4, involved in the translocation of PC lipids into the bile (Linton, 2015; Zhao et al., 2015). The interplay between lipids and ABC transporters is rather complex, being dependent, for example, as shown for ABCB4, on specific lipid compositions and on the presence of other lipid transporters (Zhao et al., 2015). Although lipid transport is mainly assessed using modified fluorescent lipids for P-gp and ABCB4, mass spectrometry studies have provided further details on the binding of different lipids to P-gp and their presence inside its cavity together with drugs (Marcoux et al., 2013). In this context, our study describes the molecular details of lipids’ access to the main cavity of the transporter. Although we focused our simulations on mouse P-gp, our findings can be translated to related ABC transporters, as the residues we identified as important for lipid–protein interactions are conserved in human P-gp as well as in human and mouse ABCB4 (Fig. 4, C and D). Conclusions In this work, we identify lipid pathways for P-gp in the inward-facing conformation in bilayers with different PC/PE lipid ratios. By performing equilibrium MD simulations on a timescale of microseconds, we show that the lipid binding events in the cavity and at the portals of P-gp are accessible and measurable. This approach reveals that lipid access events involve lipids from the lower leaflet only and identifies key lipid-binding residues at the portals and inside the cavity of P-gp. We find no selectivity for PC versus PE lipid entry or lipid binding in P-gp. Future directions include lipid interactions with the outward-facing conformation, for which there is currently no experimental structure, interactions with less prevalent lipids, and ultimately, a molecular understanding of the full transport process of physiological substrates in a realistic lipid environment. Acknowledgments This work was supported by the Canadian Institutes of Health Research. Additional support came from Alberta Innovates - Health Solutions (AIHS) and Alberta Innovates - Technology Futures (AITF). R.-X. Gu is supported by fellowships from AIHS and the Canadian Institutes for Health Research (funding reference number MFE-140949). D.P. Tieleman is an AIHS Scientist and AITF Strategic Chair in (Bio)Molecular Simulation. Simulations were run on Compute Canada machines, supported by the Canada Foundation for Innovation and their partners. This work was undertaken, in part, thanks to funding from the Canada Research Chairs program. The authors declare no competing financial interests. Author contributions: E. Barreto-Ojeda carried out simulations and analyses and wrote an initial draft of the paper with contributions from V. Corradi and D.P. Tieleman. V. Corradi carried out analyses. V. Corradi and D.P. Tieleman conceived the study. All authors contributed to the interpretation of the results and the writing of the final manuscript. José D. Faraldo-Gómez served as editor. ==== Refs Aller, S.G., J. Yu, A. Ward, Y. Weng, S. Chittaboina, R. Zhuo, P.M. Harrell, Y.T. Trinh, Q. Zhang, I.L. Urbatsch, and G. Chang. 2009. Structure of P-glycoprotein reveals a molecular basis for poly-specific drug binding. Science. 323 :1718–1722. 10.1126/science.1168750 19325113 Ambudkar, S.V., C. Kimchi-Sarfaty, Z.E. Sauna, and M.M. Gottesman. 2003. P-glycoprotein: from genomics to mechanism. Oncogene. 22 :7468–7485. 10.1038/sj.onc.1206948 14576852 Berendsen, H.J.C., J.P.M. Postma, W.F. van Gunsteren, A. DiNola, and J.R. Haak. 1984. Molecular dynamics with coupling to an external bath. J. Chem. Phys. 81 :3684–3690. 10.1063/1.448118 Borst, P., N. Zelcer, and A. van Helvoort. 2000. ABC transporters in lipid transport. Biochim. Biophys. Acta. 1486 :128–144. 10.1016/S1388-1981(00)00053-6 10856718 Bosch, I., K. Dunussi-Joannopoulos, R.L. Wu, S.T. Furlong, and J. Croop. 1997. Phosphatidylcholine and phosphatidylethanolamine behave as substrates of the human MDR1 P-glycoprotein. Biochemistry. 36 :5685–5694. 10.1021/bi962728r 9153408 Bussi, G., D. Donadio, and M. Parrinello. 2007. Canonical sampling through velocity rescaling. J. Chem. Phys. 126 :014101. 10.1063/1.2408420 17212484 Chavent, M., A.L. Duncan, and M.S. Sansom. 2016. Molecular dynamics simulations of membrane proteins and their interactions: from nanoscale to mesoscale. Curr. Opin. Struct. Biol. 40 :8–16. 10.1016/j.sbi.2016.06.007 27341016 Chen, C.J., J.E. Chin, K. Ueda, D.P. Clark, I. Pastan, M.M. Gottesman, and I.B. Roninson. 1986. Internal duplication and homology with bacterial transport proteins in the mdr1 (P-glycoprotein) gene from multidrug-resistant human cells. Cell. 47 :381–389. 10.1016/0092-8674(86)90595-7 2876781 Chen, Z., T. Shi, L. Zhang, P. Zhu, M. Deng, C. Huang, T. Hu, L. Jiang, and J. Li. 2016. Mammalian drug efflux transporters of the ATP binding cassette (ABC) family in multidrug resistance: A review of the past decade. Cancer Lett. 370 :153–164. 10.1016/j.canlet.2015.10.010 26499806 Chufan, E.E., H.-M. Sim, and S.V. Ambudkar. 2015. Molecular basis of the polyspecificity of P-glycoprotein (ABCB1): recent biochemical and structural studies. Adv. Cancer Res. 125 :71–96.25640267 de Jong, D.H., G. Singh, W.F.D. Bennett, C. Arnarez, T.A. Wassenaar, L.V. Schäfer, X. Periole, D.P. Tieleman, and S.J. Marrink. 2013. Improved Parameters for the Martini Coarse-Grained Protein Force Field. J. Chem. Theory Comput. 9 :687–697. 10.1021/ct300646g 26589065 Dolghih, E., C. Bryant, A.R. Renslo, and M.P. Jacobson. 2011. Predicting binding to p-glycoprotein by flexible receptor docking. PLOS Comput. Biol. 7 :e1002083. 10.1371/journal.pcbi.1002083 21731480 Eckford, P.D.W., and F.J. Sharom. 2006. P-glycoprotein (ABCB1) interacts directly with lipid-based anti-cancer drugs and platelet-activating factors. Biochem. Cell Biol. 84 :1022–1033. 10.1139/o06-196 17215888 Eckford, P.D.W., and F.J. Sharom. 2009. ABC efflux pump-based resistance to chemotherapy drugs. Chem. Rev. 109 :2989–3011. 10.1021/cr9000226 19583429 Esser, L., F. Zhou, K.M. Pluchino, J. Shiloach, J. Ma, W.K. Tang, C. Gutierrez, A. Zhang, S. Shukla, J.P. Madigan, 2017. Structures of the Multidrug Transporter P-glycoprotein Reveal Asymmetric ATP Binding and the Mechanism of Polyspecificity. J. Biol. Chem. 292 :446–461. 10.1074/jbc.M116.755884 27864369 Ferreira, R.J., M.-J.U. Ferreira, and D.J.V.A. Dos Santos. 2012. Insights on P-glycoprotein’s efflux mechanism obtained by molecular dynamics simulations. J. Chem. Theory Comput. 8 :1853–1864. 10.1021/ct300083m 26593820 Ferreira, R.J., M.-J.U. Ferreira, and D.J.V.A. Dos Santos. 2015. Do drugs have access to the P-glycoprotein drug-binding pocket through gates? J. Chem. Theory Comput. 11 :4525–4529. 10.1021/acs.jctc.5b00652 26574244 Fletcher, J.I., M. Haber, M.J. Henderson, and M.D. Norris. 2010. ABC transporters in cancer: more than just drug efflux pumps. Nat. Rev. Cancer. 10 :147–156. 10.1038/nrc2789 20075923 Gottesman, M.M. 1993. How cancer cells evade chemotherapy: sixteenth Richard and Hinda Rosenthal Foundation Award Lecture. Cancer Res. 53 :747–754.8094031 Gros, P., and E. Buschman. 1993. The mouse multidrug resistance gene family: structural and functional analysis. Int. Rev. Cytol. 137C :169–197.8098020 Hess, B., C. Kutzner, D. van der Spoel, and E. Lindahl. 2008. GROMACS 4: Algorithms for Highly Efficient, Load-Balanced, and Scalable Molecular Simulation. J. Chem. Theory Comput. 4 :435–447. 10.1021/ct700301q 26620784 Higgins, C.F., and M.M. Gottesman. 1992. Is the multidrug transporter a flippase? Trends Biochem. Sci. 17 :18–21. 10.1016/0968-0004(92)90419-A 1374941 Hrycyna, C.A., L.E. Airan, U.A. Germann, S.V. Ambudkar, I. Pastan, and M.M. Gottesman. 1998. Structural flexibility of the linker region of human P-glycoprotein permits ATP hydrolysis and drug transport. Biochemistry. 37 :13660–13673. 10.1021/bi9808823 9753453 Humphrey, W., A. Dalke, and K. Schulten. 1996. VMD: visual molecular dynamics. J. Mol. Graph. 14 :33–38: 27–28.8744570 Ingólfsson, H.I., C.A. Lopez, J.J. Uusitalo, D.H. de Jong, S.M. Gopal, X. Periole, and S.J. Marrink. 2014 a. The power of coarse graining in biomolecular simulations. Wiley Interdiscip. Rev. Comput. Mol. Sci. 4 :225–248. 10.1002/wcms.1169 25309628 Ingólfsson, H.I., M.N. Melo, F.J. van Eerden, C. Arnarez, C.A. Lopez, T.A. Wassenaar, X. Periole, A.H. de Vries, D.P. Tieleman, and S.J. Marrink. 2014 b. Lipid organization of the plasma membrane. J. Am. Chem. Soc. 136 :14554–14559. 10.1021/ja507832e 25229711 Jin, M.S., M.L. Oldham, Q. Zhang, and J. Chen. 2012. Crystal structure of the multidrug transporter P-glycoprotein from Caenorhabditis elegans. Nature. 490 :566–569. 10.1038/nature11448 23000902 Karo, J., P. Peterson, and M. Vendelin. 2012. Molecular dynamics simulations of creatine kinase and adenine nucleotide translocase in mitochondrial membrane patch. J. Biol. Chem. 287 :7467–7476. 10.1074/jbc.M111.332320 22241474 Katoh, K., and D.M. Standley. 2013. MAFFT multiple sequence alignment software version 7: improvements in performance and usability. Mol. Biol. Evol. 30 :772–780. 10.1093/molbev/mst010 23329690 Katoh, K., K. Misawa, K. Kuma, and T. Miyata. 2002. MAFFT: a novel method for rapid multiple sequence alignment based on fast Fourier transform. Nucleic Acids Res. 30 :3059–3066. 10.1093/nar/gkf436 12136088 Leong, M.K., H.-B. Chen, and Y.-H. Shih. 2012. Prediction of promiscuous p-glycoprotein inhibition using a novel machine learning scheme. PLoS One. 7 :e33829. 10.1371/journal.pone.0033829 22439003 Li, J., K.F. Jaimes, and S.G. Aller. 2014. Refined structures of mouse P-glycoprotein. Protein Sci. 23 :34–46. 10.1002/pro.2387 24155053 Linton, K.J. 2015. Lipid flopping in the liver. Biochem. Soc. Trans. 43 :1003–1010. 10.1042/BST20150132 26517915 Liu, F., Z. Zhang, L. Csanády, D.C. Gadsby, and J. Chen. 2017. Molecular Structure of the Human CFTR Ion Channel. Cell. 169 :85–95.e8. 10.1016/j.cell.2017.02.024 28340353 Loo, T.W., and D.M. Clarke. 1994. Reconstitution of drug-stimulated ATPase activity following co-expression of each half of human P-glycoprotein as separate polypeptides. J. Biol. Chem. 269 :7750–7755.7907331 Loo, T.W., and D.M. Clarke. 1995. Covalent modification of human P-glycoprotein mutants containing a single cysteine in either nucleotide-binding fold abolishes drug-stimulated ATPase activity. J. Biol. Chem. 270 :22957–22961. 10.1074/jbc.270.39.22957 7559432 Marcoux, J., S.C. Wang, A. Politis, E. Reading, J. Ma, P.C. Biggin, M. Zhou, H. Tao, Q. Zhang, G. Chang, 2013. Mass spectrometry reveals synergistic effects of nucleotides, lipids, and drugs binding to a multidrug resistance efflux pump. Proc. Natl. Acad. Sci. USA. 110 :9704–9709. 10.1073/pnas.1303888110 23690617 Marrink, S.J., and D.P. Tieleman. 2013. Perspective on the Martini model. Chem. Soc. Rev. 42 :6801–6822. 10.1039/c3cs60093a 23708257 Marrink, S.J., A.H. de Vries, and A.E. Mark. 2004. Coarse Grained Model for Semiquantitative Lipid Simulations. J. Phys. Chem. B. 108 :750–760. 10.1021/jp036508g Marrink, S.J., H.J. Risselada, S. Yefimov, D.P. Tieleman, and A.H. de Vries. 2007. The MARTINI force field: coarse grained model for biomolecular simulations. J. Phys. Chem. B. 111 :7812–7824. 10.1021/jp071097f 17569554 Martinez, L., O. Arnaud, E. Henin, H. Tao, V. Chaptal, R. Doshi, T. Andrieu, S. Dussurgey, M. Tod, A. Di Pietro, 2014. Understanding polyspecificity within the substrate-binding cavity of the human multidrug resistance P-glycoprotein. FEBS J. 281 :673–682. 10.1111/febs.12613 24219411 McGibbon, R.T., K.A. Beauchamp, M.P. Harrigan, C. Klein, J.M. Swails, C.X. Hernández, C.R. Schwantes, L.-P. Wang, T.J. Lane, and V.S. Pande. 2015. MDTraj: A Modern Open Library for the Analysis of Molecular Dynamics Trajectories. Biophys. J. 109 :1528–1532. 10.1016/j.bpj.2015.08.015 26488642 Mehmood, S., V. Corradi, H.G. Choudhury, R. Hussain, P. Becker, D. Axford, S. Zirah, S. Rebuffat, D.P. Tieleman, C.V. Robinson, and K. Beis. 2016. Structural and Functional Basis for Lipid Synergy on the Activity of the Antibacterial Peptide ABC Transporter McjD. J. Biol. Chem. 291 :21656–21668.27555327 Mittra, R., M. Pavy, N. Subramanian, A.M. George, M.L. O’Mara, I.D. Kerr, and R. Callaghan. 2017. Location of contact residues in pharmacologically distinct drug binding sites on P-glycoprotein. Biochem. Pharmacol. 123 :19–28. 10.1016/j.bcp.2016.10.002 27729218 Monticelli, L., S.K. Kandasamy, X. Periole, R.G. Larson, D.P. Tieleman, and S.-J. Marrink. 2008. The MARTINI Coarse-Grained Force Field: Extension to Proteins. J. Chem. Theory Comput. 4 :819–834. 10.1021/ct700324x 26621095 Neumann, J., D. Rose-Sperling, and U.A. Hellmich. 2017. Diverse relations between ABC transporters and lipids: An overview. Biochim. Biophys. Acta - Biomembr. 1859 :605–618.27693344 Nicklisch, S.C.T., S.D. Rees, A.P. McGrath, T. Gökirmak, L.T. Bonito, L.M. Vermeer, C. Cregger, G. Loewen, S. Sandin, G. Chang, and A. Hamdoun. 2016. Global marine pollutants inhibit P-glycoprotein: Environmental levels, inhibitory effects, and cocrystal structure. Sci. Adv. 2 :e1600001. 10.1126/sciadv.1600001 27152359 Periole, X., M. Cavalli, S.-J. Marrink, and M.A. Ceruso. 2009. Combining an Elastic Network With a Coarse-Grained Molecular Force Field: Structure, Dynamics, and Intermolecular Recognition. J. Chem. Theory Comput. 5 :2531–2543. 10.1021/ct9002114 26616630 Picchianti-Diamanti, A., M.M. Rosado, M. Scarsella, B. Laganà, and R. D’Amelio. 2014. P-glycoprotein and drug resistance in systemic autoimmune diseases. Int. J. Mol. Sci. 15 :4965–4976. 10.3390/ijms15034965 24658440 Quazi, F., and R.S. Molday. 2011. Lipid transport by mammalian ABC proteins. Essays Biochem. 50 :265–290. 10.1042/bse0500265 21967062 Sato, T., A. Kodan, Y. Kimura, K. Ueda, T. Nakatsu, and H. Kato. 2009. Functional role of the linker region in purified human P-glycoprotein. FEBS J. 276 :3504–3516. 10.1111/j.1742-4658.2009.07072.x 19490125 Schmidt, M.R., P.J. Stansfeld, S.J. Tucker, and M.S.P. Sansom. 2013. Simulation-based prediction of phosphatidylinositol 4,5-bisphosphate binding to an ion channel. Biochemistry. 52 :279–281. 10.1021/bi301350s 23270460 Sengupta, D., and A. Chattopadhyay. 2012. Identification of cholesterol binding sites in the serotonin1A receptor. J. Phys. Chem. B. 116 :12991–12996. 10.1021/jp309888u 23067252 Sharom, F.J. 1997. The P-glycoprotein efflux pump: how does it transport drugs? J. Membr. Biol. 160 :161–175. 10.1007/s002329900305 9425600 Sharom, F.J. 2011. The P-glycoprotein multidrug transporter. Essays Biochem. 50 :161–178. 10.1042/bse0500161 21967057 Sharom, F.J. 2014. Complex Interplay between the P-Glycoprotein Multidrug Efflux Pump and the Membrane: Its Role in Modulating Protein Function. Front. Oncol. 4 :41. 10.3389/fonc.2014.00041 24624364 Subramanian, N., K. Condic-Jurkic, A.E. Mark, and M.L. O’Mara. 2015. Identification of Possible Binding Sites for Morphine and Nicardipine on the Multidrug Transporter P-Glycoprotein Using Umbrella Sampling Techniques. J. Chem. Inf. Model. 55 :1202–1217. 10.1021/ci5007382 25938863 Subramanian, N., K. Condic-Jurkic, and M.L. O’Mara. 2016. Structural and dynamic perspectives on the promiscuous transport activity of P-glycoprotein. Neurochem. Int. 98 :146–152. 10.1016/j.neuint.2016.05.005 27180050 Syed, S.B., H. Arya, I.-H. Fu, T.-K. Yeh, L. Periyasamy, H.-P. Hsieh, and M.S. Coumar. 2017. Targeting P-glycoprotein: Investigation of piperine analogs for overcoming drug resistance in cancer. Sci. Rep. 7 :7972. 10.1038/s41598-017-08062-2 28801675 Szakács, G., J.K. Paterson, J.A. Ludwig, C. Booth-Genthe, and M.M. Gottesman. 2006. Targeting multidrug resistance in cancer. Nat. Rev. Drug Discov. 5 :219–234. 10.1038/nrd1984 16518375 Szewczyk, P., H. Tao, A.P. McGrath, M. Villaluz, S.D. Rees, S.C. Lee, R. Doshi, I.L. Urbatsch, Q. Zhang, and G. Chang. 2015. Snapshots of ligand entry, malleable binding and induced helical movement in P-glycoprotein. Acta Crystallogr. D Biol. Crystallogr. 71 :732–741. 10.1107/S1399004715000978 25760620 Tarling, E.J., T.Q. de Aguiar Vallim, and P.A. Edwards. 2013. Role of ABC transporters in lipid transport and human disease. Trends Endocrinol. Metab. 24 :342–350. 10.1016/j.tem.2013.01.006 23415156 The UniProt Consortium. 2017. UniProt: the universal protein knowledgebase. Nucleic Acids Res. 45 (D1 ):D158–D169. 10.1093/nar/gkw1099 27899622 Thiebaut, F., T. Tsuruo, H. Hamada, M.M. Gottesman, I. Pastan, and M.C. Willingham. 1987. Cellular localization of the multidrug-resistance gene product P-glycoprotein in normal human tissues. Proc. Natl. Acad. Sci. USA. 84 :7735–7738. 10.1073/pnas.84.21.7735 2444983 van der Walt, S., S.C. Colbert, and G. Varoquaux. 2011. The NumPy Array: A Structure for Efficient Numerical Computation. Comput. Sci. Eng. 13 :22–30. 10.1109/MCSE.2011.37 van Helvoort, A., A.J. Smith, H. Sprong, I. Fritzsche, A.H. Schinkel, P. Borst, and G. van Meer. 1996. MDR1 P-glycoprotein is a lipid translocase of broad specificity, while MDR3 P-glycoprotein specifically translocates phosphatidylcholine. Cell. 87 :507–517. 10.1016/S0092-8674(00)81370-7 8898203 Wassenaar, T.A., H.I. Ingólfsson, R.A. Böckmann, D.P. Tieleman, and S.J. Marrink. 2015. Computational Lipidomics with insane: A Versatile Tool for Generating Custom Membranes for Molecular Simulations. J. Chem. Theory Comput. 11 :2144–2155. 10.1021/acs.jctc.5b00209 26574417 Yu, J., J. Ge, J. Heuveling, E. Schneider, and M. Yang. 2015. Structural basis for substrate specificity of an amino acid ABC transporter. Proc. Natl. Acad. Sci. USA. 112 :5243–5248. 10.1073/pnas.1415037112 25848002 Zhang, J., T. Sun, L. Liang, T. Wu, and Q. Wang. 2014. Drug promiscuity of P-glycoprotein and its mechanism of interaction with paclitaxel and doxorubicin. Soft Matter. 10 :438–445. 10.1039/C3SM52499J 24652302 Zhang, J., D. Li, T. Sun, L. Liang, and Q. Wang. 2015. Interaction of P-glycoprotein with anti-tumor drugs: the site, gate and pathway. Soft Matter. 11 :6633–6641. 10.1039/C5SM01028D 26205623 Zhao, Y., M. Ishigami, K. Nagao, K. Hanada, N. Kono, H. Arai, M. Matsuo, N. Kioka, and K. Ueda. 2015. ABCB4 exports phosphatidylcholine in a sphingomyelin-dependent manner. J. Lipid Res. 56 :644–652. 10.1194/jlr.M056622 25601960

PMC005xxxxxx/PMC5839723.txt

==== Front J Gen Physiol J Gen Physiol jgp jgp The Journal of General Physiology 0022-1295 1540-7748 Rockefeller University Press 29374022 201711938 10.1085/jgp.201711938 Research Articles Research Article 503 512 Molecular and functional identification of a novel photopigment in Pecten ciliary photoreceptors Go-coupled rhodopsin Arenas Oscar 1 http://orcid.org/0000-0001-6015-902X Osorno Tomás 1 Malagón Gerardo 1 http://orcid.org/0000-0002-5648-066X Pulido Camila 1 Gomez María del Pilar 13 http://orcid.org/0000-0002-8319-0289 Nasi Enrico 23 1 Departamento de Biología, Universidad Nacional de Colombia, Bogotá, Colombia 2 Instituto de Genética, Universidad Nacional de Colombia, Bogotá, Colombia 3 Marine Biological Laboratory, Woods Hole, MA Correspondence to Enrico Nasi: enasil@unal.edu.co O. Arenas’s present address is Dept. of Neuroscience, Northwestern University, Chicago, IL. T. Osorno’s present address is Dept. of Neurobiology, Harvard Medical School, Boston, MA. G. Malagón’s present address is Laboratoire de Physiologie Cérébrale, Université Renée Descartes, Paris, France. C. Pulido’s present address is Dept. of Biochemistry, Cornell University, Ithaca, NY. 05 3 2018 150 3 401415 30 10 2017 20 12 2017 © 2018 Arenas et al. 2018 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). The mollusk Pecten irradians possesses ciliary photoreceptors that operate with an atypical mechanism. Arenas et al. reveal that a recently uncovered opsin type is the functional visual pigment in these photoreceptors and couples to Go, in contrast to other types of photoreceptor. The two basic animal photoreceptor types, ciliary and microvillar, use different light-transduction schemes: their photopigments couple to Gt versus Gq proteins, respectively, to either mobilize cyclic nucleotides or trigger a lipid signaling cascade. A third class of photoreceptors has been described in the dual retina of some marine invertebrates; these present a ciliary morphology but operate via radically divergent mechanisms, prompting the suggestion that they comprise a novel lineage of light sensors. In one of these organisms, an uncommon putative opsin was uncovered that was proposed to signal through Go. Orthologues subsequently emerged in diverse phyla, including mollusks, echinoderms, and chordates, but the cells in which they express have not been identified, and no studies corroborated their function as visual pigments or their suggested signaling mode. Conversely, in only one invertebrate species, Pecten irradians, have the ciliary photoreceptors been physiologically characterized, but their photopigment has not been identified molecularly. We used the transcriptome of Pecten retina to guide the cloning by polymerase chain reaction (PCR) and rapid amplification of cDNA ends (RACE) extensions of a new member of this group of putative opsins. In situ hybridization shows selective transcription in the distal retina, and specific antibodies identify a single band of the expected molecular mass in Western blots and distinctly label ciliary photoreceptors in retina sections. RNA interference knockdown resulted in a reduction in the early receptor current—the first manifestation of light transduction—and prevented the prolonged aftercurrent, which requires a large buildup of activated rhodopsin. We also obtained a full-length clone of the α-subunit of a Go from Pecten retina complementary DNA and localized it by in situ hybridization to the distal photoreceptors. Small interfering RNA targeting this Go caused a specific depression of the photocurrent. These results establish this novel putative opsin as a bona fide visual pigment that couples to Go to convey the light signal. Departamento Administrativo de Ciencia, Tecnología e Innovación https://doi.org/10.13039/100007637 FP44842-010-2015 Fund for Science ==== Body pmcIntroduction Two canonical classes of primary visual cells, dubbed ciliary and rhabdomeric photoreceptors, respectively, have been established on the basis of the structure of their light-sensing cellular specializations: either modified ciliary appendages, as in vertebrate rods and cones (Tokuyasu and Yamada, 1959), or infoldings of the plasma membrane in the form of actin-packed microvilli, as in arthropods (de Couet et al., 1984; Arikawa et al., 1990). More recently, sequence analysis of opsins from a wide range of organisms has shown that animal photopigments can also be grouped into distinct categories; two of them largely follow the two aforementioned morphological classes of photoreceptors and were therefore named C-opsins and R-opsins (Arendt et al., 2004). Because the former convey the light signal via a Gt and stimulate a phosphodiesterase to control cGMP levels (reviewed by Luo et al., 2008), whereas the latter activate Gq and trigger an inositol-lipids pathway (reviewed by Hardie and Raghu, 2001), these two opsin groups have also been labeled “Gt-coupled” and “Gq-coupled” (Shichida and Yamashita, 2003; Terakita, 2005). A third class of photoreceptor cells, first described in the distal retina of marine bivalve mollusks of the Pectinidae family (scallops), have a ciliary morphology (Miller, 1958; Barber et al., 1967) and a hyperpolarizing receptor potential like that of vertebrate rods and cones (Gorman and McReynolds, 1969). However, they operate via profoundly different ionic mechanisms that involve light-triggered opening of ion channels (Gorman and McReynolds, 1978; Gomez and Nasi, 1994a); this calls for a different phototransduction cascade, prompting the suggestion that such cells comprise a separate lineage of light sensors (Gomez and Nasi, 2000). In fact, in a survey of signaling molecules in the eye of Mizuhopecten (a.k.a., Patinopecten), the giant scallop of northern Japan, the distal retina layer was found to be devoid of both conventional opsin types; instead, a novel putative photopigment was discovered and dubbed Scop2 (NCBI accession no. BAA22218). Furthermore, the only heterotrimeric G-protein that colocalized with this unusual opsin was a Go, which was proposed to mediate the light response (Kojima et al., 1997). There were no precedents for Go participation in any phototransduction process. Indirect support for the notion that Scop2 may couple to Go was provided by in vitro assays of chimeric bovine rhodopsin: replacement of the third intracellular loop with that of Scop2 decreased its competence to activate transducin (Gt), while enhancing its ability to stimulate Go (Terakita et al., 2002). Orthologues of Scop2 have been subsequently identified in another mollusk (Crassostrea gigas, accession no. EKC29416), in a primitive chordate (Branchiostoma belcheri, Koyanagi et al., 2002; accession nos, AB050606 and AB050607), and in an echinoderm (Strongylocentrotus purpuratus, accession no. XM_011676116), indicating a rather widespread phylogenetic distribution. In spite of the accumulated molecular information, much remains to be learned about these putative novel light-sensing molecules, as the cells that natively express them have never been examined physiologically, and in most cases have not even been identified. Moreover, to date, only one of these opsins (from Branchiostoma) has been heterologously expressed and shown in vitro to bind 11-cis-retinal as a chromophore (Koyanagi et al., 2002), but, as yet, no functional reconstitution has been attained. Conclusive evidence that these molecules indeed form functional visual pigments is therefore still lacking, and so is experimental support to elucidate their mode of signaling. The physiology of the light response of nonvertebrate ciliary hyperpolarizing photoreceptors has been investigated in detail in only one species, the bay scallop, Pecten irradians (McReynolds and Gorman, 1970a,b; Gomez and Nasi, 1994b,a, 1997a,b, 2005; del Pilar Gomez and Nasi, 1995, 2005; Nasi and del Pilar Gomez, 1999). However, its photopigment had not been molecularly identified. Pharmacological clues do implicate a Go in Pecten photoresponsiveness, as was proposed for Mizuhopecten: the light response is susceptible to inhibition by pertussis toxin, which selectively targets Gi/Go, and the photoconductance can be activated by mastoparan peptides, which are preferential activators of Go (Gomez and Nasi, 2000). Direct evidence supported by molecular data, however, is still missing. The purpose of the present work was to ascertain the molecular identity of the photopigment of Pecten ciliary photoreceptors and the G-protein that conveys the light signal. In addition to helping elucidate this third light-signaling pathway, there is an additional facet that makes this endeavor appealing: biophysical studies have established that the rhodopsin of Pecten—like those of several arthropods—is bistable and undergoes a large spectral shift with photoisomerization; this allows manipulation (and quantitative analysis) of its states (Cornwall and Gorman, 1983). The photopigment of Pecten ciliary visual cells could therefore be a useful biotechnological tool, a molecule capable of turning on—or off—a G-protein cascade by simply manipulating the chromatic content of a light stimulus. Materials and methods Specimens of Pecten irradians were obtained from the Aquatic Resources Division of the Marine Biological Laboratory (Woods Hole, MA). Cell dissociation and recording Retinae of Pecten were enzymatically and mechanically dissociated as previously described (Gomez and Nasi, 1994a). Patch pipettes fabricated from borosilicate glass were fire-polished and filled with an intracellular solution containing 100 mM KCl, 200 mM K-glutamate, 5 mM MgCl2, 5 mM Na2ATP, 20 mM NaCl, 1 mM EGTA, 300 mM sucrose, 10 mM Hepes, and 0.2 mM GTP, pH 7.3. Electrode resistance in ASW was 2–4 MΩ. An Optopatch amplifier (Cairn Research) was used to measure membrane currents. Data were digitized with an analogue–digital interface (DT9834; Data Translation), which served also to generate stimuli under the control of software developed in-house. Light stimulation Light flashes were generated either by a tungsten-halogen light source equipped with an electromechanical shutter (Uniblitz), as previously described (Gomez and Nasi, 1994a), or by a high-intensity 470-nm LED (Thorlabs). To visualize cells during experimental manipulations, deep red light illumination was used (λ > 630 nm). To accurately estimate response latencies, the raw traces were corrected after measuring with a photodiode the actual opening delay of the shutter (when applicable), as well as the time shift introduced in the recorded current trace by the analogue low-pass filter. Light intensity was measured with a calibrated radiometer (UDT model 370) and converted to photon flux. For experiments that used broadband light, its intensity was converted to effective photons at 500 nm via in vivo calibration, as previously described (Gomez and Nasi, 1994b,a). Immunochemistry Polyclonal antibodies raised against the Scop2 orthologue of Pecten were generated by Biosynthesis (Lewisville, TX). To this end, a synthetic peptide (C-TRRNETRTRQGYMPRYIQD) derived from the predicted amino acid sequence was conjugated to keyhole limpet hemocyanin and used to immunize rabbits. The antiserum was affinity purified. For Western blots, retinae were homogenized (Teflon/glass) in lysis buffer (30 mM Tris HCl, 100 mM NaCl, 5 mM EDTA, 1% SDS, and 20% glycerol, pH 7.4) in the presence of protease inhibitors (0.25% Sigma protease inhibitor cocktail and 250 µM PMSF), acetone-precipitated at −20°C for 1 h, and centrifuged for 10 min at 12,000 g. The pellet was air-dried, resuspended in sample buffer, and separated by SDS PAGE (8%). Proteins were then electrotransferred (Mini Trans-Blot; Bio-Rad) onto a nitrocellulose membrane, which was blocked overnight with 3% BSA. The membrane was sequentially incubated with primary antibodies (1:1,000, 3 h), washed, treated with alkaline phosphatase (AP)-conjugated anti-rabbit secondary antibodies (1:2,000, 1 h; Promega), and developed in Western Blue (Promega). For immunohistochemistry, excised eyes were fixed in 4% paraformaldehyde overnight at 4°C. After washing, permeabilization (0.2% Triton X-100), and cryoprotection by sucrose impregnation (10%→20%, 2 h each, then 30% overnight at 4°C), the eyes were embedded in gelatin (7.5% + 15% sucrose). The blocks were frozen at −80°C, equilibrated in the cryostat (−25°C), and sectioned at 7–12 µM. The sections were mounted onto subbed glass slides (Weaver solution). After the gelatin was dissolved in PBS at 36°C, the slides were incubated with primary antibodies (1:500, 2.5 h), followed by Alexa Fluor 546–conjugated secondary goat anti-rabbit antibodies (1 h, 1:200; Molecular Probes) before mounting in 50% glycerol/PBS, covering, and sealing with transparent nail polish. The slides were viewed in a fluorescence microscope (Zeiss Axio Observer). For single-cell immunocytochemistry, isolated photoreceptors plated onto concanavalin A–treated coverslips were dipped in 2% paraformaldehyde for 10 min, permeabilized, and treated as described above. Generation of the transcriptome of the retina of Pecten The eyes of nine animals were used to microdissect ~300 retinae, and total RNA was extracted (Ambion RNAqueous-Micro kit), pooled, and EtHO-precipitated. A library of cDNA was generated and normalized (Bio S&T) and subjected to Roche 454 massively parallel pyrosequencing with GS FLX Titanium-series reagents (Cornell Core Laboratory Center–Genomics, Ithaca, NY); this produced ~7 × 108 primary reads (mean length, 325 bp). The collection was assembled by Roche GS De Novo Assembler at the Bay Paul Center (Marine Biological Laboratory, Woods Hole, MA), resulting in ~44 × 103 isotigs, which, as a preliminary annotation step, were examined by BlastX against the NCBI Reference Sequence nonredundant protein database. Molecular cloning For PCR amplifications, Pecten retina cDNA was obtained from poly-A mRNA isolated by oligo dT–conjugated magnetic microbeads (Dynal). For reverse transcription, we used the SMART RACE kit (Clontech), which yields 5′ and 3′ cDNAs ready for rapid amplification of cDNA ends (RACE). PCR reactions were performed on an M&J Research DNA Engine cycler; amplicons were purified as needed (Qiagen). For product extension, we used nested RACE reactions. Upon obtaining overlapping sequences that jointly spanned full-length clones (as assembled with the CAP contig assembly routine, BioEdit) a single end-to-end run was conducted to rule out spurious amplification of noncontiguous cDNAs. PCR products were either gel-purified and sequenced directly (Genewiz) or ligated into a sequencing vector (Promega pGEM-T Easy TA); in the latter case, white colonies of chemically transformed Escherichia coli (JM-109) grown in ampicillin agar dishes containing isopropyl β-d-1-thiogalactopyranoside and X-Gal were cultured in selective LB medium and checked by colony PCR before extracting and purifying plasmid DNA, which was subjected to restriction analysis before sequencing. In situ hybridization Probe generation Digoxigenin-labeled RNA probes were synthesized by two approaches. (a) A PCR product obtained with Taq polymerase was purified, restricted, and directionally ligated into a dual-promoter expression vector (either pGEM-T Easy [Promega] or pBluescript II SK [Stratagene]), and used to transform E. coli. The plasmid was then linearized with either SpeI or SacII. (b) Alternatively, DNA template was directly created by PCR with modified primers in which either a T7 or a Sp6 promoter sequence was added as a 5′ overhang. All the products were verified by sequencing. Labeled sense or antisense riboprobes were synthesized in vitro, by incubating the template in digoxigenin labeling mix (Roche) with T7 or SP6 RNA polymerase (2 h at 42°C). The probe was EtOH-precipitated, dried, and resuspended in diethyl pyrocarbonate H2O; the concentration was determined spectrophotometrically before dilution in hybridization buffer as a 10× stock, to be stored at −20°C. A dot-blot was performed in a nylon membrane to determine labeling efficiency; to this end, dots of varying dilutions of the probe were incubated with anti-digoxigenin antibodies and developed in Western Blue (Promega). Probe detection The procedure for in situ hybridization is based on Shimamura et al. (1994) with few modifications. In brief, samples were sequentially washed in PBS, methanol, and PBS + Tween 20 (PTw), and treated with proteinase K (20 µg/ml). After refixing (0.2% gluteraldehyde + 4% paraformaldehyde) and prehybridization (45 min at 60°C), the probe was added (2 µg/ml), and the slides were coverslipped (hybrislips; Grace Bio Labs) and incubated overnight at 60°C in a hybridization oven (Boekel). Samples were subsequently washed repeatedly and incubated with 100 µg/ml RNase A and 100 U/ml RHase T1 in NTE (0.5 M NaCl, 10 mM Tris, pH 8.0, and 1 mM EDTA). Finally, after additional washes, they were blocked (0.1% Tween, 2 mM levamisole, and 2% Boehringer Mannheim Blocking Reagent, 30 min) and incubated overnight with anti-DIG AP antibody (1:2,000, room temperature), before detection with AP substrate solution (Roche). RNA interference Double-stranded RNAs were synthesized by Invitrogen; two target sequences were selected for the putative photopigment of ciliary photoreceptors (5′-GAAGATCCGTCAGACGTCCAATCAA-3′ and 5′-CAACTCGACATTTCGTGATTCCTTA-3′), which were used in a 1:1 mix, and one for the α-subunit of the Go heterotrimeric G-protein (5′-UCCUAUCGUGUAUGCUCUCTT-3′). These were selected on the basis of the complete coding sequence of the two target transcripts, after scrutinizing several dozens of candidates for the presence of suitable motifs (Reynolds et al., 2004; Jagla et al., 2005) and ascertaining their uniqueness by a BLAST search. The probes were fluorescently labeled at the 5′ end with either fluorescein (Go) or Alexa Fluor 546 (pScop2). To introduce the small interfering RNA (siRNA) into photoreceptor cells, excised retinae were placed in 1-mm-gap electroporation cuvettes in the presence of 0.5–1 µM siRNA (omitted in controls, and replaced with 10 kD dextran-rhodamine; Sigma-Aldrich) and allowed to equilibrate for 5 min. The electroporation solution contained 590 mM sucrose, 72.7 mM NaCl, 14.5 mM KCl, 23.64 mM MgCl2, 3.64 mM Hepes, 118.2 mM Tris, and 9.1 µM CaCl2, pH 7.5. The high concentration of sucrose was empirically determined and served a dual purpose: (a) increasing the density so that the retinae did not sink to the bottom, for better control of exposure to the electric field, and (b) decreasing conductivity to reduce Joule heating. A BTX ECM 830 electroporator delivered either two 30-ms square pulses, 40 V, 5 s apart (for pScop2-siRNA) or three 30-ms pulses, 25 V, 1 Hz (Go-siRNA). After 5 min, the retinae were transferred to culture medium (DMEM with added salts to mimic the ionic composition of seawater, supplemented with 1 g/liter glucose, 100 U/ml penicillin, and 100 µg/ml streptomycin) and cultured at 15°C. For testing, retinae were enzymatically treated as described in the Cell dissociation and recording section, and dissociated ciliary photoreceptors were screened by fluorescence microscopy for the incorporation of the probe before performing whole-cell recordings. Results Cloning of the opsin of distal photoreceptors of Pecten irradians The first goal was to obtain the molecular identity of the photopigment of ciliary photoreceptors of Pecten. Exploiting the phylogenetic proximity between Pecten irradians and Mizuhopecten yessoensis, the predicted protein sequence of Scop2 (Kojima et al., 1997) was used as a query in a tBlastn search of the primary reads of the transcriptome of Pecten retina. At this stage, we chose to rely on the collection of primary reads—rather than assembled contigs—in order to identify regions of high representation (six or more overlapping reads) and 100% concordance, thus minimizing uncertainty in primer design. 11 high-scoring initial hits were assembled into a single contig; the end regions were in turn used to search the database again, eventually extending the set to 29 overlapping sequences that apparently represent a single transcript. Suitable stretches were selected and fed to Primer3 for oligonucleotide design, and candidate sequences were further scrutinized for melting point matching, secondary structure, and dimerization (OligoCalculator). Nested PCR reactions yielded the expected products, as confirmed by gel electrophoresis (Fig. 1 A) and subsequent sequencing; these were followed by design of new, outward-facing primers and nested RACE extensions (Fig. 1 B), which resulted in a full-length clone. The nucleotide sequence obtained includes a long open reading frame, translating to a predicted polypeptide of 407 aa (Fig. 1 C). A ClustalW alignment with the sequence of Mizuhopecten Scop2 shows a high degree of similarity at both the amino acid level (61% identity) and the nucleotide level (66%). A Blastp search of the NCBI nonredundant protein database yielded, in addition to Scop2 (e-value 2e-171), the other members of the novel opsin group that has been proposed to couple to Go (namely, those of sea urchin, oyster, and the two isoforms of amphioxus). Hydrophobicity analysis of the amino acid sequence using the Kyte–Doolittle algorithm (13-aa window width) indicated the presence of seven transmembrane domains (Fig. 1 D). As was anticipated, the obtained sequence exhibits a critical residue for the binding of the chromophore: by aligning its sequence with canonical photopigments such as bovine rhodopsin, NinaE (the main rhodopsin of Drosophila), and its close relative Mizuhopecten Scop2 (UniProt accession nos. P02699, P06002, and O15974, respectively), a lysine was found in a highly conserved stretch of the seventh transmembrane domain (Fig. 1 D). This corresponds to lysine 296 of the bovine form (319 in NinaE, 286 in Scop2 of M. yessoensis, and 288 in pScop2 of P. irradians), known to establish a Schiff base linkage with 11-cis-retinal. The results therefore indicate the presence of an orthologue of Scop2 in the retina of Pecten, henceforth referred to as pScop2 (GenBank accession no. MG674154). Figure 1. Cloning the photopigment of the ciliary photoreceptors of the distal retina of Pecten. (A) Ethidium bromide–stained agarose gels of the products of PCR reactions with gene-specific primers, which yielded amplicons of the predicted size. (B) Nested RACE reactions provided extensions to the 5′ and 3′ ends. (C) The full-length transcript obtained has 1647 bp, with a open reading frame that translates to a 407-aa polypeptide. (D) Hydrophobicity profile suggests the presence of seven putative membrane-spanning regions. (E) Alignment of the chromophore-binding region of bovine rhodopsin and NinaE (Drosophila rhodopsin), with Mizuhopecten Scop2 and Pecten pScop2. The critical lysine in the seventh transmembrane domain, which forms a Schiff-base linkage with retinal, is conserved (highlighted in red). pScop2 localization Localization of the transcript was assayed by in situ hybridization. Digoxigenin-labeled RNA probes, 1354 bp in length, were generated with primers 169fwd and 1523rev incorporating promoter sites, and directly synthesized from PCR templates. Hybridization assays were conducted on 12-µm-thick formaldehyde-fixed transverse cryosections of isolated retinae. Fig. 2 A shows that a distinct staining pattern was obtained, with the antisense probe targeting selectively the distal retinal layer, which is comprised of the ciliary photoreceptors; labeling was entirely absent when sense riboprobe was used. Figure 2. Localization of pScop2 in the double retina of Pecten. (A) In situ hybridization in cryosections of isolated, fixed retinae incubated with digoxigenin-labeled sense or antisense riboprobes. After hybridization, transcripts were detected using AP-conjugated anti-DIG antibody. Left: The antisense-treated retinae show selective staining in the distal layer (DR), where the ciliary photoreceptors are located, but no labeling in the proximal retina (PR), where microvillar photoreceptors are found. Right: No staining at all appeared in the retinae incubated with the sense probe. (B) Western blot of retinal lysate using anti-pScop2 antibody. Proteins separated by SDS-PAGE (8%) before transferring to nitrocellulose membrane. A prominent band is visible near 50 kD, the predicted weight of pScop2. (C) Immunohistochemistry to localize pScop2 in eye slices; secondary antibodies were conjugated to Alexa Fluor 546. The lefthand panels show Nomarski images at two magnifications, and the righthand panels show the corresponding fluorescence microscopy images. Immunofluorescence is confined to discrete cells of the distal retina. DIC, differential interference contrast. (D) Immunocytochemistry in isolated cells. The retina was enzymatically dissociated, and the cell suspension was plated onto concanavalin A–treated coverslips before incubation with antibodies. The two classes of photoreceptors are recognizable by their distinct morphology; only the ciliary photoreceptors are immunopositive for anti-pScop2 antibody and display an accumulation of the opsin in the region of the ciliary appendages. Having localized pScop2 mRNA transcripts in the distal retina, the presence of the protein was ascertained by immunodetection. To this end, custom-made, affinity-purified anti-pScop2 antibodies were first assayed in Western blots of Pecten retinal lysate. Fig. 2 B shows that a single prominent band was obtained, with an apparent mass of ~50 kD, a figure that compares favorably with the molecular weight calculated from the predicted amino acid sequence (46.2 kD). The slightly larger size and somewhat diffuse appearance of the band may indicate different states of posttranslational modification, such as glycosylations, common among opsins (Murray et al., 2009). This antibody was subsequently used to localize the protein in immunohistochemical assays, using secondary antibodies conjugated to Alexa Fluor 546. Fig. 2 C shows both Nomarski and fluorescence micrographs of Pecten eye cryosections in which only the cells of the distal retina are distinctly labeled. Control sections, in which the primary antibody was omitted, were completely devoid of labeling (not depicted). To garner further information on the expression pattern, retinae were enzymatically dissociated and processed for immunocytochemistry, allowing unambiguous morphological recognition of the cell types. Fig. 2 D (left) shows an isolated, identified ciliary photoreceptor decorated by anti-pScop2 antibodies, the staining being especially strong in the cilia; in contrast, an isolated microvillar photoreceptor (right) is devoid of label. Functional demonstration that pScop2 is the ciliary photopigment The clone that was obtained from Pecten ciliary photoreceptors is closely related to the one reported by Kojima et al. (1997) and groups with the other members of this novel opsin subclass that were subsequently uncovered by genome sequence analysis. However, to this date there have been no functional data attesting that such molecules indeed function as photopigments. Exploiting the fact that Pecten constitutes a well-developed system for single-cell physiological measurements, we sought such supporting evidence by examining the consequences of interfering with pScop2 in native ciliary photoreceptors. The initial strategy was to dialyze intracellularly the anti-pScop2 antibodies while monitoring the photocurrent elicited by repetitive flashes. This effort seemed justified because the epitope maps to residues 221–239 in the predicted protein sequence; according to the hydropathicity profile, this stretch is part of the connecting loop between α-helices 5 and 6, a likely site of interaction between opsins and effector G-proteins (reviewed by Shichida and Yamashita, 2003). However, we failed to observe a consistent depression of the light response amplitude compared with control cells (not depicted). Two factors—not mutually exclusive—could account for the negative results: one is that the antibodies may not be functional; another is that the photocurrent may be a poor criterion to gauge the amount of functional photopigment, because in visual receptors rhodopsin expresses at a vast molar excess with respect to all other signaling elements (e.g., Pugh and Lamb, 1993), and the message from the activated photopigment is subject to great amplification. As a consequence, even substantial interference with the pigment population may not significantly alter the amplitude attained by the photocurrent. We turned therefore to siRNA as a means to reduce pScop2, and to a physiological response that scales linearly with the amount of photopigment present. The maximum charge displacement during photoisomerization fills the requirement (see Discussion). To quantify this parameter, one must measure the early receptor current (ERC). Fig. 3 A illustrates the membrane current elicited by an intense broadband light stimulus in an isolated ciliary photoreceptor of Pecten voltage-clamped to −40 mV. The large outward photocurrent (IL) is mediated by the opening of K-selective ion channels (Gomez and Nasi, 1994a) and is preceded by a small, rapid hump (rectangle, shown at an enlarged scale in the inset). Identification of such early transient as a photopigment displacement current is supported by the fact that (a) it is very rapid: latency 0.3 ± 0.08 ms SEM, time to peak 3.1 ± 0.92 ms, n = 6 (after correcting for actual shutter opening time and photocurrent time shift caused by the analogue filter; see Materials and methods); (ii) it appears only at light intensities that are saturating for the late photocurrent, i.e., a regimen that causes massive, synchronous stimulation of rhodopsin; (iii) it survived application of the K-channel blocker 4-aminopyridine (20 µM; Fig. 3, B and C), which is an effective antagonist of the light-activated conductance of these cells (Gomez and Nasi, 1994b); and (iv) it is largely unaffected by changes in ionic driving forces, whereas, as Vm is hyperpolarized, the late photocurrent is gradually reduced until vanishing at −80 mV, close to EK (Fig. 3 D). In addition, because of the large spectral shift between rhodopsin (R, λmax ≈500 nm; McReynolds and Gorman, 1970b; Cornwall and Gorman, 1983) and metarhodopsin (M, λmax ≈576 nm), photopigment state distribution can be manipulated by chromatic adaptation before the test flash, and this is reflected in the size and polarity of the early component (Fig. 3 E). This feature was also exploited to reset photopigment state with a red light (λ > 630 nm, 3 s) applied 1 min before each ERC recording, to ensure reproducibility of the ERC over repeated trials. Figure 3. Identification of the ERC arising from photopigment isomerization in ciliary photoreceptors. (A) Whole-cell recording of membrane current in an isolated cell voltage-clamped at −40 mV and stimulated with an intense light step (4.1 × 1017 photons ⋅ s−1 ⋅ cm−2). Illumination evoked the characteristic outward current mediated by K-channels, but its activation was preceded by a brief, much smaller transient (rectangle and inset) that partially overlaps with the photocurrent rising phase. (B) Progressive reduction of the photocurrent evoked by moderate flashes (7.3 × 1015 photons ⋅ s−1 ⋅ cm−2, 100 ms), during superfusion with the K-channel blocker 4-aminopyridine (20 µM). (C) The early component can be isolated in the presence of 4-aminopyridine (4-AP). Under these conditions, its falling phase is fully resolved, with little contamination by the late photocurrent. (D) Manipulations of membrane voltage greatly impact the late light–evoked current but leave the early component largely unaffected. The membrane potential was set at the specified value a few seconds before delivering the light stimulus. At −80 mV, near the calculated equilibrium potential for K ions, the late photocurrent virtually disappears. Duration of light flashes, 30 ms. (E) The size and polarity of the early component is determined by prior chromatic adaptation. The cell was illuminated with monochromatic light (three-cavity filters, intensities 1014 to 2.5 × 1015 photons ⋅ s−1 ⋅ cm−2) for several seconds to reach photoequilibrium of the pigment state, before being stimulated with a fixed high-intensity test stimulus of white light. When the adapting light was of long wavelength, the early component had an outward direction, but became inward with chromatic adaptation to progressively shorter wavelengths. The holding potential was kept at −40 mV throughout. Light intensity for test flashes in C–E, 5.8 × 1017 photons ⋅ s−1 ⋅ cm−2. To assess the effects of pScop2 RNA interference (RNAi), ERCs were elicited by brief test flashes (6 ms) of 470-nm peak wavelength (near the optimum value in the photo-equilibrium spectrum to yield the largest net R→M transition; Cornwall and Gorman, 1983). The total amount displaceable charge can be gauged from the maximum value of the integral of the ERC. Fig. 4 A illustrates a family of ERC traces elicited by flashes of increasing intensity in a control cell. Each current record was integrated, and the resulting values are plotted in Fig. 4 B; the ensemble clearly shows a saturating behavior. Data points were least-squares fitted by an exponential function, the asymptote of which yields the desired estimate for the maximal photoisomerization charge, which reflects the total amount of photopigment. After siRNA introduction, a suitable temporal window must be interposed, before a phenotype becomes manifest. To preserve cell viability during that time period, organotypic culture proved far more robust than culturing dissociated cells: an excised retina can be maintained for 2–3 d with few special precautions, and upon enzymatic dispersion the photoreceptors look indistinguishable from freshly obtained cells and retain their normal light responsiveness. In contrast, cultured dissociated photoreceptors begin to de-differentiate within 1 d, and many of them subsequently die (unpublished data). Therefore, we opted for electroporating the intact retinae and dissociating them only before the electrophysiological test (see also Matsuda and Cepko, 2004). Before evaluating pScop2-RNAi, pilot experiments were conducted to optimize electroporation parameters, using 10 kD dextran-conjugated rhodamine (Molecular Probes) to mimic siRNA (~13 kD); 24–48 h later, photocurrents were examined by whole-cell patch recording. We found conditions that met two indispensable criteria: (1) a fraction of the cells showed incorporation of the fluorescent probe, attesting the efficacy of the electroporation, and (2) labeled photoreceptors remained light-sensitive (amplitude and the light intensities producing half-maximal responses were not altered, compared with control cells), indicating that the treatment is not unduly harsh. We then examined the effects of pScop2-siRNA on the ERC, compared with control retinae electroporated in the presence of dextran-rhodamine; we also exploited the unique dual nature of Pecten retina, which presents a second, proximal layer of conventional microvillar photoreceptors (Barber et al., 1967; Gorman and McReynolds, 1969), and assessed the impact of the pScop2 siRNA on the ERC of these cells, which express an entirely different rhodopsin (Kojima et al., 1997) and thus serve as ideal negative controls. Post-electroporation culturing periods, before enzymatic dispersion, were typically 24–48 h; longer incubation times up to 72 h were tested, but did not yield any additional benefit. Target cells were selected by fluorescence before conducting the electrophysiological tests. Fig. 4 C shows traces of the ERC recorded in a control photoreceptor and in one treated with pScop2 siRNA; in the latter, the depression of the photoisomerization current is dramatic. The bar graph in Fig. 4 D summarizes the data, pooling the results for several experimental versus control ciliary and rhabdomeric cells. pScop2-siRNA treatment significantly decreased the photodisplaceable charge in ciliary photoreceptors (P < 0.0029, t test, n = 8–11 per group), whereas it produced no discernible effect on the ERC of microvillar photoreceptors. This indicates that pScop2 is isomerized by light. Figure 4. pScop2 is photoisomerized by light and underlies the ERC. (A) ERCs evoked by flashes of increasing intensity (470 nm, 6 ms in duration, 1.3 × 1017 to 9.1 × 1017 photons ⋅ s−1 ⋅ ⋅cm−2) in a ciliary photoreceptor voltage-clamped at −40 mV. Between trials, a 5-s period of red illumination (620 nm) was interposed 1 min before the subsequent test flash, to reset the photopigment to the R state. (B) Saturation of the charge displaced by illumination. The amount of charge displaced at each stimulus intensity was obtained by integrating the corresponding current traces of A. An exponential function was least-squares fitted to the data points to extract the estimated asymptote; this represents the maximum photoisomerization charge, which in turn reflects the total amount of photopigment in the cell (r2 = 0.992). (C) Comparison of the ERC obtained in a control ciliary photoreceptor and one previously treated with pScop2-siRNA; the treatment greatly attenuated the ERC. Test flash, 7.9 × 1017 photons ⋅ s−1 ⋅ cm−2, 6-ms duration. (D) Mean total photoisomerization charge under control and RNAi conditions, for the two photoreceptor types in the double-retina of Pecten. Before dissociation and electrophysiological testing, the retinae were subjected to electroporation in the presence of either pScop2-siRNA or fluorescent dextran as a control and cultured for 24–48 h. Only in ciliary photoreceptors did the RNAi treatment depress the amount of photoisomerization charge (*, P < 0.01, t test). Error bars indicate SEM. We next sought evidence that pScop2 engages the light-signaling cascade (as there are groups of opsins that photoisomerize but are not involved in signaling; Terakita, 2005). The main hurdle was that the downstream physiological process to be measured must also be strongly sensitive to absolute amounts of photopigment, and this rules out the regular light response. In contrast, the prolonged aftercurrent (PA) that occurs in photoreceptors endowed with a bistable photopigment serves the purpose. This phenomenon has been documented in various species (reviewed by Hillman et al., 1983; see Discussion), in which intense chromatic illumination causes a sustained light response, thought to arise from the net accumulation of metarhodopsin, causing photo-excitation shutoff mechanisms to saturate. In the case of Pecten ciliary photoreceptors, such prolonged aftereffects can be elicited by short-wavelength illumination (Cornwall and Gorman, 1983; Gomez and Nasi, 1994b; Gomez et al., 2011). Fig. 5 A illustrates the time course of the light response evoked by a blue light (bandpass interference filter 400–500 nm); although with moderate light intensities, the photocurrent decays shortly after the termination of the stimulus, as intensity is raised a progressively more sustained tail appears. To corroborate that the phenomenon is a consequence of accumulation of activated photopigment, Fig. 5 (B and C) show repetitively evoked PAs that are terminated in an intensity-dependent fashion by the application of a long-wavelength light, which induced the reverse M→R transition. We reasoned that if pScop2 stimulates the light-transduction cascade, its knockdown must be reflected in a reduced likelihood of evoking PAs because shutoff mechanisms ought to be able to cope with the decreased amount of photopigment activated by a saturating light. Fig. 5 D shows the effect of a supersaturating flash of blue light (470 nm) in a control electroporated cell, causing the expected sustained activation of an outward photocurrent that persisted with only a modest decay for the 10-s duration of the recording. In contrast, similar chromatic stimulation applied to a photoreceptor electroporated with pScop2-siRNA evoked a photocurrent that rapidly declined toward baseline. To provide a quantitative index of the effect, we calculated the ratio between the amplitude of the photocurrent at 10 s versus at the peak, for a group of experimental and control photoreceptors; the obtained averages (0.11 ± 0.03 SEM, n = 12, and 0.36 ± 0.05, n = 9, respectively) differed in a statistically significant way (P = 0.012, t test). These observations attest the competence of pScop2 to activate the light-signaling pathway. Collectively, the observations that pScop2 knockdown depresses both ERCs (the earliest manifestation of photoexcitation) and PAs (which engage light-dependent ion channels) strongly support the notion that this molecule forms the functional visual pigment of Pecten ciliary photoreceptors. Figure 5. PAs reveal that pScop2 couples to the light-signaling cascade. (A) Progressive development of prolonged outward aftercurrents as the intensity of stimulating blue light (delivered every 2 min) is increased. Unattenuated light intensity, 5.3 × 1017 photons ⋅ s−1 ⋅ cm−2; attenuation factor, achieved by neutral-density filters, indicated at the right. The cell was voltage-clamped at −30 mV. After each trial, a 3-s period of red adaptation was applied to ensure constancy of initial distribution of photopigment state. (B) Termination of the PA by photoconversion of activated rhodopsin to the quiescent state. A bright blue flash (λ < 500 nm, 500 ms) of constant intensity (5.3 × 1017 effective photons ⋅ s−1 ⋅ cm−2) was presented every 3 min, evoking a PA with nearly identical time course. In each case, a second flash of red light (λ > 630 nm; 2.9 × 1016 photons ⋅ s−1 ⋅ cm−2) was presented after a variable delay, causing a rapid reset of the sustained photocurrent back to the baseline level. (C) Intensity dependence of the PA shutoff. A 500-ms blue light (5.3 × 1017 photons ⋅ s−1 ⋅ cm−2) was applied every 2 min, followed 4.5 s later by a 500-ms red light, the intensity of which was progressively increased, as indicated (unattenuated intensity, 2.9 × 1016 photons ⋅ s−1 ⋅ cm−2). The degree of suppression of the sustained photocurrent was monotonically related to the amount of red stimulation, as expected from the extent of the concomitant M→R conversion. (D) Treatment with pScop2 siRNA depresses the PA. A chromatic stimulus that in a control photoreceptor produces a sustained activation of the photocurrent (100 ms, 3.1 × 1017 photons ⋅ s−1 ⋅ cm−2), evokes only a transient response in a cell electroporated with siRNA (100 ms, 6 × 1017 photons ⋅ s−1 ⋅ cm−2). pScop2 operates via Go Having garnered evidence for the molecular identity of the bistable photopigment of ciliary photoreceptors of Pecten, we sought to strengthen the notion that this novel class of light-sensing molecules indeed signals through Go. This proposition was previously based only on colocalization within the retina (Kojima et al., 1997) and pharmacology (Gomez and Nasi, 2000). A more direct confirmation would entail identifying such Go and assessing the effects of selectively manipulating it. To this end, primers were initially designed on the basis of the Go of several invertebrates (M. yessoensis, BAA22220; Octopus vulgaris AB025781; Limnaea stagnalis Z15094.1; Helisoma trivolvis L18921). A first PCR amplification yielded a stretch of 114 nucleotides which served to generate additional primers for nested RACE reactions, which eventually yielded a full-length clone (GenBank accession no. MG674155). The predicted protein is comprised of 357 aa and is 99% identical to that of Mizuhopecten (Kojima et al., 1997), with only three discrepancies. Fig. 6 A illustrates a ClustalW alignment against the top-ranking Goα protein sequences. A cysteine in the fourth position from the carboxy terminus is the hallmark of susceptibility to ADP ribosylation, which dovetails with prior observations that the light response of Pecten ciliary photoreceptors is susceptible to inhibition by pertussis toxin (Gomez and Nasi, 2000). Analysis of the transcriptome indicated that the obtained clone is the sole isoform of Go expressed in the retina. Figure 6. Goα cloned a from Pecten retina cDNA. (A) ClustalW alignment of the predicted amino acid sequence of the full-length clone obtained by PCR, against other known Goα of mollusks. (B) In situ hybridization with digoxigenin-labeled riboprobes targeting Goα. The figure shows a cryosection of an isolated retina. The antisense probe distinctly labeled the distal retinal layer. No discernible staining was obtained with the control sense probe. PR, proximal retina; DR, distal retina. In situ hybridization was subsequently used to localize Go. We tested a whole-mount approach using entire eyecups (cornea and lens removed) in which the retina remains attached by the optic nerve. This was designed to reduce losses resulting from the detachment of sections from the glass slides during the hybridization procedure; only after development with the chromogenic substrate were the eyecups embedded, sectioned, and mounted on microscope slides. Probes were synthesized from linearized plasmids; because the sequence contained no suitable restriction sites for directional ligation into the vector, a 320-bp PCR fragment was generated with EcoRI and HindIII sites introduced at the 5′ ends of the modified primers. Fig. 6 B shows that a distinct staining pattern was observed with the antisense probe, with the distal layer being selectively decorated, whereas there was no discernible staining by the sense probe. Functional evidence for the specific involvement of Go in light transduction was sought by an RNAi approach, similarly to the experiments performed with pScop2, described in the previous section. Fig. 7 A shows bright-field and fluorescence micrographs of an isolated ciliary photoreceptor 24 h after electroporation with fluorescein-labeled Go-siRNA. In Fig. 7 B, superimposed traces of photocurrents evoked by flashes of increasing intensity are compared for a control cell electroporated with dextran-rhodamine and an experimental photoreceptor treated with Go siRNA; the photocurrent was substantially attenuated. In Fig. 7 C, the bar graph summarizes the results; siRNA treatment (n = 23) significantly depressed the maximum amplitude of the light-evoked current with respect to both untreated, cultured controls (by 74%; n = 7) and dextran-electroporated controls (by 68%; n = 8). The effect was statistically highly significant (P < 0.01 in both cases; t test); although in the latter group the mean light response amplitude was somewhat smaller than in cells not subjected to electroporation, the difference turned out not to be statistically significant (P = 0.18). Figure 7. siRNA against Go reduces light responsiveness. (A) Transmitted-light micrograph (left) and fluorescence image (right) of a dissociated ciliary photoreceptor that was electroporated with fluorescein-labeled siRNA. The small spherical protuberances (arrow) are the light-sensitive ciliary appendages. (B) Comparison of the photocurrent evoked by progressively brighter flashes (100 ms) in a control cell electroporated with rhodamine-conjugated 10-kD dextran (left) and a photoreceptor electroporated with siRNA targeting Go (right). Light intensity linearly increased from 1.4 × 1014 to 3.8 × 1015 photons ⋅ s−1 ⋅ cm−2; the two dimmest flashes were skipped in the siRNA-treated cell. (C) Mean maximal light response amplitude in pooled cells dissociated from retinae treated with siRNA (n = 23), cultured untreated retinae (n = 7), or retinae electroporated with rhodamine-dextran (n = 8). The first group differed significantly from the other two (P < 0.001, t test), whereas the differences between the two control groups did not attain statistical significance. U, untreated; D, dextran electroporated. Discussion A novel class of putative visual opsins that diverge from the canonical forms of both vertebrates and invertebrates has recently been established by cloning and genome analysis and shown to have a wide phylogenetic distribution. However, the identity of the cells that express these opsins remained uncertain (or, for some species, completely unknown), and no direct evidence had been garnered that such molecules actually form functional photopigments, in either native light-sensitive cells or heterologous expression systems. In the present work, a new member of this group was molecularly identified in the eye of the bay scallop, Pecten irradians, and positively localized to the distal retinal layer, which is comprised of ciliary, hyperpolarizing photoreceptors. This model system is unique in that it has been the object of a substantial number of functional studies, which have led to a detailed physiological characterization of its light response (e.g., Gorman and McReynolds, 1969, 1978; Gomez and Nasi, 1994b,a, 2000; del Pilar Gomez and Nasi, 1995). Our results provide strong evidence that the molecule in question, pScop2, indeed constitutes the photopigment that underlies the receptor potential of Pecten hyperpolarizing visual cells, linking together for the first time molecular and physiological observations on these uncommon photoreceptors and novel opsins. To provide the necessary functional support, a criterion sensitive to even moderate variations in the amount of photopigment was needed. Because of the presence of an amplification cascade and the large excess of rhodopsin over other light-signaling molecules, the photocurrent is not ideal: its peak amplitude can remain largely unaffected by a reduction of the level of rhodopsin—unless this is extreme—whereas shifts in sensitivity are subject to variability. We focused first on the charge movement that accompanies photoisomerization, which is linearly related to the photopigment content of the cell. This phenomenon underlies the early receptor potential (ERP), and its counterpart, the ERC, that was first described in mammals (Brown and Murakami, 1964; Cone, 1967) and subsequently reported in amphibians (Hodgkin and Obryan, 1977; Hestrin and Korenbrot, 1990; Makino et al., 1991) and Gq-coupled rhodopsins of arthropods (Lisman and Sheline, 1976) and of prechordates (Ferrer et al., 2012). A robust ERP had also been demonstrated in Pecten ciliary photoreceptors by Cornwall and Gorman (1983), who used it to characterize in detail the reversible transitions of the bistable photopigment. Because the ERP reports only rhodopsin conformational changes, it is impervious to the state of all downstream links of the light transduction cascade and of the cell in general; in fact, it survives anoxia (Brown and Murakami, 1964) and formaldehyde fixation (Brindley and Gardner-Medwin, 1966) and can even be measured in artificial bilayers incorporating rhodopsin (Trissl et al., 1977). Our results showed that siRNA targeting pScop2 caused a significant reduction in the total amount of charge that can be displaced by light, the effect being circumscribed specifically to ciliary photoreceptors. Although this outcome indicates that pScop2 undergoes photoisomerization, it does not demonstrate that it is capable of triggering the light-signaling cascade. For example, certain opsin subgroups are found in nonphotosensitive tissues and/or perform nonsensory functions (e.g., photoisomerase), rather than light transduction (Terakita, 2005). To clarify this point, we exploited another useful property of the model system, its ability to produce PAs in response to chromatic light stimulation. This phenomenon is a consequence of both the thermal stability of the rhodopsin (common to all invertebrates) and the condition that the inactive and active states differ in peak absorption wavelength (which occurs in several, but not all, species; Hillman et al., 1983). Narrow-band illumination can change the pigment distribution, which will asymptotically tend toFM(λ,∞)=PR(λ)/[PR+(λ)+PM(λ)], where F is the limiting fractional distribution, λ is the wavelength, P is the photosensitivity (i.e., extinction coefficient × quantum efficiency), and the subscripts M and R denote metarhodopsin and rhodopsin, respectively (Hillman et al., 1983). In Pecten, short-wavelength illumination can cause a large accumulation of metarhodopsin, resulting in a sustained photocurrent that can last for tens of seconds. Evidence has been provided that saturation of arrestin—which is molarly underrepresented with respect to the photopigment—is the main culprit of the occurrence of the aftercurrent (Gomez et al., 2011). For the present purposes, the important consideration is that PAs (a) are necessarily sensitive to the total amount of photopigment, (b) provide a readout that is distinct from the amplitude of the light response, and (c) implicate the activation of the phototransduction pathway. The reduction of PA after RNAi treatment supports the conclusion that pScop2 is responsible for initiating the signaling activity that leads to the opening of light-dependent ionic channels. In both sets of experiments using siRNA, we did not attempt to estimate protein knockdown (e.g., using the anti-pScop2 antibodies), for the following reasons: (a) the efficiency of transfection, as gauged by fluorescence microscopy, varied across preparations but did not exceed 20%; stronger electroporation pulses intended to increase the yield proved detrimental, significantly compromising cell viability; and (b) within the transfected cells, the mean decrease in photo-isomerization charge was ~60–70% (Fig. 3 E). Under such circumstances, a macroscopic comparison of protein levels between experimental and control preparations would likely fall below the resolution of a Western blot, as the total decrease of pScop2 (i.e., the product of transfection efficiency and the decrease of photoisomerization charge in transfected cells) would be only on the order of 12–14%. The establishment of pScop2 as a functional photopigment strongly implies that the founding member of this class, which Kojima et al. (1997) identified in the eye of another bivalve mollusk, is a bona fide visual pigment, and lends credence to the notion that more distantly related molecules in echinoderms and prechordates likely perform photoreceptive functions too. However, those animals are devoid of differentiated eyes, and information on the cells that express these opsins is completely lacking, so the particular role of such photopigments in other organisms remains to be investigated. We also obtained the molecular identity of a Go of Pecten retina, localized it to the ciliary photoreceptor layer, and demonstrated the specific depression of the light response after administration of siRNA. These observations provide strong evidence that, as originally proposed by Kojima et al. (1997), such novel opsins signal through a different subtype of heterotrimeric G-proteins than those that convey the rhodopsin signal in canonical photoreceptors. The results dovetail with previous clues garnered from pharmacological studies (Gomez and Nasi, 2000). A role of Go has also been proposed in the photoresponse of lizard parietal eye photoreceptors (Su et al., 2006) and extra-ocular receptors of the mollusk Onchidium verrunculatum (Gotow and Nishi, 2007), pointing to a relatively widespread representation across taxa. Little is known about the downstream effectors, which may turn out to be assorted: the mobilization of cGMP that underlies photoexcitation in Pecten (del Pilar Gomez and Nasi, 1995; Gomez and Nasi, 1997b, 2005) and Onchidium (Nishi and Gotow, 1998; Gotow and Nishi, 2002) seemingly reflects light-dependent regulation of its synthesis, suggesting a Go-regulated guanylate cyclase (Gomez and Nasi, 2000; Gotow and Nishi, 2007), which remains to be identified. In contrast, in the lizard a phosphodiesterase has been implicated (Su et al., 2006). Note that, unlike the well-delimited targets of Gs, Gq, or Gt, the effectors of Go can be quite diverse (Jiang and Bajpayee, 2009). Because nothing is known about the cells that express this type of opsin in amphioxi and sea urchins, it is unclear whether this divergence of effectors is representative of the split between protostomia and deuterostomia. In the case of both pScop2 and Go, the functional consequences of siRNA were evident at 24 h of incubation and failed to increase further at 48 h. This is consistent with observations in other photoreceptors, demonstrating that the cycling of light transduction proteins can be rapid: in fact, in invertebrate photoreceptors a massive turnover of the photosensitive membrane (up to 70%) is paced by the circadian cycle (Chamberlain and Barlow, 1979, 1984). Our findings, obtained in a system that proved amenable to both molecular and physiological analysis, provide definitive functional support for a third lineage of animal visual pigments with their distinct cognate effectors, after the well-characterized C-opsins and R-opsins. According to sequence analysis, this ancient subgroup diverged very early from other opsins (Terakita, 2005) and can help better understand the evolutionary history of light-sensing molecules in animals. The experimental confirmation that such photopigments couple to a Go could also result in a useful new tool for functionally investigating signaling mediated by this heterotrimeric G-protein, which is the most abundant in the nervous system (Jiang and Bajpayee, 2009), but also the least understood in terms of targets, mechanisms, and regulation. Acknowledgments We wish to express our gratitude to Dr. David Mark Welch, Bay Paul Center, Marine Biological Laboratory, for newbler assembly of the transcriptome, to Fabio Echeverry for the massive RNA extraction, and to Juan Manuel Angueyra and Francisca Silva for pilot work to develop the electroporation and the in situ hybridization assays. Supported by Departamento Administrativo de Ciencia, Tecnología e Innovación (Colcencias; grant FP44842-010-2015) and by Fund for Science. The authors declare no competing financial interests. Author contributions: O. Arenas, T. Osorno, G. Malagón, and C. Pulido helped plan the experiments, performed experiments, and contributed to data analysis. MdelP Gomez and E. Nasi were responsible for overseeing the project, planned and performed experiments, contributed to data analysis, and wrote the manuscript. Anna Menini served as guest editor. ==== Refs Arendt, D., K. Tessmar-Raible, H. Snyman, A.W. Dorresteijn, and J. Wittbrodt. 2004. Ciliary photoreceptors with a vertebrate-type opsin in an invertebrate brain. Science. 306 :869–871. 10.1126/science.1099955 15514158 Arikawa, K., J.L. Hicks, and D.S. Williams. 1990. Identification of actin filaments in the rhabdomeral microvilli of Drosophila photoreceptors. J. Cell Biol. 110 :1993–1998. 10.1083/jcb.110.6.1993 2112548 Barber, V.C., E.M. Evans, and M.F. Land. 1967. The fine structure of the eye of the mollusc Pecten maximus. Z. Zellforsch. Mikrosk. Anat. 76 :25–312. 10.1007/BF00339290 5590644 Brindley, G.S., and A.R. Gardner-Medwin. 1966. The origin of the early receptor potential of the retina. J. Physiol. 182 :185–194. 10.1113/jphysiol.1966.sp007817 5937411 Brown, K.T., and M. Murakami. 1964. A new receptor potential of the monkey retina with no detectable latency. Nature. 201 :626–628. 10.1038/201626a0 14160664 Chamberlain, S.C., and R.B. Barlow Jr. 1979. Light and efferent activity control rhabdom turnover in Limulus photoreceptors. Science. 206 :361–363. 10.1126/science.482946 482946 Chamberlain, S.C., and R.B. Barlow Jr. 1984. Transient membrane shedding in Limulus photoreceptors: Control mechanisms under natural lighting. J. Neurosci. 4 :2792–2810.6502204 Cone, R.A. 1967. Early receptor potential: Photoreversible charge displacement in rhodopsin. Science. 155 :1128–1131. 10.1126/science.155.3766.1128 6021913 Cornwall, M.C., and A.L.F. Gorman. 1983. Colour dependence of the early receptor potential and late receptor potential in scallop distal photoreceptor. J. Physiol. 340 :307–334. 10.1113/jphysiol.1983.sp014764 6887052 de Couet, H.G., S. Stowe, and A.D. Blest. 1984. Membrane-associated actin in the rhabdomeral microvilli of crayfish photoreceptors. J. Cell Biol. 98 :834–846. 10.1083/jcb.98.3.834 6538203 del Pilar Gomez, M., and E. Nasi. 1995. Activation of light-dependent K+ channels in ciliary invertebrate photoreceptors involves cGMP but not the IP3/Ca2+ cascade. Neuron. 15 :607–618. 10.1016/0896-6273(95)90149-3 7546740 del Pilar Gomez, M., and E. Nasi. 2005. Calcium-independent, cGMP-mediated light adaptation in invertebrate ciliary photoreceptors. J. Neurosci. 25 :2042–2049. 10.1523/JNEUROSCI.5129-04.2005 15728844 Ferrer, C., G. Malagón, M.P. Gomez, and E. Nasi. 2012. Dissecting the determinants of light sensitivity in amphioxus microvillar photoreceptors: Possible evolutionary implications for melanopsin signaling. J. Neurosci. 32 :17977–17987. 10.1523/JNEUROSCI.3069-12.2012 23238714 Gomez, M.P., and E. Nasi. 1994 a. The light-sensitive conductance of hyperpolarizing invertebrate photoreceptors: A patch-clamp study. J. Gen. Physiol. 103 :939–956. 10.1085/jgp.103.6.939 7931139 Gomez, M.D., and E. Nasi. 1994 b. Blockage of the light-sensitive conductance in hyperpolarizing photoreceptors of the scallop. Effects of tetraethylammonium and 4-aminopyridine. J. Gen. Physiol. 104 :487–505. 10.1085/jgp.104.3.487 7807058 Gomez, M.P., and E. Nasi. 1997 a. Antagonists of the cGMP-gated conductance of vertebrate rods block the photocurrent in scallop ciliary photoreceptors. J. Physiol. 500 :367–378. 10.1113/jphysiol.1997.sp022027 9147324 Gomez, M.P., and E. Nasi. 1997 b. Light adaptation in Pecten hyperpolarizing photoreceptors. J. Gen. Physiol. 109 :371–384. 10.1085/jgp.109.3.371 9089443 Gomez, M.P., and E. Nasi. 2000. Light transduction in invertebrate hyperpolarizing photoreceptors: Possible involvement of a Go-regulated guanylate cyclase. J. Neurosci. 20 :5254–5263.10884309 Gomez, M., and E. Nasi. 2005. On the gating mechanisms of the light-dependent conductance in Pecten hyperpolarizing photoreceptors. J. Gen. Physiol. 125 :455–464. 10.1085/jgp.200509269 15824193 Gomez, M.P., L. Espinosa, N. Ramirez, and E. Nasi. 2011. Arrestin in ciliary invertebrate photoreceptors: molecular identification and functional analysis in vivo. J. Neurosci. 31 :1811–1819. 10.1523/JNEUROSCI.3320-10.2011 21289191 Gorman, A.L.F., and J.S. McReynolds. 1969. Hyperpolarizing and depolarizing receptor potentials in the scallop eye. Science. 165 :309–310. 10.1126/science.165.3890.309 5787990 Gorman, A.L.F., and J.S. McReynolds. 1978. Ionic effects on the membrane potential of hyperpolarizing photoreceptors in scallop retina. J. Physiol. 275 :345–355. 10.1113/jphysiol.1978.sp012193 633125 Gotow, T., and T. Nishi. 2002. Light-dependent K+ channels in the mollusc Onchidium simple photoreceptors are opened by cGMP. J. Gen. Physiol. 120 :581–597. 10.1085/jgp.20028619 12356858 Gotow, T., and T. Nishi. 2007. Involvement of a Go-type G-protein coupled to guanylate cyclase in the phototransduction cGMP cascade of molluscan simple photoreceptors. Brain Res. 1144 :42–51. 10.1016/j.brainres.2007.01.068 17320058 Hardie, R.C., and P. Raghu. 2001. Visual transduction in Drosophila. Nature. 413 :186–193. 10.1038/35093002 11557987 Hestrin, S., and J.I. Korenbrot. 1990. Activation kinetics of retinal cones and rods: Response to intense flashes of light. J. Neurosci. 10 :1967–1973.2355261 Hillman, P., S. Hochstein, and B. Minke. 1983. Transduction in invertebrate photoreceptors: Role of pigment bistability. Physiol. Rev. 63 :668–772. 10.1152/physrev.1983.63.2.668 6340134 Hodgkin, A.L., and P.M. Obryan. 1977. Internal recording of the early receptor potential in turtle cones. J. Physiol. 267 :737–766. 10.1113/jphysiol.1977.sp011836 874877 Jagla, B., N. Aulner, P.D. Kelly, D. Song, A. Volchuk, A. Zatorski, D. Shum, T. Mayer, D.A. De Angelis, O. Ouerfelli, 2005. Sequence characteristics of functional siRNAs. RNA. 11 :864–872. 10.1261/rna.7275905 15923373 Jiang, M., and N.S. Bajpayee. 2009. Molecular mechanisms of Go signaling. Neurosignals. 17 :23–41. 10.1159/000186688 19212138 Kojima, D., A. Terakita, T. Ishikawa, Y. Tsukahara, A. Maeda, and Y. Shichida. 1997. A novel Go-mediated phototransduction cascade in scallop visual cells. J. Biol. Chem. 272 :22979–22982. 10.1074/jbc.272.37.22979 9287291 Koyanagi, M., A. Terakita, K. Kubokawa, and Y. Shichida. 2002. Amphioxus homologs of Go-coupled rhodopsin and peropsin having 11-cis- and all-trans-retinals as their chromophores. FEBS Lett. 531 :525–528. 10.1016/S0014-5793(02)03616-5 12435605 Lisman, J.E., and Y. Sheline. 1976. Analysis of the rhodopsin cycle in limulus ventral photoreceptors using the early receptor potential. J. Gen. Physiol. 68 :487–501. 10.1085/jgp.68.5.487 11271 Luo, D.-G., T. Xue, and K.-W. Yau. 2008. How vision begins: An odyssey. Proc. Natl. Acad. Sci. USA. 105 :9855–9862. 10.1073/pnas.0708405105 18632568 Makino, C.L., W.R. Taylor, and D.A. Baylor. 1991. Rapid charge movements and photosensitivity of visual pigments in salamander rods and cones. J. Physiol. 442 :761–780. 10.1113/jphysiol.1991.sp018818 1818565 Matsuda, T., and C.L. Cepko. 2004. Electroporation and RNA interference in the rodent retina in vivo and in vitro. Proc. Natl. Acad. Sci. USA. 101 :16–22. 10.1073/pnas.2235688100 14603031 McReynolds, J.S., and A.L.F. Gorman. 1970 a. Photoreceptor potentials of opposite polarity in the eye of the scallop, Pecten irradians. J. Gen. Physiol. 56 :376–391. 10.1085/jgp.56.3.376 5476388 McReynolds, J.S., and A.L.F. Gorman. 1970 b. Membrane conductances and spectral sensitivities of Pecten photoreceptors. J. Gen. Physiol. 56 :392–406. 10.1085/jgp.56.3.392 5476389 Miller, W.H. 1958. Derivatives of cilia in the distal sense cells of the retina of Pecten. J. Biophys. Biochem. Cytol. 4 :227–228. 10.1083/jcb.4.2.227 13525437 Murray, A.R., S.J. Fliesler, and M.R. Al-Ubaidi. 2009. Rhodopsin: The functional significance of asn-linked glycosylation and other post-translational modifications. Ophthalmic Genet. 30 :109–120. 10.1080/13816810902962405 19941415 Nasi, E., and M. del Pilar Gomez. 1999. Divalent cation interactions with light-dependent K channels. Kinetics of voltage-dependent block and requirement for an open pore. J. Gen. Physiol. 114 :653–672. 10.1085/jgp.114.5.653 10532963 Nishi, T., and T. Gotow. 1998. Light-increased cGMP and K+ conductance in the hyperpolarizing receptor potential of Onchidium extra-ocular photoreceptors. Brain Res. 809 :325–336. 10.1016/S0006-8993(98)00913-5 9853128 Pugh, E.N. Jr., and T.D. Lamb. 1993. Amplification and kinetics of the activation steps in phototransduction. Biochim. Biophys. Acta. 1141 :111–149. 10.1016/0005-2728(93)90038-H 8382952 Reynolds, A., D. Leake, Q. Boese, S. Scaringe, W.S. Marshall, and A. Khvorova. 2004. Rational siRNA design for RNA interference. Nat. Biotechnol. 22 :326–330. 10.1038/nbt936 14758366 Shichida, Y., and T. Yamashita. 2003. Diversity of visual pigments from the viewpoint of G protein activation: Comparison with other G protein-coupled receptors. Photochem. Photobiol. Sci. 2 :1237–1246. 10.1039/B300434A 14717216 Shimamura, K., S. Hirano, A.P. McMahon, and M. Takeichi. 1994. Wnt-1-dependent regulation of local E-cadherin and alpha N-catenin expression in the embryonic mouse brain. Development. 120 :2225–2234.7925023 Su, C.Y., D.G. Luo, A. Terakita, Y. Shichida, H.W. Liao, M.A. Kazmi, T.P. Sakmar, and K.-W. Yau. 2006. Parietal-eye phototransduction components and their potential evolutionary implications. Science. 311 :1617–1621. 10.1126/science.1123802 16543463 Terakita, A. 2005. The opsins. Genome Biol. 6 :213. 10.1186/gb-2005-6-3-213 15774036 Terakita, A., T. Yamashita, N. Nimbari, D. Kojima, and Y. Shichida. 2002. Functional interaction between bovine rhodopsin and G protein transducin. J. Biol. Chem. 277 :40–46. 10.1074/jbc.M104960200 11606568 Tokuyasu, K., and E. Yamada. 1959. The fine structure of the retina studied with the electron microscope. IV. Morphogenesis of outer segments of retinal rods. J. Biophys. Biochem. Cytol. 6 :225–230. 10.1083/jcb.6.2.225 13838675 Trissl, H.W., A. Darszon, and M. Montal. 1977. Rhodopsin in model membranes: Charge displacements in interfacial layers. Proc. Natl. Acad. Sci. USA. 74 :207–210. 10.1073/pnas.74.1.207 13363

PMC005xxxxxx/PMC5881497.txt

==== Front J Cell Biol J Cell Biol jcb jcb The Journal of Cell Biology 0021-9525 1540-8140 Rockefeller University Press 29440514 201707027 10.1083/jcb.201707027 Research Articles Article 39 23 34 SLAMF1 is required for TLR4-mediated TRAM-TRIF–dependent signaling in human macrophages SLAMF1 regulates TLR4–TRAM–TRIF–mediated signaling http://orcid.org/0000-0002-0516-4747 Yurchenko Maria 13 Skjesol Astrid 1 http://orcid.org/0000-0002-7127-6958 Ryan Liv 1 Richard Gabriel Mary 1 http://orcid.org/0000-0002-0894-1222 Kandasamy Richard Kumaran 1 http://orcid.org/0000-0002-6063-9290 Wang Ninghai 2 http://orcid.org/0000-0001-6233-7029 Terhorst Cox 2 http://orcid.org/0000-0002-8828-1059 Husebye Harald 13* http://orcid.org/0000-0003-0354-5068 Espevik Terje 13* 1 Centre of Molecular Inflammation Research, Norwegian University of Science and Technology, Trondheim, Norway 2 Division of Immunology, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA 3 The Central Norway Regional Health Authority, St. Olavs Hospital HF, Trondheim, Norway Correspondence to Maria Yurchenko: mariia.yurchenko@ntnu.no * H. Husebye and T. Espevik contributed equally to this paper. 02 4 2018 217 4 14111429 06 7 2017 31 10 2017 20 12 2017 © 2018 Yurchenko et al. 2018 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). Yurchenko et al. discover that the Ig-like receptor molecule SLAMF1 enhances production of type I interferon induced by Gram-negative bacteria through modulation of MyD88-independent TLR4 signaling. This makes SLAMF1 a potential target for controlling inflammatory responses against Gram-negative bacteria. Signaling lymphocytic activation molecule family 1 (SLAMF1) is an Ig-like receptor and a costimulatory molecule that initiates signal transduction networks in a variety of immune cells. In this study, we report that SLAMF1 is required for Toll-like receptor 4 (TLR4)-mediated induction of interferon β (IFNβ) and for killing of Gram-negative bacteria by human macrophages. We found that SLAMF1 controls trafficking of the Toll receptor–associated molecule (TRAM) from the endocytic recycling compartment (ERC) to Escherichia coli phagosomes. In resting macrophages, SLAMF1 is localized to ERC, but upon addition of E. coli, it is trafficked together with TRAM from ERC to E. coli phagosomes in a Rab11-dependent manner. We found that endogenous SLAMF1 protein interacted with TRAM and defined key interaction domains as amino acids 68 to 95 of TRAM as well as 15 C-terminal amino acids of SLAMF1. Interestingly, the SLAMF1–TRAM interaction was observed for human but not mouse proteins. Overall, our observations suggest that SLAMF1 is a new target for modulation of TLR4–TRAM–TRIF inflammatory signaling in human cells. Graphical Abstract Research Council of Norway https://doi.org/10.13039/501100005416 223255/F50 Norwegian University of Science and Technology https://doi.org/10.13039/100009123 Norwegian University of Science and Technology https://doi.org/10.13039/100009123 ==== Body pmcIntroduction Toll-like receptors (TLRs) are pivotal for the defense against multiple pathogens by recognizing pathogen-associated molecular patterns. TLR4 recognizes lipopolysaccharide (LPS) from Gram-negative bacteria in complex with the coreceptors myeloid differentiation factor 2 and CD14, and it recruits signaling adapters myeloid differentiation primary response gene 88 (MyD88) and MyD88 adapter–like (Mal). This results in an immediate activation of nuclear factor κB (NF-κB) and production of proinflammatory cytokines. TLR4 is also present on endosomes and phagosomes to which the signaling adapter Toll receptor–associated molecule (TRAM) is recruited (Husebye et al., 2006, 2010; Kagan et al., 2008). The mechanism controlling TRAM recruitment remains unclear but seems to be Rab11 dependent (Husebye et al., 2010; Klein et al., 2015). TRAM is crucial for the subsequent recruitment of Toll/interleukin (IL)-1 receptor (TIR) domain–containing adapter-inducing IFN-β (TRIF) and other downstream molecules, leading to IFNβ secretion (Fitzgerald et al., 2003b; Oshiumi et al., 2003; Yamamoto et al., 2003; Husebye et al., 2010). The role of endogenous type I IFNs in host defense against bacterial infections could be either beneficial or detrimental. Type I IFNs make macrophages more sensitive to cell death–inducing stimuli that could favor bacterial replication and release (Trinchieri, 2010). At the same time, type I IFNs are required for the host resistance to group B streptococci, pneumococci, and Escherichia coli (Mancuso et al., 2007). Assembly of the TLR4–TRAM–TRIF complex followed by the activation of TANK-binding kinase 1 (TBK1) results not only in the induction of type I IFNs but also is required for maintenance of the integrity of pathogen-containing vacuoles and restriction of bacterial proliferation in the cytosol (Radtke et al., 2007; Thurston et al., 2016). Moreover, TBK1 activates the Akt–mTOR–HIF1α signaling axis, which orchestrates metabolic reprogramming to aerobic glycolysis in immune cells (Krawczyk et al., 2010; Everts et al., 2014). Glycolysis provides ATP for driving phagocytosis, proinflammatory cytokine production, and NADPH for the NADPH oxidase 2 (NOX2) enzyme to generate reactive oxygen species (ROS; Kelly and O’Neill, 2015). Signaling lymphocytic activation molecule family 1 (SLAMF1)/CD150 is a type I glycoprotein belonging to the SLAM subfamily of the CD2-like family of proteins (Sidorenko and Clark, 1993; Cocks et al., 1995). SLAMF1 acts as a coreceptor that can modulate signaling via the TNF family and antigen receptors (Mikhalap et al., 1999; Wang et al., 2004; Réthi et al., 2006; Makani et al., 2008). SLAMF1 is involved in the regulation of innate immune responses. Slamf1−/− bone marrow–derived macrophages (BMDMs) are deficient in bacterial killing as they produce less ROS in response to Escherichia coli. Mouse SLAMF1 positively regulates NOX2 activity by forming a complex with beclin-1–Vps34–Vps15–UVRAG (Berger et al., 2010; Ma et al., 2012). Thus, it was essential to explore the contribution of SLAMF1 to TLR4-mediated signaling in human cells. In this study, we show that in human macrophages, SLAMF1 acts as a critical regulator of TLR4-mediated signaling from the phagosome by interacting with TRAM adapters and class I Rab11 family interacting proteins (FIPs) and recruiting the adapter to the TLR4 signaling complex. Results SLAMF1 is expressed in human macrophages and localized to the Rab11-positive endocytic recycling compartment (ERC) Previous studies have suggested that SLAMF1 is found in intracellular compartments of human primary dendritic cells and glioblastoma cells (Avota et al., 2011; Romanets-Korbut et al., 2015). Human peripheral blood monocytes do not express SLAMF1 on the plasma membrane (Farina et al., 2004; Romero et al., 2004). Therefore, we first analyzed the cellular distribution of SLAMF1 in human monocytes, macrophages, and THP-1 cells by confocal microscopy. In all the cell types examined, the major pool of SLAMF1 was located in a perinuclear area negative for the Golgi marker GM130 (Fig. 1, A and B). To further define SLAMF1 localization, monocytes were costained with markers for different types of endosomes: recycling (Rab11a; Fig. 1 C), early (EEA1), and late endosomes (LAMP1; Fig. 1, D and E). Rab11a also defines the ERC, a condensed perinuclear region containing tubular membrane structures that originate from the microtubule organizing center (Yamashiro et al., 1984). TRAM and TLR4 are also present in ERCs of human monocytes and macrophages (Husebye et al., 2010; Klein et al., 2015). Figure 1. SLAMF1 is enriched in the Rab11-positive ERCs in unstimulated macrophages, and SLAMF1 expression is induced by LPS and several other TLR ligands in primary human monocytes and macrophages. (A) Monocytes, macrophages, and differentiated THP-1 cells stained with antibodies against SLAMF1 (green) and GM130 (red) and imaged by confocal microscopy. (B) 3D model of cis-Golgi (GM130) and SLAMF1 in THP-1 cells. Z stacks from the GM130 and SLAMF1 channels were obtained using high-resolution confocal microscopy followed by 3D modeling in IMARIS software. (C) Macrophages stained for SLAMF1 and Rab11 (ERC marker). Representative image. Overlapping pixels for SLAMF1 and Rab11 are shown in the white overlap. tM1 = 0.683 ± 0.08 (mean with SD) for z stacks of ERCs as ROIs (30 ROIs analyzed per donor) where tM1 was the Manders’s colocalization coefficient with thresholds calculated in the Coloc 2 Fiji plugin with anti-SLAMF1 staining as first channel. (D) Macrophages costained for SLAMF1 and EEA1. (E) Macrophages costained for SLAMF1 and LAMP1. Colocalization accessed for z stacks for at least 30 cells for each experiment (four total) showing no colocalization for markers in both D and E. (F) Flow cytometry analysis of SLAMF1 surface expression by primary macrophages and differentiated THP-1 cells. Cells were costained for SLAMF1 and CD14 and gated for CD14-positive cells (primary cells) or stained for SLAMF1 (THP-1 cells). (G) Flow cytometry analysis of SLAMF1 surface expression by human macrophages stimulated by ultrapure K12 LPS (100 ng/ml) for 2, 4, and 6 h. (H) Western blot analysis of lysates from primary human macrophages stimulated by LPS for 2, 4, and 6 h. Graphs present mean values for three biological replicates with SD. Molecular weight is given in kilodaltons. (I and J) Quantification of SLAMF1 mRNA expression by qPCR in monocytes (I) and macrophages (J) stimulated by TLRs’ ligands FSL-1 (20 ng/ml), K12 LPS (100 ng/ml), and CL075 (1 μg/ml; both I and J) as well as R848 (1 μg/ml), Pam3Cys (P3C; 1 μg/ml), or K12 E. coli particles (20/cell; I only). Results are presented as means with SD. Statistical significance between groups was evaluated by a two-tailed t test. *, P < 0.01. Results are representative of at least four independent experiments/donors (A–H) or combined data for at least three donors (I and J). A marked colocalization was found between SLAMF1 and Rab11 in ERCs of resting cells with a Manders’s colocalization coefficient of tM = 0.683 ± 0.08 (Fig. 1 C), whereas there was no colocalization with the other endosomal markers (Fig. 1, D and E). As determined by flow cytometry, only 1% of the monocytes and 4% of macrophages showed surface expression of SLAMF1, whereas 40% of the differentiated THP-1 cells were SLAMF1 positive (Fig. 1 F). LPS stimulation increased the surface expression of SLAMF1 in primary macrophages by >50% after 6 h of LPS stimulation, with an increase in the total SLAMF1 protein expression (Fig. 1, G and H; and Fig. S1 A). Moreover, various TLR ligands such as Pam3Cys (TLR1/2), FSL-1 (TLR2/6), R848 (TLR7 and -8), and CL075 (TLR8) increased SLAMF1 mRNA expression in monocytes and macrophages (Fig. 1, I and J), with E. coli being the most potent stimulator (Fig. 1 I). These results indicate that several TLRs control SLAMF1 expression in human cells. In summary, resting macrophages showed very low SLAMF1 surface level expression, and the major cellular pool of SLAMF1 was found to be in the ERC. THP-1 cells had more surface SLAMF1, but the major cellular pool was still located in the ERC. These observations suggest that ERC-located SLAMF1 may have a yet-undefined function in macrophages. SLAMF1 is required for TLR4-mediated IFNβ production, but its expression is not regulated by the IFNα/β receptor (IFNAR) Next, we used siRNAs to target SLAMF1 in THP-1 cells (Fig. 2 A) and human macrophages (Fig. 2 B). We found that SLAMF1 silencing caused consistent reduction in LPS-mediated IFNβ mRNA levels (Fig. 2, A and B) and IFNβ secretion (Fig. 2, C and D). In contrast, TNF mRNA amounts were only reduced at late time points of LPS stimulation, and secretion was affected only in THP-1 cells (Fig. 2, A–D). SLAMF1 silencing impaired both IL-6 and CXCL10 secretion but did not affect the secretion of IL-1β and IL-8 in THP-1 cells and human macrophages (Fig. 2, E and F). The phosphorylation of STAT1 and the initiation of transcription of IFN-inducible genes like CXCL10 are readouts of IFNβ binding to the IFNAR (Toshchakov et al., 2002). IFNAR-dependent STAT1 phosphorylation (Y701) and CXCL10 mRNA expression in response to LPS were both significantly decreased in THP-1 cells pretreated by anti-IFNAR α/β chain 2 mAbs (Fig. S1, B and C). However, SLAMF1 mRNA expression was not altered by blocking IFNAR (Fig. S1 D). Thus, SLAMF1 mRNA expression was not driven by IFNβ-mediated signaling. Figure 2. Knockdown of SLAMF1 in macrophages results in strongly reduced TLR4-mediated IFNβ mRNA expression and protein secretion as well as some decrease of TNF, IL-6, and CXCL10 secretion. (A and B) Quantification of SLAMF1, IFNβ, and TNF mRNA expression by qPCR in THP-1 cells (A) and macrophages (B) treated by 100 ng/ml ultrapure K12 LPS. (C and D) IFNβ and TNF secretion levels by THP-1 cells (C) and macrophages (D) in response to LPS (4 and 6 h) assessed by ELISA. (E and F) Secretion levels of IL-1β, IL-6, IL-8, and CXCL-10 (6 h LPS) analyzed by multiplex assays. Data are presented as means with SD for combined data from three independent experiments (A, C, and E), for three biological replicates from one of six donors (B and D), or one of three donors (F). *, P < 0.01. Upon stimulation with E. coli particles, SLAMF1-silenced THP-1 cells also showed a consistent reduction in IFNβ and TNF mRNA (Fig. S2 A). However, SLAMF1 silencing in macrophages had no effect on IFNβ or TNF mRNA expression in response to polyinosinic-polycytidylic acid (poly I:C) with or without transfection (RIG-I/MDA5 or TLR3) or to the TLR8 ligand CL075 (Fig. S2, B–D). SLAMF1 regulates TLR4-mediated signaling upstream of TBK1 and IRF3 Phosphorylation of IFN regulatory factor 3 (IRF3) transcription factor is critical for the regulation of early IFNβ transcription in macrophages (Sakaguchi et al., 2003). TBK1 acts upstream of IRF3 and phosphorylates IRF3 by itself or together with inhibitor of NF-κB kinase subunit ε (IKKε; Fitzgerald et al., 2003a). Macrophages silenced for SLAMF1 showed decreased levels in both LPS-induced TBK1 and IRF3 phosphorylation (Fig. 3, top). This was also observed in SLAMF1-silenced THP-1 cells stimulated with LPS or E. coli particles (Fig. S3, A and B). Figure 3. SLAMF1 silencing in macrophages impairs TLR4-mediated phosphorylation of TBK1, IRF3, and TAK1. Western blotting of lysate macrophages treated with a control nonsilencing oligonucleotide or SLAMF1-specific siRNA oligonucleotides and stimulated with 100 ng/ml LPS. The antibodies used are indicated in the figure. An antibody toward SLAMF1 was used to control for SLAMF1 silencing, and GAPDH was used as an equal loading control. Same GAPDH controls are presented for pTBK1, total TBK1, and phospho-p38MAPK, for total IRF3 and total TAK1, and for pTAK1 and pIκBα because they were probed on the same membranes. Western blots are representative of one of five donors. Molecular weight is given in kilodaltons. Graphs (right) show quantifications of protein levels relative to GAPDH levels obtained with Odyssey software. Transcription of IFNβ is coordinately regulated by several transcription factor families such as IRFs, NF-κB, and ATF-2–c-Jun (Ford and Thanos, 2010). To explore events upstream of ATF-2–c-Jun activation, we analyzed the effect of SLAMF1 silencing on LPS-mediated activation of MAPKs. SLAMF1 silencing resulted in decreased phosphorylation of MAPK 7 (MAP3K)/TAK1 and downstream MAPKs (p38MAPK and JNK1/2; Figs. 3 and S3 A). Both p38MAPK and JNK1/2 positively regulate the transcriptional activity of AP1 (ATF-2–c-Jun; Chang and Karin, 2001), and it is therefore likely that the observed reduction in MAPK phosphorylation upon SLAMF1 silencing may contribute to decreased AP-1 activity. The total level and phosphorylation of IκBα protein were not affected by LPS stimulation in the SLAMF1-depleted macrophages (Figs. 3 and S3 A, bottom), suggesting that SLAMF1 is not involved in the early NF-κB activation. This is consistent with our data showing that SLAMF1 silencing affected TNF levels only at late time points (Fig. 2, A–D). To further support the hypothesis that SLAMF1 regulates signaling from the endosome leading to IFNβ expression, we transduced primary macrophages with lentiviruses encoding SLAMF1. LPS-mediated IFNβ mRNA expression was significantly higher in SLAMF1-transduced cells, with only a modest effect on TNF mRNA expression (Fig. 4 A). Western blot analysis showed that the upregulation of IFNβ mRNA expression in SLAMF1-transduced cells was accompanied by higher amounts of phosphorylated TBK1, IRF3, and MAPK phosphorylation (Fig. 4 B). Thus, these results also suggest that SLAMF1 acts as a positive regulator of endosomal TLR4–TRAM–TRIF signaling. Figure 4. Lentiviral transduction of SLAMF1 in macrophages results in the increase of IRF3 and TBK1 phosphorylation in response to LPS and upregulation of IFNβ and TNF expression. (A) Quantification of SLAMF1, IFNβ, and TNF mRNA expression by qPCR in macrophages transduced by Flag-tagged SLAMF1 coding or control virus and treated by LPS. The qPCR data are presented as means and SD for three biological replicates of one of three experiments. Significance was calculated by two-tailed t tests. *, P < 0.01. (B) Western blots showing LPS-induced phosphorylation of signaling molecules in cells transduced with the SLAMF1-expressing virus versus control virus. Dividing lines were added where the time point of 4 h was excised. The same GAPDH controls are presented for total IRF3 and total IκBα and for pIκBα and pTAK1 because they were probed on the same membranes. Molecular weight is given in kilodaltons. Graphs (right) show quantifications of protein levels relative to GAPDH levels obtained with Odyssey software. SLAMF1 regulates TRAM recruitment to E. coli phagosomes in a Rab11-dependent manner Both TLR4 and TRAM are rapidly recruited to E. coli phagosomes after phagocytosis and are required for induction of IFNβ (Husebye et al., 2010). Because SLAMF1 was needed for TRAM-TRIF signaling, we tested whether SLAMF1 was recruited to E. coli phagosomes containing TRAM. We found that TRAM and SLAMF1 were recruited to early (EEA1-positive) and late (LAMP1-positive) E. coli phagosomes (Fig. 5 A). This was consistent with data published for SLAMF1 in mouse macrophages, where SLAMF1 was found both on EEA1- and LAMP1-positive E. coli phagosomes (Berger et al., 2010). Moreover, we did not detect SLAMF1 on Staphylococcus aureus phagosomes (Fig. S3, A and B), similar to data reported for mouse macrophages (Berger et al., 2010). Figure 5. SLAMF1 regulates TRAM recruitment to E. coli phagosomes. (A) SLAMF1 costaining with TRAM, EEA1, or LAMP1 in primary macrophages coincubated with E. coli pHrodo particles for indicated time points. SLAMF1 (green), E. coli (blue), and TRAM, EEA1, or LAMP1 (red) are shown. The data shown are representative of one out of four donors. Bars, 10 μm. (B and C) TRAM and SLAMF1 MIs on E. coli phagosomes upon SLAMF1 silencing (B) or simultaneous Rab11a and Rab11b silencing (C) in primary human macrophages quantified from xyz images. The scatter plots are presented as median values of TRAM voxel intensity, and numbers of phagosomes are shown at the top. The nonparametric Mann-Whitney test was used to evaluate statistical significance. *, P < 0.01; ***, P ≤ 0.0001. Human macrophages were incubated with E. coli particles for indicated time points, fixed, and costained for SLAMF1 and TRAM, normal rabbit (rIgG), or mouse IgG (mIgG). The data shown are representative for one out of five (B) or four (C) donors. We hypothesized that SLAMF1 could be involved in the transport of TRAM to E. coli phagosomes as this is a crucial step for TLR4-dependent IFNβ induction. Control and SLAMF1-silenced macrophages were pulsed with E. coli pHrodo particles for 15 min followed by 15 min chase in particle-free medium. The mean voxel intensities (MIs) for TRAM, SLAMF1, and pHrodo fluorescence on the phagosomes were calculated from z stacks obtained by confocal microscopy using 3D image analysis software (Fig. 5 B). We found that uptake of E. coli particles was not significantly affected by SLAMF1 silencing (Fig. S4 C). However, acidification of the E. coli phagosomes was significantly decreased upon SLAMF1 silencing (Fig. S4 D). Remarkably, we found that TRAM recruitment to E. coli phagosomes was markedly decreased upon SLAMF1 silencing (Fig. 5 B, left). As expected, SLAMF1-silenced cells showed decreased amounts of SLAMF1 on E. coli phagosomes (Fig. 5 B, right). Thus, SLAMF1 seems to positively regulate TRAM recruitment to E. coli phagosomes. Transport of TRAM to phagosomes is known to be Rab11 dependent (Husebye et al., 2010). Moreover, SLAMF1 was located to the Rab11-positive compartment in resting cells (Fig. 1 C), positively regulated transport of TRAM to phagosomes (Fig. 5 B), and relocalized from the ERC in monocytes upon addition of E. coli or LPS (Fig. S4, E–G). Based on these observations, we tested whether SLAMF1 recruitment to phagosomes was Rab11 dependent. Two members of Rab11 subfamily, Rab11a and Rab11b, were simultaneously silenced in human macrophages. After silencing, macrophages were stimulated with E. coli for 15 and 30 min, and recruitment of TRAM and SLAMF1 to the phagosomes was quantified by evaluating MIs for TRAM and SLAMF1 staining (Fig. 5 C). Rab11 silencing significantly reduced the amounts of SLAMF1 and TRAM at the phagosomes (Fig. 5 C). SLAMF1 interacts with the N-terminal part of the TRAM TIR domain To investigate whether TRAM recruitment to E. coli phagosomes could be regulated by a physical interaction between SLAMF1 and TRAM, we performed endogenous immunoprecipitations (IPs) using anti-SLAMF1 and anti-TRAM antibodies. Endogenous SLAMF1 coprecipitated with TRAM in macrophages, and this interaction was enhanced upon LPS stimulation (Fig. 6 A). In contrast, the TIR-adapter MyD88 did not coprecipitate with SLAMF1 (Fig. 6 A, right), supporting the specificity of the SLAMF1–TRAM interaction. Endogenous TRAM also coprecipitated with SLAMF1 both before and after LPS treatment (Fig. 6 B). The bands detected by anti-TRAM antibody were specific as a similar band could be observed in coimmunoprecipitations (co-IPs) with TLR4 upon LPS stimulation (Fig. S5 A). Overall, we conclude that endogenous SLAMF1 interacts with TRAM in human macrophages and that the interaction is enhanced upon LPS stimulation. Figure 6. SLAMF1 interacts with TRAM protein. (A) Endogenous IPs using specific anti-SLAMF1 mAbs from macrophages stimulated by LPS. (B) Endogenous IPs using anti-TRAM polyclonal antibodies from macrophages stimulated by LPS. (C) TRAMFlag-precipitated SLAMF1 and SLAMF1ct was needed for interaction with TRAM. (D) Coprecipitation of TRAM deletion mutants: TIR domain (68–235), short TRAM TIR domain (68–176 aa), and N-terminal (1–68 aa) or C-terminal (158–235 aa) domains with SLAMF1 protein. (E) Coprecipitation of TRAM deletion mutants containing the N-terminal part of TRAM TIR domain with SLAMF1. (F) Coprecipitation of SLAMF1Flag deletion mutants with TRAMYFP. (G) Coprecipitation of human SLAMF1Flag with human TRAMYFP and of mouse SLAMF1Flag with mouse TRAMEGFP. Black dashed lines indicate that intervening lanes have been spliced out. (H) Human SLAMF1 cytoplasmic tail coprecipitation with TRAMYFP with or without amino acid substitutions at 321–324. Graphs under C–F summarize the IPs’ results. Indicated constructs were transfected to HEK293T cells, and anti-Flag agarose was used for the IPs. For endogenous IPs, specific SLAMF1 or TRAM antibodies were covalently coupled to beads. At least three independent experiments were carried out for anti-Flag IPs, and five independent experiments were carried out for the endogenous IPs, and one representative experiment is shown for each. Molecular weight is given in kilodaltons. WB, Western blot; WCL, whole-cell lysate. Furthermore, HEK293T cells were transiently transfected with Flag-tagged TRAM (TRAMFlag) and full-length SLAMF1 or deletion mutant lacking the C terminus of SLAMF1 (SLAMF1Δct). We found that full-length SLAMF1 but not SLAMF1Δct coprecipitated with TRAMFlag, suggesting that the TRAM interaction site was located at the C terminus of SLAMF1 protein (Fig. 6 C). To map the TRAM region responsible for the interaction with SLAMF1, we generated several Flag-tagged TRAM deletion mutants. The mutants contained the N terminus (1–68), C terminus (158–225), TIR domain (68–176), or the TIR domain plus the C terminus (68–235) of TRAM (1–235; UniProtKB, Q86XR7). Of all these mutants, only TRAM 68–235 and TRAM 68–176 coprecipitated with SLAMF1 (Fig. 6 D). To further define the subdomain of TRAM involved in the TRAM–SLAMF1 interaction, we made a series of TRAM deletion mutants, which contained the N-terminal part of TRAM with 10–20-aa increments. Although TRAM 1–68 mutant did not bind SLAMF1 (Fig. 6, D and E), a weak interaction was found with TRAM 1–79 that increased markedly for TRAM 1–90 and further for TRAM 1–100. Collectively, these results suggest that the SLAMF1 binding site in TRAM is located within the first 30–35 aa of the TRAM TIR domain (68–95 aa; Fig. 6 E). TRAM interacts with the C-terminal part of SLAMF1, and the interaction occurs for human but not mouse proteins We used a similar strategy to establish the TRAM-interacting subdomain of SLAMF1 by deleting amino acids from the C-terminal part of SLAMF1. Cells were cotransfected with Flag-tagged SLAMF1 WT or deletion mutants along with TRAMYFP (Fig. 6 F). A deletion mutant (1–330) lacking the last five C-terminal amino acids did not interact with TRAMYFP (Fig. 6 F), pinpointing the TRAM interaction site at the very C terminus of SLAMF1. There are two tyrosine residues in the cytoplasmic tail of human SLAMF1 in the signaling motifs designated as immunoreceptor tyrosine–based switch motifs (Shlapatska et al., 2001). SLAMF1 tyrosine phosphorylation and interaction with other proteins via immunoreceptor tyrosine–based switch motifs could potentially alter SLAMF1–TRAM interaction. Point mutations Y281F, Y327F, and double mutation Y281/327F did not alter the interaction with TRAM (Fig. S5 B). Endogenous SLAMF1 was tyrosine phosphorylated in resting macrophages and subsequently dephosphorylated within the 45 min of LPS stimulation (Fig. S5, C and D). Furthermore, both nonphosphorylated and tyrosine-phosphorylated (pY) recombinant SLAMF1ct GST fusion proteins effectively pulled out TRAM from lysates of untreated or LPS-treated macrophages (Fig. S5 E). Thus, SLAMF1–TRAM interaction was not altered by tyrosine phosphorylation of SLAMF1. Regulation of the LPS-induced IFNβ response seems to differ between humans and mice. Human macrophages respond to LPS with at least 10-fold higher IFNβ mRNA expression compared with mouse BMDMs and thioglycolate-elicited peritoneal macrophages (Schroder et al., 2012). Thus, we wanted to check whether SLAMF1 and TRAM interaction was conserved across species, and we tested murine SLAMF1 and TRAM proteins for interaction. Indeed, mouse TRAMEGFP did not coprecipitate with mouse SLAMF1Flag (Fig. 6 G). Interestingly, three amino acids in human SLAMF1ct upstream of potential TRAM-binding site TNSI (321–324; UniProtKB, Q13291) are different from mouse SLAMF1ct, containing PNPT (329–332; UniProtKB, Q9QUM4; Fig. S6 A). Substitution of the TNSI sequence with PNPT in human SLAMF1 abrogated its interaction with TRAMYFP (Fig. 6 H). Thus, these amino acids are crucial for interaction. However, the sequence in human TRAM involved in the interaction with human SLAMF1 is not fully conserved in murine TRAM (Fig. S6 B), which also could explain why murine TRAM and SLAMF1 do not interact. SLAMF1 has been shown to regulate E. coli phagosome maturation in mouse BMDMs, but it did not modify the response to ultrapure LPS (Berger et al., 2010). However, regulation of mRNA expression or secretion of type I IFNs by SLAMF1 have not been previously addressed in BMDMs. We stimulated C57BL/6 Slamf1−/− and control BMDMs with 100 ng/ml of ultrapure LPS or E. coli particles and tested for Ifnβ and Tnf mRNA expression and cytokine secretion (Fig. S6, C–E). Indeed, Slamf1−/− BMDMs showed comparable Ifnβ and Tnf mRNA levels to control BMDMs (Fig. S5 C), and the amounts of IFNβ and TNF secreted from Slamf1−/− BMDMs stimulated with LPS or E. coli were not significantly altered (Fig. S6, D and E). Thus, mouse SLAMF1 does not interact with TRAM protein. Therefore, mouse SLAMF1 did not affect TRAM-TRIF–mediated IFNβ secretion in murine macrophages. Rab11 interacts with SLAMF1 via class I FIPs As SLAMF1 recruitment to E. coli phagosomes was found to be Rab11 dependent (Fig. 5 C), we investigated whether SLAMF1 could form a complex with Rab11 via effector proteins such as Rab11 FIPs (Horgan and McCaffrey, 2009). Individual FIPs (FIP1–5) were coexpressed together with SLAMF1 and Rab11aFlag proteins in HEK cells followed by co-IP with Rab11aFlag (Fig. 7 A). All three members of class I FIPs were found to form a complex with SLAMF1 and Rab11a (Fig. 7 A). All class I FIPs are characterized by a phospholipid-binding C2 domain (Fig. 7 B), which is located between aa 1–129 in FIP2 (Lindsay and McCaffrey, 2004). We found that the ΔC2 mutant of FIP2 (lacking aa 1–128) could still bind SLAMF1 (Fig. 7 D). Protein sequence alignment between the class I FIPs showed a highly conserved domain between aa 117–191 in FIP2 with undefined function (Fig. 7, B and C). To figure out whether this domain in FIP2 could be responsible for interaction with SLAMF1, several Flag-tagged FIP2 deletion mutants were tested with or without Rab11CFP overexpression. The 1–192-aa deletion mutant was the minimal deletion mutant found to interact with SLAMF1 (Fig. 7, D–F). Both this mutant and the ΔC2 mutant contains a common 62-aa motif that could be important for interaction with SLAMF1 (Fig. 7, D–F). Figure 7. SLAMF1 interacts with all class I Rab11 FIPs. (A) Anti-Flag IPs for Rab11aFlag with EGFP-tagged Rab11FIPs (1–5) and SLAMF1. (B) Schematic figure for class I and class II Rab11 FIPs domain structure. C2, phospholipid-binding C2 domain; EF, EF-hand domain; PRR, proline-rich region; RBD, Rab11 binding domain. (C) Homologous protein sequence in class I FIPs, which follow the C2 domain. Identical amino acids in all three class I FIPs are highlighted. (D) Coprecipitation of SLAMF1Flag with FIP2EGFP WT or FIP2 deletion mutant lacking the C2 domain (ΔC2). (E and F) Coprecipitation of untagged SLAMF1 with FIP2Flag (1–512 aa) and Flag-tagged FIP2 deletion mutants in anti-Flag IPs in the absence (E) or presence (F) of overexpressed Rab11CFP. (G) Quantification of coprecipitations in E and F between SLAMF1 and FIP2Flag variants correlated with the amount of Flag-tagged protein on the blot and Flag-tagged protein sizes. Error bars represent means ± SD for three independent experiments. (H) Coprecipitation of FIP2Flag with SLAMF1 and Rab11a WT, Rab11a Q70L mutant (QL), or Rab11a S25N mutant (SN). (I) Coprecipitation of SLAMF1Flag deletion mutants with FIP2EGFP. Molecular weight is given in kilodaltons. WB, Western blot; WCL, whole-cell lysate. (J) Scheme for FIP2- and TRAM-interacting domains in SLAMF1ct. The results are representative of at least three independent experiments. The tested deletion mutants more efficiently precipitated SLAMF1 than the full-length FIP2 (Fig. 7 E), but coexpression of Rab11 with full-length FIP2 and SLAMF1 markedly increased its binding to SLAMF1 (Fig. 7, F and G). All FIP2 deletion mutants in IPs lacked a C-terminal Rab11 binding domain (Fig. 7, B, E, and F) and showed better coprecipitation with SLAMF1 without Rab11 overexpression, which suggested that Rab11 has a critical role in controlling FIP2–SLAMF1 interactions. Next, we examined whether FIPs could coprecipitate SLAMF1 efficiently only in the presence of GTP-bound active Rab11. Rab11 GTPase functions as a molecular switch, being active in the GTP-bound state and inactive in the GDP-bound state. It has been shown that FIPs only interact with activated Rab11 (Junutula et al., 2004; Gidon et al., 2012). FIP2Flag was precipitated from cells, which coexpressed SLAMF1 with either Rab11a WT, GTP-bound Rab11Q70L, or GDP-bound Rab11a S25N (Fig. 7 G). FIP2Flag coprecipitated with SLAMF1 only in the presence of Rab11 WT and Rab11Q70L but not Rab11 serine mutant (Fig. 7 G). We also found that the interaction domain for FIP2 in SLAMF1ct was distinctly different from the TRAM interaction domain (Figs. 6 F and 7, I and J). Our results suggest that TLR4-induced activation of Rab11 is a signal for the recruitment of SLAMF1 and TRAM via the FIPs to E. coli phagosomes and that class I FIPs may link the SLAMF1–TRAM complex to Rab11. SLAMF1 is recruited to the TLR4–TRAM–TRIF complex We defined the SLAMF1-interacting site in TRAM as the N-terminal part of TRAM TIR domain (68–95 aa; Fig. 6 C). This raised an important question of whether SLAMF1 could regulate the subsequent formation of TLR4–TRAM–TRIF complex needed for LPS-mediated signaling. To address this question, HEK cells were cotransfected by SLAMF1Flag, TRAMYFP, and TLR4Cherry or TRIFHA, and their interactions were monitored by co-IP with SLAMF1Flag. Both TLR4Cherry and TRIFHA coprecipitated with SLAMF1Flag in the presence of TRAMYFP (Fig. 8, A and B). In addition, SLAMF1 did not coprecipitate with TLR4Flag in the absence of TRAMYFP (Fig. 8 C). Overexpression of SLAMF1 did not alter the ability of TLR4 to attract TRIF via TRAM despite that SLAMF1 also coprecipitated with the complex in the presence of TRAM and TRIF (Fig. 8 D). Furthermore, TRIF overexpression strongly enhanced SLAMF1 coprecipitation with TLR4Flag (Fig. 8 E). In summary, SLAMF1 binding to TRAM seems to be unique and outside of the TIR-TIR dimerization domain as it does not interfere with TRAM–TLR4 interaction and subsequent TRIF recruitment. Figure 8. TRAM acts as a bridge between the SLAMF1 and TLR4 signaling complex. (A and B) Coprecipitations of SLAMF1Flag with TLR4Cherry (A) or TRIFHA (B) with or without TRAMYFP overexpression. (C) Coprecipitation of TLR4Flag with SLAMF1 with or without TRAMYFP overexpression. (D) TLR4Flag interaction with TRAMYFP and TRIFHA with or without SLAMF1 coexpression. (E) Coprecipitation of SLAMF1 with or without TRIFHA in the presence of TRAMYFP by TLR4Flag. Indicated constructs were transfected to HEK293T cells. pDuo-CD14/MD-2 vector was cotransfected to all wells (A and C–E). Anti-Flag agarose was used for IPs. At least three independent experiments were performed. Molecular weight is given in kilodaltons. WB, Western blot; WCL, whole-cell lysate. TRAM and SLAMF1 positively regulate bacterial killing by macrophages SLAMF1 controls killing of Gram-negative bacteria by mouse BMDMs through generation of ROS (Berger et al., 2010). We tested E. coli–mediated ROS generation by SLAMF1-silenced human macrophages and found that SLAMF1 also acts as a positive regulator of ROS generation in human cells (Fig. 9 A). Figure 9. TRAM and SLAMF1 are essential for the killing of E. coli by human macrophages. (A) Flow cytometry analysis of dihydrorhodamine 123 (DHR-123) fluorescence to access ROS activation in control siRNA or SLAMF1 siRNA human macrophages upon stimulation by E. coli red pHrodo particles. One of three experiments shown. (B and C) Bacterial killing assays by SLAMF1-silenced and control THP-1 cells (B) as well as TRAM KO and control THP-1 cells (C) infected with a DH5α strain at MOI 40. (D and E) Western blot analysis of pAkt (S473) and pIRF3 (S396) levels induced by E. coli particles in THP-1 WT and TRAM KO cells (D) as well as SLAMF1-silenced or control oligonucleotide–treated cells (E). Graphs (right) on Western blotting show quantification of protein levels relative to β-tubulin obtained with Odyssey software. (F) Western blot showing phospho-(S396) IRF3 and phospho-(S473) Akt levels in lysates of THP-1 cells coincubated with E. coli particles for 1 h in the presence or absence of TBK1-IKKε inhibitor (MRT67307), pan-Akt allosteric inhibitor (MK2206), or DMSO. Molecular weight is given in kilodaltons. (G) Bacterial killing assays by THP-1 cells with DMSO (<0.01%), 1 μM Akt inhibitor MK2206, or 2 μM TBK1-IKKε inhibitor MRT67307 upon infection by DH5α at MOI 40. Percent killing was calculated as 100 − (number of CFUs at time X/number of CFUs at time 0) × 100 for average values of technical replicates, and each dot on the graphs in B, C, and G represents a biological replicate from three independent experiments. Median values are shown by lines. Statistical significance was calculated by a Mann-Whitney nonparametric test. **, P < 0.01; ***, P < 0.001. Slamf1−/− BMDMs demonstrated reduced bacterial killing at 6 h of E. coli infection (Berger et al., 2010). To test the effect of SLAMF1 on bacterial killing by human cells, SLAMF1-silenced THP-1 or TRAM knockout (KO) cells with respective control cells were incubated with live DH5α E. coli. Bacterial killing was strongly decreased already at early time points (1 h and 1.5 h) in SLAMF1-silenced cells (Fig. 9 B) and was almost completely abolished in TRAM KO cells (Fig. 9 C). Negative values of the percentage of killing in TRAM KO cells pointed to intracellular bacterial replication in these cells (Fig. 9 C). TRIF-dependent signaling activates TBK1-IKKε kinases that regulate the integrity of pathogen-containing vacuoles and restrict bacterial proliferation in the cytosol (Radtke et al., 2007; Thurston et al., 2016). TRAM is a crucial adapter for TRIF recruitment to activated TLR4, leading to the activation of TBK1 and IKKε (Oshiumi et al., 2003; Yamamoto et al., 2003; Fitzgerald et al., 2004). Upon TLR4 ligation, TBK1 and IKKε phosphorylate Akt kinase (S473), resulting in Akt activation (Krawczyk et al., 2010; Everts et al., 2014). In turn, the Akt-mTORC1 signaling axis can drive phagocytosis, phagosome maturation, and ROS production, which are essential for bacterial killing (Kelly and O’Neill, 2015). As expected, TRAM KO cells had no detectable IRF3 phosphorylation in response to E. coli particles (Fig. 9 D). In control cells, E. coli–mediated Akt S473 phosphorylation underwent the similar kinetics as IRF3 phosphorylation and was completely abolished in TRAM KO cells (Fig. 9 D) and strongly decreased in SLAMF1-depleted cells (Fig. 9 E). The TBK1-IKKε inhibitor MRT67307 decreased TLR4-mediated Akt phosphorylation in THP-1 cells, whereas the Akt inhibitor MK2206 completely abrogated Akt phosphorylation (Fig. 9 F). Both compounds inhibited bacterial killing in THP-1 cells, especially at the earliest time point (Fig. 9 G). Moreover, coincubation with MRT67307 resulted in the increase of intracellular bacterial number as could be seen by negative values in the percentage of bacterial killing (Fig. 9 G). Hence, TBK1-IKKε activity and subsequent E. coli–mediated Akt phosphorylation directly correlated with the ability of cells to kill bacteria and restrict intracellular replication. Activation of PI3K and subsequent Akt phosphorylation downstream of TLR2 and TLR4 has been extensively explored in many model systems (Laird et al., 2009; Troutman et al., 2012). It was previously reported that TLR2- and TLR4-mediated Akt S473 phosphorylation is MyD88 dependent in murine model systems (Laird et al., 2009). Our data on MyD88 silencing in primary human macrophages showed that E. coli–induced Akt S473 phosphorylation was not dependent on MyD88 but was dependent on TRAM (Fig. S7 A). In human macrophages, the kinetics of Akt phosphorylation induced by E. coli particles were much faster and robust than those induced by the TLR2 ligand FSL-1 or the TLR4 ligand LPS (Fig. S7 B). TLR2 and TLR4 ligands were not inducing pAkt much over the background level in THP-1 cells, which were used for bacterial killing assays, and pAkt levels were only modestly affected by SLAMF1 or TRAM silencing (Fig. S7 C). In contrast, THP-1 cells stimulated with E. coli particles showed a 15–20-fold increase in pAkt that were almost lost in cells depleted for TRAM or SLAMF1 (Fig. 9, D and E). Thus, TRAM and SLAMF1 are involved in regulation of E. coli–mediated but not pure TLR ligand–mediated Akt phosphorylation in THP-1 cells. Discussion Despite that SLAMF1 has been reported to control inflammatory responses and defense against Gram-negative bacteria in mice, the underlying mechanisms are elusive (Theil et al., 2005; van Driel et al., 2012, 2016). Moreover, little has been shown about the role of SLAMF1 in modulating the inflammatory response against Gram-negative bacteria in human macrophages. In this study, we show for the first time that human SLAMF1 regulates TLR4-mediated TRAM-TRIF–dependent signaling by the unique interaction with the signaling adapter TRAM. Mouse BMDMs express high levels of SLAMF1 on the plasma membrane, whereas resting human monocytes and human monocyte–derived macrophages have been considered to be SLAMF1 negative (Farina et al., 2004; Romero et al., 2004). In contrast, we found that human monocytes and macrophages largely expressed SLAMF1 in the intracellular Rab11+ ERC compartment; however, they had weak or no expression on the cell surface. After stimulation by E. coli, SLAMF1 relocalized from ERCs to E. coli– or LPS-containing phagosomes that resembled the previously reported Rab11a-dependent transport of TLR4 and TRAM from ERCs to E. coli phagosomes (Husebye et al., 2010). Endogenous SLAMF1 was already bound to TRAM before stimulation, and upon E. coli phagocytosis, both proteins were recruited to phagosomes by Rab11 GTPases, with class I Rab11 FIPs as effector molecules. It is known that Rab11 functions as a molecular switch, which cycles between two conformational states: a GTP-bound “active” form and a GDP-bound “inactive” form (Guichard et al., 2014). Surface TLR4 interaction with LPS on the E. coli outer membrane induces fast intracellular complex formation, resulting in multiple posttranslational modifications of signaling molecules (Mogensen, 2009). We hypothesize that TLR4 signaling results in a shift from a Rab11 GDP-bound to a Rab11 GTP-bound active state. Moreover, previous research has demonstrated that FIPs prefer binding to GTP-bound Rab11 (Junutula et al., 2004). Thus, after the Rab11 GDP/GTP ratio shifts to the GTP-bound state, FIPs would connect cargo to Rab11 vesicles, which would enhance delivery of SLAMF1 and TRAM via class I FIPs from ERCs to E. coli phagosomes. Mouse SLAMF1 was shown to be a bacterial sensor by itself, recognizing porins in the outer bacterial membrane (Berger et al., 2010). The regulatory role of mouse SLAMF1 upon TLR4 ligation by LPS is directly dependent on the porins present in crude LPS preparations or porins in the bacterial outer membrane (Berger et al., 2010). Unlike its mouse orthologue, human SLAMF1 did not require interaction with bacterial porins to elicit its effects on TLR4-mediated IFNβ production as similar data were obtained with both E. coli bioparticles and ultrapure LPS. The delivery of TRAM to endosomes and phagosomes is crucial for the activation of the IRF3 signaling pathway and IFNβ induction (Kagan et al., 2008; Husebye et al., 2010). We found that SLAMF1 silencing caused a significant decrease in TRAM accumulation around E. coli phagosomes. Moreover, endogenous TRAM coimmunoprecipitated with SLAMF1. These data demonstrate that SLAMF1 is a critical regulator of TRAM recruitment to the phagosomes. We were able to map the domain in SLAMF1 involved in the interaction with TRAM to 15 C-terminal amino acids as both deletion of five C-terminal amino acids and substitution of amino acids at positions 321–324 abrogated interaction of SLAMF1 protein with TRAM. The interaction domain in TRAM was located outside the BB loop as TIR-TIR dimerization is not affected, and it was mapped to the N-terminal part of TRAM TIR domain between aa 68–95. We found that mouse SLAMF1ct contains different amino acid sequences at positions corresponding with human 321–324 residues. This resulted in the absence of interaction between mouse SLAMF1 and TRAM, and consequently, LPS- or E. coli–induced IFNβ expression was not altered in Slamf1−/− BMDMs when compared with WT cells. The failure to activate TBK1-IKKε kinase observed upon SLAMF1 silencing may affect the antibacterial functions of TBK1-IKKε (Radtke et al., 2007; Thurston et al., 2016). The Akt kinase, which is activated by TBK1-IKKε upon TLR4 ligation, is directly involved in the TLR4-mediated switch to glycolysis by phosphorylating crucial downstream target proteins (Krawczyk et al., 2010; Kelly and O’Neill, 2015). Moreover, Akt is involved in the activation of NADPH oxidase by the phosphorylating p47 subunit (Chen et al., 2003; Hoyal et al., 2003), which may result in ROS generation needed for bacterial killing (West et al., 2011). We demonstrate that SLAMF1 and TRAM were required for E. coli–mediated Akt phosphorylation via TBK1-IKKε as well as for the efficient bacterial killing. It is known that TLR2 and TLR4 ligands activate Akt S473 in a MyD88-dependent manner (Laird et al., 2009; Troutman et al., 2012). It should be noted that in contrast with these studies, we have used E. coli particles and found that Akt S473 phosphorylation was TRAM and SLAMF1 dependent. Akt phosphorylation induced by pure TLR2 and TLR4 ligands could be more dependent on MyD88, but it was not dependent on SLAMF1 or TRAM in THP-1 macrophages. There is accumulating evidence on divergent regulation of TLR4 signaling and gene expression in different species (Schroder et al., 2012; Vaure and Liu, 2014). It is known that humans and old-world monkey species are highly sensitive to LPS with physiological changes induced by a dose at nanograms per kilogram, whereas rodents are highly insensitive to LPS with physiological changes only induced by a dose at milligrams per kilogram (Vaure and Liu, 2014). Many therapeutic agents that reduce inflammation and mortality in mouse septic shock models show no clinical benefit for humans (Poli-de-Figueiredo et al., 2008). Human monocyte-derived macrophages express much higher levels of IFNβ mRNA in response to LPS than mouse BMDM or thioglycolate-elicited peritoneal macrophages (Schroder et al., 2012). Our findings support higher LPS-induced secretion of IFNβ by human macrophages compared with BMDMs. Thus, during evolution, human macrophages must have acquired mechanisms to enhance TLR4-mediated IFNβ production in response to LPS or, vice versa, mouse cells developed less sensitive response to bacterial LPS. We suggest that SLAMF1-regulated transport of TRAM to the TLR4 signaling complex on bacterial phagosomes could be one of the features specific for human cells, which amplifies the IFNβ secretion. Thus, human SLAMF1 could potentially be targeted to regulate TLR4-mediated cytokine production in inflammatory conditions. Materials and methods Primary cells and cell lines Use of human monocytes from blood donors was approved by the Regional Committees from Medical and Health Research Ethics at the Norwegian University of Science and Technology. Human monocytes were isolated from buffycoat by adherence as previously described (Husebye et al., 2010). In brief, freshly prepared buffycoat (St. Olavs Hospital) was diluted by 100 ml of PBS and applied on top of Lymphoprep (Axis-Shield) according to the manufacturer’s instructions. PBMCs were collected and washed by HBSS (Sigma-Aldrich) four times with low-speed centrifugation (150–200 g). Cells were counted using Z2 Coulter particle count and size analyzer (Beckman Coulter) on program B, resuspended in RPMI 1640 (Sigma-Aldrich) supplemented with 5% of pooled human serum at a concentration of 8 × 106 per ml, and seeded to six-well (1 ml per well) or 24-well (0.5 ml per well) cell culture dishes. After a 45-min incubation allowing surface adherence of monocytes, the dishes were washed three times by HBSS to remove nonadherent cells. Monocytes were maintained in RPMI 1640 supplemented with 10% pooled human serum (St. Olavs Hospital) and used within 24 h after isolation. Monocyte-derived macrophages were obtained by differentiating cells for 8–10 d in RPMI 1640 with 10% human serum and 25 ng/ml rhM-CSF (216-MC-025; R&D Systems). THP-1 cells (ATCC) were cultured in RMPI 1640 supplemented by 10% heat-inactivated FCS, 100 nM penicillin/streptomycin (Thermo Fisher Scientific), and 5 μM β-mercaptoethanol (Sigma-Aldrich). THP-1 cells were differentiated with 50 ng/ml of PMA (Sigma-Aldrich) for 72 h, followed by 48 h in medium without PMA. HEK293T cells (ATCC) were cultured in DMEM with 10% FCS. For making the TRAM KO THP-1 cell line, LentiCRISPRv2 plasmid (a gift from F. Zhang; 52961; Addgene; Sanjana et al., 2014) was ligated with 5′-CACCGATGACTTTGGTATCAAACC-3′ and 5′-AAACGGTTTGATACCAAAGTCATC-3′ for TRAM. Packaging plasmids pMD2.G and psPAX2 were used for producing lentivirus (provided by D. Trono, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland; 12260 and 12259; Addgene). HEK293T cells were cotransfected with the packaging and lentiCRISPRv2 plasmids and then washed after 16 h. The lentivirus-containing supernatants were collected after 48 h and used for transduction of THP-1 cells along with 8 µg/ml protamine sulphate. Transduced THP-1 cells were then selected with puromycin (1 μg/ml) for 1 mo and tested for TRAM protein expression by Western blotting. All cell lines were regularly checked for mycoplasma contamination. Reagents and cell stimulation pHrodo red E. coli and S. aureus as well as AF488-conjugated E. coli bioparticles were purchased from Thermo Fisher Scientific. Ultrapure 0111:B4, K12 LPS from E. coli, poly I:C, imidazoquinoline compound R848 (Resiquimod), thiazoloquinoline compound CL075, and synthetic diacylated lipoproteins FSL-1 (Pam2CGDPKHPKSF) and Pam3CSK4 (P3C) were from InvivoGen. Ultrapure K12 LPS or 0111:B4 LPS (InvivoGen) were used at concentrations of 100 ng/ml. E. coli bioparticles were reconstituted in 2 ml PBS, and 50 µl/well (1.5 × 107 particles) in 1 ml of media was used for cells in six-well plates (Nunc) or 35-mm glass-bottomed tissue cell dishes (MatTek Corporation), and 15 µl/well (0.45 × 107 particles) in 0.5 ml of media were used for 24-well plates (Nunc). The pan-Akt inhibitor MK2206 (1032350-13-2; Axon Medchem) and the TBK1-IKKε inhibitor MRT67307 (from P. Cohen, University of Dundee, Dundee, Scotland, UK; Clark et al., 2011) were diluted in DMSO at a concentration of 20 mM and stored at −80°C, and working solutions were prepared in cell culture media immediately before use. Antibodies The following primary antibodies were used: rabbit anti–TICAM-2/TRAM (GTX112785) from Genetex; rabbit mAb anti–human SLAMF1/SLAMF1 (10837-R008-50) from Sino Biological Inc.; mouse anti-GAPDH (ab9484) and rabbit anti–phospho-IRF3 Ser386 (ab76493) from Abcam; rabbit anti–phospho-Akt Ser473 (D9E; 4060), phospho-IRF3 Ser396 (4D4G; 4947), IkB-α (44D4; 4812), phospho–IkB-α (14D4; 2859), p38MAPK (9212), phospho-p38MAPK (Thr180/Tyr182; D3F9; 4511), TBK1/NAK (D1B4; 3504), phospho-TBK1/NAK (Ser172; D52C2; 5483), phospho-TAK1 (T184/187; 90C7; 4508), TAK1 5206, phospho–stress-activated protein kinase (SAPK)/JNK (Thr183/Tyr185; 81E11; 4668), anti-DYKDDDDK tag (D6W5B)/Flag tag (14793), anti-MyD88 (D80F5; 4283), and phospho-STAT1 (Tyr701; D4A7; 7649) from Cell Signaling Technology; rabbit anti–total IRF3 (FL-425; sc-9082) and proliferating cell nuclear antigen (PCNA; FL-261; sc-7907) were from Santa Cruz Biotechnology, Inc.; Living Colors rabbit anti–full-length GFP polyclonal antibodies (632592) from Takara Bio Inc.; 4G10 platinum antiphosphotyrosine antibody biotin conjugated (16-452) from EMD Millipore; and mouse anti-GST antibodies (SAB4200237) and monoclonal mouse ANTI-FLAG M2 antibodies (F1804-200UG) from Sigma-Aldrich. Secondary antibodies (HRP linked) for Western blotting were swine anti–rabbit (P039901-2) and goat anti–mouse (P044701-2) from Dako/Agilent Technologies. The following antibodies were used for staining and/or IPs: rabbit anti-LAMP1 (ab24170) and GM130 antibody (EP892Y) cis-Golgi marker (ab52649) from Abcam; rabbit anti-EEA1 (H-300; sc-33585), TICAM2/TRAM (H-85), TLR4 (H-80; sc-10741), normal rabbit IgG (sc-2027), and normal mouse IgG (sc-2025) were from Santa Cruz Biotechnology, Inc.; rabbit anti-Rab11, low endotoxin, azide-free (LEAF)-purified mouse IgG1 isotype control (MOPC-21; 400124) and LEAF-purified mouse anti-CD150 (SLAMF1) A12 (7D4; 306310) were from BioLegend; and anti-SLAMF1 IgG1 (IPO-3) was provided by S.P. Sidorenko (Natural Academy of Sciences, Kiev, Ukraine; Sidorenko and Clark, 1993). Secondary antibodies for confocal microscopy were goat anti–mouse IgG (H+L) Alexa Fluor 405 conjugate (A-31553), Alexa Fluor 488 conjugate (A-11001), Alexa Fluor 647 conjugate (A-21235), goat anti–rabbit IgG (H+L) Alexa Fluor 405 conjugate (A-31556), Alexa Fluor 488 conjugate (A-11008), and DNA stain Hoechst 33342 (62249) from Thermo Fisher Scientific. Imaging and image analysis Confocal images were captured using either an LSM510 META (ZEISS) equipped with a Plan Apochromat 63× 1.4 NA oil immersion objective (images presented in Fig. 1 A and used for 3D modeling in Fig. 1 B) or a TCS SP8 (Leica Microsystems) equipped with a high-contrast Plan Apochromat 63× 1.40 NA oil CS2 objective. Fluorescence was captured by standard photomultiplier tube detectors (LSM 510 META; ZEISS), stimulated emission depletion hybrid detector, or photomultiplier tube detectors (TCS SP8). Acquisition software for the LSM 510 META was Zen microscope software (2012; ZEISS) and for the TCS SP8 was LAS AF software (4.0.0.11706; Leica Microsystems). Before imaging, cells were fixed with 2% paraformaldehyde in PBS on ice, and then immunostaining was performed as described previously (Husebye et al., 2010). In brief, upon fixation, the cells were permeabilized with PEM buffer (80 mM K-Pipes, pH 6.8, 5 mM EGTA, 1 mM MgCl2, and 0.05% saponin) for 15 min on ice, quenched of free aldehyde groups in 50 mM NH4Cl with 0.05% saponin for 5 min, and blocked in PBS with 20% human serum and 0.05% saponin. The cells were incubated with primary antibody in PBS with 2% human serum and 0.05% saponin overnight at 4°C or for 2 h at RT. Alexa Fluor–labeled secondary antibodies (Invitrogen/Thermo Fisher Scientific) were incubated for 15 min at RT after three washes in PBS with 0.05% saponin. If double staining was made, cells were sequentially stained by primary antibodies, specific secondary Alexa Fluor–conjugated antibodies, second primary antibodies, and then specific secondary Alexa Fluor–conjugated antibodies. Images of stained cells, washed in PBS with 0.05% saponin and left in PBS, were captured at RT. 3D data were captured with identical settings, which were also adjusted to avoid saturation of voxel (3D pixels) intensities. For colocalization analysis, the Coloc 2 plugin with thresholds in Fiji (ImageJ; National Institutes of Health) application was applied (Schindelin et al., 2012). The pHrodo fluorescence was used to spot or surface render the volume of individual phagosomes when E. coli pHrodo, S. aureus pHrodo red, or AF488-conjugated E. coli particles were used. A binary mask was created around bacterial particles (Process/Make Binary function) and used to define the regions for quantification of MIs for TRAM and SLAMF1 voxels in original images and to quantify E. coli pHrodo particles MI when redirected to the original image. E. coli pHrodo MI was evaluated to quantify acidification of E. coli–containing phagosomes in cells treated by control or SLAMF1 siRNA. For analysis of the sum of voxel intensities of SLAMF1 inside Golgi rings, GM130 staining was used to define the region to evaluate SLAMF1 intensities for individual 3D Golgi ring structure. Using ImageJ/Fiji software, 3D Golgi ring structures were selected as a region of interest (ROI) and used as a mask to obtain a numerical value of the relative amount of SLAMF1 as a sum of voxel intensities for SLAMF1 staining in Golgi ring ROIs from original image. ImarisXT software (Bitplane) was used to surface render the imaged GM130-positive structures, giving one surface for each. The values for voxel intensities did not follow a Gaussian distribution, and therefore we used median as a measure of average intensities and the nonparametric Mann-Whitney test to evaluate statistical significance in Prism (5.03; GraphPad Software). siRNA treatment Oligonucleotides used for silencing were AllStars negative control siRNA (SI03650318), FlexiTube siRNA Hs_SLAMF1_2 (SI00047250), Hs_MYD88_2 (SI00038297), Hs_TICAM2_2 (SI00130893), Hs_RAB11A_5 (SI00301553), and Hs_RAB11B_6 (SI02662695; QIAGEN). On day 7, cells were transfected by silencing oligonucleotides (20 nM final concentration) using Lipofectamine 3000 (L3000008) from Invitrogen/Thermo Fisher Scientific as suggested by the manufacturer. Cells were stimulated by LPS or E. coli particles 96 h after transfection. For THP-1 cells, cells were seeded in six-well plates (Nunc) 0.4 × 106 per well in antibiotic-free media supplemented by 40 ng/ml of PMA. Transfection of siRNA was performed for 24 h, media was changed to PMA-free for 72 h, and cells were kept for another 48 h before stimulation by LPS or E. coli particles. Quantitative PCR (qPCR) Total RNA was isolated form the cells using Qiazol reagent (79306; QIAGEN), and chloroform extraction followed by purification was performed on RNeasy Mini columns with DNase digestion step (QIAGEN). cDNA was prepared with the Maxima first strand cDNA synthesis kit for RT-qPCR (Thermo Fisher Scientific) according to the manufacturer’s protocol. qPCR was performed using the PerfeCTa qPCR FastMix (Quanta Biosciences) in replicates and was cycled in a StepOnePlus real-time PCR cycler. The following TaqMan gene expression assays (Applied Biosystems) were used: IFNβ (Hs01077958_s1), TNF (Hs00174128_m1), SLAMF1 (Hs00900288_m1), TBP (Hs00427620_m1), CXCL10 (Hs01124251_g1), Rab11a (Hs00366449_g1), and Rab11b (Hs00188448_m1) for human cells; and Ifnβ (Mm00439552_s1), Tnf (Mm00443258_m1), and Tbp (Mm01277042_m1) for mouse cells. No–reverse transcription controls were negative. The level of TBP mRNA was used for normalization, and results are presented as relative expression compared with the control untreated sample. Relative expression was calculated using the Pfaffl’s mathematical model (Pfaffl, 2001). Results are presented as means and SD of expression fold change for biological replicates relative to nonstimulated cells. Statistical significance was evaluated with Prism software. Data distribution was assumed to be normal, but this was not formally tested. The difference between the two groups was determined by the two-tailed t test. Cloning, expression vectors, and DNA transfection Phusion high-fidelity DNA polymerase and respective Fast Digest enzymes (Thermo Fisher Scientific) were used for cDNA recloning. Plasmids we purified by the Endofree plasmid maxi kit (QIAGEN). Sequencing of plasmids was done at the Eurofins genomics facility. Primers for cloning are listed below. SLAMF1 coding sequence was subcloned from retroviral vector (from A. Taranin, Siberian Branch of the Russian Academy of Sciences, Novosibirsk, Russia) to pcDNA3.1 (Invitrogen), C-terminal DYKDDDDK (Flag tag) vector (Takara Bio Inc.), deletion mutants of SLAMF1 made in pcDNA3.1 vector, or C-terminal DYKDDDDK vector. Human TRIFHA and TRAMYFP from K. Fitzgerald (University of Massachusetts Medical School, Worcester, MA) were used for transfections or as templates for subcloning and making TRAM deletion mutants. Rab11aFlag coding construct was described previously (Klein et al., 2015); Rab11FIP1, Rab11FIP2 ΔC2, Rab11FIP2, Rab11FIP3, Rab11FIP4, and Rab11FIP5 in pEGFPC1 vector were from M. McCaffrey (University College Cork, Cork, Ireland). Rab11aQ70L and Rab11aS25N were PCR amplified from pEGFP-Rab11Q70L and pEGFP-Rab11aS25N (Husebye et al., 2010), respectively. The amplified fragments were inserted into SalI- and BamHI-restricted pECFP-C1 vector. TLR4 was subcloned from a TLR4Cherry construct (Husebye et al., 2010) to a C-terminal DYKDDDDK vector. Mouse SLAMF1 was subcloned to a C-terminal DYKDDDDK vector, and mouse TRAM was subcloned from a GeneScript ORF clone (OMu22478D) to pEGFP-N1 vector (Takara Bio Inc.). pDUO-hMD-2/CD14 (Invivogen) was coexpressed with TLR4 to ensure TLR4 dimer formation. HEK 293T cells in six-well plates were transfected by 0.2–0.4 µg of vector/well using Genejuice transfection reagent (EMD Millipore). Lysates were prepared 48 h after transfection. Primers used for cloning are listed in Table 1. Table 1. Primers used for cloning of full-size or deletion mutants of SLAMF1, TRAM, TLR4, and Rab11 FIP2. Construct For/Rev Primer sequence (5′–3′) Restriction enzyme PCR product mapped on nucleotide sequence SLAMF1 ( NM_003037.3 ) pcDNA3.1 and pLVX-EF1α-IRES-ZsGreen1 For TTCGAATTCTGATGGGATCCCAAGGGGCTCC EcoRI 367–1,374 Rev GCAGCGGCCGCTCAGCTCTCTGGAAGTGTCA NotI C-terminal DYKDDDDK For TTAGAATTCATGGATCCCAAGGGGCTCC EcoRI 367–1,371 Rev GGTACCTCGAGAGCTCTCTGGAAGTGTCACAC XhoI pGEX-2TK For TATGGATCCCAGTTGAGAAGAAGAGGTAAAACG BamHI 1,141–1,374 Rev ATAGAATTCTCAGCTCTCTGGAAGTGTCAC EcoRI SLAMF1 ( NM_003037.3 ) deletion mutants to C-terminal DYKDDDDK All deletion mutants For TTAGAATTCATGGATCCCAAGGGGCTCC EcoRI N/a 1–265 aa Rev GGTACCTCGAGATTTACCTCTTCTTCTCAACTGTAG XhoI 367–1,161 1–326 aa Rev GTACCTCGAGAGACTGTGATGGAATTTGTTTCCTG XhoI 367–1,344 1–330 aa Rev GTACCTCGAGACACACTAGCATAGACTGTGATG XhoI 367–1,356 SLAMF1 ( NM_003037.3 ) deletion mutant to pcDNA3.1 1–265 aa For TTCGAATTCTGATGGGATCCCAAGGGGCTCC EcoRI 367–1,161 Rev TTAAGCGGCCGCTCATTTACCTCTTCTTCTCAACTG NotI SLAMF1 ( NM_003037.3 ) mutants with amino acid substitutions to C-terminal DYKDDDDK SLAMF1 Y281F For TTAGAATTCATGGATCCCAAGGGGCTCC EcoRI 367–1,225 Rev GTTTCTGGACTTGGGCAAAGATCGTAAGGC For GCCTTACGATCTTTGCCCAAGTCCAGAAAC 1,196–1,371 Rev GGTACCTCGAGATTTACCTCTTCTTCTCAACTGTAG XhoI SLAMF1 Y327F For TTAGAATTCATGGATCCCAAGGGGCTCC EcoRI 367–1,371 Rev GGTACCTCGAGAGCTCTCTGGAAGTGTCACACTAGCAAAGACTGTG XhoI SLAMF1 TNSI/PNPT For TTAGAATTCATGGATCCCAAGGGGCTCC EcoRI 367–1,341 Rev TGTGGTGGGGTTTGGTTCCTGGACAGACTCTGG For GAACCAAACCCCACCACAGTCTATGCTAGTGTGACACTTC 1,324–1,371a Revb CTTGTCATCGTCGTCCTTG TRAM/TICAM-2 ( NM_021649.7 ) C-terminal DYKDDDDK For CATGAATTCATGGGTATCGGGAAGTCTAAA EcoRI 443–1,147 Rev TTAACTCGAGCGGCAATAAATTGTCTTTGTACC XhoI TRAM deletion mutants to C-terminal DYKDDDDK 1–68 aa For CATGAATTCATGGGTATCGGGAAGTCTAAA EcoRI 443–646 Rev TTAACTCGAGCCATCTCTTCCACGCTCTGAGC XhoI 1–79 aa For CATGAATTCATGGGTATCGGGAAGTCTAAA EcoRI 443–679 Rev TTACCTCGAGAGAGGAACACCTCTTCTTCAGC XhoI 1–90 aa For CATGAATTCATGGGTATCGGGAAGTCTAAA EcoRI 443–712 Rev TTACCTCGAGATGTGTCATCTTCTGCATGCAATATC XhoI 1–100 aa For CATGAATTCATGGGTATCGGGAAGTCTAAA EcoRI 443–742 Rev TTACCTCGAGATAGCAGATTCTGGACTCTGAGG XhoI 1–120 aa For CATGAATTCATGGGTATCGGGAAGTCTAAA EcoRI 443–802 Rev TTACCTCGAGACTGTCTGCCACATGGCATCTC XhoI 68–235 aa For CATGAATTCATGTTTGAAGAAGAAGCTGAA EcoRI 644–1,147 Rev TTAACTCGAGCGGCAATAAATTGTCTTTGTACC XhoI 68–176 aa For CATGAATTCATGTTTGAAGAAGAAGCTGAA EcoRI 644–970 Rev TTAACTCGAGCCAGGGGCCGCATGGGTATAACAG XhoI 158–235 aa For CATGAATTCATGAACTCCGTTAACAGGCAGC EcoRI 914–1,147 Rev TTAACTCGAGCGGCAATAAATTGTCTTTGTACC XhoI TLR4 ( NM_138554.4 ) C-terminal DYKDDDDK For CGGTCGACCGAGATCTCATGATGTCTGCCTCGCGCCTGG N/a 299–2,815 Rev CTTGTAGTCGCCGGTACCGATAGATGTTGCTTCCTGCCAATTG N/a Mus Musculus TRAM/Ticam-2 ( NM_173394.3 ) to EGFP-N1 1–232 aa For ACTAAGCTTATGGGTGTTGGGAAGTCTAAAC HindIII 477–1,175 Rev ATATGGATCCCGGGCAATGAACTGTTTCTGCGAC BamHI M. Musculus Slamf1 ( NM_013730.4 ) from pDisplay vector to C-terminal DYKDDDDK 30–343 aa For TTGGAATTCGGCTTGGGGATATCCACCATGG EcoRI 183–1,127 Rev TATACCTCGAGAGCTCTCTGGCAGTGTCACACTG XhoI Rab11 FIP2 ( NM_014904.2 ) and deletion mutants to N-terminal DYKDDDDK All constructs For GCCCGAATTCGGCTGTCCGAGCAAGCCCAAAAG EcoRI 1–512 aa Rev ATAGCGGCCGCTCATTAACTGTTAGAGAATTTGCCAGC NotI 446–1,980 1–327 aa Rev ATAGCGGCCGCTCATTCGCTGCTTTCTTCAAATGG NotI 446–1,429 1–290 aa Rev ATAGCGGCCGCTTACACAATGCTGTCAGGTTGG NotI 446–1,310 1–254 aa Rev ATAGCGGCCGCTTATCCGAGAAGATGTGTTTGACC NotI 446–1,199 1–192 aa Rev ATAGCGGCCGCTTAGTGAGTACTTGGAATGATTGC NotI 446–1,016 Rab11a ( NM_004663.4 ) QL and SN mutants to pECFP-C1 1–216 aa For ATCAGTCGACATGGGCACCCGCGACGAC SalI 128–145 Rev TTAAGGATCCTTATATGTTCTGACAGCACTG BamHI 758–774 For, forward; QL, Rab11a Q70L mutant; Rev, reverse; SN, Rab11a S25N mutant. a Partially on vector sequence. b On C-terminal DYKDDDDK vector. IPs HEK293T cells expressing Flag-tagged proteins or macrophages for endogenous IPs were lysed using 1× lysis buffer (150 mM NaCl, 50 mM Tris-HCl, pH 8.0, 1 mM EDTA, and 1% NP-40) or 1× lysis buffer with high salt (400 mM NaCl, 50 mM Tris-HCl, pH 7.5, 1% Triton X-100, and 5 mM EDTA) for antiphosphotyrosine IPs supplemented with EDTA-free Complete mini protease inhibitor cocktail tablets and PhosSTOP phosphatase inhibitor cocktail (Roche), 50 mM NaF, and 2 mM Na3VO3 (Sigma-Aldrich). IPs were carried out by rotation at 4°C for 2 h of cell lysates with either anti-Flag (M2) agarose (Sigma-Aldrich) or specific antibodies coupled to Dynabeads (M-270 Epoxy; Thermo Fisher Scientific) or phosphotyrosine-biotinylated antibodies on streptavidin beads (Invitrogen). Agarose, Sepharose, or Dynabeads were washed five times by respective lysis buffers and heated for 5 min with 1× NuPAGE LDS sample buffer (Thermo Fisher Scientific) for agarose and Sepharose beads or eluted by elution buffer (from Dynabeads co-IP kit; 14321D; Thermo Fisher Scientific) for Dynabeads before analysis by Western blotting. Western blotting Cell lysates other than those used as controls in IPs were prepared by simultaneous extraction of proteins and total RNA using Qiazol reagent as suggested by the manufacturer. Protein pellets were dissolved by heating protein pellets for 10 min at 95°C in buffer containing 4 M urea, 1% SDS (Sigma-Aldrich), and NuPAGE LDS sample buffer (4×; Thermo Fisher Scientific). Otherwise, lysates were made using 1× RIPA lysis buffer (150 mM NaCl, 50 mM Tris-HCl, pH 7.5, 1% Triton X-100, 5 mM EDTA, protease inhibitors, and phosphatase inhibitors). For Western blot analysis, we used precast protein gels NuPAGE, Novex, iBlot transfer stacks, and the iBlot gel transfer device (Thermo Fisher Scientific). Lentiviral transduction For making the TRAM KO cell line, LentiCRISPRv2 plasmid (gift from F. Zhang; 52961; Addgene; Sanjana et al., 2014) was ligated with 5′-CACCGATGACTTTGGTATCAAACC-3′ and 5′-AAACGGTTTGATACCAAAGTCATC-3′ for TRAM. The second-generation packaging plasmids pMD2.G and psPAX2 were used for producing lentivirus (provided by D. Trono; 12260 and 12259; Addgene). HEK293T cells were cotransfected with the packaging and lentiCRISPRv2 plasmids and then washed after 16 h. The lentivirus-containing supernatants were collected after 48 h and used for transduction of THP-1 WT cells along with protamine sulphate (8 µg/ml final concentration). Transduced THP-1 cells were selected with puromycin (1 μg/ml) for 1 mo and then tested for TRAM protein expression by Western blotting. Lentivirus construct of SLAMF1 was prepared by cloning full-size SLAMF1 with or without Flag tag to the bicistronic lentiviral expression vector pLVX-EF1α-IRES-ZsGreen1 (Takara Bio Inc.; primers are listed in Table 1). The construct was sequenced and cotransfected with packaging plasmids (psPAX2 and pMD2.G provided by D. Trono; 12260 and 12259; Addgene) to produce pseudoviral particles in HEK293T cells. Supernatants were collected at 48 and 72 h, combined, and concentrated using Lenti-X concentrator (631231; Takara Bio Inc.). Viral particles were titrated in HEK293T cells. Titrated virus particles, which gave 90–100% of transduction efficiencies, were subsequently used for transduction of primary human macrophages (resulting in 30–40% of ZsGreen-positive cells). Macrophages were infected on day 6 of differentiation, media was changed after 24 h, and stimulation by LPS (100 ng/ml) was performed 72 h after transduction. Cell lysates for simultaneous RNA/protein isolation were prepared using Qiazol reagent. Flow cytometry Untreated or LPS-stimulated monocyte-derived macrophages were detached using accutase (A6964; Sigma-Aldrich) and stained with a cocktail of antibodies against human CD14 (MΦP9) FITC conjugated from BD and SLAMF1 IgG1 (IPO-3) mAbs labeled by AF647 using the Alexa Fluor 647 protein labeling kit (A20173) for 30 min on ice. Flow cytometry was performed using LSR II (BD) with FACS Diva software (BD). Samples were analyzed with FlowJo software (7.6; TreeStar). ELISA and multiplex cytokine assay TNF in supernatants was detected using a human TNF-α DuoSet ELISA (DY210-05; R&D Systems), and IFNβ level was detected using VeriKine-HS human IFNβ serum ELISA kit (41415; PBL Assay Science). Supernatants were also analyzed by multiplex cytokine assay (Bio-Plex; Bio-Rad Laboratories) for IL-1β, IL-6, IL-8, and CXCL-10/IP-10. TNF in BMDM supernatants was detected using mouse TNF-α DuoSet ELISA (DY410-05; R&D Systems), and IFNβ level was assessed using a VeriKine-HS mouse IFNβ serum ELISA kit (42410-1; PBL Assay Science). Results are presented as means and SD for biological replicates for representative donors (primary human macrophages) or at least three independent experiments for model cell line THP-1. Statistical significance was evaluated in Prism software. Data distribution was assumed to be normal, but this was not formally tested. The difference between the two groups was determined by a two-tailed t test. Blocking IFN receptors by specific antibodies Differentiated THP-1 cells in six-well plates were incubated for 30 min at 37oC with 2.5 μg/ml anti-IFNAR chain 2 antibodies, clone MMHAR-2 isotype IgG2a (MAB1155; EMD Millipore), or control mAbs LEAF-purified mouse IgG2a κ isotype MOPC-173 (400224; BioLegend). After preincubation with mAbs, cells were stimulated with LPS (100 ng/ml) and lysed using Qiazol reagent for simultaneous extraction of proteins and total RNA. GST pulldown assays The GST fusion protein construct of SLAMF1ct (GST-SLAMF1ct) was prepared by cloning SLAMF1ct (corresponding with 259–335 aa of SLAMF1 protein; UniProtKB, Q13291) to pGEX-2TK vector (GE Healthcare) with the help of V. Kashuba (Karolinska Institute, Stockholm, Sweden). The sequenced plasmid was transformed into the BL21 DE3 bacterial strain (New England Biolabs, Inc.) or to the TKX1 strain (Agilent Technologies) for production of tyrosine-phosphorylated GST-SLAMF1ct-antiphosphotyrosine. Expression and purification of GST fusion proteins were performed as described previously (Shlapatska et al., 2001). For protein purification and pulldown assays, we used Glutathione high-capacity magnetic agarose beads (G0924; Sigma-Aldrich). Cell lysates of untreated and LPS-stimulated macrophages were prepared in 1× lysis buffer (0.5% NP-40, 150 mM NaCl, and 50 mM Tris-HCl, pH 8.0). Pulldowns were performed as described previously (Shlapatska et al., 2001). Mouse BMDM differentiation and stimulation All protocols on animal work were approved by the Norwegian National Animal Research Authorities and were carried out in accordance with Norwegian and European regulations and guidelines. BMDM cultures were generated from bone marrow aspirates extracted from the femurs of C57BL/6 8–10-wk-old male control mice or from Slamf1−/− C57BL/6 mice (Wang et al., 2004). Cells were cultured in complete RPMI 1640 medium containing 20% of L929-conditioned media produced in-house for 8–10 d in sterile Petri dishes; cells were counted and seeded to 24-well cell culture plates in RPMI 1640 with 10% FCS at concentrations of 0.3 × 106 per well in triplicate, left overnight, and treated the next day in fresh media by 100 ng/ml ultrapure LPS (Invivogen) or 50 µg/well for six-well plates or 20 µg/well for 24-well plates of E. coli particles. Cell lysates for RNA isolation were made using Qiazol reagent. ROS activation assay Primary human macrophages (in six-well plates) were treated by siRNA as described in the siRNA treatment section, and treated by E. coli red pHrodo bacterial particles (excitation wavelength, 561 nm) in 0.5 ml RPMI 1640 containing 10% human serum on water bath (for 20 min incubation) or in a CO2 incubator (for 120 min incubation). Freshly dissolved in washing buffer (reagent A) or dihydrorhodamine 123 (DHR-123; excitation wavelength, 488 nm; reagent E), both from the PHAGOBURST kit (Glycotope Biotechnology), was added to the wells (except for control well, to which washing buffer was added) for the last 10 min of incubation. After stimulation, cells were placed on ice, washed by cold PBS, incubated with accutase (Sigma-Aldrich) on ice for 5 min, and scraped using cell scrapers. Cells were washed by flow wash (PBS with 0.5% FCS), fixed by fixation buffer (BD), washed by PBS, and analyzed by flow cytometry using LSR II with FACS Diva software. In FlowJo software, cells were gated for pHrodo-positive cells, and DHR-123 fluorescence was presented on the graphs for this gate. Bacterial killing assay THP-1 cells were plated at 2 × 105 cells/well in 24-well plates and differentiated as described in the Primary cells and cell lines section for 5 d. Cells were washed and transferred to serum-free RPMI medium. Live DH5α E. coli were added at an MOI of 40. E. coli were centrifuged onto differentiated THP-1 monolayers at 2,000 rpm for 5 min at 4°C. Plates were warmed to 37°C for 15 min in a water bath. Each well was then washed 3× with ice-cold PBS and incubated with warm 10% FCS RPMI medium containing 100 μg/ml of gentamycin for 30 min at 37°C to remove extracellular bacteria. If inhibitors were used in the assay, either DMSO or inhibitors were added to the media at designated concentrations. Cells were washed again 2× with PBS. This time point (45 min after adding bacteria) was designated as time 0. To measure colony-forming units (CFUs) at the end of incubation time, triplicate wells were washed and lysed in 1 ml sterile water. Plates for time points 1 h and 1.5 h were further incubated at 37°C in a CO2 incubator in medium with 10% FCS without antibiotics and with or without kinase inhibitors. At each time point, triplicate wells were washed 3× with PBS before lysing the cells. Viable counts were determined by plating 10 μl of 10-fold dilutions, 1:100 and 1:1,000, onto Luria-Bertani agar (in triplicates to account for technical pipetting error). CFUs were counted at each time point including time 0. Percent killing was calculated as 100 − (number of CFUs at time X/number of CFUs at time 0) × 100 for average values of technical replicates. Statistical significance was calculated in Prism software for biological replicates using unpaired two-tailed tests. Online supplemental material Fig. S1 shows that LPS treatment induces SLAMF1 expression in human cells, resulting in its surface localization, and that the increase in SLAMF1 expression is not dependent on signaling from IFNAR. Fig. S2 shows that SLAMF1 is involved in regulation of E. coli– or LPS-mediated but not TLR3-, TLR8-, or RIG-I/MDA5-mediated IFNβ or TNF mRNA expression. Fig. S3 shows that knockdown of SLAMF1 in THP-1 cells impairs TLR4-mediated phosphorylation of TBK1, IRF3, and TAK1 in response to LPS or E. coli particles. Fig. S4 shows that SLAMF1 relocalizes from ERCs to early and late E. coli phagosomes but not to S. aureus phagosomes and that SLAMF1 is required for E. coli phagosome acidification in human cells. Fig. S5 shows that SLAMF1 interaction with TRAM is independent from SLAMF1 tyrosine phosphorylation. Fig. S6 shows that TLR4-mediated IFNβ and TNF mRNA expression and corresponding cytokine secretion are not altered in Slamf1−/− BMDMs and provides human and murine SLAMF1 and TRAM proteins sequence alignments. Fig. S7 shows that E. coli–mediated Akt phosphorylation in human macrophages is not dependent on MyD88 expression and that TLR2- and TLR4-induced phosphorylation of Akt is weak and not much dependent on SLAMF1 or TRAM expression. Supplementary Material Supplemental Materials (PDF) Acknowledgments We thank S. Sidorenko, V. Kashuba, and M. McCaffrey for providing reagents and V. Boyartchuk for expert advice. Confocal imaging was performed at the Cellular and Molecular Imaging Core Facility of the Norwegian University of Science and Technology. This work was supported by the Research Council of Norway through its Centers of Excellence funding scheme grant 223255/F50 (to T. Espevik), by the Norwegian University of Science and Technology’s Onsager Fellowship (to R.K. Kandasamy), by grants from the Liaison Committee for Education, Research and Innovation in Central Norway (to T. Espevik), and by the Joint Research Committee between St. Olavs Hospital and Faculty of Medicine and Health Science of the Norwegian University of Science and Technology (to H. Husebye). The authors declare no competing financial interests. Author contributions: Conceptualization: M. Yurchenko, H. Husebye, and T. Espevik; Methodology: M. Yurchenko, H. Husebye, A. Skjesol, and T. Espevik; Investigation: M. Yurchenko, A. Skjesol, L. Ryan, H. Husebye, G.M. Richard, and N. Wang; Writing, original draft: M. Yurchenko; Writing, review and editing: M. Yurchenko, A. Skjesol, H. Husebye, C. Terhost, and T. Espevik; Supervision: H. Husebye and T. Espevik; Resources: R.K. Kandasamy, C. Terhorst, and N. Wang. ==== Refs Avota, E., E. Gulbins, and S. Schneider-Schaulies. 2011. DC-SIGN mediated sphingomyelinase-activation and ceramide generation is essential for enhancement of viral uptake in dendritic cells. PLoS Pathog. 7 :e1001290. 10.1371/journal.ppat.1001290 21379338 Berger, S.B., X. Romero, C. Ma, G. Wang, W.A. Faubion, G. Liao, E. Compeer, M. Keszei, L. Rameh, N. Wang, 2010. SLAM is a microbial sensor that regulates bacterial phagosome functions in macrophages. Nat. Immunol. 11 :920–927. 10.1038/ni.1931 20818396 Chang, L., and M. Karin. 2001. Mammalian MAP kinase signalling cascades. Nature. 410 :37–40. 10.1038/35065000 11242034 Chen, Q., D.W. Powell, M.J. Rane, S. Singh, W. Butt, J.B. Klein, and K.R. McLeish. 2003. Akt phosphorylates p47phox and mediates respiratory burst activity in human neutrophils. J. Immunol. 170 :5302–5308. 10.4049/jimmunol.170.10.5302 12734380 Clark, K., M. Peggie, L. Plater, R.J. Sorcek, E.R. Young, J.B. Madwed, J. Hough, E.G. McIver, and P. Cohen. 2011. Novel cross-talk within the IKK family controls innate immunity. Biochem. J. 434 :93–104. 10.1042/BJ20101701 21138416 Cocks, B.G., C.C. Chang, J.M. Carballido, H. Yssel, J.E. de Vries, and G. Aversa. 1995. A novel receptor involved in T-cell activation. Nature. 376 :260–263. 10.1038/376260a0 7617038 Everts, B., E. Amiel, S.C. Huang, A.M. Smith, C.H. Chang, W.Y. Lam, V. Redmann, T.C. Freitas, J. Blagih, G.J. van der Windt, 2014. TLR-driven early glycolytic reprogramming via the kinases TBK1-IKKɛ supports the anabolic demands of dendritic cell activation. Nat. Immunol. 15 :323–332. 10.1038/ni.2833 24562310 Farina, C., D. Theil, B. Semlinger, R. Hohlfeld, and E. Meinl. 2004. Distinct responses of monocytes to Toll-like receptor ligands and inflammatory cytokines. Int. Immunol. 16 :799–809. 10.1093/intimm/dxh083 15096475 Fitzgerald, K.A., S.M. McWhirter, K.L. Faia, D.C. Rowe, E. Latz, D.T. Golenbock, A.J. Coyle, S.M. Liao, and T. Maniatis. 2003 a. IKKepsilon and TBK1 are essential components of the IRF3 signaling pathway. Nat. Immunol. 4 :491–496. 10.1038/ni921 12692549 Fitzgerald, K.A., D.C. Rowe, B.J. Barnes, D.R. Caffrey, A. Visintin, E. Latz, B. Monks, P.M. Pitha, and D.T. Golenbock. 2003 b. LPS-TLR4 signaling to IRF-3/7 and NF-kappaB involves the toll adapters TRAM and TRIF. J. Exp. Med. 198 :1043–1055. 10.1084/jem.20031023 14517278 Fitzgerald, K.A., D.C. Rowe, and D.T. Golenbock. 2004. Endotoxin recognition and signal transduction by the TLR4/MD2-complex. Microbes Infect. 6 :1361–1367. 10.1016/j.micinf.2004.08.015 15596121 Ford, E., and D. Thanos. 2010. The transcriptional code of human IFN-beta gene expression. Biochim. Biophys. Acta. 1799 :328–336. 10.1016/j.bbagrm.2010.01.010 20116463 Gidon, A., S. Bardin, B. Cinquin, J. Boulanger, F. Waharte, L. Heliot, H. de la Salle, D. Hanau, C. Kervrann, B. Goud, and J. Salamero. 2012. A Rab11A/myosin Vb/Rab11-FIP2 complex frames two late recycling steps of langerin from the ERC to the plasma membrane. Traffic. 13 :815–833. 10.1111/j.1600-0854.2012.01354.x 22420646 Guichard, A., V. Nizet, and E. Bier. 2014. RAB11-mediated trafficking in host-pathogen interactions. Nat. Rev. Microbiol. 12 :624–634. 10.1038/nrmicro3325 25118884 Horgan, C.P., and M.W. McCaffrey. 2009. The dynamic Rab11-FIPs. Biochem. Soc. Trans. 37 :1032–1036. 10.1042/BST0371032 19754446 Hoyal, C.R., A. Gutierrez, B.M. Young, S.D. Catz, J.H. Lin, P.N. Tsichlis, and B.M. Babior. 2003. Modulation of p47PHOX activity by site-specific phosphorylation: Akt-dependent activation of the NADPH oxidase. Proc. Natl. Acad. Sci. USA. 100 :5130–5135. 10.1073/pnas.1031526100 12704229 Husebye, H., Ø. Halaas, H. Stenmark, G. Tunheim, Ø. Sandanger, B. Bogen, A. Brech, E. Latz, and T. Espevik. 2006. Endocytic pathways regulate Toll-like receptor 4 signaling and link innate and adaptive immunity. EMBO J. 25 :683–692. 10.1038/sj.emboj.7600991 16467847 Husebye, H., M.H. Aune, J. Stenvik, E. Samstad, F. Skjeldal, O. Halaas, N.J. Nilsen, H. Stenmark, E. Latz, E. Lien, 2010. The Rab11a GTPase controls Toll-like receptor 4-induced activation of interferon regulatory factor-3 on phagosomes. Immunity. 33 :583–596. 10.1016/j.immuni.2010.09.010 20933442 Junutula, J.R., E. Schonteich, G.M. Wilson, A.A. Peden, R.H. Scheller, and R. Prekeris. 2004. Molecular characterization of Rab11 interactions with members of the family of Rab11-interacting proteins. J. Biol. Chem. 279 :33430–33437. 10.1074/jbc.M404633200 15173169 Kagan, J.C., T. Su, T. Horng, A. Chow, S. Akira, and R. Medzhitov. 2008. TRAM couples endocytosis of Toll-like receptor 4 to the induction of interferon-beta. Nat. Immunol. 9 :361–368. 10.1038/ni1569 18297073 Kelly, B., and L.A. O’Neill. 2015. Metabolic reprogramming in macrophages and dendritic cells in innate immunity. Cell Res. 25 :771–784. 10.1038/cr.2015.68 26045163 Klein, D.C., A. Skjesol, E.D. Kers-Rebel, T. Sherstova, B. Sporsheim, K.W. Egeberg, B.T. Stokke, T. Espevik, and H. Husebye. 2015. CD14, TLR4 and TRAM Show Different Trafficking Dynamics During LPS Stimulation. Traffic. 16 :677–690. 10.1111/tra.12274 25707286 Krawczyk, C.M., T. Holowka, J. Sun, J. Blagih, E. Amiel, R.J. DeBerardinis, J.R. Cross, E. Jung, C.B. Thompson, R.G. Jones, and E.J. Pearce. 2010. Toll-like receptor-induced changes in glycolytic metabolism regulate dendritic cell activation. Blood. 115 :4742–4749. 10.1182/blood-2009-10-249540 20351312 Laird, M.H., S.H. Rhee, D.J. Perkins, A.E. Medvedev, W. Piao, M.J. Fenton, and S.N. Vogel. 2009. TLR4/MyD88/PI3K interactions regulate TLR4 signaling. J. Leukoc. Biol. 85 :966–977. 10.1189/jlb.1208763 19289601 Lindsay, A.J., and M.W. McCaffrey. 2004. The C2 domains of the class I Rab11 family of interacting proteins target recycling vesicles to the plasma membrane. J. Cell Sci. 117 :4365–4375. 10.1242/jcs.01280 15304524 Ma, C., N. Wang, C. Detre, G. Wang, M. O’Keeffe, and C. Terhorst. 2012. Receptor signaling lymphocyte-activation molecule family 1 (Slamf1) regulates membrane fusion and NADPH oxidase 2 (NOX2) activity by recruiting a Beclin-1/Vps34/ultraviolet radiation resistance-associated gene (UVRAG) complex. J. Biol. Chem. 287 :18359–18365. 10.1074/jbc.M112.367060 22493499 Makani, S.S., K.Y. Jen, and P.W. Finn. 2008. New costimulatory families: signaling lymphocytic activation molecule in adaptive allergic responses. Curr. Mol. Med. 8 :359–364. 10.2174/156652408785161005 18691062 Mancuso, G., A. Midiri, C. Biondo, C. Beninati, S. Zummo, R. Galbo, F. Tomasello, M. Gambuzza, G. Macrì, A. Ruggeri, 2007. Type I IFN signaling is crucial for host resistance against different species of pathogenic bacteria. J. Immunol. 178 :3126–3133. 10.4049/jimmunol.178.5.3126 17312160 Mikhalap, S.V., L.M. Shlapatska, A.G. Berdova, C.L. Law, E.A. Clark, and S.P. Sidorenko. 1999. CDw150 associates with src-homology 2-containing inositol phosphatase and modulates CD95-mediated apoptosis. J. Immunol. 162 :5719–5727.10229804 Mogensen, T.H. 2009. Pathogen recognition and inflammatory signaling in innate immune defenses. Clin. Microbiol. Rev. 22 :240–273. 10.1128/CMR.00046-08 19366914 Oshiumi, H., M. Sasai, K. Shida, T. Fujita, M. Matsumoto, and T. Seya. 2003. TIR-containing adapter molecule (TICAM)-2, a bridging adapter recruiting to toll-like receptor 4 TICAM-1 that induces interferon-beta. J. Biol. Chem. 278 :49751–49762. 10.1074/jbc.M305820200 14519765 Pfaffl, M.W. 2001. A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Res. 29 :e45. 10.1093/nar/29.9.e45 11328886 Poli-de-Figueiredo, L.F., A.G. Garrido, N. Nakagawa, and P. Sannomiya. 2008. Experimental models of sepsis and their clinical relevance. Shock. 30 (Suppl 1 ):53–59. 10.1097/SHK.0b013e318181a343 18704008 Radtke, A.L., L.M. Delbridge, S. Balachandran, G.N. Barber, and M.X. O’Riordan. 2007. TBK1 protects vacuolar integrity during intracellular bacterial infection. PLoS Pathog. 3 :e29. 10.1371/journal.ppat.0030029 17335348 Réthi, B., P. Gogolák, I. Szatmari, A. Veres, E. Erdôs, L. Nagy, E. Rajnavölgyi, C. Terhorst, and A. Lányi. 2006. SLAM/SLAM interactions inhibit CD40-induced production of inflammatory cytokines in monocyte-derived dendritic cells. Blood. 107 :2821–2829. 10.1182/blood-2005-06-2265 16317102 Romanets-Korbut, O., A.M. Najakshin, M. Yurchenko, T.A. Malysheva, L. Kovalevska, L.M. Shlapatska, Y.A. Zozulya, A.V. Taranin, B. Horvat, and S.P. Sidorenko. 2015. Expression of CD150 in tumors of the central nervous system: identification of a novel isoform. PLoS One. 10 :e0118302. 10.1371/journal.pone.0118302 25710480 Romero, X., D. Benítez, S. March, R. Vilella, M. Miralpeix, and P. Engel. 2004. Differential expression of SAP and EAT-2-binding leukocyte cell-surface molecules CD84, CD150 (SLAM), CD229 (Ly9) and CD244 (2B4). Tissue Antigens. 64 :132–144. 10.1111/j.1399-0039.2004.00247.x 15245368 Sakaguchi, S., H. Negishi, M. Asagiri, C. Nakajima, T. Mizutani, A. Takaoka, K. Honda, and T. Taniguchi. 2003. Essential role of IRF-3 in lipopolysaccharide-induced interferon-beta gene expression and endotoxin shock. Biochem. Biophys. Res. Commun. 306 :860–866. 10.1016/S0006-291X(03)01049-0 12821121 Sanjana, N.E., O. Shalem, and F. Zhang. 2014. Improved vectors and genome-wide libraries for CRISPR screening. Nat. Methods. 11 :783–784. 10.1038/nmeth.3047 25075903 Schindelin, J., I. Arganda-Carreras, E. Frise, V. Kaynig, M. Longair, T. Pietzsch, S. Preibisch, C. Rueden, S. Saalfeld, B. Schmid, 2012. Fiji: an open-source platform for biological-image analysis. Nat. Methods. 9 :676–682. 10.1038/nmeth.2019 22743772 Schroder, K., K.M. Irvine, M.S. Taylor, N.J. Bokil, K.A. Le Cao, K.A. Masterman, L.I. Labzin, C.A. Semple, R. Kapetanovic, L. Fairbairn, 2012. Conservation and divergence in Toll-like receptor 4-regulated gene expression in primary human versus mouse macrophages. Proc. Natl. Acad. Sci. USA. 109 :E944–E953. 10.1073/pnas.1110156109 22451944 Shlapatska, L.M., S.V. Mikhalap, A.G. Berdova, O.M. Zelensky, T.J. Yun, K.E. Nichols, E.A. Clark, and S.P. Sidorenko. 2001. CD150 association with either the SH2-containing inositol phosphatase or the SH2-containing protein tyrosine phosphatase is regulated by the adaptor protein SH2D1A. J. Immunol. 166 :5480–5487. 10.4049/jimmunol.166.9.5480 11313386 Sidorenko, S.P., and E.A. Clark. 1993. Characterization of a cell surface glycoprotein IPO-3, expressed on activated human B and T lymphocytes. J. Immunol. 151 :4614–4624.8409422 Theil, D., C. Farina, and E. Meinl. 2005. Differential expression of CD150 (SLAM) on monocytes and macrophages in chronic inflammatory contexts: abundant in Crohn’s disease, but not in multiple sclerosis. J. Clin. Pathol. 58 :110–111. 10.1136/jcp.2004.019323 15623499 Thurston, T.L., K.B. Boyle, M. Allen, B.J. Ravenhill, M. Karpiyevich, S. Bloor, A. Kaul, J. Noad, A. Foeglein, S.A. Matthews, 2016. Recruitment of TBK1 to cytosol-invading Salmonella induces WIPI2-dependent antibacterial autophagy. EMBO J. 35 :1779–1792. 10.15252/embj.201694491 27370208 Toshchakov, V., B.W. Jones, P.Y. Perera, K. Thomas, M.J. Cody, S. Zhang, B.R. Williams, J. Major, T.A. Hamilton, M.J. Fenton, and S.N. Vogel. 2002. TLR4, but not TLR2, mediates IFN-beta-induced STAT1alpha/beta-dependent gene expression in macrophages. Nat. Immunol. 3 :392–398. 10.1038/ni774 11896392 Trinchieri, G. 2010. Type I interferon: friend or foe? J. Exp. Med. 207 :2053–2063. 10.1084/jem.20101664 20837696 Troutman, T.D., W. Hu, S. Fulenchek, T. Yamazaki, T. Kurosaki, J.F. Bazan, and C. Pasare. 2012. Role for B-cell adapter for PI3K (BCAP) as a signaling adapter linking Toll-like receptors (TLRs) to serine/threonine kinases PI3K/Akt. Proc. Natl. Acad. Sci. USA. 109 :273–278. 10.1073/pnas.1118579109 22187460 van Driel, B., G. Liao, X. Romero, M.S. O’Keeffe, G. Wang, W.A. Faubion, S.B. Berger, E.M. Magelky, M. Manocha, V. Azcutia, 2012. Signaling lymphocyte activation molecule regulates development of colitis in mice. Gastroenterol. 143 :1544–1554. van Driel, B.J., G. Liao, P. Engel, and C. Terhorst. 2016. Responses to Microbial Challenges by SLAMF Receptors. Front. Immunol. 7 :4. 10.3389/fimmu.2016.00004 26834746 Vaure, C., and Y. Liu. 2014. A comparative review of toll-like receptor 4 expression and functionality in different animal species. Front. Immunol. 5 :316. 10.3389/fimmu.2014.00316 25071777 Wang, N., A. Satoskar, W. Faubion, D. Howie, S. Okamoto, S. Feske, C. Gullo, K. Clarke, M.R. Sosa, A.H. Sharpe, and C. Terhorst. 2004. The cell surface receptor SLAM controls T cell and macrophage functions. J. Exp. Med. 199 :1255–1264. 10.1084/jem.20031835 15123745 West, A.P., I.E. Brodsky, C. Rahner, D.K. Woo, H. Erdjument-Bromage, P. Tempst, M.C. Walsh, Y. Choi, G.S. Shadel, and S. Ghosh. 2011. TLR signalling augments macrophage bactericidal activity through mitochondrial ROS. Nature. 472 :476–480. 10.1038/nature09973 21525932 Yamamoto, M., S. Sato, H. Hemmi, S. Uematsu, K. Hoshino, T. Kaisho, O. Takeuchi, K. Takeda, and S. Akira. 2003. TRAM is specifically involved in the Toll-like receptor 4-mediated MyD88-independent signaling pathway. Nat. Immunol. 4 :1144–1150. 10.1038/ni986 14556004 Yamashiro, D.J., B. Tycko, S.R. Fluss, and F.R. Maxfield. 1984. Segregation of transferrin to a mildly acidic (pH 6.5) para-Golgi compartment in the recycling pathway. Cell. 37 :789–800. 10.1016/0092-8674(84)90414-8 6204769

PMC005xxxxxx/PMC5881500.txt

==== Front J Cell Biol J Cell Biol jcb jcb The Journal of Cell Biology 0021-9525 1540-8140 Rockefeller University Press 29382700 201708044 10.1083/jcb.201708044 Research Articles Article 47 19 34 Tom70 enhances mitochondrial preprotein import efficiency by binding to internal targeting sequences iMTS-L sequences facilitate mitochondrial import http://orcid.org/0000-0002-0032-7690 Backes Sandra 1* Hess Steffen 1* Boos Felix 1 Woellhaf Michael W. 1 http://orcid.org/0000-0002-5043-1357 Gödel Sabrina 2 http://orcid.org/0000-0002-1482-7020 Jung Martin 3 http://orcid.org/0000-0003-3925-6778 Mühlhaus Timo 2 http://orcid.org/0000-0003-2081-4506 Herrmann Johannes M. 1 1 Cell Biology, University of Kaiserslautern, Kaiserslautern, Germany 2 Computational Systems Biology, University of Kaiserslautern, Kaiserslautern, Germany 3 Medical Biochemistry, Saarland University, Homburg, Germany Correspondence to Johannes M. Herrmann: hannes.herrmann@biologie.uni-kl.de; Timo Mühlhaus: muehlhaus@bio.uni-kl.de * S. Backes and S. Hess contributed equally to this paper. 02 4 2018 217 4 13691382 07 8 2017 12 12 2017 17 1 2018 © 2018 Backes et al. 2018 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). N-terminal matrix-targeting signals (MTSs) are critical for mitochondrial protein import. Backes et al. identified additional internal MTS-like sequences scattered along the sequences of mitochondrial proteins. By binding to Tom70 on the mitochondrial surface, these sequences support the import process. The biogenesis of mitochondria depends on the import of hundreds of preproteins. N-terminal matrix-targeting signals (MTSs) direct preproteins to the surface receptors Tom20, Tom22, and Tom70. In this study, we show that many preproteins contain additional internal MTS-like signals (iMTS-Ls) in their mature region that share the characteristic properties of presequences. These features allow the in silico prediction of iMTS-Ls. Using Atp1 as model substrate, we show that iMTS-Ls mediate the binding to Tom70 and have the potential to target the protein to mitochondria if they are presented at its N terminus. The import of preproteins with high iMTS-L content is significantly impaired in the absence of Tom70, whereas preproteins with low iMTS-L scores are less dependent on Tom70. We propose a stepping stone model according to which the Tom70-mediated interaction with internal binding sites improves the import competence of preproteins and increases the efficiency of their translocation into the mitochondrial matrix. Graphical Abstract Deutsche Forschungsgemeinschaft https://doi.org/10.13039/501100001659 He2803/9-1 IRTG 1830 ==== Body pmcIntroduction The intracellular sorting of newly synthesized proteins relies on targeting information encoded in their amino acid sequences. Proteins that are translocated as unfolded polypeptides typically use N-terminal targeting sequences that can be decoded during their synthesis on the ribosome. Well-studied examples for such N-terminal targeting sequences are the signal sequences of proteins of the ER (Blobel and Dobberstein, 1975; Schibich et al., 2016), leader peptides of proteins of the bacterial periplasm (Wickner et al., 1978), transit peptides of chloroplast proteins (Lubben et al., 1988), and presequences or matrix-targeting sequences (MTSs), which direct proteins to mitochondria (Hartl et al., 1986; Hurt et al., 1986; von Heijne, 1986a). Receptors on the surface of the target compartment recognize these signals and pass them on to protein-conducting channels through which the precursor proteins are threaded. Finally, processing peptidases remove the targeting sequences, and the mature proteins are folded with assistance of chaperones. Proteins that are transported in a folded conformation (such as in the case of nuclear proteins) often use more complex internal signals, which are displayed on the 3D protein surface (De Robertis et al., 1978; Lee et al., 2006). These signals are part of the mature protein structure and are normally not removed by proteases. Mitochondria are comprised of 800–1,500 different proteins (Mootha et al., 2003; Sickmann et al., 2003; Rhee et al., 2013; Morgenstern et al., 2017). About two thirds of these proteins are synthesized as precursors with N-terminal presequences that are both necessary and sufficient for their import (Wiedemann and Pfanner, 2017). These signals form amphipathic helices with one positively charged and one hydrophobic surface (von Heijne, 1986a). They are of variable length, typically between 8 and 70 amino acids, cleaved by the matrix processing peptidase (MPP), and degraded by the presequence peptidase PreP (Vögtle et al., 2009; Alikhani et al., 2011; Mossmann et al., 2014). Presequences are recognized by Tom20 and Tom22, receptors of the translocase of the outer membrane of mitochondria (TOM) complex, and are directed into the translocation pore formed by the β-barrel protein Tom40 (Rimmer et al., 2011; Shiota et al., 2011). The inner membrane translocase or TIM23 complex together with the import motor completes protein translocation into the matrix (Malhotra et al., 2013; Banerjee et al., 2015; Ramesh et al., 2016; Backes and Herrmann, 2017; Schendzielorz et al., 2017). The inner membrane harbors a second, independent translocase, the TIM22 complex, that inserts hydrophobic carrier proteins into the inner membrane (Rehling et al., 2003; Hasson et al., 2010; Wrobel et al., 2013). TIM22 substrates lack N-terminal presequences but carry internal targeting signals that are recognized on the surface of the mitochondria by a dedicated TOM receptor called Tom70 (Sirrenberg et al., 1996). Tom70, such as its paralog Tom71, has a tetratricopeptide structure and can, together with cytosolic Hsp70 and Hsp90 chaperones, recruit and stabilize its substrate proteins on the mitochondrial surface (Young et al., 2003; Fan et al., 2011; Hoseini et al., 2016; Zanphorlin et al., 2016; Xue et al., 2017). Tom70 and Tom20/Tom22 partially overlap in their substrate spectrum so that Tom70 is not essential as long as Tom20/Tom22 receptors are present (Ramage et al., 1993). Some recent studies suggest that the mature parts of mitochondrial precursor proteins play a critical role in the efficiency of the translocation reaction that cannot simply be explained by the absence or presence of tightly folded translocation-resisting regions (Yamamoto et al., 2009; Schendzielorz et al., 2017). In this study, we show that sequences with MTS-like features are not confined to the N termini of mitochondrial proteins but are also frequently present in the mature parts of a subset of precursor proteins. These internal MTS-like signals (iMTS-Ls) show affinity for the Tom70 receptor and increase the efficiency of protein translocation. Our study points to a novel as yet unknown category of mitochondrial import signals that mimic the sequence properties of classical mitochondrial targeting sequences and which are present in the majority of matrix proteins. Results The matrix protein Atp25 has an internal presequence-like segment that improves its import efficiency Atp25 is a composite matrix protein that employs three MPP-processing sites to generate two mature matrix-localized fragments referred to as the Rsf and M domains (Woellhaf et al., 2016). Deletion of the internal region in Atp25 that contains the two MPP cleavage sites led to the generation of one matured fusion protein (Fig. 1 A, m). To our knowledge, this was a very exceptional organization of a precursor protein because only very few tandem proteins were identified thus far: established examples are Arg6-Arg7 of Saccharomyces cerevisiae (Boonchird et al., 1991) and Rsm22-Cox11, Cox15-Yah1, YKR070w-Crd1, and Aco2-Mrpl49 of Schizosaccharomyces pombe (Khalimonchuk et al., 2006). To identify further tandem proteins, we screened the mitochondrial proteome of yeast for proteins that contain internal sequences that adhere to the MPP consensus R(AFLR)(FLY)↓(AKLS)(HQST) (Vögtle et al., 2009). We subcloned the coding sequences of these proteins (Nam2, Nca2, Hmi1, Kgd1, Mis1, Mrp7, Aim19, and Cbp6) as well as those of Atp25 (a tandem protein) and Oxa1 (a nontandem protein), generated radiolabeled proteins by in vitro transcription/translation reactions, and incubated them with isolated mitochondria (Fig. 1 B). Most of these precursors were imported into mitochondria, but for none of these cases did we observe internal cleavage by MPP. Thus, the tandem organization of Atp25 and Arg6-Arg7 are presumably rare exceptions. Obviously, nontandem organization of mitochondrial precursor proteins is strongly favored unless, such as in the case of Atp25, there is a good reason (Woellhaf et al., 2016) for such a composite structure. Figure 1. Atp25 is a rare example for a tandem precursor protein in yeast. (A) Atp25 and Atp25ΔMPP2ΔMPP3 (lacking amino acids 279–293; Woellhaf et al., 2016) were synthesized in the presence of [35S]methionine and incubated with isolated mitochondria for the times indicated. Nonimported material was removed by degradation with proteinase K (PK) before samples were analyzed by SDS-PAGE and autoradiography. Precursor (pre) and the M and Rsf domains as well as the N-terminally cleaved mature (m) species are indicated. 10% of the radiolabeled precursor protein used per time point was loaded for control. (B) The indicated proteins were radiolabeled and incubated with isolated mitochondria in the presence or absence of membrane potential (Δψ). The samples were split, and one fraction was treated with proteinase K. A 20% total (T) of the precursor protein used per import reaction was loaded for control. Yellow arrowheads depict precursor proteins, red arrowheads indicate N-terminally matured proteins, and green arrowheads indicate internally matured proteins. The presequences of Aim19 and Cbp6 were not cleaved by MPP (Vögtle et al., 2009); the imported species of these proteins are indicated by blue arrowheads. No protease-protected forms of Nca2 and Hmi1 were observed, indicating that these proteins were not imported in our in vitro import system. To our surprise, we realized that the internal presequence-like structure in Atp25 was not only important for its internal processing but also for its overall import into mitochondria. As obvious from Fig. 1 A, Atp25 was efficiently imported, resulting in two strong signals of the Rsf and the M domain, respectively. In contrast, the matured species of Atp25ΔMPP2ΔMPP3 that acquired a protease-inaccessible location was very faint, indicating that this protein was barely imported. Low levels were not caused by rapid degradation as the imported Atp25ΔMPP2ΔMPP3 protein was stable under the import conditions used (Fig. S1 A). Obviously, the internal region in Atp25 that contains the two internal MPP cleavage sites contributes to the import competence of the protein. We showed in a previous study that this internal region has the ability to serve as an N-terminal MTS (Woellhaf et al., 2016). Thus, the presence of the internal mitochondrial targeting-like sequence in Atp25 considerably improved the import efficiency of this protein, at least in the in vitro assay used in this study. iMTS-Ls can be predicted by TargetP Several established algorithms exist to predict the presence of N-terminal MTSs on precursor proteins. Initially, algorithms were developed that searched for specific features of targeting signals such as helicity, positive charge, amphipathy, or the presence or absence of specific amino acid residues (McGeoch, 1985; von Heijne, 1986b). More recently, neural networks such as the TargetP algorithm were trained on sets of proteins of known cellular localization to predict the presence or absence of MTSs with high confidence (Nakai and Kanehisa, 1992; Emanuelsson et al., 2000, 2007; Habib et al., 2007). To identify potential internal targeting information in proteins, we used the TargetP predictive score consecutively for each residue in a given protein (Fig. 2 A), leading to a profile reflecting MTS-like properties. This approach was able to recognize the iMTS-Ls in Atp25 and showed that this stretch of high iMTS-L profile score was absent in Atp25ΔMPP2ΔMPP3 (Fig. 2 B). Figure 2. The TargetP algorithm can be used to predict iMTS-Ls throughout protein sequences. (A) The TargetP algorithm is designed to calculate prediction scores for N-terminal sequences. To calculate scores for internal regions, we consecutively N-terminally truncated the sequences and calculated the corresponding TargetP scores for each position (orange dots), leading to a characteristic profile that shows internal regions in proteins with presequence-like properties. Profiles are smoothed using a Savitzky-Golay filter (blue line). (B) Raw TargetP profiles in Atp25 predict with accuracy the iMTS of the protein, which contains the two internal MPP cleavage sites (Woellhaf et al., 2016). In the Atp25ΔMPP2ΔMPP3 variant, the internal presequence-like region was deleted, and hence, the internal region with the very high TargetP scores is missing. (C) Smoothed TargetP profiles of mitochondrial preproteins without (Hsp60) and with (Atp1, Pim1, Oxa1, and Hmi1) iMTS-Ls. The TargetP profile of the cytosolic protein actin (Act1) is shown for comparison. We wondered whether similar iMTS-Ls are also present in other matrix proteins. In some preproteins such as Hsp60, the N-terminal MTS is the only segment of the protein with a TargetP score >0.3 (Fig. 2 C). However, most matrix proteins contained iMTS-Ls in addition to their N-terminal MTS (Table S1). As examples, profiles for the mitochondrial proteins Atp1, Pim1, Oxa1, and Hmi1 are shown. Hmi1 is an unconventional precursor that carries its targeting signal at its C terminus (Lee et al., 1999). The profile of the cytosolic protein actin (Fig. 2 C, Act1) is shown for comparison. It should be noted that internal sequences of high iMTS-L score were also found in several cytosolic proteins and thus, these sequences, unlike N-terminal MTSs, were no reliable indication of mitochondrial localization. Many matrix proteins contain iMTS-Ls in their mature part Next, we analyzed the structural features of the iMTS-Ls in more detail. We found that the iMTS-Ls are spread over the length of mature regions of mitochondrial proteins (Fig. S1 B and Table S1). We defined the width of the iMTS-Ls as the number of amino acids under a peak of scores higher than a random TargetP score. It is a priori not obvious from the nature of the TargetP algorithm that this setting captured the correct beginning and end of the relevant signals: Commonly, a high TargetP score is assigned to the first amino acid of an N-terminal sequence patch indicating the presence of an MTS. However, our current approach can predict the length of MTS as well as iMTS-Ls under the TargetP scoring peak. The width showed some variation, but most ranged between 20 and 30 residues, and thus, the lengths of iMTS-Ls resembled that of N-terminal MTSs analyzed by using the same parameters (Fig. 3 A) and the experimentally verified length determined by the published mitochondrial N proteome (Vögtle et al., 2009). To analyze the specific characteristics of iMTS-Ls, we compared different features of these sequences with those of the total yeast proteome (Fig. 3 B). We found that iMTS-Ls had low frequencies of aspartate and glutamate but high frequencies of arginine, lysine, and hydroxylated amino acids. They are predominantly helical, of amphipathic nature, and show low hydrophobicity scores. Thus, they clearly mimic N-terminal presequences, with the only obvious difference being that they are not N-terminal. Figure 3. Many precursor proteins contain iMTS-Ls. (A) Kernel density estimation of the length distribution shows the agreement of experimentally verified MTS length (blue; Vögtle et al., 2009), TargetP-predicted MTS (orange), and the length of iMTS-Ls predicted by our approach (dotted gray line). (B) Sequence property differences between iMTS-Ls and the total yeast proteome are shown in percent change. (C) The iMTS-L propensity profile (blue) for two selected proteins, Atp1 and Hsp60, is plotted against the smoothed TargetP profile (gray). The experimentally verified end of the MTS (Vögtle et al., 2009) is marked (orange dot) and matches the x axis intersection of the corresponding iMTS-L propensity profile. (D) The iMTS-L propensity calculated for experimentally verified mitochondrial matrix proteins (blue; Vögtle et al., 2009) shows a clear difference from random iMTS-L propensity calculated based on a sequence biased by yeast proteome amino acid frequency. However, the iMTS-L propensity of the total yeast proteome is similar to that of mature regions of matrix proteins. (E and F) Atp1 and Hsp60/GroEL sequences of S. cerevisiae (NP_009453 and NP_013360), Kluyveromyces lactis (XP_454248 and XP_455510), Candida glabrata (XP_449761 and XP_448482), S. pombe (CAB11207 and NP_592894), Drosophila (NP_726243 and NP_511115), Homo sapiens (AAH11384 and P10809), Arabidopsis thaliana (P92549 and AEE76842), Chlamydomonas reinhardtii (EDP07337 and XP_001691353), E. coli (WP_021537164 and ABF67773), and Rickettsia prowazekii (WP_004599658 and AMS12509) were aligned. iMTS-L propensity scores of mitochondrial and bacterial homologues were calculated and are shown as blue and orange traces, respectively. Next, we generated 10,000 random protein sequences reflecting the mitochondrial proteome in length and amino acid frequency and calculated their iMTS-L scores. The comparison of the iMTS-L scores/profiles of mitochondrial proteins with these values allowed us to calculate a measure for each protein combining length and scores of the iMTS-Ls in their sequences, which we called iMTS-L propensity (Fig. 3 C). These propensities allowed it to compare different proteins by a single parameter. A list with mitochondrial proteins of particularly high or low iMTS-L propensities is shown in Table 1. For instance, Atp1 (the F1α subunit of the FoF1 ATPase) had an iMTS-L propensity of 1.57, whereas Hsp60 had 0.18 (Table S2). There was no obvious correlation between the iMTS-L propensities of yeast proteins with their lengths (Fig. S1 C). Most proteins contain considerably higher iMTS-L scores than randomly generated sequences of the same length and amino acid content. iMTS-L sequences were not only enriched in mitochondrial proteins but also in the total yeast proteome (Fig. 3 D), suggesting that iMTS-Ls are a structural sequence feature of more general distribution. This is also supported by the observation that the distribution and positions of iMTS-Ls were conserved not only among mitochondrial homologues of Atp1 and Hsp60/GroEL but even in their bacterial counterparts (Fig. 3, E and F, orange lines). Table 1. Proteins with the highest or lowest iMTS-L propensity among all 135 analyzed mitochondrial proteins Standard name iMTS-L propensity Mss18 2.48 Qcr2 2.35 Idh2 2.33 Mba1 2.32 Ndi1 1.86 Arg5,6 1.73 Mrs2 1.69 Pkp1 1.67 Mss51 1.65 Cox15 1.57 Mgm101 1.57 Atp1 1.57 Ism1 1.54 Yme2 1.50 Psd1 1.47 Rmd9 1.44 Coq2 1.43 Dss1 1.41 Msy1 1.39 Kgd2 1.39 Fmp16 0.00 Sdh4 0.00 Atp5 0.00 Cox4 0.00 Pdx1 0.00 Atp12 0.00 Atp14 0.00 Cox8 0.00 Cpr3 0.00 Cox5a 0.00 Grx5 0.00 Isu1 0.00 Atp15 0.00 Stf1 0.00 Trx3 0.00 Sod2 0.00 Nfu1 0.00 Atp16 0.00 Inh1 0.08 Ppa2 0.15 iMTS-Ls have the potential to function as mitochondrial targeting sequences The similarity of the properties of iMTS-Ls to canonical mitochondrial presequences inspired us to test whether iMTS-Ls can target proteins to mitochondria if placed at the N terminus of a protein. To this end, we used Atp1 as a model protein. This protein contains several iMTS-Ls, two very prominent ones starting with residues 308 and 393 (Fig. 4 A). We generated constructs for the cytosolic expression of N-terminally truncated versions of Atp1, which either started with an iMTS-L sequence (Atp1Δ307 and Atp1Δ392) or with a sequence of low TargetP score (Atp1Δ50 and Atp1Δ330) fused to GFP. As shown in Fig. 4 B, Atp1-GFP, Atp1Δ307-GFP, and Atp1Δ392-GFP colocalized with mitochondria, whereas Atp1Δ50-GFP remained in the cytosol. For Atp1Δ330-GFP, no expression was detected. The iMTS-Ls were obviously sufficient for mitochondrial targeting because we observed that fusion proteins consisting of 50-residue segments of Atp1 and GFP confirmed the localization that was found with the N-terminally truncated Atp1 versions (Fig. 4 C). Thus, the iMTS-Ls in Atp1 can efficiently target GFP to mitochondria in vivo. Figure 4. iMTS-Ls have mitochondrial targeting potential if present at the N terminus. (A) TargetP profile of Atp1. Overview and TargetP scores of N-terminally truncated Atp1 variants. (B and C) Fusion proteins of N-terminally truncated Atp1 variants and GFP were expressed in yeast cells. Representative fluorescence microscopy images of yeast cells show the intracellular distribution of GFP (green) and mitochondria stained with rhodamine B hexylester (red). Please note that Atp1-GFP, Atp1Δ307-GFP, and Atp1Δ392-GFP as well as Atp11–50-GFP, Atp1306–358-GFP, and Atp1391–443-GFP colocalize with mitochondria, whereas Atp1Δ50-GFP and Atp11–50-GFP are present in the cytosol. Atp1Δ330-GFP and Atp1339–388-GFP failed to be expressed. (D) N-terminally truncated variants of Atp1 were synthesized in reticulocyte lysate in the presence of [35S]methionine and incubated for different times with isolated WT mitochondria. For the sample labeled with –ψ, the mitochondrial membrane potential was dissipated by the addition of valinomycin. Nonimported protein was removed by treatment with proteinase K (PK) before the samples were visualized by SDS-PAGE and autoradiography. 10% of the radiolabeled protein used per import lane was shown for control. Atp1 was efficiently imported into mitochondria. Only very minor amounts of Atp1Δ307-GFP and Atp1Δ392-GFP were taken up, and Atp1Δ50-GFP was not imported. Atp1Δ330-GFP failed to be synthesized. Next, we tested whether these sequences can serve as MTSs in vitro. We synthesized Atp1 and its N-terminally truncated variants in the presence of [35S]methionine and incubated these variants with yeast mitochondria. Atp1 was efficiently imported into the mitochondria, where its presequence was cleaved off by MPP, resulting in a mature form (Fig. 4 D, m). The N-terminally truncated variants of Atp1 were either not imported (Atp1Δ50 and Atp1Δ330) or imported only with very low efficiency (Atp1Δ307 and Atp1Δ392). Even after an incubation of 30 min, most of these proteins remained accessible to added protease. This shows that iMTS-Ls, when placed to the N terminus, can efficiently mediate the targeting of proteins to mitochondria. However, they apparently are not able to drive the complete import of proteins into mitochondria, suggesting that they differ in certain critical properties. For example, iMTS-Ls might be unable to open the protein-conducting channel of the TIM23 complex or to activate the import motor, processes which still are poorly understood (Truscott et al., 2001; Okamoto et al., 2002; Chacinska et al., 2005; Meinecke et al., 2006; Longen et al., 2014; Ramesh et al., 2016; Schendzielorz et al., 2017; Ting et al., 2017). iMTS-Ls can serve as binding regions for TOM receptors A previous study by Yamamoto et al. (2009) reported that the outer membrane receptor Tom70 interacts with the mature parts of some mitochondrial precursor proteins. In particular, they observed that the import of Atp1 into Δtom70 mitochondria occurred only with reduced efficiency and that the cytosolic receptor domain of Tom70 helps to prevent the aggregation of the Atp1 precursor in vitro. We therefore wondered whether the iMTS-Ls in Atp1 might represent binding sites for Tom70. To test a potential binding of the Atp1 protein to Tom70, we incubated radiolabeled Atp1 and Atp1Δ50 with either purified recombinant GST, Tom70-GST, or Tom20-GST coupled with glutathione (GSH) beads (Hoseini et al., 2016). Atp1Δ50 lacks the N-terminal 50 residues of Atp1, including the 35 residues of the presequence (Vögtle et al., 2009). Tom20 showed a strong preference for the presequence-containing Atp1 precursor (Fig. 5 A), consistent with the well-documented function of Tom20 in presequence recognition (Brix et al., 1999; Yamamoto et al., 2011; Shiota et al., 2015). Interestingly, Tom70 efficiently bound Atp1 as well as Atp1Δ50, suggesting that the presequence of Atp1 is not essential for Tom70 binding. Neither Atp1 nor Atp1Δ50 were recovered with GST control beads. Figure 5. Tom70 binds to specific regions in the mature part of Atp1. (A) The cytosolic receptor domains of Tom70 and Tom20 were purified as GST fusion proteins and bound to GSH beads. Radiolabeled Atp1 or Atp1Δ50 was incubated with the indicated beads for 10 min at room temperature. Beads were then pelleted by centrifugation, the supernatant fractions were discarded, and the pellet fractions were analyzed by SDS-PAGE and autoradiography. The results were quantified using ImageJ. (B) 20-mer peptides shifted by three amino acids along the entire 545 residues of the Atp1 sequence were covalently bound to a membrane and incubated with purified Tom70-GST as used in A. After extensive washing, the bound Tom70 was detected by Western blotting using a specific GST antibody. Circles indicate the location of each single peptide on the membrane. White circles represent the peptides of the N-terminal presequence of Atp1 (peptides 1–7) and the iMTS-L sequences presented in Fig. 4 A (peptides 95–101 and 128–132). (C) Binding of Tom70-GST was quantified, and the relative intensity of each peptide spot was plotted (blue trace). For comparison, the profile of Atp1 TargetP scores is shown (purple trace). To identify the regions in Atp1 that are recognized by Tom70, we used a peptide scan approach that had been successfully used in the past to determine Tom70 binding sites in mitochondrial carrier proteins (Brix et al., 1999). Peptides of 20-amino-acid residues each were covalently bound to a membrane. In total, 176 peptides were synthesized, each shifted by three amino acids along the entire 545 residues of the Atp1 sequence. The membrane was incubated with purified Tom70-GST. After extensive washing, the bound Tom70 was detected by Western blotting (Fig. 5 B). We observed that several regions in Atp1 were efficiently bound by Tom70, which very well reflected the patterns of the MTS/iMTS-L sequences (Fig. 5 C). This suggests that the iMTS-Ls in Atp1, and presumably also in other mitochondrial precursor proteins, represent Tom70 binding regions in these proteins. The presence or absence of iMTS-L sequences in Atp1 and Hsp60 correlates with their Tom70 dependence Next, we tested the relevance of iMTS-Ls for protein import reactions in vitro. We produced radiolabeled Atp1 (as a protein with iMTS-Ls) and Hsp60 (as a protein without such sequences) as well as a mutated version of Atp1 in which its three iMTS-Ls were mutated (Atp1mut; Fig. 6, A and B). At least in vitro, this Atp1mut version hardly bound to Tom70, whereas Atp1 did (Fig. 6 C), supporting the idea that the iMTS-Ls support the association of precursor proteins with the Tom70 receptor. Figure 6. The presence of MTS-like sequences in the mature part of mitochondrial proteins influences the import behavior. (A and B) TargetP probabilities of Hsp60, Atp1, and an Atp1mut version of Atp1 in which the iMTS-Ls were mutated by replacing positively charged residues by negative ones. (C) Radiolabeled Atp1 or Atp1mut precursor was incubated with empty GSH beads or beads coupled with GST or GST-Tom70. Beads were pelleted by centrifugation, extensively washed, and analyzed by SDS-PAGE and autoradiography. Lanes labeled “10%” show 10% of the radiolabeled precursor proteins used per reaction. The levels of proteins bound to Tom70 relative to those bound to empty beads were quantified from three independent experiments. Shown are means and SD. (D) The indicated model proteins were radiolabeled in reticulocyte lysate and incubated for the indicated times with WT and Δtom70/Δtom71 mitochondria at 25°C. Mitochondria were incubated with 100 µg/ml proteinase K for 30 min on ice to remove nonimported material and analyzed by SDS-PAGE and autoradiography. Lanes labeled “10%” show 10% of the radiolabeled precursor proteins used per time point. (E) The experiments shown in C were repeated three times and quantified. Intensities were normalized to the 10% control of the corresponding precursor protein. Means and SEM are shown. To assess the Tom70 dependence of these proteins, Hsp60, Atp1, and Atp1mut were incubated with isolated WT and Δtom70/Δtom71 mitochondria for different times before nonimported protein was degraded (Fig. 6, D and E). Hsp60 was efficiently imported into WT and Δtom70/Δtom71 mitochondria. In contrast, efficient import of Atp1 required the presence of Tom70/Tom71, consistent with an earlier study that showed that Tom70 can prevent the aggregation of Atp1 precursor (Yamamoto et al., 2009). Interestingly, when the iMTS-Ls in Atp1 were mutated, the import competence of the precursor protein was almost completely abolished (Fig. 6 E; compare 10% of the total to the protease-inaccessible protein after 15 min of import). From this, we conclude that Atp1 requires the outer membrane receptor Tom70/Tom71 as well as its iMTS-Ls to be efficiently imported into mitochondria. iMTS-Ls maintain nonimported Atp1 precursor in an import-competent conformation The reduced efficiency by which Atp1 is imported into Δtom70/Δtom71 mitochondria suggests a direct role of Tom70 that is mediated via interaction with its iMTS-Ls. We used a pulse-chase assay that had been developed to study the relevance of Tom70 binding to precursors bound to the mitochondrial surface (Hines and Schatz, 1993). To this end, the mitochondrial membrane potential was dissipated by carbonyl cyanide chlorophenyl hydrazine (CCCP) before radiolabeled Hsp60 or Atp1 were bound to WT or Δtom70/Δtom71 mitochondria. Mitochondria were reisolated and reenergized by treatment with DTT, NADH, and ATP (Fig. 7, A–C). Atp1 was efficiently chased into mitochondria only in the presence of Tom70/Tom71 (Fig. 7 B, arrowheads), which obviously were crucial to maintain Atp1 import competence on the mitochondrial surface. In contrast, Hsp60 remained largely import competent also in Δtom70/Δtom71 mitochondria unless it was incubated with nonenergized mitochondria for prolonged time (Fig. 7 D). Figure 7. Tom70 is essential to maintain the Atp1 precursor in an import-competent conformation. (A) Model of the pulse-chase assay used in the following experiments. The membrane potential of isolated mitochondria was dissipated by addition of CCCP. Radiolabeled precursor proteins were added. After incubation of the mitochondria for 10–30 min, CCCP was quenched by DTT, and the mitochondria were reenergized. (B–E) WT or Δtom70/71 mitochondria were preincubated with CCCP for 10–30 min in the presence of radiolabeled Hsp60, Atp1, and Atp1mut (pulse). Mitochondria were reisolated and reenergized by treatment with DTT. After incubation (chase) for 10 min at 25°C, nonimported protein was removed by protease treatment, and samples were analyzed by SDS-PAGE and autoradiography. C shows means and SEM of three replicates. Arrowheads indicate the Atp1 protein that only was chased into mitochondria if Tom70/Tom71 were present. (F and G) A peptide from the rat pALDH presequence that was shown to work as a competitive inhibitor of the Tom70 receptor (Melin et al., 2015) was added to isolated mitochondria to the pulse-chase reaction with Atp1mut, Atp1, and Hsp60 precursors. The presence of 5 µM pALDH peptide prevented the import of Atp1 (red arrowheads). The Atp1 version in which the iMTS-Ls were mutated was not imported in the pulse-chase experiment. Obviously, this variant lost its minor import competence during the preincubation completely, irrespective of the presence of Tom70/Tom71 (Fig. 7, C and E). To exclude that the pronounced defect of the pulse-chase import into Δtom70/Δtom71 mitochondria is caused by pleiotropic problems of this mutant, we generated a peptide aldehyde dehydrogenase (pALDH) from which it was previously shown that it binds to the substrate-binding groove of Tom70 (Melin et al., 2015) and which hence competes for Tom70 binding with precursor proteins. Addition of excess amounts of this peptide to WT mitochondria during preincubation with precursor proteins compromised the import of Atp1 but not that of Hsp60 (Fig. 7, F and G). This nicely confirmed that Tom70 binding is critical to maintaining Atp1 in an import-competent state, for which iMTS-Ls are of critical relevance. Tom70 supports the import of proteins with high iMTS-L propensities Tom70 and its paralog Tom71 are not essential proteins, and in most genetic backgrounds, Δtom70/Δtom71 double mutants grow efficiently even under nonfermentative conditions (Fig. 8 A). To analyze the relevance of Tom70 in more detail, we purified WT and Δtom70/Δtom71 mitochondria, which contained similar levels of Tom20 as well as of the TIM23 subunits Tim44, Tim23, and Tim17 (Fig. 8 B). These mitochondria were incubated with radiolabeled precursor forms of Atp1, Atp2, Atp25, cytochrome b2, cytochrome c1, Kgd1, Mrp7, and Oxa1 for 2, 5, and 15 min, and the amounts of fully imported proteins were visualized after proteolytic digestion of nonimported proteins (Fig. 8 C). The signals of three replicates were quantified (see Fig. S2 for examples), and the amounts of maximally imported proteins were calculated. These amounts differed considerably between the different proteins, and there was no significant correlation with their respective iMTS-L propensities (Fig. 8 D). However, when we compared the ratios of the maximal imported protein of WT to that of Δtom70/Δtom71 mitochondria (Fig. 8 E), there was a convincing correlation between Tom70 dependence and the iMTS-L propensity. The larger the iMTS-L propensities of the precursor proteins tested, the more their import was improved by the presence of Tom70/Tom71. This is consistent with a role of Tom70 in promoting the import of mature parts of mitochondrial precursor proteins, presumably via its interaction with iMTS-L sequences (Fig. 8 F). Figure 8. The presence of Tom70 supports the import of precursor proteins with high iMTS-L propensity scores. (A) WT and Δtom70/Δtom71 were grown to log phase in galactose medium before tenfold serial dilutions were spotted onto plates containing glucose or glycerol medium. Plates were incubated at the indicated temperatures for 2 d. Mutants lacking Tom70 and its paralog Tom71 show growth defects at 37°C. (B) Mitochondria were isolated from WT and Δtom70/Δtom71 cells. 20, 40, and 80 µg of mitochondrial proteins were resolved by SDS-PAGE and analyzed by Western blotting. (C) The indicated proteins were radiolabeled in reticulocyte lysate and incubated for the indicated times with WT and Δtom70/Δtom71 mitochondria at 25°C. Mitochondria were reisolated, incubated with 100 µg/ml proteinase K for 30 min on ice to remove nonimported material, and analyzed by SDS-PAGE and autoradiography. Lanes labeled “20%” show 20% of the radiolabeled precursor proteins used per time point. (D) The import experiments shown in C were repeated three times. The signals were quantified and fitted with the Michaelis-Menten equation y = (x + imax)/(x + i50%) from which maximal imported amounts imax were deduced (Fig. S2). Absolute amounts of imported proteins and iMTS-L propensities show no significant relatedness (y = 5.051x − 5.399; adjusted R2 = 0.23; signal β = 0.125). (E) In contrast, regression analysis revealed a significant relationship between the iMTS-L propensity of a protein and its Tom70 relevance (y = 3.821x − 4.048; adjusted R2 = 0.66; signal β = 0.008**), which we define as the protein maximally imported into WT mitochondria divided by that of Δtom70/Δtom71 mitochondria. Error bars denote SD. (F) Hypothetical model for the role of iMTS-Ls as stepping stones for Tom70 interaction during protein import. Binding of Tom70 to the iMTS-Ls prevents premature folding or aggregation of the precursors and maintains their import competence. Discussion Protein targeting relies on the specific recognition of precursor proteins by receptors of the destination membrane. Proteins of the mitochondrial matrix are characterized by N-terminal MTSs, which bind to the Tom20/Tom22 receptor on the mitochondrial surface. These N-terminal signals are both necessary and sufficient for mitochondrial protein targeting. In this study, we describe additional as yet unidentified iMTS-L sequences that are scattered over the mature part of mitochondrial precursor proteins. Previous studies already reported the presence of latent targeting signals in cytosolic proteins (Hurt and Schatz, 1987) as well as in Escherichia coli DNA-derived sequences (Baker and Schatz, 1987). In this study, we show that iMTS-Ls mimic the structure of presequences in respect to their length, amphipathicity, positive charge distribution, and high content of hydroxylated amino acid residues. Despite their similarity to MTSs, the iMTS-Ls of Atp1 failed to target proteins to the matrix in in vitro import experiments even if they were presented N-terminally, and thus apparently are not necessarily fully functional presequences. Nevertheless, these sequences were able to target GFP fusion proteins to mitochondria in vivo, indicating that they interact with proteins on the mitochondrial surface. We identified Tom70 as an efficient and specific iMTS-L binding partner that, potentially in cooperation with the other TOM receptors, interacts with these regions during preprotein import into mitochondria (Fig. 8 F). The interaction of Tom70 with the iMTS-Ls is obviously not essential for protein translocation into mitochondria because mutants lacking Tom70 still import matrix proteins (Ramage et al., 1993; Gärtner et al., 1995; Yamamoto et al., 2009) and because a small number of matrix proteins lack iMTS-Ls completely (Tables 1 and S2). It appears conceivable that the binding of iMTS-Ls can delay the passage of precursors across the outer membrane and hence might be counterproductive for the import of proteins that are not prone to adopt import-incompetent conformations. Our understanding of the interactions of mitochondrial surface receptors with their substrates is still incomplete. However, substrate release from TOM receptors was reported to be triggered by an interaction of the cytosolic domains of Tom70 and Tom20 (Fan et al., 2011). It was shown that Tom70 improves the import of only certain precursor proteins (Yamamoto et al., 2009; Horvath et al., 2012). The import of some matrix proteins such as Atp1 into Tom70-deficient mitochondria was reported to be strongly reduced, consistent with what we observed in this study. Interestingly, the mature part but not the presequences was found to determine the Tom70 dependence of these preproteins, which were proposed to aggregate if Tom70 is absent (Yamamoto et al., 2009). Hsp90 and Hsp70 chaperones, which directly interact with Tom70, support Tom70 in this process (Young et al., 2003; Bhangoo et al., 2007; Li et al., 2009; Fan et al., 2011; Hoseini et al., 2016). Tom70 consists primarily of tetratricopeptide repeat motifs (Chan et al., 2006; Wu and Sha, 2006), a structure characteristic for many interactors of Hsp70 and Hsp90 proteins. Moreover, Δtom70 deletion mutants show strong synthetic negative growth defects with mutants lacking cytosolic chaperones such as the cytosolic J protein Djp1 (Papić et al., 2013). Tom70 might constitute a platform on the mitochondrial surface that interacts with specific precursor proteins and with different cytosolic chaperones. It appears likely that the iMTS-Ls identified in this study serve as Tom70 binding sites distributed like stepping stones in the sequence of precursor proteins, which keep precursors unfolded and import competent during their translocation into mitochondria (Fig. 8 F). According to this hypothesis, the predominant function of iMTS-Ls is not that of mitochondrial targeting signals but rather that of unfolding signatures in precursor proteins, which facilitate the import of potentially aggregation-prone or multidomain precursors. Our analysis suggests that iMTS-Ls are not a characteristic feature of mitochondrial proteins but rather are distributed generally in proteins. It appears likely that Tom70 had evolved to exploit this pervasive feature in proteins to prevent premature folding before import into mitochondria. Cytosolic chaperones and folding factors such as the recently described ubiquilins (Itakura et al., 2016) might support Tom70 in this function. Because presequences are both necessary and sufficient for mitochondrial targeting, the relevance of the mature part of proteins was not carefully analyzed thus far. It is well established that tightly folded protein domains such as those of the methotrexate-bound dihydrofolate reductase domain or titin can prevent or slow down protein translocation into mitochondria (Wienhues et al., 1991; Gaume et al., 1998; Sato et al., 2005; Yagawa et al., 2010). Moreover, it was shown that the recognition of mature stretches of precursors by the mitochondrial chaperone machinery influences their import efficiency (Okamoto et al., 2002; Schendzielorz et al., 2017). The identification of iMTS-Ls in this study is the first approach to systematically identify specific biogenesis signals in the mature parts of mitochondrial proteins. Interestingly, internal targeting sequences were recently also identified in the mature parts of secretory proteins, where they might serve a similar purpose (Chatzi et al., 2017). It will be exciting to study their specific relevance for the mitochondrial protein import in more detail in the future. Materials and methods Yeast strains and plasmids All yeast strains used in this study were based on the WT strain W303-1A (Sherman, 1963). The Δtom70/Δtom71 strain, which was a gift from D. Rapaport (University of Tübingen, Tübingen, Germany), was described previously (Kondo-Okamoto et al., 2008). All strains were grown on YP (1% yeast extract and 2% peptone) medium containing 2% galactose (Altmann et al., 2007). DNA sequences corresponding with regions coding for Atp1 or fragments thereof were amplified by PCR and cloned into the EcoRI and BamHI sites of a pYX122 vector. For confocal microscopy, 3 ml yeast culture (OD600 0.8) was incubated for 5 min with 30 µl of 10 µM rhodamine B hexylester dissolved in DMSO. The Atp25-coding region or fragment was amplified by PCR and cloned into pGEM4 (Promega) using the EcoRI and HindIII restriction sites. For construction of the Atp25ΔMPP2 ΔMPP3 version, the sequence encoding for amino acid residues 279–293 were deleted and replaced by the restriction sites of XbaI and SalI (Woellhaf et al., 2016). Import of radiolabeled proteins into isolated mitochondria Import reactions were essentially performed as described previously (Weckbecker et al., 2012) in the following import buffer: 500 mM sorbitol, 50 mM Hepes, pH 7.4, 80 mM KCl, 10 mM magnesium acetate, and 2 mM KH2PO4. Mitochondria were energized by addition of 2 mM ATP and 2 mM NADH before radiolabeled precursor proteins were added. To dissipate the membrane potential, a mixture of 1 µg/ml valinomycin, 8.8 µg/ml antimycin, and 17 µg/ml oligomycin was added to the mitochondria. Precursor proteins were incubated with mitochondria for different times at 25°C before nonimported protein was degraded by addition of 100 µg/ml proteinase K. CCCP chase experiment Isolated mitochondria were preincubated in import buffer containing 50 µM CCCP for 5 min at 25°C. Radiolabeled precursor proteins were added. After incubation for 10–30 min at 25°C, mitochondria were reisolated by centrifugation for 10 min at 30,000 g. Pellets were resuspended in import buffer containing 6% fatty acid–free BSA and 5 mM DTT. 2 mM ATP and 2 mM NADH were added to energize the mitochondria. After incubation for 10 min at 25°C, mitochondria were treated with proteinase K to remove nonimported material. Protein purification of Tom70-GST, Tom20-GST, and GST E. coli cells harboring the pGEX4T1 GST-Tom20 cytosolic domain, pGEX4T1 GST-Tom70 cytosolic domain, or pGEX4T1-GST plasmid (Hoseini et al., 2016) were grown to an OD600 of 0.8 at 37°C, and 0.5 mM IPTG was added. After incubation for 16 h, 50 ml of the culture was harvested. The cells were resuspended in lysis buffer (50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 0.1% NP-40, and 1 mM β-mercaptoethanol) and incubated with 1 mg/ml lysozyme for 30 min at room temperature. After one step of freeze thawing and 15 cycles of sonification for 1 s at 60% duty level, the lysate was cleared at 25,000 g for 5 min at 4°C. The supernatant was incubated with GSH Sepharose 4B beads (GE Healthcare) and washed 3× with PBS. Bound GST fusion proteins were eluted by incubation with elution buffer (50 mM Tris-HCl, pH 8.0, 25 mM GSH, and 1 mM DTT) for 30 min at 4°C. Purification efficiency was analyzed by SDS-PAGE and Coomassie brilliant blue staining. Binding assay N-terminally GST-tagged cytosolic domains of Tom20 or Tom70 were expressed in E. coli cells and purified with GSH beads. 15 µl of the GSH Sepharose with bound GST-tagged proteins or GST alone were washed two times with 200 µl import buffer. 200 µl import buffer containing 0.5 µl radiolabeled Atp1, Atp1Δ50, or Atp1mut was subsequently added to the GSH Sepharose and incubated 10 min on an end-over-end rotator at room temperature. The Sepharose was pelleted by centrifugation, washed with import buffer, and boiled in 25 µl sample buffer containing 50 mM DTT. Dot blot assay Peptides of Atp1 with a length of 20 amino acids each were synthesized on a cellulose membrane as described previously (Weckbecker et al., 2012). The amino acid frame was shifted by three amino acids from one spot to the next. The peptide spots covered the whole amino acid sequence of Atp1. The membrane was incubated with methanol for 2 min at room temperature, subsequently washed 2 min with H2O, and equilibrated in binding buffer (20 mM Tris-HCl, pH 7.0, and 200 mM NaCl) for 20 min. After this, the membrane was incubated for 3 h with 0.5 µM of recombinantly purified GST-Tom70 dissolved in binding buffer. The membrane was washed twice for 10 min with binding buffer and twice with TBS (10 mM Tris-HCl, pH 7.4, and 150 mM NaCl) and then was subjected to immunoblotting against GST. iMTS-L profile generation Suffix sequences of the given protein were subjected to TargetP prediction (TargetP 1.1 standalone software package; DTU Bioinformatics) in standard FastA format. The resulting mitochondrial targeting peptide probability of the suffix sequence was used as positional information and concatenated to generate the raw iMTS-L scoring sequence. A Savitzky–Golay filtering step with the successive subsets of adjacent data points was applied. A window size of the expected value of the length distribution of known MTSs and a quadratic polynomial was used to smooth the raw profile into the iMTS-L probability profile PiiMTS−L, where i is the position in the amino acid sequence. iMTS-L propensity calculation The PiiMTS−L score was normalized to allow the comparison between amino acid sequences of different lengths. The normalization uses expected value µiMTS-L and SD σiMTS-L were calculated over random sequences (with n = 5,000) using the amino acid frequencies of the whole yeast proteome as follows:P'iiMTS−L=PiiMTS−L−µiMTS−L σiMTS−L. From the P'iiMTS−Lscore, it was possible to define an overall iMTS-L propensity by summing over all the amino acids of a sequence that had an iMTS-L score higher than those of random sequences (i.e., those positions i where P'iiMTS−L≥0), excluding the MTS at sequence start, by using the equation ∑i>MTSmaxP'i iMTS−L,0. Analysis and sequence property calculation The iMTS-Ls of a given sequence were detected by searching for regions of positive P'iiMTS−L that were flanked by zero or negative regions. Afterward, the sequence properties were independently deduced from two sets of amino acid sequences containing all identified iMTSs and all yeast proteins. A composition vector for each set was generated and multiplied by the helicity index (Koehl and Levitt, 1999), amphiphilicity index (Cornette et al., 1987), coil index (Ptitsyn and Finkelstein, 1983), β sheet propensity (Crawford et al., 1973), and hydrophibicity index (Fasman, 1989). Analyses and calculations were performed using Microsoft F# functional programming language with the bioinformatics library BioFSharp (available on GitHub at https://github.com/CSBiology/BioFSharp) and the graphical library FSharp.Plotly (available on GitHub at https://github.com/muehlhaus/FSharp.Plotly). Online supplemental material The supplemental material provides additional information on the distribution of iMTS-L scores in proteins (Fig. S1) and the import kinetics corresponding with Fig. 8 (Fig. S2) as well as lists of iMTS-L propensities for mitochondrial proteins in S. cerevisiae (Tables S1 and S2). Supplementary Material Supplemental Materials (PDF) Tables S1 and S2 (Excel) Acknowledgments We thank Doron Rapaport for the Δtom70/Δtom71 deletion strain and for E. coli strains to express Tom70-GST, Tom20-GST, and GST. We are grateful to Sabine Knaus for technical assistance and to Bruce Morgan for discussion. The work was supported by grants from the Deutsche Forschungsgemeinschaft (He2803/9-1 and IRTG 1830 to J.M. Herrmann) and the Landesschwerpunkt BioComp. The authors declare no competing financial interests. Author contributions: S. Hess generated mutants, recombinant proteins, cell fractions, and antibodies. F. Boos analyzed the data and designed experiments. S. Backes purified mitochondria, synthesized radiolabeled proteins, and performed in vitro import experiments. M.W. Woellhaf, S. Gödel, and T. Mühlhaus generated the iMTS-L prediction algorithm and analyzed the targeting signals in Atp25. S. Gödel characterized the biochemical properties of iMTS-Ls and assessed the relation between iMTS-L propensity and import efficiency. M. Jung generated the peptide scans for Atp1. T. Mülhaus and J.M. Herrmann analyzed the data, designed the experiments, and wrote the draft of the manuscript to which all other authors contributed. ==== Refs Alikhani, N., A.K. Berglund, T. Engmann, E. Spånning, F.N. Vögtle, P. Pavlov, C. Meisinger, T. Langer, and E. Glaser. 2011. Targeting capacity and conservation of PreP homologues localization in mitochondria of different species. J. Mol. Biol. 410 :400–410. 10.1016/j.jmb.2011.05.009 21621546 Altmann, K., M. Dürr, and B. Westermann. 2007. Saccharomyces cerevisiae as a model organism to study mitochondrial biology. In Mitochondria. Practical Protocols. Vol. 372 . D. Leister, and J.M. Herrmann, editors. Humana Press, Totowa, New Jersey. 81–90. 10.1007/978-1-59745-365-3_6 Backes, S., and J.M. Herrmann. 2017. Protein Translocation into the Intermembrane Space and Matrix of Mitochondria: Mechanisms and Driving Forces. Front. Mol. Biosci. 4 :83. 10.3389/fmolb.2017.00083 29270408 Baker, A., and G. Schatz. 1987. Sequences from a prokaryotic genome or the mouse dihydrofolate reductase gene can restore the import of a truncated precursor protein into yeast mitochondria. Proc. Natl. Acad. Sci. USA. 84 :3117–3121. 10.1073/pnas.84.10.3117 3033634 Banerjee, R., C. Gladkova, K. Mapa, G. Witte, and D. Mokranjac. 2015. Protein translocation channel of mitochondrial inner membrane and matrix-exposed import motor communicate via two-domain coupling protein. eLife. 4 :e11897. 10.7554/eLife.11897 26714107 Bhangoo, M.K., S. Tzankov, A.C. Fan, K. Dejgaard, D.Y. Thomas, and J.C. Young. 2007. Multiple 40-kDa heat-shock protein chaperones function in Tom70-dependent mitochondrial import. Mol. Biol. Cell. 18 :3414–3428. 10.1091/mbc.E07-01-0088 17596514 Blobel, G., and B. Dobberstein. 1975. Transfer of proteins across membranes. I. Presence of proteolytically processed and unprocessed nascent immunoglobulin light chains on membrane-bound ribosomes of murine myeloma. J. Cell Biol. 67 :835–851. 10.1083/jcb.67.3.835 811671 Boonchird, C., F. Messenguy, and E. Dubois. 1991. Determination of amino acid sequences involved in the processing of the ARG5/ARG6 precursor in Saccharomyces cerevisiae. Eur. J. Biochem. 199 :325–335. 10.1111/j.1432-1033.1991.tb16128.x 1649049 Brix, J., S. Rüdiger, B. Bukau, J. Schneider-Mergener, and N. Pfanner. 1999. Distribution of binding sequences for the mitochondrial import receptors Tom20, Tom22, and Tom70 in a presequence-carrying preprotein and a non-cleavable preprotein. J. Biol. Chem. 274 :16522–16530. 10.1074/jbc.274.23.16522 10347216 Chacinska, A., M. Lind, A.E. Frazier, J. Dudek, C. Meisinger, A. Geissler, A. Sickmann, H.E. Meyer, K.N. Truscott, B. Guiard, 2005. Mitochondrial presequence translocase: switching between TOM tethering and motor recruitment involves Tim21 and Tim17. Cell. 120 :817–829. 10.1016/j.cell.2005.01.011 15797382 Chan, N.C., V.A. Likić, R.F. Waller, T.D. Mulhern, and T. Lithgow. 2006. The C-terminal TPR domain of Tom70 defines a family of mitochondrial protein import receptors found only in animals and fungi. J. Mol. Biol. 358 :1010–1022. 10.1016/j.jmb.2006.02.062 16566938 Chatzi, K.E., M.F. Sardis, A. Tsirigotaki, M. Koukaki, N. Šoštarić, A. Konijnenberg, F. Sobott, C.G. Kalodimos, S. Karamanou, and A. Economou. 2017. Preprotein mature domains contain translocase targeting signals that are essential for secretion. J. Cell Biol. 216 :1357–1369. 10.1083/jcb.201609022 28404644 Cornette, J.L., K.B. Cease, H. Margalit, J.L. Spouge, J.A. Berzofsky, and C. DeLisi. 1987. Hydrophobicity scales and computational techniques for detecting amphipathic structures in proteins. J. Mol. Biol. 195 :659–685. 10.1016/0022-2836(87)90189-6 3656427 Crawford, J.L., W.N. Lipscomb, and C.G. Schellman. 1973. The reverse turn as a polypeptide conformation in globular proteins. Proc. Natl. Acad. Sci. USA. 70 :538–542. 10.1073/pnas.70.2.538 4510294 De Robertis, E.M., R.F. Longthorne, and J.B. Gurdon. 1978. Intracellular migration of nuclear proteins in Xenopus oocytes. Nature. 272 :254–256. 10.1038/272254a0 564468 Emanuelsson, O., H. Nielsen, S. Brunak, and G. von Heijne. 2000. Predicting subcellular localization of proteins based on their N-terminal amino acid sequence. J. Mol. Biol. 300 :1005–1016. 10.1006/jmbi.2000.3903 10891285 Emanuelsson, O., S. Brunak, G. von Heijne, and H. Nielsen. 2007. Locating proteins in the cell using TargetP, SignalP and related tools. Nat. Protoc. 2 :953–971. 10.1038/nprot.2007.131 17446895 Fan, A.C., G. Kozlov, A. Hoegl, R.C. Marcellus, M.J. Wong, K. Gehring, and J.C. Young. 2011. Interaction between the human mitochondrial import receptors Tom20 and Tom70 in vitro suggests a chaperone displacement mechanism. J. Biol. Chem. 286 :32208–32219. 10.1074/jbc.M111.280446 21771790 Fasman, G.D., editor. 1989. Prediction of Protein Structure and the Principles of Protein Conformation. Springer US, Boston, MA. 10.1007/978-1-4613-1571-1 Gärtner, F., W. Voos, A. Querol, B.R. Miller, E.A. Craig, M.G. Cumsky, and N. Pfanner. 1995. Mitochondrial import of subunit Va of cytochrome c oxidase characterized with yeast mutants: Independence from receptors, but requirement for matrix hsp70 translocase function. J. Biol. Chem. 270 :3788–3795. 10.1074/jbc.270.8.3788 7876120 Gaume, B., C. Klaus, C. Ungermann, B. Guiard, W. Neupert, and M. Brunner. 1998. Unfolding of preproteins upon import into mitochondria. EMBO J. 17 :6497–6507. 10.1093/emboj/17.22.6497 9822595 Habib, S.J., W. Neupert, and D. Rapaport. 2007. Analysis and prediction of mitochondrial targeting signals. Methods Cell Biol. 80 :761–781. 10.1016/S0091-679X(06)80035-X 17445721 Hartl, F.U., B. Schmidt, E. Wachter, H. Weiss, and W. Neupert. 1986. Transport into mitochondria and intramitochondrial sorting of the Fe/S protein of ubiquinol-cytochrome c reductase. Cell. 47 :939–951. 10.1016/0092-8674(86)90809-3 3022944 Hasson, S.A., R. Damoiseaux, J.D. Glavin, D.V. Dabir, S.S. Walker, and C.M. Koehler. 2010. Substrate specificity of the TIM22 mitochondrial import pathway revealed with small molecule inhibitor of protein translocation. Proc. Natl. Acad. Sci. USA. 107 :9578–9583. 10.1073/pnas.0914387107 20457929 Hines, V., and G. Schatz. 1993. Precursor binding to yeast mitochondria. A general role for the outer membrane protein Mas70p. J. Biol. Chem. 268 :449–454.8380165 Horvath, S.E., L. Böttinger, F.N. Vögtle, N. Wiedemann, C. Meisinger, T. Becker, and G. Daum. 2012. Processing and topology of the yeast mitochondrial phosphatidylserine decarboxylase 1. J. Biol. Chem. 287 :36744–36755. 10.1074/jbc.M112.398107 22984266 Hoseini, H., S. Pandey, T. Jores, A. Schmitt, M. Franz-Wachtel, B. Macek, J. Buchner, K.S. Dimmer, and D. Rapaport. 2016. The cytosolic cochaperone Sti1 is relevant for mitochondrial biogenesis and morphology. FEBS J. 283 :3338–3352. 10.1111/febs.13813 27412066 Hurt, E.C., and G. Schatz. 1987. A cytosolic protein contains a cryptic mitochondrial targeting signal. Nature. 325 :499–503. 10.1038/325499a0 3543689 Hurt, E.C., N. Soltanifar, M. Goldschmidt-Clermont, J.-D. Rochaix, and G. Schatz. 1986. The cleavable pre-sequence of an imported chloroplast protein directs attached polypeptides into yeast mitochondria. EMBO J. 5 :1343–1350.16453686 Itakura, E., E. Zavodszky, S. Shao, M.L. Wohlever, R.J. Keenan, and R.S. Hegde. 2016. Ubiquilins Chaperone and Triage Mitochondrial Membrane Proteins for Degradation. Mol. Cell. 63 :21–33. 10.1016/j.molcel.2016.05.020 27345149 Khalimonchuk, O., M. Ott, S. Funes, K. Ostermann, G. Rödel, and J.M. Herrmann. 2006. Sequential processing of a mitochondrial tandem protein: insights into protein import in Schizosaccharomyces pombe. Eukaryot. Cell. 5 :997–1006. 10.1128/EC.00092-06 16835444 Koehl, P., and M. Levitt. 1999. Structure-based conformational preferences of amino acids. Proc. Natl. Acad. Sci. USA. 96 :12524–12529. 10.1073/pnas.96.22.12524 10535955 Kondo-Okamoto, N., J.M. Shaw, and K. Okamoto. 2008. Tetratricopeptide repeat proteins Tom70 and Tom71 mediate yeast mitochondrial morphogenesis. EMBO Rep. 9 :63–69. 10.1038/sj.embor.7401113 18007655 Lee, B.J., A.E. Cansizoglu, K.E. Süel, T.H. Louis, Z. Zhang, and Y.M. Chook. 2006. Rules for nuclear localization sequence recognition by karyopherin beta 2. Cell. 126 :543–558. 10.1016/j.cell.2006.05.049 16901787 Lee, C.M., J. Sedman, W. Neupert, and R.A. Stuart. 1999. The DNA helicase, Hmi1p, is transported into mitochondria by a C-terminal cleavable targeting signal. J. Biol. Chem. 274 :20937–20942. 10.1074/jbc.274.30.20937 10409639 Li, J., X. Qian, J. Hu, and B. Sha. 2009. Molecular chaperone Hsp70/Hsp90 prepares the mitochondrial outer membrane translocon receptor Tom71 for preprotein loading. J. Biol. Chem. 284 :23852–23859. 10.1074/jbc.M109.023986 19581297 Longen, S., M.W. Woellhaf, C. Petrungaro, J. Riemer, and J.M. Herrmann. 2014. The disulfide relay of the intermembrane space oxidizes the ribosomal subunit mrp10 on its transit into the mitochondrial matrix. Dev. Cell. 28 :30–42. 10.1016/j.devcel.2013.11.007 24360785 Lubben, T.H., S.M. Theg, and K. Keegstra. 1988. Transport of proteins into chloroplasts. Photosynth. Res. 17 :173–194. 10.1007/BF00047688 24429668 Malhotra, K., M. Sathappa, J.S. Landin, A.E. Johnson, and N.N. Alder. 2013. Structural changes in the mitochondrial Tim23 channel are coupled to the proton-motive force. Nat. Struct. Mol. Biol. 20 :965–972. 10.1038/nsmb.2613 23832274 McGeoch, D.J. 1985. On the predictive recognition of signal peptide sequences. Virus Res. 3 :271–286. 10.1016/0168-1702(85)90051-6 3000102 Meinecke, M., R. Wagner, P. Kovermann, B. Guiard, D.U. Mick, D.P. Hutu, W. Voos, K.N. Truscott, A. Chacinska, N. Pfanner, and P. Rehling. 2006. Tim50 maintains the permeability barrier of the mitochondrial inner membrane. Science. 312 :1523–1526. 10.1126/science.1127628 16763150 Melin, J., M. Kilisch, P. Neumann, O. Lytovchenko, R. Gomkale, A. Schendzielorz, B. Schmidt, T. Liepold, R. Ficner, O. Jahn, 2015. A presequence-binding groove in Tom70 supports import of Mdl1 into mitochondria. Biochim. Biophys. Acta. 1853 :1850–1859. 10.1016/j.bbamcr.2015.04.021 25958336 Mootha, V.K., J. Bunkenborg, J.V. Olsen, M. Hjerrild, J.R. Wisniewski, E. Stahl, M.S. Bolouri, H.N. Ray, S. Sihag, M. Kamal, 2003. Integrated analysis of protein composition, tissue diversity, and gene regulation in mouse mitochondria. Cell. 115 :629–640. 10.1016/S0092-8674(03)00926-7 14651853 Morgenstern, M., S.B. Stiller, P. Lübbert, C.D. Peikert, S. Dannenmaier, F. Drepper, U. Weill, P. Höß, R. Feuerstein, M. Gebert, 2017. Definition of a High-Confidence Mitochondrial Proteome at Quantitative Scale. Cell Reports. 19 :2836–2852. 10.1016/j.celrep.2017.06.014 28658629 Mossmann, D., F.N. Vögtle, A.A. Taskin, P.F. Teixeira, J. Ring, J.M. Burkhart, N. Burger, C.M. Pinho, J. Tadic, D. Loreth, 2014. Amyloid-β peptide induces mitochondrial dysfunction by inhibition of preprotein maturation. Cell Metab. 20 :662–669. 10.1016/j.cmet.2014.07.024 25176146 Nakai, K., and M. Kanehisa. 1992. A knowledge base for predicting protein localization sites in eukaryotic cells. Genomics. 14 :897–911. 10.1016/S0888-7543(05)80111-9 1478671 Okamoto, K., A. Brinker, S.A. Paschen, I. Moarefi, M. Hayer-Hartl, W. Neupert, and M. Brunner. 2002. The protein import motor of mitochondria: a targeted molecular ratchet driving unfolding and translocation. EMBO J. 21 :3659–3671. 10.1093/emboj/cdf358 12110579 Papić, D., Y. Elbaz-Alon, S.N. Koerdt, K. Leopold, D. Worm, M. Jung, M. Schuldiner, and D. Rapaport. 2013. The role of Djp1 in import of the mitochondrial protein Mim1 demonstrates specificity between a cochaperone and its substrate protein. Mol. Cell. Biol. 33 :4083–4094. 10.1128/MCB.00227-13 23959800 Ptitsyn, O.B., and A.V. Finkelstein. 1983. Theory of protein secondary structure and algorithm of its prediction. Biopolymers. 22 :15–25. 10.1002/bip.360220105 6673754 Ramage, L., T. Junne, K. Hahne, T. Lithgow, and G. Schatz. 1993. Functional cooperation of mitochondrial protein import receptors in yeast. EMBO J. 12 :4115–4123.8223428 Ramesh, A., V. Peleh, S. Martinez-Caballero, F. Wollweber, F. Sommer, M. van der Laan, M. Schroda, R.T. Alexander, M.L. Campo, and J.M. Herrmann. 2016. A disulfide bond in the TIM23 complex is crucial for voltage gating and mitochondrial protein import. J. Cell Biol. 214 :417–431. 10.1083/jcb.201602074 27502485 Rehling, P., K. Model, K. Brandner, P. Kovermann, A. Sickmann, H.E. Meyer, W. Kühlbrandt, R. Wagner, K.N. Truscott, and N. Pfanner. 2003. Protein insertion into the mitochondrial inner membrane by a twin-pore translocase. Science. 299 :1747–1751. 10.1126/science.1080945 12637749 Rhee, H.W., P. Zou, N.D. Udeshi, J.D. Martell, V.K. Mootha, S.A. Carr, and A.Y. Ting. 2013. Proteomic mapping of mitochondria in living cells via spatially restricted enzymatic tagging. Science. 339 :1328–1331. 10.1126/science.1230593 23371551 Rimmer, K.A., J.H. Foo, A. Ng, E.J. Petrie, P.J. Shilling, A.J. Perry, H.D. Mertens, T. Lithgow, T.D. Mulhern, and P.R. Gooley. 2011. Recognition of mitochondrial targeting sequences by the import receptors Tom20 and Tom22. J. Mol. Biol. 405 :804–818. 10.1016/j.jmb.2010.11.017 21087612 Sato, T., M. Esaki, J.M. Fernandez, and T. Endo. 2005. Comparison of the protein-unfolding pathways between mitochondrial protein import and atomic-force microscopy measurements. Proc. Natl. Acad. Sci. USA. 102 :17999–18004. 10.1073/pnas.0504495102 16326810 Schendzielorz, A.B., C. Schulz, O. Lytovchenko, A. Clancy, B. Guiard, R. Ieva, M. van der Laan, and P. Rehling. 2017. Two distinct membrane potential-dependent steps drive mitochondrial matrix protein translocation. J. Cell Biol. 216 :83–92. 10.1083/jcb.201607066 28011846 Schibich, D., F. Gloge, I. Pöhner, P. Björkholm, R.C. Wade, G. von Heijne, B. Bukau, and G. Kramer. 2016. Global profiling of SRP interaction with nascent polypeptides. Nature. 536 :219–223. 10.1038/nature19070 27487212 Sherman, F. 1963. Respiration-deficient mutants of yeast. I. Genetics. Genetics. 48 :375–385.13977171 Shiota, T., H. Mabuchi, S. Tanaka-Yamano, K. Yamano, and T. Endo. 2011. In vivo protein-interaction mapping of a mitochondrial translocator protein Tom22 at work. Proc. Natl. Acad. Sci. USA. 108 :15179–15183. 10.1073/pnas.1105921108 21896724 Shiota, T., K. Imai, J. Qiu, V.L. Hewitt, K. Tan, H.H. Shen, N. Sakiyama, Y. Fukasawa, S. Hayat, M. Kamiya, 2015. Molecular architecture of the active mitochondrial protein gate. Science. 349 :1544–1548. 10.1126/science.aac6428 26404837 Sickmann, A., J. Reinders, Y. Wagner, C. Joppich, R. Zahedi, H.E. Meyer, B. Schönfisch, I. Perschil, A. Chacinska, B. Guiard, 2003. The proteome of Saccharomyces cerevisiae mitochondria. Proc. Natl. Acad. Sci. USA. 100 :13207–13212. 10.1073/pnas.2135385100 14576278 Sirrenberg, C., M.F. Bauer, B. Guiard, W. Neupert, and M. Brunner. 1996. Import of carrier proteins into the mitochondrial inner membrane mediated by Tim22. Nature. 384 :582–585. 10.1038/384582a0 8955274 Ting, S.Y., N.L. Yan, B.A. Schilke, and E.A. Craig. 2017. Dual interaction of scaffold protein Tim44 of mitochondrial import motor with channel-forming translocase subunit Tim23. eLife. 6 :e23609. 10.7554/eLife.23609 28440746 Truscott, K.N., P. Kovermann, A. Geissler, A. Merlin, M. Meijer, A.J. Driessen, J. Rassow, N. Pfanner, and R. Wagner. 2001. A presequence- and voltage-sensitive channel of the mitochondrial preprotein translocase formed by Tim23. Nat. Struct. Biol. 8 :1074–1082. 10.1038/nsb726 11713477 Vögtle, F.N., S. Wortelkamp, R.P. Zahedi, D. Becker, C. Leidhold, K. Gevaert, J. Kellermann, W. Voos, A. Sickmann, N. Pfanner, and C. Meisinger. 2009. Global analysis of the mitochondrial N-proteome identifies a processing peptidase critical for protein stability. Cell. 139 :428–439. 10.1016/j.cell.2009.07.045 19837041 von Heijne, G. 1986 a. Mitochondrial targeting sequences may form amphiphilic helices. EMBO J. 5 :1335–1342.3015599 von Heijne, G. 1986 b. A new method for predicting signal sequence cleavage sites. Nucleic Acids Res. 14 :4683–4690. 10.1093/nar/14.11.4683 3714490 Weckbecker, D., S. Longen, J. Riemer, and J.M. Herrmann. 2012. Atp23 biogenesis reveals a chaperone-like folding activity of Mia40 in the IMS of mitochondria. EMBO J. 31 :4348–4358. 10.1038/emboj.2012.263 22990235 Wickner, W., G. Mandel, C. Zwizinski, M. Bates, and T. Killick. 1978. Synthesis of phage M13 coat protein and its assembly into membranes in vitro. Proc. Natl. Acad. Sci. USA. 75 :1754–1758. 10.1073/pnas.75.4.1754 273906 Wiedemann, N., and N. Pfanner. 2017. Mitochondrial Machineries for Protein Import and Assembly. Annu. Rev. Biochem. 86 :685–714. 10.1146/annurev-biochem-060815-014352 28301740 Wienhues, U., K. Becker, M. Schleyer, B. Guiard, M. Tropschug, A.L. Horwich, N. Pfanner, and W. Neupert. 1991. Protein folding causes an arrest of preprotein translocation into mitochondria in vivo. J. Cell Biol. 115 :1601–1609. 10.1083/jcb.115.6.1601 1757464 Woellhaf, M.W., F. Sommer, M. Schroda, and J.M. Herrmann. 2016. Proteomic profiling of the mitochondrial ribosome identifies Atp25 as a composite mitochondrial precursor protein. Mol. Biol. Cell. 27 :3031–3039. 10.1091/mbc.E16-07-0513 27582385 Wrobel, L., A. Trojanowska, M.E. Sztolsztener, and A. Chacinska. 2013. Mitochondrial protein import: Mia40 facilitates Tim22 translocation into the inner membrane of mitochondria. Mol. Biol. Cell. 24 :543–554. 10.1091/mbc.E12-09-0649 23283984 Wu, Y., and B. Sha. 2006. Crystal structure of yeast mitochondrial outer membrane translocon member Tom70p. Nat. Struct. Mol. Biol. 13 :589–593. 10.1038/nsmb1106 16767096 Xue, Q., H. Pei, Q. Liu, M. Zhao, J. Sun, E. Gao, X. Ma, and L. Tao. 2017. MICU1 protects against myocardial ischemia/reperfusion injury and its control by the importer receptor Tom70. Cell Death Dis. 8 :e2923. 10.1038/cddis.2017.280 28703803 Yagawa, K., K. Yamano, T. Oguro, M. Maeda, T. Sato, T. Momose, S. Kawano, and T. Endo. 2010. Structural basis for unfolding pathway-dependent stability of proteins: vectorial unfolding versus global unfolding. Protein Sci. 19 :693–702. 10.1002/pro.346 20095049 Yamamoto, H., K. Fukui, H. Takahashi, S. Kitamura, T. Shiota, K. Terao, M. Uchida, M. Esaki, S. Nishikawa, T. Yoshihisa, 2009. Roles of Tom70 in import of presequence-containing mitochondrial proteins. J. Biol. Chem. 284 :31635–31646. 10.1074/jbc.M109.041756 19767391 Yamamoto, H., N. Itoh, S. Kawano, Y. Yatsukawa, T. Momose, T. Makio, M. Matsunaga, M. Yokota, M. Esaki, T. Shodai, 2011. Dual role of the receptor Tom20 in specificity and efficiency of protein import into mitochondria. Proc. Natl. Acad. Sci. USA. 108 :91–96. 10.1073/pnas.1014918108 21173275 Young, J.C., N.J. Hoogenraad, and F.U. Hartl. 2003. Molecular chaperones Hsp90 and Hsp70 deliver preproteins to the mitochondrial import receptor Tom70. Cell. 112 :41–50. 10.1016/S0092-8674(02)01250-3 12526792 Zanphorlin, L.M., T.B. Lima, M.J. Wong, T.S. Balbuena, C.A. Minetti, D.P. Remeta, J.C. Young, L.R. Barbosa, F.C. Gozzo, and C.H. Ramos. 2016. Heat shock protein 90 kDa Hsp90 has a second functional interaction site with the mitochondrial import receptor Tom70. J. Biol. Chem. 291 :18620–18631. 10.1074/jbc.M115.710137 27402847

PMC006xxxxxx/PMC6122969.txt

==== Front J Exp Med J Exp Med jem jem The Journal of Experimental Medicine 0022-1007 1540-9538 Rockefeller University Press 30115740 20180577 10.1084/jem.20180577 Research Articles Article 314 Identification of non-mutated neoantigens presented by TAP-deficient tumors Identification of non-mutated neoantigens http://orcid.org/0000-0003-4396-0047 Marijt Koen A. 1 Blijleven Laura 1 http://orcid.org/0000-0002-4449-8707 Verdegaal Els M.E. 1 Kester Michel G. 2 Kowalewski Daniel J. 34 http://orcid.org/0000-0003-1614-2647 Rammensee Hans-Georg 34 http://orcid.org/0000-0003-1954-7762 Stevanović Stefan 34 http://orcid.org/0000-0001-6320-9133 Heemskerk Mirjam H.M. 2 http://orcid.org/0000-0002-6556-0354 van der Burg Sjoerd H. 1 http://orcid.org/0000-0002-9115-558X van Hall Thorbald 1 1 Department of Medical Oncology, Leiden University Medical Center, Leiden, Netherlands 2 Department of Hematology, Leiden University Medical Center, Leiden, Netherlands 3 Department of Immunology, Institute for Cell Biology, University of Tübingen, Tübingen, Germany 4 German Cancer Consortium, German Cancer Research Center, Tübingen, Germany Correspondence to Thorbald van Hall: T.van_Hall@lumc.nl 03 9 2018 215 9 23252337 23 3 2018 30 5 2018 10 7 2018 © 2018 Marijt et al. 2018 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). A hybrid forward-reversed immunological screen is performed to identify 16 novel HLA-A2 presented cancer antigens. These peptides are selectively presented by immune-escaped cancer cells with defects in the peptide transporter TAP. In contrast to mutated neoantigens, these “self” neoantigens are universally presented across different cancer types. Most T cell–based immunotherapies of cancer depend on intact antigen presentation by HLA class I molecules (HLA-I). However, defects in the antigen-processing machinery can cause downregulation of HLA-I, rendering tumor cells resistant to CD8+ T cells. Previously, we demonstrated that a unique category of cancer antigens is selectively presented by tumor cells deficient for the peptide transporter TAP, enabling a specific attack of such tumors without causing immunopathology in mouse models. With a novel combinatorial screening approach, we now identify 16 antigens of this category in humans. These HLA-A*02:01 presented peptides do not derive from the mutanome of cancers, but are of “self” origin and therefore constitute universal neoantigens. Indeed, CD8+ T cells specific for the leader peptide of the ubiquitously expressed LRPAP1 protein recognized TAP-deficient, HLA-Ilow lymphomas, melanomas, and renal and colon carcinomas, but not healthy counterparts. In contrast to personalized mutanome-targeted therapies, these conserved neoantigens and their cognate receptors can be exploited for immune-escaped cancers across diverse histological origins. Graphical Abstract Dutch Cancer Foundation https://doi.org/10.13039/501100004622 2013-6142 ==== Body pmcIntroduction Many T cell–based immunotherapies for cancer are based on recognition of tumor antigens presented in HLA class I (HLA-I) molecules by tumor cells (Robbins et al., 2013; Schumacher and Schreiber, 2015). Success of immune checkpoint blockade therapy is strongly correlated with mutational load and mismatch repair-deficient cancers, irrespective of tumor type (Snyder et al., 2014; Lauss et al., 2017). Point-mutated peptides indeed constitute formidable tumor antigens due to their nonself nature, for which a noncurtailed T cell repertoire is available. An absolute requirement for such T cells to exert their action against cancer is the display of HLA-I at the surface of tumor cells. However, HLA-I down-modulation on cancer cells is observed in many immune-escaped cancers, often caused by epigenetic silencing of antigen-processing components, like the transporter associated with antigen processing (TAP; Setiadi et al., 2007; Garrido et al., 2016; Ritter et al., 2017). Recent studies implicated that acquired resistance to checkpoint therapy can occur through alterations in genes relevant for antigen processing and presentation (Patel et al., 2017; Sucker et al., 2017). For instance, mutations in the JAK1/JAK2 IFN signaling pathway represented acquired and primary resistance mechanisms in cancer patients who relapsed from or did not respond at all to checkpoint therapy, respectively. Notably, these mutations resulted in the inability to respond to IFN-γ and thus to upregulate antigen processing and presentation by HLA-I (Gao et al., 2016; Zaretsky et al., 2016; Shin et al., 2017). Our group previously discovered a novel category of tumor antigens, referred to as TEIPP (T cell epitopes associated with peptide processing), that are presented at the surface of tumor cells carrying defects in antigen processing (Marijt et al., 2018). In mouse tumor models in which MHC-I display is down-modulated by defects in the peptide transporter TAP, we showed a selective presentation of TEIPP peptides and successful targeting of immune-escaped tumor variants by TEIPP-specific T cells (Doorduijn et al., 2016, 2018a). Thus, targeting TEIPP neoantigens is a potent strategy to induce antitumor responses for tumors with low MHC-I expression. TEIPPs are derived from ubiquitously expressed non-mutated “self” proteins; however, their processed peptides fail to be loaded into MHC-I in healthy cells. Their surface presentation is highly promoted by defects in the antigen-processing machinery, especially in the absence of the peptide transporter TAP. Due to this virtue, TEIPP peptides constitute tumor-specific antigens. We have shown that the CD8+ T cell repertoire against TEIPP neoantigens is positively selected in the thymus and that these cells remain naive, even in tumor-bearing mice, making this subset fully exploitable for T cell–based therapies against immune-escaped cancers without any signs of autoimmune reactivity (Doorduijn et al., 2018a). As of yet, only one human TEIPP neoantigen has been identified at the molecular level (El Hage et al., 2008; Durgeau et al., 2011). To identify multiple human TEIPP antigens, we developed a systematic hybrid forward-reversed immunology screen to identify human TEIPP antigens. This approach encompassed an in silico prediction of TEIPP neoantigen candidates from the whole humane proteome, matching candidates to the cancer-specific peptidome, and an ex vivo screen to confirm the presence of a TEIPP T cell repertoire in healthy donors. Here, we present data on 16 identified HLA-A*02:01–binding TEIPP epitopes and a full characterization of the T cell reactivity against one of them. Results Strategy for target identification from the complete human proteome To identify human TEIPP antigens that are presented by TAP-deficient cancer cells, we developed a hybrid forward-reversed immunology identification approach based on alternative antigen-processing rules in combination with cancer-specific peptidome database matching (Fig. 1 A). The whole human proteome was chosen as a starting point, since TEIPP antigens are non-mutated self antigens that are preferentially displayed on cells with deficiency in the peptide transporter TAP. This TAP-independent loading in HLA-I molecules can occur via two known alternative processing pathways: liberation of N-terminal ‘‘signal peptides” and C-terminal “tail peptides” (Martoglio and Dobberstein, 1998; Yewdell et al., 1998; Neefjes et al., 2011; Blum et al., 2013; Oliveira et al., 2013; Fig. 1, A and B). A list of signal peptide–containing proteins was selected from the human proteome with the use of a web-based algorithm that predicts the signal peptidase cleavage site (Käll et al., 2004). Most of these leader peptides need further processing by the proteolytic enzyme signal peptide peptidase in order to release them from the ER membrane and thereby generate peptides of different lengths suitable for loading in HLA-I (Martoglio and Dobberstein, 1998; Blum et al., 2013). This signal peptide search yielded 111,525 different 9- and 10-mer TEIPP candidates. Additionally, C-terminal tail peptides were selected using information on the topology of transmembrane proteins in the same web-based algorithm, knowing that liberation and TAP-independent processing of C-termini can be exerted by proteases like furin and signal peptide peptidase (Tiwari et al., 2007; Medina et al., 2009; Leonhardt et al., 2010, 2011; Oliveira et al., 2013). This search resulted in 6,674 tail peptides (Fig. 1, A and B). Next, we filtered these combined peptides for a predicted high HLA-I binding affinity (IC50 < 500 nM) to one of the so-called HLA “supertypes,” representing most alleles in the Caucasian population (HLA-A*01:01, HLA-A*02:01, HLA-A*03:01, HLA-B*07:02, HLA-B*40:01, and HLA-A*24:02; Sette and Sidney, 1999; Nielsen et al., 2003; Sidney et al., 2008; Andreatta and Nielsen, 2016). This final list of 13,731 TEIPP candidates was then cross-matched with a large peptidome database containing peptides eluted from >200 tumor samples and 120 healthy tissue control samples (Table S1). Thereby, we focused on those TEIPP candidates that were endogenously processed and exclusively presented by tumor cells (Fig. 1 C). This strategy yielded a short list of 65 TEIPP neoantigen candidates, of which 40 had a predicted binding to the HLA-A*02:01 allele (Fig. S1, A–D). These 40 HLA-A*02:01–binding peptides were chosen to explore the existence of cognate CD8+ T cells (Fig. 1 A). Figure 1. Identification approach of TEIPP neoantigens associated with immune-edited cancers. (A) A schematic overview of the applied screening approach. Human peptide sequences were selected via algorithms-based predictions on the whole human proteome for presentation by HLA-I molecules independent of the peptide transporter TAP. Two main unraveled pathways have been described for which algorithms are available: the processing of signal peptides and of C-terminal tail peptides (red). The resulting peptide selection was matched with a large database of naturally presented peptides from human cancers and healthy tissue (blue). The subsequent short list of candidates with a predicted binding to HLA-A*02:01 was subjected to functional testing for T cell immunogenicity and recognition of tumor cells deficient for the peptide transporter (green). (B) Graphical display of the two molecularly described TAP-independent processing mechanisms. Top: Signal peptides containing proteins are processed in the ER membrane by the proteolytic enzymes signal peptidase (SP) and signal peptide peptidase (SPP), resulting in the liberation of small peptides in the lumen of the ER, where they can bind to HLA-I molecules. Bottom: Proteins with their C-terminal tail protruding in the ER can be cleaved by proteases like furins or single peptide peptidase. (C) Venn diagram of in silico predicted TEIPP neoantigens matched with databases comprising naturally presented peptides from tumor tissues or healthy tissues. Detectable frequencies of CD8+ T cells against 16 out of 40 HLA-A*02:01–binding TEIPP neoantigens To validate that T cells against these TEIPP neoantigens were present in the repertoire of healthy donors, we grouped 40 different HLA-A*02:01 tetramers containing the TEIPP neoantigens in eight pools of five peptides each, based on predicted binding affinity (from high in “peptide pool 1” to low in “peptide pool 8”). Next, CD8+ T cell cultures were started with tetramer-assisted “pull-downs” from peripheral blood mononuclear cells (PBMCs). These enriched CD8+ T cell pools were stimulated with the five respective synthetic peptides of that particular pool for several rounds and analyzed for the presence of peptide-specific T cells by flow cytometry (Fig. 2 A). In an initial screen of three different HLA-A*02:01–positive healthy donors, we observed T cell repertoires for 16 out of 40 TEIPP neoantigen candidates (Fig. 2 A). Frequencies of tetramer-stained CD8+ T cells varied from 0.1% up to 78% in these short-term expanded cultures. These percentages differed between the three tested donors, indicating that this protocol enabled the detection of TEIPP-specific CD8+ T cells, but that the observed frequencies reflected variation of the in vitro steps of enrichment and expansion. Additional donors were examined for the 16 positive peptides to evaluate the consistency of this finding, and, against most peptides, CD8+ T cells were detected in the majority of the donors (Fig. 2 B and Table 1). Collectively, these data indicated that our TEIPP identification strategy had a high success rate to identify human antigens in which a CD8+ T cell repertoire existed in healthy individuals. Figure 2. Detection of CD8+ T cells against TEIPP neoantigens in PBMCs of healthy donors. PBMCs of healthy donors were screened with HLA-A*02:01 tetramers to identify whether a T cell repertoire is present for TEIPP neoantigens. HLA-A*02:01 tetramers folded with the selected 40 TEIPP neoantigen candidates were arranged in eight pools based on predicted peptide binding. (A) TEIPP-enriched T cell pools were analyzed by flow cytometry (n = 3). Representative dot plots of TEIPP neoantigen-specific CD8+ T cells are shown. (B) PBMCs of additional healthy donors were tested to validate the presence of a TEIPP T cell repertoire. All 40 TEIPP neoantigen candidates were tested in at least three different donors. Shown are data of all tested donors. Each bar represents one TEIPP neoantigen candidate. Blue indicates the percentage of T cell–positive donors based on flow cytometry analysis. Numbers above each bar indicate the fraction of positive donors out of total tested. Table 1. Overview of TEIPP neoantigen candidates # Gene name Peptide sequencea Accession numberb HLA-A2 affinity (IC50)c Expressiond T cell repertoire Tissuee Normal Cancer Ubiquitous TEIPP p1 ERGIC3 FLSELQYYL Q9Y282 2 All ++ ++ 4/7 p4 TTYH3 ALFSFVTAL Q9C0H2 10 All + + 1/3 p9 IGJ VLAVFIKAV P01591 52 All ++ ++ 1/3 p14 LRPAP1 FLGPWPAAS D6REW6 353 All ++ ++ 14/14 p29 ARSA LLALAAGLAV P15289 79 All ++ ++ 9/10 p44 EMILIN2 FLYPFLSHL Q9BXX0 4 All + ++ 4/7 p49 H2AFV ILEYLTAEV E5RJU1 30 All ++ ++ 6/9 p67 SEP15 LSEKLERI O60613 15 All + + 4/8 Tissue-specific TEIPP p2 IL12A VLLDHLSLA P29459 7 Mixed + + 3/7 p17 HPR LLWGRQLFA P00739 24 Liver +++ + 6/7 p18 FSTL4 TLLGASLPA Q6MZW2 27 Mixed + +++ 5/7 p30 IFI30 LLLDVPTAAV M0QZG3 10 Mixed ++ ++ 5/8 p32 CD79B LLLSAEPVPA P40259 41 Lymphoid + ++ 3/7 p34 PCDHGA6 LTLLGTLWGA Q9Y5G7 24 Mixed +/− +/− 8/8 p35 C20orf141 SVLWLGALGL Q9NUB4 161 Testis + + 16/16 p55 APCS VIIKPLVWV P02743 113 Liver +++ +++ 9/10 a Peptide sequences of the non-mutated neoantigens. b UniProt accession numbers. c Predicted peptide affinity binding in HLA-A2:01. d Protein expression scores are based on data from the Human Protein Atlas. e Protein expression based on the Cancer Genome Atlas database score. Most TEIPP-specific CD8+ T cells reside in the naive repertoire Previous investigations in our mouse tumor model of TEIPP showed that TEIPP-specific T cells remain naive, even in the presence of TAP-deficient tumor cells (Doorduijn et al., 2018a). This was shown to be an advantage for their exploitability in therapeutic interventions (Doorduijn et al., 2018a). To test this aspect for human TEIPP-specific T cells, we again enriched tetramer-positive CD8+ T cells from PBMCs for the five peptides (p14, p29, p34, p35, and p55) that were frequently positive in our screen. Naive CD8+ T cells were directly separated from antigen-experienced CD8+ T cells by flow cytometry sorting, based on differentiation markers CD62L and CD45RA at the cell surface (Fig. S2 A). After this ex vivo sorting, the two populations were expanded by three rounds of in vitro stimulation and analyzed for the presence of tetramer-positive CD8+ T cells (Fig. 3, A and B). Importantly, T cells against p14, p34, and p35 were exclusively found in the sorted naive population in all tested donors, indicating that these specificities had never encountered cognate antigen in these healthy individuals. In addition, T cells against p29 and p55 were found in the naive subset of some donors, while in other donors, these specificities were detected in the antigen-experienced repertoire (Fig. 3 B). These data indicated that some TEIPP-specific CD8+ T cells already had been primed in a few healthy individuals, but that the majority of them reside in a naive state. We speculated that this priming could have been triggered by bacterial species in the gut with antigens that show striking sequence similarity with peptide 29 and 55 (Fig. S2 B). In summary, these data demonstrate a universal existence of naive T cells with TEIPP specificity in healthy individuals. Figure 3. Most TEIPP-specific CD8+ T cells reside in the naive repertoire. The status and functional avidity of TEIPP-specific T cell peptide candidates p14, p29, p34, p35, and p55 were examined. CD8+ T cells to these peptides were most abundantly detected in PBMCs of healthy donors. (A) Representative dot plots of pHLA-A*02:01 tetramer–stained CD8+ T cells cultured from either the naive subset or the experienced subset. Percentages represent pHLA-A*02:01 tetramer–positive cells out of total CD8+ T cells. (B) Pie charts for each of the TEIPP neoantigen candidates. Five donors were tested, and each slice represents one donor. The relative percentage of naive cells (blue) and percentage of experienced cells (red) are calculated out of total pHLA-A*02:01 tetramer–positive CD8+ cells. (C) T cell clones against these peptides were generated by single-cell sorting on HLA-A*02:01 tetramer–positive cells and subsequent expansion. Functional T cell avidity was examined by measuring cytokine production of activated T cells recognizing TEIPP peptide–pulsed, HLA-A*02:01–positive T2 cells with different concentrations of peptide. EC50 was determined by calculating the peptide concentration (nanomolar) corresponding to the 50% cytokine production of the maximum response. Individual clones were tested at least three times. Functional avidity of clonal CD8+ T cell cultures To functionally study TEIPP-specific CD8+ T cells, we generated clonal cultures against the five TEIPP peptides p14, p29, p34, p35, and p55. Four of these peptides were derived from N-terminal signal sequences of proteins, and one was derived from a C-terminal tail peptide (Fig. S2 C). We evaluated the binding affinity and stability of these five peptides to HLA-A*02:01 since these determinants are critically involved in immunogenicity for T cells (van der Burg et al., 1996; Müllbacher et al., 1999). Four of the five peptides displayed an intermediate affinity and formed complexes with HLA-A*02:01 for at least 4 h; however, p29 surprisingly failed to show binding in these cellular assays (Fig. S2, D and E). Multiple T cell clones were generated using an antigen-independent expansion protocol of tetramer-positive CD8+ T cells after single-cell sorting. Since the functional avidity correlates with high affinity TCRs and thus for in vivo efficacy (Dutoit et al., 2001; Snyder et al., 2003; McKee et al., 2005; Viganò et al., 2012), we assessed the peptide concentrations resulting in half-maximum cytokine response (EC50) of the several TEIPP-specific T cell clones when exposed to APCs pulsed with titrated concentrations of cognate peptide (Fig. 3 C). All isolated T cell clones exhibited a functional avidity between 0.4 and 66 nM, with the highest avidity for clones recognizing p34 and p55. The other clones, like those against p14, were considered to display moderate avidity. Of note, all clones with the same peptide specificity used different TCRβ V-segments, pointing at their independent origin and the availability of a broad repertoire (Fig. S2 F). These results demonstrated that the broad repertoire of TEIPP-specific T cells in healthy individuals exhibits a usual functional avidity for their cognate peptide and is competent to respond upon antigen encounter. CD8+ T cells against p14 recognize TAP-deficient tumors across diverse histological origin In search for TEIPP antigens on immune-escaped cancers, we selected the p14 peptide-epitope with sequence FLGPWPAAS from the ubiquitously expressed LRPAP1 protein (from now on called LRPAP121-30) for in-depth examination (Table 1). First, we confirmed specificity for the LRPAP121-30 epitope for T cell clone 1A8 by testing different peptide length variants of the LRPAP1 signal peptide (Fig. S3 A). Although unconventional, this examination confirmed the serine amino acid as C-terminus for this HLA-A*02:01 allele. Next, LRPAP1 gene expression data were collected from the Cancer Genome Atlas database for different tumor types, revealing that its expression is ubiquitous with exceptionally high expression levels in skin melanomas (Fig. 4 A). We therefore selected the HLA-A*02:01–positive skin melanoma cell line 518A2 and confirmed the expression of the LRPAP1 gene (Fig. 4 A and Fig. S3 B). For this cell line, a TAP1 KO variant, a CRISPR/CAS9 control TAP1-WT variant, and a TAP1/antigen double KO variant were generated by CRISPR/CAS9 technology; these gene-edited variants were tested for recognition by the LRPAP121-30-specific CD8+ T cell clone 1A8 (Fig. 4 B and Fig. S3, C and D). Surface display of HLA-A*02:01 decreased after knocking out the gene for the peptide transporter TAP1 (Fig. 4 C). The LRPAP121-30-specific CD8+ T cell clone selectively recognized the TAP1-deficient variant of 518A2. This suggested that the LRPAP121-30 signal peptide was only presented by the TAP-deficient tumor variant, despite equal expression of the LRPAP1 gene and lower HLA-A*02:01 levels on these cells (Fig. 4 D). Furthermore, selective KO of the epitope (LRPAP121-30 Ag-KO) resulted in complete abrogation of T cell recognition (Fig. 4 D). To control for artificial CRISPR/CAS9-induced immunogenicity, we included a previously established T cell clone (HSS1 PRAME) that recognizes the TAP-dependent PRAME (SLLQHLIGL) tumor antigen. This PRAME425-433-specific CD8+ T cell only recognized 518A2 WT cells, whereas it completely failed to respond toward the TAP KO variant. In contrast, LRPAP121-30-specific T cells displayed a clear preference for the TAP KO variant (Fig. 4 E). Collectively, these data confirmed that the applied tetramer-based selection approach allows the isolation of functional CD8+ T cells with a bona fide specificity for the antigen from which the TEIPP was derived. Moreover, these data excluded potential cross-reactivity of T cell clone 1A8 against other peptides presented by the 518A2 melanoma cells. Finally, antibodies against HLA-I molecules were able to block the tumor recognition by T cells, demonstrating dependency on peptide/HLA-I (Fig. 4 F). Figure 4. CD8+ T cells against p14 recognize TAP-deficient tumors of different histological origin. T cell clones were tested for their specificity to selectively recognize the LRPAP121-30 TEIPP epitope presented on TAP-impaired tumors. (A) Whisker plots of LRPAP1 gene expression in human cancers collected from the Cancer Genome Atlas database. (B) The TAP1 gene in the melanoma 518A2 was knocked out (TAP KO) using CRISPR/CAS9. In addition, the LRPAP121-30 epitope was knocked out in these cells (ag KO). mRNA expression of LRPAP1 was examined by qPCR. Data are shown as mean ± SD of triplicates. Significance testing: unpaired Student’s t test; n.s., not significant; ***, P < 0.001 (n = 3). (C) HLA-A*02:01 surface expression on 518A2 melanoma variants measured by flow cytometry. Plots are representative of three independent experiments. (D) GM-CSF production of T cell clone 1A8 upon 518A2 TAP KO recognition. When LRPAP121-30 epitope was knocked out (ag KO), recognition by the T cells was abolished. Representative data of three independent experiments are shown. Each dot represents a technical replicate. Significance testing: Student’s unpaired t test; ***, P < 0.001. (E) To control for artificial CRISPR/CAS9-mediated immunogenicity, we cocultured T cell clones specific for LRPAP121-30 (TAP-independent epitope) or PRAME425-433 (TAP-dependent epitope) together with the 518A2 melanoma (WT versus TAP KO). GM-CSF cytokine response was measured. Each dot represents one independent experiment with technical triplicates. (F) HLA dependency was tested using HLA-ABC– or HLA-A*02:01–blocking antibodies. Representative data of two independent experiments are shown. Each dot represents a technical replicate. Significance testing: unpaired Student’s t test; n.s., not significant; **, P < 0.01; ***, P < 0.001. (G) TEIPP T cell specificity was determined by coculturing three different T cell clones together with TAP-proficient T1 cells and TAP-deficient T2 lymphoma cells. Cytokine responses were measured to detect T cell recognition. Representative data of three independent experiments with technical triplicates are shown. (H) HLA-A*02:01 surface expression of the TAP-impaired tumor panels as measured by flow cytometry (SKCM, melanoma; KIRC, renal carcinoma). Representative data of three independent experiments are shown. (I) Cytokine production of indicated T cell clones upon antigen recognition of four melanomas and two renal carcinomas. Each dot represents one T cell clone specific for the LRPAP121-30 epitope. Representative data of three independent experiments with technical triplicates are shown. The ubiquitous expression of the LRPAP1 gene in diverse tumor types prompted us to test T cell recognition of other HLA-A*02:01–positive melanomas and tumors of other histological origin, including renal cell carcinoma and lymphoma. The panel of T1 and T2 lymphoma cells is known for its characterized genomic TAP status (Fig. S3 E; Wei and Cresswell, 1992; Anderson et al., 1993). We confirmed equal LRPAP1 gene expression by quantitative PCR (qPCR; Fig. S3 F) and examined recognition by multiple LRPAP121-30-specific T cell clones (Fig. 4 G). All tested CD8+ T cell clones selectively recognized the TAP-deficient T2 cells, again displaying a typical TEIPP specificity. Next, a panel of HLA-A*02:01–positive TAP-deficient melanomas and renal cell carcinomas was generated by genetically disrupting the TAP1 gene. Tumor cell mRNA levels of the LRPAP1 gene were again comparable as quantified by PCR (Fig. S3 G). Flow cytometry analysis revealed that HLA-A*02:01 levels varied between the lines, but that these decreased after TAP1 gene KO (Fig. 4 H). As observed for the 518A2 melanoma and the T2 lymphoma, multiple LRPAP121-30-specific T cell clones exhibited a preferential recognition of the TAP1-deficient tumor lines (Fig. 4 I). Interestingly, some parental tumors were already recognized even before silencing the TAP1 gene (e.g., melanoma 93.04), implying that low endogenous TAP1 levels in some primary tumors already allow presentation of the LRPAP121-30 antigen. Indeed, qPCR confirmed low TAP1 gene expression in the melanoma 93.04 (Fig. S3 H). In line with this, complete abolishment of the TAP1 gene in tumor line 93.04 did not result in much improved T cell recognition (Fig. 4 I). To further corroborate the shared character across multiple tumor types, we also tested T cell recognition of our 1A8 clone against two colon carcinomas (Fig. S3, I and J). Interestingly, we observed specific T cell responses toward the TAP KO variants of these two colon carcinoma cell lines (Fig. S3 K). Overall, these data clearly demonstrated that the LRPAP121-30 TEIPP epitope is universally presented in HLA-A*02:01 by cancers of diverse histology when the peptide transporter TAP is not functional. TEIPP-specific CD8+ T cells enable targeting of this shared, non-mutated peptide on immune-escaped HLA-Ilow cancers. Targeting TEIPP antigens appears safe despite ubiquitous expression of their proteins in healthy cells The fact that human TEIPP antigens derive from ubiquitously expressed non-mutated proteins may raise concerns for autoimmune pathology. Indeed, the LRPAP1 protein is clearly expressed in several critical organs of our body, as revealed by tissue slide staining images available from the Human Protein Atlas (Fig. 5 A and Fig. S4 A; Uhlén et al., 2015). Therefore, two primary melanocyte cultures and two immortalized kidney cell lines were tested for CD8+ T cell recognition. These nontransformed cells contained similar high levels of LRPAP1 transcripts as the 518A2 tumor (Fig. 5 B) and displayed sufficient levels of HLA-A*02:01 at their cell surface (Fig. 5 C). Despite these optimal conditions, LRPAP121-30-specific T cells did not respond to these healthy cells (Fig. 5 D). Exogenous loading of synthetic LRPAP1 peptide to the cells, however, did lead to LRPAP1-specific T cell response, suggesting that healthy cells were capable of presenting peptides (Fig. 5 E). These data confirmed a crucial role for low TAP function for the presentation of TEIPP antigens and suggested that TEIPP targeting might be considered a safe therapy, even though these proteins are present in healthy tissues as well. Figure 5. Targeting TEIPP antigens is safe despite ubiquitous expression of their proteins in healthy cells. To assess the safety of targeting non-mutated TEIPP neoantigens, we examined the T cell recognition of healthy cells. (A) Histological tissue stainings for LRPAP1 were collected from an online-accessible database (the Human Protein Atlas). Positive protein detection was observed in various healthy organs. (B) LRPAP1 mRNA levels were determined of two primary melanocyte cultures and two nontransformed kidney epithelial cell lines by qPCR. Data are shown as mean ± SD from one out of two experiments. (C) Surface HLA-A*02:01 expression of the healthy cells was determined by flow cytometry. Representative data of three independent experiments are shown. (D) No cytokine responses were detected when T cells specific for LRPAP121-30 were cocultured with these healthy cells. (E) To assess the ability of the healthy cells to present the epitope in HLA-A*02:01, peptide was exogenously loaded. LRPAP121-30-specific T cell clone 1A8 was cocultured with peptide-pulsed healthy cells, and GM-CSF cytokine production was measured. (D and E) Representative triplicate data of two independent experiments are shown. In summary, our data reveal an array of novel non-mutated tumor antigens displayed by tumors with (partial) TAP deficiency (Table 1). Some of these are tissue restricted, but others, like the LRPAP1 signal peptide, are ubiquitously expressed and represent universal HLA-presented tumor antigens on cancers with impaired antigen processing. Discussion We developed a screening approach to identify human antigens belonging to the novel category of non-mutated neoantigens of self origin called TEIPP. Our screen yielded 65 potential TEIPP candidates, which are exclusively present in the peptidome of cancer cells. Analysis of the 40 HLA-A*02:01–binding TEIPP candidates revealed a consistent CD8+ T cell repertoire to 16 of these 40 peptides. Additional analyses demonstrated that these CD8+ T cells were present in the naive pool for five peptide candidates and that they were not affected by central or peripheral tolerance. Monoclonal T cell cultures against the TEIPP epitope LRPAP121-30 selectively recognized TAP1-deficient cancers from different histological origins, including lymphoma, melanoma, colon carcinoma, and renal cell carcinoma, whereas TAP-proficient primary cells were spared from T cell recognition. This emphasizes the attractiveness of targeting TEIPP neoantigens on HLA-Ilow cancers, namely for its shared character on those tumor cells. Thus far, we focused on HLA-A*02:01–binding TEIPP peptides, but our screen also predicted a broad array of antigens presented by other HLA types. Furthermore, we realize that this proof-of-concept approach is far from complete. Most likely, many more undiscovered pathways enable the presentation of TAP-independent processed TEIPP neoantigens. We observed that knocking out the TAP1 gene rendered all tested cancer cell lines vulnerable for TEIPP-specific T cells. Importantly, TEIPP-specific T cells exhibited moderate to high recognition of some WT skin melanomas and renal cell carcinomas (e.g., 93.04, mz1257), and CRISPR/CAS9-mediated KO of the TAP1 gene led to only modest improvement of cancer cell recognition. In other cancer lines (e.g., melanoma 518A2, 08.11), minimal T cell responses were detected to WT cancer cells, and knockdown of TAP1 was necessary to induce strong T cell recognition. These data suggested that some cancer cell lines already down-modulated TAP1 gene expression to such a degree that it was already sufficient for the presentation of TEIPP neoantigens. Although only 1–2% of melanomas have deleterious mutations in TAP1 or TAP2, a high frequency of metastatic melanomas displays low TAP1 expression due to epigenetic silencing (Garrido et al., 2016; Ritter et al., 2017). At this moment, we have not determined how low TAP1 function needs to be for TEIPP neoantigens to be presented on the cell surface of cancer cells. The exact mechanisms of selective TEIPP neoantigen presentation on antigen-processing defective cells are still not completely known. Our previous experiments in mouse models suggested a competition between peptide antigens in the ER (Oliveira et al., 2011; Viganò et al., 2012). The peptide-loading complex (PLC) is a multisubunit membrane complex orchestrating the loading and editing of peptides in HLA-I molecules, facilitated by the TAP1/TAP2 channel, and thereby responsible for the translocation of peptides from the cytosol into the PLC (Blees et al., 2017). The TAP peptide transporter is incorporated in this efficient PLC, in close proximity to HLA-I molecules, which could explain why peptides processed through different mechanisms, independent of TAP, will not have the chance to bind in HLA-I molecules. Besides TAP levels, the amount of available antigenic protein also needs to be taken into account. Interestingly, massive overexpression of a TEIPP antigen makes it possible to present its TEIPP peptide in the cell surface even in TAP-proficient cells (Oliveira et al., 2011). We theorize that these two factors govern the efficiency of TEIPP neoantigen presentation. Low TAP function allows for low antigen expression, whereas moderate TAP function needs higher amounts of antigen expression (Durgeau et al., 2011; Oliveira et al., 2011). Of utmost importance is that TAP function in healthy cells of our organs is high enough to prevent recognition by TEIPP-specific T cells (Doorduijn et al., 2016, 2018a). The presence of a TEIPP T cell repertoire in healthy donors indicates that no negative selection has occurred in the thymus, even though gene expression in the thymus was observed for most of the TEIPP neoantigen candidates (Fig. S4 B). During T cell education in the thymus, TEIPP-specific T cells are positively selected for low affinity interaction with peptide/MHC in the cortex of the thymus. To prevent autoimmune responses, negative selection of T cells in the medulla of the thymus is essential to delete self-reactive T cells. This strengthens the hypothesis that TAP-proficient medullary thymic epithelial cells might express the target proteins but do not present TEIPP antigens by their HLA-I, and thus TEIPP T cells will undergo normal thymic selection. Indeed, TEIPP-specific T cells are deleted from the repertoire of TAP-deficient mice (Doorduijn et al., 2016). Moreover, most of the evaluated TEIPP-specific T cells were only found in the naive pool in the circulation and have not encountered antigen yet, signifying that TAP-proficient healthy cells did not trigger TEIPP-specific immunity. Although not tested, we presume this T cell repertoire to be naive in cancer patients as well, since tumor cells are not real APCs and harbor an immune-suppressive environment. Moreover, some viruses that down-modulate TAP (e.g., HSV and CMV) might also manipulate other pathways and therefore not allow TEIPP T cell priming. Ex vivo T cell activation or in vivo vaccination strategies might be needed to recruit the TEIPP T cell repertoire for immunotherapy. These assumptions are based on our earlier observations in a mouse model for TEIPP (Doorduijn et al., 2018a). It is noteworthy that, in some donors, we observed that T cells against p29 and p55 were derived from the antigen-experienced pool. We speculate that these TEIPP specificities had been triggered by cross-reactivity to microbiota-derived epitopes leaked from the gut, which can occur after the use of antibiotics or other intestinal injury (Lee et al., 2011; Zitvogel et al., 2016). Since the tested healthy donors did not suffer from severe autoimmune reactivity, even antigen-experienced TEIPP-specific T cells apparently do not target healthy TAP-proficient tissues. The safety of TEIPP-targeting therapies was thoroughly evaluated in mouse tumor models (van Hall et al., 2006; Chambers et al., 2007; Doorduijn et al., 2016, 2018a,b). Vaccine-induced TEIPP-specific T cells via synthetic peptides or dendritic cells, or adoptive transfer of TEIPP-specific T cells, all reduced outgrowth of TAP-deficient tumors without any sign of autoimmune reactivity. TEIPP neoantigens are selectively presented in cancer cells with deficient function of MHC-I–processing machinery. Therefore, we anticipate a complementary role for these antigens to those of other categories, like patient-specific mutanome peptides or cancer-testis antigens. A combination of these might prevent immune escape via shutdown of HLA-I expression due to defects in the antigen-processing machinery. A great asset of TEIPP antigens is their non-mutated status and the fact that the tumor antigens are broadly shared across many cancer types. A few validated TEIPP antigens or gene transfer of their cognate-cloned T cell receptors might be sufficient for increased efficacy of immunotherapy through prevention of tumor escape. Future investigations need to reveal safety and applicability of vaccination-based strategies and TCR-based gene transfer approaches for TEIPP. Together, we propose that TEIPP neoantigens are an interesting way to reestablish antitumor responses in human immune-escaped cancers. Materials and methods Study design The aim of this study was to test a systematic, hybrid, forward-reversed immunology approach to identify TEIPP-specific T cells and their cognate antigen (Table 1). For this, we collected the whole human proteome dataset from UniProt (ID: 9606, Release: 2014_06). For predictions of the N-terminal signal cleavage location and the topology prediction for C-terminal tail peptides, the web-based topology algorithm Phobius was used. For N-terminal peptide candidates, proteins with a predicted signal peptide probability of >0.5 were included. For C-terminal peptide candidates, proteins with a predicted C-terminal tail located in the ER were included. Peptide binding affinity to HLA-I molecules was predicted using MHCnet4.0 (http://www.cbs.dtu.dk/services/NetMHC/). Peptides with a predicted binding affinity <500 nM were included. We used the peptidome elution library (release 2014_10) of the Tübingen group (Department of Immunology, University of Tübingen) to select for biologically processed peptides. Next, in vitro analysis was performed on 40 HLA-A*02:01 TEIPP candidates using healthy donor PBMCs. In-depth analysis was performed on five TEIPP candidates. To guarantee statistical power, we performed all experiments on at least three different PBMC batches isolated from different donors. Tumor cell lines Human tumor cell lines were obtained through ATCC or derived from patient surgery material. The clinical protocol was approved by the Medical Ethics Committee of the Leiden University Medical Center and conducted in accordance with the Declaration of Helsinki. All tumor cell lines were cultured at 37°C, 5% CO2, in DMEM (Invitrogen) containing 8% heat-inactivated FCS, 100 U/ml penicillin, and 100 µg/ml streptomycin (Life Technologies) and supplemented with 2 mM glutamine (Invitrogen). Generation of TEIPP-specific T cell clones Buffy coats from healthy donors were obtained from Sanquin (Sanquin blood facility). Informed consent was given in writing by all participants. PBMCs were isolated using a ficoll gradient (density 1.077 g/ml). Approximately 500 × 106 PBMCs were incubated with PE-pHLA-A*02:01 tetramers for 30 min at 4°C. Anti-PE magnetic beads (MACS) were used to pull out pHLA-A*02:01 tetramer–positive cells over a MACS LS column as instructed by the manufacturer (Miltenyi Biotech). T cells were cultured in complete IMDM containing 8% human serum (Sanquin blood facility) and 100 U/ml IL-2 (Proleukine; Novartis). Bulk T cells were stimulated every 2 wk with a mixture of T cells (106), 10 µM synthetic peptide, irradiated PBMCs (106 cells, 80 Gy), and EBV-JY (105 cells, 100 Gy) in complete T cell culture medium supplemented with 100 U/ml IL-2 in 24-well plates (Costar). Culture medium was replenished every 2–3 d with fresh complete T cell medium. After two rounds of specific stimulation, the polyclonal T cell bulk was incubated with PE-labeled tetramers and single-cell sorted using flow cytometry to obtain monoclonal T cells. After monoclonal cell sorting, T cell clones were stimulated with a PHA mixture in 1 well of a 96-well plate: complete T cell medium supplemented with 800 ng/ml PHA (Murex Biotech), 100 U/ml IL-2, irradiated PBMCs (50 × 105 cells, 80 Gy), and EBV-JY cells (10 × 104 cells, 100 Gy). Generation of pHLA-A*02:01 tetramers Ultra-pure peptides were synthesized by JPT Peptide Technologies. Recombinant HLA-A*02:01 and human β2m proteins were in-house synthesized in Escherichia coli (Garboczi et al., 1992; Burrows et al., 2000). Biotinylated HLA-A*02:01 molecules were produced containing a UV-sensitive peptide to enable easy transfer of specific peptide, as previously described (Bakker et al., 2008). In brief, HLA-A*02:01 complexes were purified by gel filtration using fast protein liquid chromatography. pHLA-A*02:01 complexes were exposed to UV light to facilitate peptide exchange, resulting in peptide-specific monomers. pHLA-A*02:01 tetramers were formed by coupling biotinylated pHLA-A*02:01 monomers with streptavidin-coupled R-PE (Thermo Fisher). Tetramers were stored at −80°C for long-term storage or 4°C for short-term storage. Coculture T cell reactivity assay Dependent on the T cell clone, GM-CSF (T cell clone 1A8 and PRAME) or IFN-γ (all other tested T cell clones) ELISA assays (GM-CSF, MabTech; IFN-γ, PeliKine) were used to measure the cytokine production of activated T cells. ELISA plates were coated with human IFN-γ or GM-CSF primary antibody overnight at 4°C. T cell–conditioned medium was collected and incubated in the precoated ELISA plates for 2 h at room temperature. Next, ELISA plates were washed with PBS with 0.1% Tween-20 followed by incubation with the HRP-conjugated secondary antibody for 1 h. ELISA plates were developed with TMB (3,3,5,5-tetramethyl-benzidine liquid substrate supersensitive, for ELISA; Sigma-Aldrich) and measured at 450 nm using a plate reader (Spectramax id3; Molecular Devices). Lower cut-off values for cytokine production were calculated as medium values plus three times the SD. Values below this threshold were considered negative. All plotted values were within the upper and lower detection limits of the ELISAs. Affinity and stability assay Peptide affinity and stability were determined as previously described (van der Burg et al., 1996; Harndahl et al., 2012). In brief, HLA-A*02:01+ T2 cells were pulsed with increasing concentrations of peptide (1–100 μM) in serum-free RPMI culture medium for 24 h at 37°C. For MHC:peptide affinity, the maximum HLA-A*02:01 expression was measured by flow cytometry to calculate the EC50 concentration. For MHC:peptide stability, HLA-A*02:01–positive T2 cells were pulsed with 100 μM peptide for 24 h at 37°C. Maximum HLA-A*02:01 expression was determined and followed over time to calculate the t1/2 value. Generation of gene KOs in tumor cells using CRISPR/CAS9 Single guide RNAs (sgRNAs) were designed to target exon 1 of the human TAP1 gene (sgTAP1: 5′-GCTGCTACTTCTCGCCGACT-3′) or the LRPAP121-30 epitope located in exon 1 (LRPAP121-30: 5′-AGGGTCAGGTCGTTTCTGCG-3′). The sgRNA target sequence was cloned into the lentiCRISPR v2 vector (Sanjana et al., 2014). Virus particles were generated by cotransfecting sgRNA/CAS9 containing plasmid together with PAX2/pMD2.G packaging vectors into HEK293T cells using Lipofectamine 2000 (Thermo Fisher). Tumor cells were incubated with medium containing the virus particles for 24 h, and transduced tumor cells were selected with 1–5 μg/ml puromycin (Gibco). TAP1 KO efficiency was analyzed by measuring surface HLA-ABC (w6/32; BioLegend) expression using flow cytometry. Polyclonal TAP KO cell lines were generated by FACS sorting the HLA-Ilow cell population. CRISPR/CAS9 TAP-WT control cells were generated by FACS sorting the HLA-Ihigh cell population of the polyclonal bulk. TAP1 expression was verified by Western blot using anti-TAP1 antibodies (Cell Signaling). Western blot Proteins from cell cultures were extracted in lysis buffer (radioimmunoprecipitation assay; Cell Signaling), and protein concentrations were assessed using the Bradford Protein Assay Kit (Thermo Fisher). Protein samples were separated on a 4–12% SDS-PAGE gel and transferred onto nitrocellulose membranes using semi-dry transfer (Bio-Rad). To prevent nonspecific binding, membranes were blocked with 5% milk in Tris-buffered saline with 0.1% Tween-20 for 1 h at room temperature. The following antibodies were used: anti-TAP1 (dilution 1:1,000, 4°C; Cell Signaling), anti–β-actin (dilution 1:1,000, 4°C; Cell Signaling), and HRP-linked mouse anti-rabbit IgG (dilution 1:5,000, room temperature; Cell Signaling). Flow cytometry analysis CD3 (SK7), CD4 (SK3), CD8 (SK1), CD45RO (UCHL1), HLA-A*02:01 (bb7.2; BD Biosciences), and HLA-ABC (MCA81A), CD45RA (HI100), CD62L (DREG-56; BioLegend) anti-human mAbs were used for flow cytometry analysis. Cells were incubated with mAbs for 30 min at 4°C and washed three times. Unstained cells were used as negative controls. The Vβ repertoire analysis kit was used to determine the TCR Vβ of the different monoclonal T cells (Beckman Coulter). For T cell specificity analysis, we incubated T cells with pHLA-A*02:01 tetramers for 30 min at room temperature and washed three times. T cells were considered tetramer positive when mean fluorescence intensity was at least 102 times higher than unstained control. Cells were measured using an LSR-II flow cytometer (BD Biosciences) and analyzed using FlowJo software version 10 (Tree Star). mRNA isolation, cDNA synthesis, and qPCR analysis Cell pellets were washed twice with PBS and snap-frozen in liquid nitrogen. RNA was isolated using the RNAeasy Kit (Qiagen) according to the manufacturer’s protocol. cDNA was synthesized using the High Capacity RNA-to-cDNA Kit (Applied Biosystems). qPCR was done using SYBR Green supermix (Bio-Rad). For qPCR analysis, we used the following primers: GAPDH_f: 5′-GTGCTGAGTATGTCGTGGAGTCTAC-3′; GAPDH_r: 5′-GGCGGAGATGATGACCCTTTTGG-3′; TAP1_f: 5′-CTTGCAGGGAGAGGTGTTTG-3′; TAP1_r: 5′-GAGCATGATCCCCAAGAGAC-3′; LRPAP1_f: 5′-GACCGAAGAAATCCACGAGA-3′; LRPAP1_r: 5′-AGGTCCCACAGGTCAATCAC-3′. Statistical analysis All data are presented as means and SD. Statistical analysis was done using a paired Student’s t test (two-tailed) to determine the statistical significance of the differences. A minimum of three technical replicates were used in all experiments and were repeated at least two times. Online supplemental material Table S1 shows an overview of the number of histological tissue samples and cancer samples that are used to create the peptide elution database. In Fig. S1, we plotted the results of the predicted TEIPP candidate peptide binding in HLA-I. Fig. S2 shows the characterization of TEIPP T cells, sequence, binding affinity and stability of the TEIPP antigens, and homology with bacterial peptides. In Fig. S3, mRNA and protein expression analysis of LRPAP1 and TAP1 on the tumor panel and functional T cell assays are shown to support the findings in Fig. 4 in the main text. Fig. S4 illustrates the protein expression analysis of TEIPP candidates in healthy and cancer tissue. Supplementary Material Supplemental Materials (PDF) Acknowledgments We thank the peptide synthesis core facility and the flow cytometry core facility of the Leiden University Medical Center for their contribution to this project. The peptide elution database was made available by the Department of Immunology, University of Tübingen. This work was supported by the Dutch Cancer Foundation (2013-6142 to K.A. Marijt). The authors declare no competing financial interests. Author contributions: T. van Hall and S.H. van der Burg designed the project. K.A. Marijt and L. Blijleven performed and designed all experiments. M.G. Kester and K.A. Marijt synthesized the tetramers. E.M.E. Verdegaal provided all cancer cell lines. K.A. Marijt and T. van Hall wrote the manuscript. D.J. Kowalewski performed peptide matching analysis. H.-G. Rammensee and S. Stevanović shared their peptidome databases. K.A. Marijt, T. van Hall, and S.H. van der Burg interpreted data. S.H. van der Burg, H.-G. Rammensee, S. Stevanović, and M.H.M. Heemskerk reviewed and edited the manuscript. ==== Refs Anderson, K.S., J. Alexander, M. Wei, and P. Cresswell. 1993. Intracellular transport of class I MHC molecules in antigen processing mutant cell lines. J. Immunol. 151 :3407–3419.8376783 Andreatta, M., and M. Nielsen. 2016. Gapped sequence alignment using artificial neural networks: application to the MHC class I system. Bioinformatics. 32 :511–517. 10.1093/bioinformatics/btv639 26515819 Bakker, A.H., R. Hoppes, C. Linnemann, M. Toebes, B. Rodenko, C.R. Berkers, S.R. Hadrup, W.J. van Esch, M.H. Heemskerk, H. Ovaa, and T.N. Schumacher. 2008. Conditional MHC class I ligands and peptide exchange technology for the human MHC gene products HLA-A1, -A3, -A11, and -B7. Proc. Natl. Acad. Sci. USA. 105 :3825–3830. 10.1073/pnas.0709717105 18308940 Blees, A., D. Januliene, T. Hofmann, N. Koller, C. Schmidt, S. Trowitzsch, A. Moeller, and R. Tampé. 2017. Structure of the human MHC-I peptide-loading complex. Nature. 551 :525–528.29107940 Blum, J.S., P.A. Wearsch, and P. Cresswell. 2013. Pathways of antigen processing. Annu. Rev. Immunol. 31 :443–473. 10.1146/annurev-immunol-032712-095910 23298205 Burrows, S.R., N. Kienzle, A. Winterhalter, M. Bharadwaj, J.D. Altman, and A. Brooks. 2000. Peptide-MHC class I tetrameric complexes display exquisite ligand specificity. J. Immunol. 165 :6229–6234. 10.4049/jimmunol.165.11.6229 11086057 Chambers, B., P. Grufman, V. Fredriksson, K. Andersson, M. Roseboom, S. Laban, M. Camps, E.Z. Wolpert, E.J. Wiertz, R. Offringa, 2007. Induction of protective CTL immunity against peptide transporter TAP-deficient tumors through dendritic cell vaccination. Cancer Res. 67 :8450–8455. 10.1158/0008-5472.CAN-07-1092 17875682 Doorduijn, E.M., M. Sluijter, B.J. Querido, C.C. Oliveira, A. Achour, F. Ossendorp, S.H. van der Burg, and T. van Hall. 2016. TAP-independent self-peptides enhance T cell recognition of immune-escaped tumors. J. Clin. Invest. 126 :784–794. 10.1172/JCI83671 26784543 Doorduijn, E.M., M. Sluijter, K.A. Marijt, B.J. Querido, S.H. van der Burg, and T. van Hall. 2018 a. T cells specific for a TAP-independent self-peptide remain naïve in tumor-bearing mice and are fully exploitable for therapy. OncoImmunology. 7 :e1382793. 10.1080/2162402X.2017.1382793 29399388 Doorduijn, E.M., M. Sluijter, B.J. Querido, U.J.E. Seidel, C.C. Oliveira, S.H. van der Burg, and T. van Hall. 2018 b. T Cells Engaging the Conserved MHC Class Ib Molecule Qa-1b with TAP-Independent Peptides Are Semi-Invariant Lymphocytes. Front. Immunol. 9 :60. 10.3389/fimmu.2018.00060 29422902 Durgeau, A., F. El Hage, I. Vergnon, P. Validire, V. de Montpréville, B. Besse, J.C. Soria, T. van Hall, and F. Mami-Chouaib. 2011. Different expression levels of the TAP peptide transporter lead to recognition of different antigenic peptides by tumor-specific CTL. J. Immunol. 187 :5532–5539. 10.4049/jimmunol.1102060 22025554 Dutoit, V., V. Rubio-Godoy, P.Y. Dietrich, A.L. Quiqueres, V. Schnuriger, D. Rimoldi, D. Liénard, D. Speiser, P. Guillaume, P. Batard, 2001. Heterogeneous T-cell response to MAGE-A10(254-262): high avidity-specific cytolytic T lymphocytes show superior antitumor activity. Cancer Res. 61 :5850–5856.11479225 El Hage, F., V. Stroobant, I. Vergnon, J.F. Baurain, H. Echchakir, V. Lazar, S. Chouaib, P.G. Coulie, and F. Mami-Chouaib. 2008. Preprocalcitonin signal peptide generates a cytotoxic T lymphocyte-defined tumor epitope processed by a proteasome-independent pathway. Proc. Natl. Acad. Sci. USA. 105 :10119–10124. 10.1073/pnas.0802753105 18626012 Gao, J., L.Z. Shi, H. Zhao, J. Chen, L. Xiong, Q. He, T. Chen, J. Roszik, C. Bernatchez, S.E. Woodman, 2016. Loss of IFN-γ Pathway Genes in Tumor Cells as a Mechanism of Resistance to Anti-CTLA-4 Therapy. Cell. 167 :397–404.e9. 10.1016/j.cell.2016.08.069 27667683 Garboczi, D.N., D.T. Hung, and D.C. Wiley. 1992. HLA-A2-peptide complexes: refolding and crystallization of molecules expressed in Escherichia coli and complexed with single antigenic peptides. Proc. Natl. Acad. Sci. USA. 89 :3429–3433. 10.1073/pnas.89.8.3429 1565634 Garrido, F., N. Aptsiauri, E.M. Doorduijn, A.M. Garcia Lora, and T. van Hall. 2016. The urgent need to recover MHC class I in cancers for effective immunotherapy. Curr. Opin. Immunol. 39 :44–51. 10.1016/j.coi.2015.12.007 26796069 Harndahl, M., M. Rasmussen, G. Roder, I. Dalgaard Pedersen, M. Sørensen, M. Nielsen, and S. Buus. 2012. Peptide-MHC class I stability is a better predictor than peptide affinity of CTL immunogenicity. Eur. J. Immunol. 42 :1405–1416. 10.1002/eji.201141774 22678897 Käll, L., A. Krogh, and E.L. Sonnhammer. 2004. A combined transmembrane topology and signal peptide prediction method. J. Mol. Biol. 338 :1027–1036. 10.1016/j.jmb.2004.03.016 15111065 Lauss, M., M. Donia, K. Harbst, R. Andersen, S. Mitra, F. Rosengren, M. Salim, J. Vallon-Christersson, T. Törngren, A. Kvist, 2017. Mutational and putative neoantigen load predict clinical benefit of adoptive T cell therapy in melanoma. Nat. Commun. 8 :1738. 10.1038/s41467-017-01460-0 29170503 Lee, Y.K., J.S. Menezes, Y. Umesaki, and S.K. Mazmanian. 2011. Proinflammatory T-cell responses to gut microbiota promote experimental autoimmune encephalomyelitis. Proc. Natl. Acad. Sci. USA. 108 (Suppl 1 ):4615–4622. 10.1073/pnas.1000082107 20660719 Leonhardt, R.M., D. Fiegl, E. Rufer, A. Karger, B. Bettin, and M.R. Knittler. 2010. Post-endoplasmic reticulum rescue of unstable MHC class I requires proprotein convertase PC7. J. Immunol. 184 :2985–2998. 10.4049/jimmunol.0900308 20164418 Leonhardt, R.M., N. Vigneron, C. Rahner, and P. Cresswell. 2011. Proprotein convertases process Pmel17 during secretion. J. Biol. Chem. 286 :9321–9337. 10.1074/jbc.M110.168088 21247888 Marijt, K.A., E.M. Doorduijn, and T. van Hall. 2018. TEIPP antigens for T-cell based immunotherapy of immune-edited HLA class Ilow cancers. Mol. Immunol.. 10.1016/j.molimm.2018.03.029 Martoglio, B., and B. Dobberstein. 1998. Signal sequences: more than just greasy peptides. Trends Cell Biol. 8 :410–415. 10.1016/S0962-8924(98)01360-9 9789330 McKee, M.D., J.J. Roszkowski, and M.I. Nishimura. 2005. T cell avidity and tumor recognition: implications and therapeutic strategies. J. Transl. Med. 3 :35. 10.1186/1479-5876-3-35 16174302 Medina, F., M. Ramos, S. Iborra, P. de León, M. Rodríguez-Castro, and M. Del Val. 2009. Furin-processed antigens targeted to the secretory route elicit functional TAP1-/-CD8+ T lymphocytes in vivo. J. Immunol. 183 :4639–4647. 10.4049/jimmunol.0901356 19752221 Müllbacher, A., M. Lobigs, J.W. Yewdell, J.R. Bennink, R. Tha Hla, and R.V. Blanden. 1999. High peptide affinity for MHC class I does not correlate with immunodominance. Scand. J. Immunol. 50 :420–426. 10.1046/j.1365-3083.1999.00619.x 10520183 Neefjes, J., M.L. Jongsma, P. Paul, and O. Bakke. 2011. Towards a systems understanding of MHC class I and MHC class II antigen presentation. Nat. Rev. Immunol. 11 :823–836. 10.1038/nri3084 22076556 Nielsen, M., C. Lundegaard, P. Worning, S.L. Lauemøller, K. Lamberth, S. Buus, S. Brunak, and O. Lund. 2003. Reliable prediction of T-cell epitopes using neural networks with novel sequence representations. Protein Sci. 12 :1007–1017. 10.1110/ps.0239403 12717023 Oliveira, C.C., B. Querido, M. Sluijter, J. Derbinski, S.H. van der Burg, and T. van Hall. 2011. Peptide transporter TAP mediates between competing antigen sources generating distinct surface MHC class I peptide repertoires. Eur. J. Immunol. 41 :3114–3124. 10.1002/eji.201141836 21898382 Oliveira, C.C., B. Querido, M. Sluijter, A.F. de Groot, R. van der Zee, M.J. Rabelink, R.C. Hoeben, F. Ossendorp, S.H. van der Burg, and T. van Hall. 2013. New role of signal peptide peptidase to liberate C-terminal peptides for MHC class I presentation. J. Immunol. 191 :4020–4028. 10.4049/jimmunol.1301496 24048903 Patel, S.J., N.E. Sanjana, R.J. Kishton, A. Eidizadeh, S.K. Vodnala, M. Cam, J.J. Gartner, L. Jia, S.M. Steinberg, T.N. Yamamoto, 2017. Identification of essential genes for cancer immunotherapy. Nature. 548 :537–542. 10.1038/nature23477 28783722 Ritter, C., K. Fan, A. Paschen, S. Reker Hardrup, S. Ferrone, P. Nghiem, S. Ugurel, D. Schrama, and J.C. Becker. 2017. Epigenetic priming restores the HLA class-I antigen processing machinery expression in Merkel cell carcinoma. Sci. Rep. 7 :2290. 10.1038/s41598-017-02608-0 28536458 Robbins, P.F., Y.C. Lu, M. El-Gamil, Y.F. Li, C. Gross, J. Gartner, J.C. Lin, J.K. Teer, P. Cliften, E. Tycksen, 2013. Mining exomic sequencing data to identify mutated antigens recognized by adoptively transferred tumor-reactive T cells. Nat. Med. 19 :747–752. 10.1038/nm.3161 23644516 Sanjana, N.E., O. Shalem, and F. Zhang. 2014. Improved vectors and genome-wide libraries for CRISPR screening. Nat. Methods. 11 :783–784. 10.1038/nmeth.3047 25075903 Schumacher, T.N., and R.D. Schreiber. 2015. Neoantigens in cancer immunotherapy. Science. 348 :69–74. 10.1126/science.aaa4971 25838375 Setiadi, A.F., M.D. David, R.P. Seipp, J.A. Hartikainen, R. Gopaul, and W.A. Jefferies. 2007. Epigenetic control of the immune escape mechanisms in malignant carcinomas. Mol. Cell. Biol. 27 :7886–7894. 10.1128/MCB.01547-07 17875943 Sette, A., and J. Sidney. 1999. Nine major HLA class I supertypes account for the vast preponderance of HLA-A and -B polymorphism. Immunogenetics. 50 :201–212. 10.1007/s002510050594 10602880 Shin, D.S., J.M. Zaretsky, H. Escuin-Ordinas, A. Garcia-Diaz, S. Hu-Lieskovan, A. Kalbasi, C.S. Grasso, W. Hugo, S. Sandoval, D.Y. Torrejon, 2017. Primary Resistance to PD-1 Blockade Mediated by JAK1/2 Mutations. Cancer Discov. 7 :188–201. 10.1158/2159-8290.CD-16-1223 27903500 Sidney, J., B. Peters, N. Frahm, C. Brander, and A. Sette. 2008. HLA class I supertypes: a revised and updated classification. BMC Immunol. 9 :1. 10.1186/1471-2172-9-1 18211710 Snyder, A., V. Makarov, T. Merghoub, J. Yuan, J.M. Zaretsky, A. Desrichard, L.A. Walsh, M.A. Postow, P. Wong, T.S. Ho, 2014. Genetic basis for clinical response to CTLA-4 blockade in melanoma. N. Engl. J. Med. 371 :2189–2199. 10.1056/NEJMoa1406498 25409260 Snyder, J.T., M.A. Alexander-Miller, J.A. Berzofskyl, and I.M. Belyakov. 2003. Molecular mechanisms and biological significance of CTL avidity. Curr. HIV Res. 1 :287–294. 10.2174/1570162033485230 15046253 Sucker, A., F. Zhao, N. Pieper, C. Heeke, R. Maltaner, N. Stadtler, B. Real, N. Bielefeld, S. Howe, B. Weide, 2017. Acquired IFNγ resistance impairs anti-tumor immunity and gives rise to T-cell-resistant melanoma lesions. Nat. Commun. 8 :15440. 10.1038/ncomms15440 28561041 Tiwari, N., N. Garbi, T. Reinheckel, G. Moldenhauer, G.J. Hämmerling, and F. Momburg. 2007. A transporter associated with antigen-processing independent vacuolar pathway for the MHC class I-mediated presentation of endogenous transmembrane proteins. J. Immunol. 178 :7932–7942. 10.4049/jimmunol.178.12.7932 17548631 Uhlén, M., L. Fagerberg, B.M. Hallström, C. Lindskog, P. Oksvold, A. Mardinoglu, Å. Sivertsson, C. Kampf, E. Sjöstedt, A. Asplund, 2015. Tissue-based map of the human proteome. Science. 347 :1260419. 10.1126/science.1260419 25613900 van der Burg, S.H., M.J. Visseren, R.M. Brandt, W.M. Kast, and C.J. Melief. 1996. Immunogenicity of peptides bound to MHC class I molecules depends on the MHC-peptide complex stability. J. Immunol. 156 :3308–3314.8617954 van Hall, T., E.Z. Wolpert, P. van Veelen, S. Laban, M. van der Veer, M. Roseboom, S. Bres, P. Grufman, A. de Ru, H. Meiring, 2006. Selective cytotoxic T-lymphocyte targeting of tumor immune escape variants. Nat. Med. 12 :417–424. 10.1038/nm1381 16550190 Viganò, S., D.T. Utzschneider, M. Perreau, G. Pantaleo, D. Zehn, and A. Harari. 2012. Functional avidity: a measure to predict the efficacy of effector T cells? Clin. Dev. Immunol. 2012 :153863. 10.1155/2012/153863 23227083 Wei, M.L., and P. Cresswell. 1992. HLA-A2 molecules in an antigen-processing mutant cell contain signal sequence-derived peptides. Nature. 356 :443–446. 10.1038/356443a0 1557127 Yewdell, J.W., H.L. Snyder, I. Bacik, L.C. Antón, Y. Deng, T.W. Behrens, T. Bachi, and J.R. Bennink. 1998. TAP-independent delivery of antigenic peptides to the endoplasmic reticulum: therapeutic potential and insights into TAP-dependent antigen processing. J. Immunother. 21 :127–131. 10.1097/00002371-199803000-00006 9551364 Zaretsky, J.M., A. Garcia-Diaz, D.S. Shin, H. Escuin-Ordinas, W. Hugo, S. Hu-Lieskovan, D.Y. Torrejon, G. Abril-Rodriguez, S. Sandoval, L. Barthly, 2016. Mutations Associated with Acquired Resistance to PD-1 Blockade in Melanoma. N. Engl. J. Med. 375 :819–829. 10.1056/NEJMoa1604958 27433843 Zitvogel, L., M. Ayyoub, B. Routy, and G. Kroemer. 2016. Microbiome and Anticancer Immunosurveillance. Cell. 165 :276–287. 10.1016/j.cell.2016.03.001 27058662

PMC006xxxxxx/PMC6168275.txt

==== Front J Cell Biol J Cell Biol jcb jcb The Journal of Cell Biology 0021-9525 1540-8140 Rockefeller University Press 30154187 201705127 10.1083/jcb.201705127 Research Articles Article 37 21 10 Cdt1 stabilizes kinetochore–microtubule attachments via an Aurora B kinase–dependent mechanism Cdt1, a new kinetochore–microtubule binding factor Agarwal Shivangi 1 Smith Kyle Paul 1 http://orcid.org/0000-0003-1903-0798 Zhou Yizhuo 2 Suzuki Aussie 3 McKenney Richard J. 4 http://orcid.org/0000-0002-0972-8669 Varma Dileep 1 1 Department of Cell and Molecular Biology, Feinberg School of Medicine, Northwestern University, Chicago, IL 2 Department of Biochemistry and Biophysics, The University of North Carolina at Chapel Hill, Chapel Hill, NC 3 Department of Biology, The University of North Carolina at Chapel Hill, Chapel Hill, NC 4 Department of Molecular and Cellular Biology, University of California, Davis, Davis, CA Correspondence to Dileep Varma: dileep.varma@northwestern.edu 01 10 2018 217 10 34463463 18 5 2017 06 2 2018 17 7 2018 © 2018 Agarwal et al. 2018 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). Cdt1 is a novel kinetochore–microtubule binding protein. The middle and the C-terminal winged-helix domains of Cdt1 are involved in microtubule binding. Aurora B kinase phosphorylation of Cdt1 influences its microtubule binding in vitro and is necessary for kinetochore–microtubule stability and mitotic progression in cells. Robust kinetochore–microtubule (kMT) attachment is critical for accurate chromosome segregation. G2/M-specific depletion of human Cdt1 that localizes to kinetochores in an Ndc80 complex–dependent manner leads to abnormal kMT attachments and mitotic arrest. This indicates an independent mitotic role for Cdt1 in addition to its prototypic function in DNA replication origin licensing. Here, we show that Cdt1 directly binds to microtubules (MTs). Endogenous or transiently expressed Cdt1 localizes to both mitotic spindle MTs and kinetochores. Deletion mapping of Cdt1 revealed that the regions comprising the middle and C-terminal winged-helix domains but lacking the N-terminal unstructured region were required for efficient MT binding. Mitotic kinase Aurora B interacts with and phosphorylates Cdt1. Aurora B–phosphomimetic Cdt1 exhibited attenuated MT binding, and its cellular expression induced defective kMT attachments with a concomitant delay in mitotic progression. Thus we provide mechanistic insight into how Cdt1 affects overall kMT stability in an Aurora B kinase phosphorylation-dependent manner; which is envisioned to augment the MT-binding of the Ndc80 complex. National Cancer Institute https://doi.org/10.13039/100000054 R00CA178188 National Institute of Neurological Disorders and Stroke https://doi.org/10.13039/100000065 R00NS089428 National Institute of General Medical Sciences https://doi.org/10.13039/100000057 R35GM124889 ==== Body pmcIntroduction Accurate chromosome segregation during mitosis is accomplished by the concerted function of the bipolar mitotic spindle and kinetochores (Westhorpe and Straight, 2013). In this process, mitotic cells inevitably confront a challenge to maintain robust kinetochore–microtubule (kMT) attachment despite dynamic instability generated by rapid polymerization (growth) and depolymerization (shrinkage) of microtubules (MTs; Joglekar et al., 2010; DeLuca and Musacchio, 2012). How this dynamic process of kMT coupling is accomplished and regulated in vertebrates is not well understood. A highly conserved network of protein complexes, called the KMN (Knl1, Mis12, Ndc80) network, is the core interface that links spindle MTs to the kinetochores (Cheeseman et al., 2006; Cheeseman and Desai, 2008; Afreen and Varma, 2015). The N-terminal region of the Ndc80 complex, containing the calponin homology (CH) domain and positively charged tail domain of the Hec1 subunit, constitutes an essential MT-binding site (Varma and Salmon, 2012). The N-terminal region of Knl1 has also been shown to be necessary for MT binding (Cheeseman et al., 2006; Welburn et al., 2010; Espeut et al., 2012). Recently, an unprecedented mitotic role for human Cdt1, a well-established DNA replication licensing factor, was discovered (Varma et al., 2012; Pozo and Cook, 2016). The Hec1 loop domain that generates a flexible hinge in an otherwise rigid Ndc80 complex (Wang et al., 2008) has been shown to recruit Cdt1 to kinetochores by interacting with Cdt1’s N-terminal region (Varma et al., 2012). Precise, high-resolution separation measurement (delta) analysis between the extreme N and C termini of Ndc80 revealed that the Ndc80 complex bound to Cdt1 maintains an extended conformation that serves to stabilize kMT attachments via an unknown mechanism (Varma et al., 2012). Thus, while substantial research has provided insights into the structural and mechanistic aspects of the “canonical” licensing function of Cdt1, how Cdt1 influences kMT attachments during mitosis remains unclear. Besides Cdt1, the loop region of Hec1 also serves as a docking site for several other microtubule-associated proteins (MAPs) such as Dis1 (vertebrate homologue of chTOG) and Dam1 in yeast (Hsu and Toda, 2011; Maure et al., 2011; Schmidt and Cheeseman, 2011). In fact, in budding yeast, the Ndc80 and Dam1 complexes function synergistically to bind to MTs (Tien et al., 2010). Similarly, in vertebrates, the loop region has been reported to recruit the Ska complex (Zhang et al., 2012) that also interacts with MTs through the unique winged-helix domain (WHD) of the Ska1 subunit. Further, the Ndc80 complex increases the affinity of the Ska1 subunit for MTs by eightfold (Schmidt et al., 2012; Abad et al., 2014). These studies suggest that although the Ndc80 complex is critical for kMT-binding, other factors such as the Dam1 and Ska complexes are required to efficiently orchestrate kMT attachments and chromosome segregation. The present study was thus undertaken to address critical outstanding questions surrounding the role of Cdt1 at kinetochores in stabilizing kMT attachments (Varma et al., 2012). These include (1) whether Cdt1 directly binds to MTs and (2) how Cdt1 is regulated for its mitotic function. Using several biochemical, biophysical, and cell biological approaches, we demonstrate that human Cdt1 can directly interact with MTs of the mitotic spindle. We further show that Cdt1 is a novel target for Aurora B kinase and that Aurora B–mediated phosphorylation of Cdt1 regulates its MT-binding properties, which in turn influence mitotic progression. Results Cdt1 directly binds to MTs in vitro We had previously demonstrated that Cdt1 localizes to mitotic kinetochores, dependent on the loop domain of the Hec1 subunit of the Ndc80 complex. Further, using a novel RNAi-mediated knockdown approach and microinjection of a function-blocking Cdt1 antibody, we showed that perturbation of Cdt1 function specifically during mitosis led to unstable kMT attachments, culminating in a late prometaphase arrest (Varma et al., 2012). Moreover, high-resolution microscopic analysis suggested that in the absence of Cdt1, the coiled coil of the Ndc80 complex assumed a bent conformation and the complex was not able to make a full extension along the kMT axis (Varma et al., 2012). But how Cdt1 contributed to this mechanism and imparted kMT stability was unclear. To investigate this mechanism systematically, we began by analyzing the structure of Cdt1, but since no high-resolution structures of full-length Cdt1 have been published, we first subjected the human Cdt1 amino acid sequence to at least 10 different secondary structure prediction algorithms that predict disordered regions. Although the consensus readily identified previously crystallized WHDs as being well folded, the N-terminal 93 aa were convincingly predicted to be at least ∼35% intrinsically disordered, in addition to the linker region between the two WHDs (Fig. 1 A). Homology models generated using three modeling programs reliably identified the winged-helix middle (WHM) and C-terminal (WHC) domains, while the N-terminal region (1–160) was consistently predicted to have only limited secondary structure elements in the absence of other binding partners (Fig. 1 B). The NMR solution structure (PDB ID: 2KLO) determined for the C-terminal 420–557 aa of Mus musculus, abbreviated as mCdt1, revealed that this region adopts a WHD conformation (Khayrutdinov et al., 2009) in addition to the other WHD in the middle of Cdt1 (172–368 aa) responsible for Geminin binding (Lee et al., 2004). The WHD has traditionally been implicated primarily as a DNA binding motif, but emerging evidence suggests that it could also serve as a protein–protein interaction domain (Khayrutdinov et al., 2009). In addition to the canonical three-helix bundles and antiparallel β-sheets, the Cdt1 WHC contains an additional helix (H4) at the C terminus (Fig. S1 A). Superimposition of the homology model for human Cdt1 fragment 410–546 aa (generated using Phyre2; Kelley et al., 2015) with the available NMR structure of an mCdt1 fragment showed an overlap with root mean square deviation of 1.038 Å, indicating that the 410–546 aa of human Cdt1 indeed conforms to a WHD (Fig. 1 B). Figure 1. Cdt1 is a novel MT-binding protein. (A) Disorder prediction in human Cdt1. Disordered prediction (sum of 10 programs) is plotted as a function of residue number. (B) Seven homology models of full-length Cdt1; colored from N terminus (blue) to C terminus (red) with regions of low confidence or low overlap shown in translucent. Conserved WHM and WHC domains shown and aligned in gray and black, respectively, for clarity. Magnified is the PyMol rendered cartoon representation showing superimposition of the C-terminal human Cdt1 (410–546 aa, in red) generated using Phyre2 server with available NMR structure (PDB ID: 2KLO; in green) of mouse C-terminal Cdt1 (420–557 aa). (C) Constructs used in this study, full-length Cdt1 (1–546 aa), its deletion variant devoid of N-terminal 92 aa (92–546), and N- and C-terminal deletion variants generated in fusion with an N-terminal GFP tag for expression in bacteria are shown as black bars. Disordered domain shown in blue, WHM domain shown in yellow, and WHC domain shown in red. (D) SDS-PAGE (15%) of the indicated purified recombinant proteins along with native tubulin purified from porcine brain. M, molecular mass standard; *, degradation bands (confirmed by Western blot) in Cdt192–410. Arrows point toward the band corresponding to the protein of interest. (E) Western blots showing MT cosedimentation for Cdt192–546 (GFP-tagged) and full-length Cdt11–546 (without any tag), 1 µM each; purified from bacteria with the indicated concentrations of taxol-stabilized MTs (in μM). Samples fractionated as supernatant (S) and pellet (P) were analyzed by Western blot and probed with antibodies against 6×-His tag or anti-Cdt1 as indicated in each case to detect Cdt1. Ponceau S–stained tubulin. (F) Quantification (mean ± SD, n = 3) of the Cdt192–546 and Cdt11–546 blots using ImageJ. The 95% CI values obtained for the fit were 0.60–0.94 for Bmax, 0.37–4.04 for Kd, and 0.07–0.29 for the background for the former construct; and 0.63–0.79 for Bmax, 1.33–3.25 for Kd, and 0.08–0.18 for the background for the latter construct. (G) Representative Western blots of the binding of indicated deletion fragments to MTs in a cosedimentation assay with the indicated MT concentrations (top). (H) Quantification of binding (mean ± SD, n = 3) of the deletion fragments to MTs. (I) Quantification of MT-binding of Cdt1’s smallest C-terminal fragment, Cdt1410-546, and a representative Western blot (top) from three independent experiments (mean ± SD, n = 3). (J) Blot overlay assay to study Cdt1–Hec1 interaction. Indicated proteins (0.5 µg each) were loaded as bait on 18% SDS-PAGE gel, transferred to nitrocellulose membrane, and blocked with 5% SM-TBST. Hec1::Nuf2-His dimeric complex (1 µg; a generous gift from A. Desai and D. Cheerambathur, University of California, San Diego, La Jolla, CA) was overlaid as prey protein on the membrane for 12 h at 4°C. The blot was washed and probed with anti-Hec1 antibody (1:2,000; 9G3; Abcam) followed by chemiluminescence. The same blot was probed with Ponceau S. The presence of a central and a C-terminal WHD in human Cdt1 similar to the MT-binding domain of the Ska complex (Abad et al., 2014) prompted us to posit a new paradigm in which Cdt1 bound to the loop may directly bind to MTs to create an additional MT-attachment site for the Ndc80 complex and contribute to the robustness of kMT attachments. To test our hypothesis, we first evaluated the ability of Cdt1 to directly bind to MTs using in vitro assays. Proteins with large disordered regions are generally characterized by rapid degradation (Vavouri et al., 2009). Indeed, challenges in purifying the full-length human Cdt1 from bacterial system were circumvented by the deletion of the 92-aa disordered stretch from its N terminus, which likely contributed to its degradation, reduced expression, and poor solubility. The Cdt1 protein was successfully purified with an N-terminal GFP fusion (abbreviated as Cdt192–546; Fig. 1, C and D) and was evaluated for its ability to bind to MTs in a copelleting assay. The Cdt192–546 (1 µM) was found to copellet efficiently with taxol-stabilized MTs (Fig. 1 E). This interaction with MTs is specific, as Cdt1 was not enriched in the pellet fraction in the absence of MTs. To determine the apparent MT-binding affinity of Cdt192–546, we varied the MT concentration and measured the Cdt1 concentration required for half-maximal binding. In this assay, the apparent dissociation constant (Kd) was determined to be 2.20 ± 0.87 µM (Fig. 1 F). Because the construct used in this assay was devoid of the first 92 aa, full-length (FL) Cdt11–546 was purified with a GST tag that was subsequently cleaved using Factor Xa, and the untagged Cdt11–546 was tested for its ability to bind to MTs. As expected, the full-length protein also bound to MTs (Fig. 1 E) with an affinity (2.29 ± 0.45 µM) similar to that of Cdt192–546 (Fig. 1 F), indicating that the first disordered 92 aa are dispensable for the MT-binding function of Cdt1. As controls, neither purified GFP nor GST proteins fractionated with MTs, substantiating the specificity of interaction (Fig. S1, B and C). To map the Cdt1-MT interaction, deletion fragments within Cdt192–546 were generated (Fig. 1, C and D). Interestingly, neither of the N-terminal fragments, Cdt192–232 or Cdt192–410, cosedimented with MTs (Fig. 1, G and H). However, the smallest C-terminal fragment, Cdt1410–546, did show MT binding, although its efficacy was reduced by ∼50% compared with Cdt192–546 (Fig. 1 I). Thus, we conclude that the C terminus of Cdt1 (410–546 aa) is necessary but not sufficient for MT binding. We further tested the ability of Cdt1 deletions to bind to Hec1 in an in vitro blot overlay assay. In accordance with the previous study (Varma et al., 2012), our data show that although the N-terminal fragments, Cdt192–232 and Cdt192–410, were incompetent to sediment with MTs (Fig. 1 G); they interacted with Hec1 in this assay (Fig. 1 J). On the other hand, the Cdt1410–546 that did bind MTs to some extent was not able to bind Hec1 (Fig. 1 J). GFP loaded as control bait protein was unable to bind Hec1 prey. Cdt1 preferentially binds to straight MTs independent of the E-hook tails At the C terminus of tubulin are negatively charged acidic tails of ∼20 aa, known as E-hooks, that represent a critical recognition site for many MT-binding proteins including Ndc80Bonsai (Ciferri et al., 2008) and Dam1 (Ramey et al., 2011). The ability of Cdt1 to bind to MTs devoid of E-hooks was thus assessed in a cosedimentation assay. Similar to Ska1 (Abad et al., 2014) and chTOG (Spittle et al., 2000), Cdt192–546 did not show any appreciable reduction in its ability to bind to MTs lacking E-hooks (Fig. 2, A and B). This suggests that the critical contacts involved in this interaction might be the structured regions of tubulin monomers rather than the unstructured acidic tails. The highly basic nature of Cdt1 (predicted isoelectric point = 9.9) could be indicative of MT recognition through electrostatic/ionic interactions. Indeed, the binding of Cdt192–546 with MTs was substantially reduced to ∼50% in the presence of 250 mM NaCl (Fig. 2 C), suggesting that the binding may be at least partially electrostatic in nature. Figure 2. Cdt1 diffuses on MTs in vitro and decorates spindle MTs in vivo. (A) Representative Western blot of Cdt192–546 (1 µM) cosedimentation with intact (MT) or subtilisin-treated (ST-MT) microtubules at 1 and 4 µM concentrations. (B) Quantification of A (mean ± SD, n = 3). Significance assessed by the two-sided unpaired nonparametric Student’s t test. (C) Quantification of cosedimentation of Cdt192–546 (1 µM) with 16 µM MTs in the absence (+MT) and presence (+MT+S; top) of 250 mM NaCl; mean ± SD, n = 3. P < 0.005 represents statistical significance as assessed by the two-sided unpaired nonparametric Student’s t test. Representative Western blot shown (bottom). (D) Single-channel and merged TIR-FM images showing surface immobilized Dylight405-labeled MTs (blue) and GFP-tagged Cdt192–546 (green). (E) Kymograph analysis of Cdt192–546 interactions with the MT lattice. (F) Cumulative frequency plot and exponential fit of Cdt192–546 dwell times. n = 304 molecules, two independent protein preparations. (G) Diffusion coefficient analysis of Cdt192–546 on MTs performed using DiaTrack particle tracking Software. Inset graph shows the histogram distribution of number of particles of Cdt192–546 that diffused on MTs within the range of 0–1 µm2/s. (H) Untransfected mitotic HeLa cell stained with anti-Cdt1 and either anti-Hec1 antibody (top, panels i and ii) or anti-tubulin antibody (bottom, panels iii–vi). Scale bar, 5 µm. Shown are the representative images for Cdt1 kinetochore (anti-Cdt1/red and anti-Hec1/green) and spindle (anti-Cdt1/red and anti-tubulin/green) staining at different stages of mitosis in HeLa cells as indicated. Orange arrows in panels iv–vi depict Cdt1 staining at the spindle poles; yellow arrow in panel v depicts the Cdt1 staining at the cleavage furrow in late anaphase, and green arrows in panel vi depict Cdt1 localization to the reforming nucleus in telophase. (I and J) Immunostaining of stably transfected (I) or transiently transfected (J) HeLa cells with anti-HA, anti-Hec1, and anti-tubulin antibodies, as indicated. Scale bar, 5 µm. The middle and bottom panels in J depict confocal images of cells in cytokinesis and interphase, respectively, with DAPI staining of chromosomes in blue. (K) Insets from the image stack of the mitotic cell in the top panels of J (left; also see Video 2) with an example from another cell in Video 3 (right). Scale bar, 1 µm. (L) Western blot of mock and transfected lysates to probe for Cdt1 (using anti-HA or anti-Cdt1 as indicated) and mouse tubulin as loading control. Expressed, ectopically expressed (with C-terminal 2×-HA, S- and 12×-His-tags; abbreviated as HASH tags); endogenous, endogenous (without tag) Cdt1. Further, we assessed the preference for Cdt1 to bind to straight versus curved MTs. Unlike the CH domains of Ndc80 that bind to the dimeric tubulin interface and prefer a straight MT protofilament conformation, Ska accesses MTs by interacting with the structured regions within tubulin monomers independent of any conformation (Abad et al., 2014). The data revealed that even though MT binding was considerably reduced when the MTs were not straight, Cdt1 did retain the capacity to bind to curved MT protofilaments (Fig. S1, D–F), which may have important implications for the function of Cdt1 at the kMT interface in conjunction with the Ndc80 complex. To test the preference of Cdt1 to bind to MT ends, polymerized MTs were sheared to increase the number of ends. We did not observe any significant increase in the association of Cdt1 with the MTs in this scenario (Fig. S1, D–F). This suggests that Cdt1 might not be merely a MT plus/minus end–specific protein, but that its binding sites are likely distributed along the length throughout the MT lattice. Cdt1 diffuses on MTs in vitro and decorates mitotic spindles in vivo To substantiate Cdt1-MT binding further, we directly visualized Cdt192–546 interaction with MTs using total internal reflection fluorescence microscopy (TIR-FM). The Cdt192–546 bound to Dylight405-labeled MTs at concentrations as low as 1 nM (Fig. 2 D) and started decorating the entire MT lattice at concentrations of ∼20 nM or higher (Fig. S1 G). In our conditions, individual molecules either bound statically or diffused along the MT lattice (Fig. 2 E). The dwell time distribution for Cdt192–546 on MTs was well described by a single exponential decay function yielding a characteristic dwell time (τ) of 2.37 s (Fig. 2 F). The calculated association rate (Κon, 10−6 s−1 ⋅ nM−1) was 1.07 ± 0.38, and the dissociation constant (Κd, μM) was 0.46 ± 0.17 (Fig. 2 F). A large fraction of Cdt192–546 (∼65%) was found to undergo short-range diffusive motion along the MT lattice to a moderate extent that falls in the range of 0–0.5 µm2/s (Fig. 2, G and E; and Video 1). A smaller fraction (∼20%) did not apparently diffuse at all, while an even smaller fraction (∼15%) exhibited relatively fast diffusion (>1 µm2/s). The average diffusion coefficient for Cdt192–546 was ∼0.16 µm2/s under our experimental conditions (Fig. 2 G). Both Ndc80 and Ska complexes have also been shown to exhibit diffusion along the MT lattice, albeit at variable rates comparable to that observed for Cdt1 under our assay conditions (Schmidt et al., 2012; Zaytsev et al., 2015). Although the in vitro experiments provide evidence that Cdt1 binds to MTs, we sought further evidence for this activity in cells. We had previously observed Cdt1 localization at mitotic kinetochores in paraformaldehyde-fixed LLCPK1 and PTK2 cells (Varma et al., 2012). We now show Cdt1 localization to kinetochores in HeLa cells using identical fixation conditions (Fig. 2 H). While Cdt1 transiently localized to the kinetochores predominantly during late prometaphase and metaphase (Fig. 2 H, panels i and ii), the detection dropped to almost negligible levels in anaphase and telophase (not depicted). Immunofluorescence staining of Cdt1 in methanol-fixed HeLa cells, which removed most of the kinetochore staining, allowed us to visualize clear mitotic spindle staining (Fig. 2 H). The variability observed in our previous (Varma et al., 2012) and present studies with respect to Cdt1 localization to kinetochores or spindle MTs is attributed to different antibodies, fixation conditions, and types of cells used. To evaluate the spindle localization of Cdt1 comprehensively, HeLa cells were examined at different stages of mitosis. Anti-Cdt1 antibody staining substantially overlapped with the mitotic spindle, but individual spindle MT staining was often hard to discern. Cdt1 staining on spindles was intermittent, with areas of brighter or dimmer intensities, with maximum intensity observed at the spindle poles. Cdt1 spindle staining started to appear in late prometaphase after a majority of the chromosomes were aligned, with no evident MT staining during early or mid-prometaphase (not depicted). Cdt1 spindle staining was the brightest in metaphase (Fig. 2 H, panel iii). A low-magnification image with several cells demonstrating similar Cdt1 colocalization with spindle MTs during metaphase is shown (Fig. S1 H). In anaphase and telophase, Cdt1 MT staining is predominantly restricted to the spindle poles as discrete crescent-shaped structures that are brighter in anaphase compared with telophase (Fig. 2 H, panels iv–vi). No discernable Cdt1 localization to interpolar MTs or the MT bundles of the central spindle regions in anaphase or telophase was observed. Interestingly (but unexplainably as of now), Cdt1 staining at the cell periphery became prominent at the beginning of anaphase and in telophase, with moderately bright staining appearing at the regressing cleavage furrow (Fig. 2 H, panels v and vi). Faint Cdt1 colocalization was also observed in the condensing nuclei in late telophase, consistent with its G1 nuclear function (Fig. 2 H, panel vi). To lend further support to these observations, full-length Cdt1 was transiently overexpressed with an HA tag in HeLa cells, and immunostaining was performed to assess its localization. In agreement with the results so far, staining with anti-HA clearly reveals localization of Cdt1-HA on the spindle MTs and also on kinetochores (overlap with Hec1, a kinetochore marker required for Cdt1 recruitment) in mitotic cells (Fig. 2 J, top panels; and Videos 2 and 3). As shown in Fig. 2 K, magnification of selected kinetochores from multiple cells further validates Cdt1 localization to kinetochores and kMTs (Fig. 2 K and Videos 2 and 3). Strikingly, interphase cells expressing high levels of Cdt1 showed an MT bundling phenotype (Fig. 2 J). This was not evident in cells with low or moderate expression of Cdt1, which is similar to the previous findings of Welburn et al. (2009), where a bundling phenotype was observed only when GFP-Ska1 was overexpressed in vivo or when a higher concentration of Ska complex was used in vitro. Further, in cells that were undergoing cytokinesis, Cdt1 was localized to MT bundles at the midbody region (Fig. 2 J). Western blotting of transfected HeLa cell lysates with anti-HA (representing the exogenously expressed Cdt1 with HA tag) confirmed expression of Cdt1 as a full-length protein in fusion with the HA tag (Fig. 2 L). Similar MT-staining (spindle/spindle pole) was also obtained in HeLa cells that stably expressed HA-tagged Cdt1, representing a scenario between overexpression and endogenous expression (Fig. 2 I). Aurora B kinase phosphorylates Cdt1 and affects Cdt1-MT binding in vitro We then sought to understand how the MT binding of Cdt1 could be regulated. Aurora B phosphorylates multiple substrates at kinetochores to negatively regulate kMT attachments and is also instrumental in correcting erroneous attachments (Welburn et al., 2010; Chan et al., 2012; Schmidt et al., 2012). In vitro studies with purified Ndc80 complex demonstrated that introduction of either multiple phosphorylations or phosphomimetic substitutions within the Hec1 tail reduced Ndc80-MT binding, while a lack of phosphorylation promoted stronger binding (Cheeseman et al., 2006; Umbreit et al., 2012). Similarly, Ska1, another kinetochore-localized MAP, is a substrate for Aurora B, and its phosphorylation at four sites located within the MT-binding domain drastically reduces its kinetochore localization (Welburn et al., 2010; Chan et al., 2012; Schmidt et al., 2012). Mutating two of these, S185 and S242, to aspartate dramatically reduced the MT-binding activity of the Ska complex in vitro and resulted in a mitotic delay and reduced kMT stability in vivo (Welburn et al., 2010; Chan et al., 2012; Schmidt et al., 2012). These aforementioned studies, coupled with the fact that Cdt1 closely resembles the Ndc80 and Ska complexes in being recruited to kinetochores where it binds to spindle MTs and contributes to kMT stabilization (Varma et al., 2012), prompted us to insinuate that Aurora B might be one of the essential kinases involved in regulating Cdt1-MT binding during mitosis. A close examination of the Cdt1 sequence indeed revealed at least seven putative Aurora B phosphorylation sites (T7, S73, S101, T102, S143, S265, and T358) organized in a canonical Aurora B consensus motif ((R/K)1-3-X-(S/T); Meraldi et al., 2004; Fig. 3 A). Five of these seven sites (S73, S101, T102, S143, and S265) have been reported as phosphorylation sites in the literature-curated database (http://www.phosphosite.org and http://www.phosida.org). Because highly conserved S159 in Ska3 was clearly phosphorylated by Aurora B even though it did not match the canonical consensus (Chan et al., 2012), we also included three such sites (T270, T287, and T329) in Cdt1 that were conserved across the species but did not conform to the classic Aurora B sequence motif (Fig. 3 A). Figure 3. Aurora B kinase targets Cdt1 for phosphorylation in vitro and affects Cdt1-MT binding. (A) Sequence alignment of Cdt1 from the indicated species using Clustal Omega. Highlighted in the red font are the potential Ser/Thr Aurora B sites. On the top, mentioned are the residues and their positions with respect to the full-length human Cdt1 (1–546 aa). (B) In vitro kinase assay with Aurora B alone or kinase plus ZM447439 inhibitor (10 µM) on HEK 293–purified Cdt1 or hVimentin (as a positive control). The autoradiogram (top) and Coomassie-stained gel (bottom) are shown. M, migration of molecular mass standards in kilodaltons on a 12% SDS-PAGE gel. (C) Pulldown of Aurora B by HA/His-tagged Cdt1-WT or cy-Cdt1 mutant that blocks cyclin A/Cdk binding (RRL [68–70] AAA), from thymidine synchronized and nocodazole arrested mitotic HeLa cell extracts. The pulldown was performed using Ni+2-NTA agarose beads followed by immunoblotting with either anti-HA or anti–Aurora B (AurB-K) antibodies. 1% of the lysate was loaded as total protein. HEK cells expressing GFP were used as a control. (D) HeLa cells were either treated with Aurora B inhibitor, ZM447439 (5 µM), for 1 h or left untreated (control) followed by release into MG132 (10 µM) containing medium in each case. The cells were immunostained with anti-Cdt1 (red), anti-tubulin (green) antibodies and counterstained with DAPI to mark the chromosomes (blue); scale bar, 5 µm. (E) n = 10 cells were quantified to obtain normalized Cdt1 fluorescence intensities on spindle MTs as depicted in D. P < 0.0001 represents statistical significance as assessed by the two-sided unpaired nonparametric Student’s t test. (F) Schematic representation showing the full-length Cdt1 (546 aa) and its deletion variant devoid of N-terminal 92 aa generated in fusion with GFP tag for expression in bacteria. The following two proteins generated represent Aurora B phosphomimetic (8D, Ser/Thr substituted with Asp) and phospho-deficient (8A, Ser/Thr substituted with Ala) Cdt1 mutants. Black asterisk, confirmation of phosphosites by mass spectrometry; red asterisk, residues that were inherently phosphorylated in Cdt1 obtained from HEK 293 cells. (G) Representative Western blot is shown. Samples fractionated as supernatant (S) and pellet (P) analyzed by Western blot probed with antibody against 6×-His tag to detect Cdt1 and stained with Ponceau S for tubulin. (H) Quantification (mean ± SD, n = 3) showing MT cosedimentation of purified Cdt192–546 or the mutant proteins (1 µM each) with the indicated concentrations of taxol-stabilized MTs (in μM). The 95% CI values obtained for each fit were: Bmax = 0.34–0.65, Kd = 0.43–10.57 for 8D; Bmax = 0.36–0.67, Kd = 0–3.22 for 8A. To validate this hypothesis, we first sought to test if Cdt1 is a substrate for Aurora B kinase using an in vitro phosphorylation assay. Under the assay conditions, both myc-tagged full-length hekCdt1 (purified from HEK 293 cells and procured commercially) and vimentin (an established Aurora B substrate [Goto et al., 2003]) were readily phosphorylated by Aurora B (Fig. 3 B). This phosphorylation was markedly reduced upon addition of ZM447439, an Aurora B inhibitor, attesting to its specificity. In tandem, hekCdt1 phosphorylated by Aurora B (using a nonradioactive ATP) was subjected to mass spectroscopy (MS) for identification of phosphosites. The tandem MS (MS/MS) spectra conclusively detected phosphorylation on four Aurora B consensus sites, T7, T102, S143, and T358, uniquely/exclusively in the kinase-treated hekCdt1; with 90% sequence coverage in two independent runs (Fig. S2). S73 and T270 were detected as phosphorylated even in the control hekCdt1 (without the Aurora B kinase), suggesting that these residues can potentially acquire phosphorylation in vivo. Although the S101 and S265 localized on the peptides that were phosphorylated, the exact location of the modified residues could not be designated. The remaining two noncanonical sites, T287 and T329, were unmodified. Interestingly, Ni+2-NTA agarose-mediated affinity precipitation of HA/His-tagged Cdt1 (from thymidine-synchronized and nocodazole-arrested mitotic cell extracts), but not GFP, led to the coprecipitation of Aurora B kinase as an interacting partner (Fig. 3 C). Moreover, the kinase was also coprecipitated with the HA-tagged Cdt1 mutant (RRL(68–70)AAA), which was incompetent to bind Cyclin/Cdk, equally efficiently as with the WT Cdt1, implying that the interaction between Cdt1 and Aurora B in mitosis is independent of Cdk phosphorylation (Fig. 3 C). To test the significance of Aurora B phosphorylation on the MT-binding of Cdt1 in vivo, we treated HeLa cells with Aurora B inhibitor (ZM 447439) and assessed the localization of Cdt1 on mitotic spindle MTs. We observed a 3.6-fold increase in spindle MT-localized Cdt1 in ZM-treated cells compared with the untreated control cells (Fig. 3, D and E). Consistent with this finding, upon mutating the eight putative Aurora B Ser/Thr residues (8D) to Asp in Cdt192–546 in an attempt to mimic constitutively phosphorylated Cdt1 (Fig. 3 F), we observed ∼2.5-fold reduction in the apparent MT binding as evidenced by an increase in the Kd from 2.03 ± 0.99 µM for the Cdt192–546 to 5.50 ± 2.41 µM for the 8D mutant (Fig. 3, G and H). Substitution of the residues to nonphosphorylatable Ala (8A) did not affect the MT binding, with Kd value comparable to the parental Cdt192–546. Aurora B–phosphomimetic mutations in Cdt1 affect kMT stability and mitotic progression To assess the role of Aurora B–mediated phosphorylation for Cdt1-MT binding in vivo, we investigated the function of the Aurora B phosphomimetic and nonphosphorylatable Cdt1 mutants in HeLa cells using an experimental strategy that had been established previously. This approach involves siRNA-based knockdown and rescue of Cdt1 function in double thymidine-synchronized cells specifically in the G2/M phase of the cell cycle to ensure that Cdt1 function in replication licensing during G1 phase is unperturbed (Varma et al., 2012). We first generated HeLa cells stably expressing HA-tagged WT and phosphomutant Cdt1 using a retroviral transduction-based approach. The stability of kMTs in cells that were depleted of endogenous Cdt1 but rescued with Cdt1 WT or phosphorylation site mutants was tested. Upon cold treatment, the expression of Cdt1-WT rescued kMT stability as opposed to that of the phosphomimetic mutant (Cdt1-10D) or the vector control (simulates Cdt1-depleted state; Fig. 4, A and B). However, not only did the nonphosphorylatable mutant (Cdt1-10A) maintain stable kMT attachments, but the quantitation of fluorescence intensities of spindle MTs suggested that the kMTs were also hyperstable in this case (Fig. 4, A and B). Under our conditions, efficient knockdown of endogenous Cdt1 was achieved while the expressed siRNA resistant HA-tagged Cdt1 proteins were retained (Fig. 4 C, top and bottom panels). The cells expressing either the WT or mutant proteins after endogenous Cdt1 depletion had apparently normal spindle structures at 37°C (Fig. S3 A) and no defect in localization of Hec1 to kinetochores (Fig. S3 B). Figure 4. Aurora B phosphorylation of Cdt1 regulates kMT stability and mitotic progression. (A) Synchronized HeLa cells treated with siRNA against endogenous Cdt1 in G2/M phase and rescued with stably expressing siRNA-resistant Cdt1-WT, 10D (phosphomimetic), 10A (unphosphorylated) proteins or vector control (abbreviated as Vec, simulating Cdt1 depleted state); were stained with antibodies against Zwint1 in red (1:400 dilution, kinetochore marker), MTs in green (1:500 dilution, spindle marker), and DAPI in black and white marking the chromosomes. Cells were placed at 4°C for 15 min before being fixed and stained. Scale bar, 5 µm. (B) Bar graph showing the quantification of MT staining intensity in each case after background subtraction (n = 10 cells). P < 0.0001 represents statistical significance as assessed by the two-sided unpaired nonparametric Student’s t test. (C) HeLa cells treated with siRNA against endogenous Cdt1 rescued with stably expressing siRNA-resistant Cdt1-WT, 10D (phosphomimetic), 10A (phosphodefective) proteins (with C-terminal HASH tags) or vector control (Vec) were lysed using 2× Laemmli buffer and electrophoresed on 10% SDS-PAGE gel. Representative Western blots from three independent experiments of synchronized HeLa cell lysates probed for expression levels of the indicated proteins with (bottom) or without (top) Cdt1 siRNA (200 nM). Expressed, ectopically expressed (with HASH tags); endogenous, endogenous (without tag) Cdt1; * in the bottom panel, detection of a nonspecific band reactive with the Cdt1 antibody. The same blot was reprobed with GAPDH (1:8,000) as a sample recovery control. (D) Quantification of the percentage of bi-attached, bi-oriented kinetochores in each of the conditions as indicated. Kinetochore marker, Zwint1, is in red, and MT is in green. (E) Interkinetochore (K-K) stretch was calculated between the two kinetochore pairs in each case and is plotted (n = 115 pairs; 10 cells). P < 0.001 represents statistical significance as assessed by the two-sided unpaired nonparametric Student’s t test. Cropped images of representative kinetochore pairs are also provided to indicate the status of their kMT attachment and K-K stretch; scale bars, 1 µm. (F) STLC washout assay was performed in double thymidine synchronized HeLa cells, treated with siRNA against endogenous Cdt1, and rescued with either Cdt1-WT or Cdt1-10A followed by treatment with STLC for 2 h. The cells were washed with and released into medium containing MG132 and fixed either immediately after the initiation of the washout (t = 0) or after 1 h (t = 60), followed by fixation and staining. DAPI is pseudocolored in red to mark chromosomes, and tubulin is in green. (G) Bar graph showing the quantification of monopolar (early prometaphase) versus bipolar spindle (metaphase) structures at the indicated time points after STLC washout. Cells from two coverslips in two independent experiments were counted; n = 2,114 and 2,659 for WT and 10A, respectively, for t = 60; n = 565 and 785 for WT and 10A, respectively, for t = 0. The data are plotted as percentage mitotic cells with monopolar or bipolar spindle structures on the y-axis. (H) Mitotic progression (metaphase-to-anaphase transition) was assessed in HeLa cells stably expressing vector, Cdt1-WT, Aurora B Cdt1-phosphomimetic (10D) or phospho-defective (10A) mutants at 11.5 and 12.5 h after release from thymidine treatment in the presence of Cdt1 siRNA. The number of cells in each stage of mitosis was counted and plotted from three different coverslips of two independent experiments to illustrate the nature of mitotic delay/arrest in Cdt1-10D mutant–expressing cells compared with cells rescued with Cdt1-WT or Cdt1-10A expression (n = 600 cells; a and b represent the statistical significance obtained using two-sided unpaired nonparametric Student’s t test performed on the WT versus 10D in the indicated stages of mitosis). (I) Representative images are shown from H; DAPI pseudocolored in red marks the chromosomes, and antibody against tubulin stained the MTs green. Approximately 60% of the cells analyzed in Cdt1-10D and 80% in vector control had no discernible MT structures after cold exposure; in the remaining cells that showed extant MT traces, the MTs were not attached to the kinetochores (Fig. 4 A, last panels in each vector and 10D). Quantification of the bi-oriented kinetochores that made contact with cold-stable MTs from opposite spindle poles revealed an ∼90% decrease in cells expressing phosphomimetic Cdt1 compared with cells expressing WT or phospho-defective Cdt1 (Fig. 4 D). The interkinetochore distance (K-K stretch), which serves as another indicator of the strength of kMT attachment, was also measured in each case as another parameter. The average K-K stretch in cells expressing Cdt1-WT and Cdt1-10A was 1.11 ± 0.09 and 1.16 ± 0.15 µm, respectively (Fig. 4 E), implying that kMTs from opposite spindle poles form robust contact with sister kinetochores. In contrast, the Cdt1-10D mutant exhibited a significant reduction in the distance between sister kinetochores (0.73 ± 0.10 µm) similar to the vector control, 0.72 ± 0.08 µm, indicating weakened load-bearing kMT attachments between kinetochore pairs (Fig. 4 E). Magnified images of kinetochore pairs are also shown to underscore the observed effects (i.e., reduction of the K-K distance and occurrence of mono-oriented or unattached KT pairs) in Aurora B phosphomimetic mutant and the vector control (Fig. 4 E). To conclusively address if expression of Cdt1-10A led to the production of hyperstable kMT attachments, we performed STLC (analogous to Monastrol, an inhibitor of Kinesin Eg5) washout assay to test the ability of cells to correct syntelic attachments (Kapoor et al., 2000). The data revealed that 1 h after drug washout, only ∼38% of cells expressing nonphosphorylatable Cdt1 mutant (10A) had attained bipolar spindle structures, in contrast to ∼76% of cells rescued with the WT-Cdt1 protein (Fig. 4, F and G). Based on the substantial delay in bipolarization of the mitotic spindles and accumulation of either monopolar spindles or early prometaphase-like spindle structures in 10A rescued cells, it is evident that there is some degree of kMT hyperstability associated with the expression of the Cdt1-10A mutant compared with cells rescued with WT-Cdt1. Progression through mitosis was monitored in each case by fixed-cell analyses. Cells were fixed at two different time points after release from thymidine arrest: mitotic onset (11.5 h) and anaphase onset (12.5 h; Varma et al., 2012). The mitotic progression of cells expressing Cdt1-10D was severely disrupted. Similar to that observed for vector control after the depletion of endogenous Cdt1 (Varma et al., 2012), the Cdt1-10D–expressing cells exhibited a phenotype resembling a mid- to late pro-metaphase arrest at the latter time point (Fig. 4, H and I). In this case, majority of the cells had chromosomes that were partially aligned around a vaguely defined metaphase plate (Fig. S3 A), while a smaller fraction of cells exhibited substantially higher levels of chromosome misalignment (not depicted). These phenotypes were strikingly different from the cells expressing the Cdt1-10A mutant, which underwent anaphase onset and cytokinesis similar to the cells rescued with Cdt1-WT. Although we did not demonstrate that these 10 sites in Cdt1 could be phosphorylated by Aurora B kinase in vivo, MS data unequivocally confirmed at least four target residues (T7, T102, S143, and T358) that were phosphorylated by Aurora B in vitro. Therefore, to evaluate whether mutating the four identified sites is sufficient to recapitulate some or all of the observed phenotypes (especially perturbation of kMT stability and mitotic progression), we generated HeLa cells stably expressing siRNA resistant versions of HA-tagged phosphomimetic (4D) and nonphosphorylatable (4A) mutant Cdt1 proteins. Using the aforementioned siRNA-based knockdown-rescue approach, kMT cold stability and fixed-cell mitotic progression analyses were conducted for these mutants as well. Results from the cold stability experiment indicate that cells rescued with the expression of Cdt1-4D mutant after the depletion of endogenous Cdt1 still exhibited substantial loss of kMT stability compared with the cells rescued with WT-Cdt1, even though the effects were not as severe as those observed in cells expressing Cdt1-10D mutant (Fig. 5 A). Consequently, the incidence of unattached or mono-oriented kinetochores after Cdt1-4D expression was intermediate/partial with respect to Cdt1-WT– and Cdt1-10D–expressing cells. Interestingly, the cells expressing Cdt1-4A also displayed a significant increase in the stability of kMTs compared with the Cdt1-WT, but to a significantly lesser extent than the cells expressing the Cdt1-10A mutant (Fig. 5 B). Mitotic progression analysis corroborates the above data, as cells rescued with Cdt1-4A/4D exhibited phenotypes of intermediate severity between the Cdt1-WT expression and the more penetrant, 10A/10D mutant expression (Fig. 5, C and D). As previously described, the cells were fixed at two different time points, mitotic onset (11.5 h) and anaphase onset (12.5 h). Although fewer mitotic cells expressing Cdt1-4A were able to enter anaphase compared with those expressing Cdt1-10A at 12.5 h, there was a significant increase in the number of mitotic cells expressing Cdt1-4D that were able to enter anaphase compared with the Cdt1-10D expressing cells (Fig. 5, C and D). In support of the in vivo analysis, an MT copelleting assay was also performed with the 3D mutant in which three of four identified sites in Cdt192–546 were substituted with Asp to mimic constitutively phosphorylated Cdt1 (Fig. 5 E). The apparent Kd of the 3D mutant for MTs was calculated to be 3.27 ± 0.66 µM, as opposed to 1.85 ± 0.34 µM for Cdt192–546, indicating an ∼1.5-fold reduction in the MT-binding ability of the 3D mutant (Fig. 5, F and G). Figure 5. Aurora B phospho mutants of Cdt1 at the four sites identified by mass spectroscopic analysis partially affect kMT stability and mitotic progression. Synchronized HeLa cells treated with siRNA against endogenous Cdt1 in G2/M phase and rescued with stably expressing siRNA-resistant Cdt1-WT, 4D (phosphomimetic), 4A (unphosphorylated) proteins or vector control (simulating Cdt1 depleted state) were subjected to 4°C treatment for 15 min before being fixed and stained. (A) Representative images of cells from C, immunostained with antibodies against Zwint1 (kinetochore marker) in red, tubulin (MTs) in green or in black and white on the left of the merged images, and DAPI in blue marking the chromosomes. Scale bar, 5 µm. (B) Bar graph showing the quantification of spindle MT staining intensity after cold treatment in each case after background subtraction (n = 7). P < 0.05 represents statistical significance as assessed by the two-sided unpaired nonparametric Student’s t test. (C) Mitotic progression (metaphase-to-anaphase transition) was gauged in HeLa cells stably expressing Cdt1-WT, Cdt1-10D, Cdt1-10A, Cdt1-4D, and Cdt1-4A mutants at 11.5 and 12.5 h after release from thymidine treatment in the presence of Cdt1 siRNA (200 nM). Representative images are shown from E; DAPI pseudocolored in red marks the chromosomes, and antibody against tubulin stained the MTs green; scale bar, 10 µm. (D) The number of cells in each stage of mitosis was counted and plotted from three different coverslips to illustrate the difference in mitotic delay in Cdt1-4A/4D mutant–expressing cells compared with cells rescued with Cdt1-WT or Cdt1-10A/10D expression (n = 300); * and #, P values showing the statistical significance among the indicated groups obtained using two-sided unpaired nonparametric Student’s t test. (E) Schematic representation showing the three indicated Aurora B sites mutated to generate a 3D phosphomimetic mutant in the Cdt192–546 parental background. (F) Representative Western blot from a MT-cosedimentation experiment performed with Cdt192–546 and Cdt1-3D mutant is also shown. Samples fractionated as supernatant (S) and pellet (P) analyzed by Western blot probed with antibody against 6×-His tag to detect Cdt1 (WT or 3D mutant) and stained with Ponceau S for tubulin. (G) Quantification (mean ± SD, n = 3) showing MT cosedimentation of purified Cdt192–546 or the mutant proteins (1 µM each) with the indicated concentrations of taxol-stabilized MTs (in μM). The 95% CI values obtained for each fit were: Bmax = 0.8–1.0, Kd = 1.22–2.90 for Cdt192–546; Bmax = 0.67–0.85, Kd = 2.25–4.9 for 3D mutant. We performed live-cell imaging to substantiate our data from fixed-cell mitotic progression. However, for this analysis, we chose Cdt1-10A and -10D mutants since they yielded a more complete and potent phenotype. The vector control and cells rescued with Cdt1-WT or -10A or -10D Aurora B mutant versions were subjected to live imaging after the knockdown of endogenous Cdt1 (Fig. 6 A and Videos 4, 5, 6, and 7). Hoechst was used to mark the chromosomes before the initiation of live imaging. Because there was no noticeable delay in the alignment of the majority of chromosomes in any of the four samples, we quantified the time elapsed between the initial establishment of the metaphase plate and the final fate of each of the imaged mitotic cells. We found that ∼100% of the Cdt1-WT rescued mitotic cells underwent normal anaphase onset without any delays (n = 26/27; Video 4), while >80% of the vector control (n = 22/27; Video 5) and the Cdt1-10D rescued cells (n = 23/28; Video 6) exhibited prolonged mitotic arrest with a metaphase-like chromosome alignment. In the latter two conditions, the aligned state of the chromosomes was (a) maintained for an extended period of time until the last time point of live imaging (outcome 1); (b) followed by anaphase onset with chromosome missegregation events (outcome 2); or (c) disrupted by chromosomal decondensation and reversion to an interphase-like stage (outcome 3). More importantly, vector control and the Cdt1-10D–expressing mitotic cells, on average, took approximately three times longer (∼200 min) to reach their final fate (outcomes 1, 2, and 3 above), with ∼15% cells never progressing into anaphase (outcome 1; Fig. 6 B and Videos 5 and 6). Even though 10A mutants progressed through mitosis comparably to the WT-rescued cells, >50% of these cells (n = 15/28; Video 7) underwent chromosome missegregation events during anaphase, which we speculate is a consequence of hyperstable kMT attachments. Figure 6. Live-cell imaging analysis of cells rescued with Aurora B phosphomimetic Cdt1 mutant reveals a severe delay in mitotic progression. (A) Scheme of the time course and still images obtained at the indicated time points during the imaging of live mitotic progression in different stable cell lines rescued with either the plasmid vector (analogous to Cdt1 depletion) or by WT or Aurora B mutant Cdt1 expression. Chromosomes are marked with Hoechst dye. (See also Videos 4, 5, 6, and 7.) (B) Quantification of A. Time (in min) elapsed between the establishment of a metaphase plate and the final fate of each of the imaged mitotic cells, which included delayed anaphase onset or a prolonged mitotic arrest, was plotted on the y-axis for each of the samples imaged live as indicated. Statistical significance assessed by the two-sided unpaired nonparametric Student’s t test. (C) Comparative tabulation of MT-binding parameters and phospho-regulation of relevant kinetochore-associated proteins from vertebrates. (D) Working model depicting the role of Cdt1 in providing an additional kMT-attachment site besides the CH and the N-terminal tail domain of the Hec1 subunit of the Ndc80 complex. The CH domains in Hec1 and Nuf2 along with the Hec1 unstructured tail bind MTs. The unstructured ∼40-aa loop region of Hec1 recruits Cdt1 by interacting with the N-terminal 1–320 aa of Cdt1 (Varma et al., 2012). This induces a conformational shift in the Ndc80 complex in a manner that is dependent on Cdt1 binding to MTs. In prometaphase, when kinetochores are not bi-oriented, Aurora B kinase is active to destabilize the erroneous attachments by phosphorylating several kinetochore proteins including Ndc80 (1), Ska (2), and Cdt1 (3) shown in the present study, all of which bind to MTs. Upon phosphorylation, these proteins exhibit reduced affinity for MTs. As observed for the Ska complex, which cannot interact with the KMN network effectively upon its phosphorylation by Aurora B (4); whether Cdt1~P can dock on to the Hec1 loop (5) with similar efficiency to unphosphorylated Cdt1 is still undetermined. In metaphase, when the Aurora B gradient diminishes at kinetochores, Cdt1 becomes proficient to bind to MTs (6), thus providing an additional site for kMT attachment besides the CH and the tail domains of Ndc80. Whether Cdt1 can interact with the Ska complex in its free or loop-bound state and if this interaction is dependent on their phosphorylation status (7 and 8) is yet to be determined. The model also depicts the possibility of interaction of other MAPs with the Hec1 loop, either alone or with the help of Cdt1 or Ska or both (9), as shown using dashed arrows. Discussion In this study, we demonstrate that the DNA replication licensing protein Cdt1, which is recruited to mitotic kinetochores by the Ndc80 complex (Varma et al., 2012), also binds to MTs directly in a manner that its affinity is controlled by Aurora B kinase phosphorylation. The affinity of Cdt1 for MTs was found to be comparable with other relevant kinetochore MAPs such as chTOG (Spittle et al., 2000), Ska (Schmidt et al., 2012), the KMN network, and the Ndc80 complex (Cheeseman et al., 2006; Ciferri et al., 2008) computed using similar approaches. Lack of contribution from the Cdt1’s unstructured N-terminal 92 aa in MT binding indicates that the mechanism by which Cdt1 binds to MTs is distinct from the Hec1 subunit of Ndc80, where a positively charged N-terminal 80-aa-long unstructured tail is essential for proper MT binding of the Ndc80 complex (Guimaraes et al., 2008). The fact that the C terminus of Cdt1 (410–546 aa) was not able to bind to MTs alone as effectively as in the presence of N terminus is reminiscent of the pattern obtained with the Ska complex, wherein the removal of C termini from either Ska1 or Ska3 impaired MT binding, but at the same time, the purified C-terminal domains of Ska1 or Ska3 by themselves did not associate with MTs (Jeyaprakash et al., 2012). This suggests that the N terminus of Cdt1 (1–410 aa) may be inert for MT binding by itself but might facilitate the folding or positioning of the C terminus of Cdt1 to allow MT binding. Moreover, we predict that WHC or WHM domains may be required in combination for efficient binding of Cdt1 to MTs. These observations also suggest that the WHD domain could serve as one of the conserved MT-binding domains in such proteins, as has already been demonstrated for the Ska complex (Schmidt et al., 2012; Abad et al., 2014). Further, it is interesting that although the N-terminal region of Cdt1 could not bind to MTs, it was competent enough to interact with Hec1; thus it seems that Cdt1 modular functions (Hec1 and MT binding) are independent of each other even though Ndc80 loop is essential for recruiting Cdt1 to the kinetochores. Additionally, our study demonstrates for the first time that Cdt1 is a substrate for Aurora B kinase and provides evidence for a physical interaction between the two proteins in vivo. However, in the pulldown experiment, we found that the amount of tagged Cdt1 retrieved/bound (Fig. 3 C, WB: HA, lanes 5 and 6) was much less compared with the “total” protein expressed (lanes 2 and 3). This is because the polyhistidine tag used to retrieve Cdt1 binds to nickel resin very specifically but extremely poorly in the incompatible buffer conditions used, essentially to preserve and retain Cdt1 binding partners. Nonetheless, endogenous Aurora B kinase was associated significantly above the background control in each case. In addition to Aurora B kinase, Cdt1 has been shown to be phosphorylated by cyclin-dependent kinase 1 (Cdk1) during the S-phase at threonine 29 (Fujita, 2006) and by MAPK p38 and JNK during G2/M phase (Chandrasekaran et al., 2011). Fairly recently, Ska3 has also been shown to undergo Cdk1-mediated phosphorylation that influences its interaction with the Ndc80 loop domain (Zhang et al., 2017). Mps1 kinase was shown to phosphorylate both Ska3 (Maciejowski et al., 2017) and the yeast Ndc80 complex (Kemmler et al., 2009), in addition to a physical interaction with human Ndc80 (Nilsson, 2015). For Hec1, although Aurora B is regarded as the master regulator, the involvement of Aurora A kinase in late mitosis has lately been uncovered (DeLuca et al., 2018). With this ever-growing evidence of spatio-temporal integration of several mitotic kinases to fine-tune kMT stability, attachment error correction, and mitotic progression, we envision a similar interplay of kinases in regulating mitotic functions of Cdt1, besides Aurora B. Nevertheless, we extended our study to understand the implications of Aurora B kinase–mediated phosphorylation on MT-binding ability of Cdt1. Indeed, our data demonstrated a correlation between the phosphorylation of Cdt1 by Aurora B kinase and MT binding. While the phosphomimetic Cdt1-10D mutant was severely compromised to sustain stable kMT attachments, the Cdt1-10A mutant, on the contrary, exhibited enhanced attachment stability. However, this hyperstability did not translate into any discernible chromosome alignment defects, unlike those observed after Hec1-9A mutant expression (DeLuca et al., 2011). This could be attributed to the specific role of Cdt1 during later stages of mitosis, i.e., at metaphase, where it facilitates stabilization of kMT attachments. This is in contrast to Hec1, where Aurora B–mediated Hec1 phosphorylation is actively involved in error correction during early mitosis. Thus, it is reasonable that hyperstable kMT attachments would lead to defective chromosome alignment in Hec1-9A mutant, but in the case of Cdt1, only translate to erroneous chromosome segregation later during anaphase onset. In an attempt to identify the exact functional Aurora B sites that can sufficiently mimic the observed phenotypes, we focused on only the four residues that were unambiguously identified by MS analysis. While our results confirm the vital contribution of those four Aurora B sites, phosphorylation on some or all of the other remaining sites seems requisite to attain the complete gamut of the phenotype pertinent to kMT stability. In light of the severity of phenotype obtained with the 10D mutation, the possibility of a complete loss of mitotic function for this protein cannot be completely ruled out. It is intriguing that these 10 predicted Aurora B sites are located within the N-terminal unstructured and central WHM (1–410 aa) of Cdt1 that did not show direct association with MTs but somehow seem to contribute to efficient Cdt1-MT interaction (Fig. 1, G and I). Because many of these sites are clustered in regions that are predicted to be disordered, we surmise that Aurora B phosphorylation might induce a conformational change in Cdt1 to regulate its MT-binding ability. This mode of regulation can exist in addition to or in parallel to the manner in which Aurora B phosphorylation generally modulates the affinity of binding; i.e., by interfering with the electrostatic interactions through introduction of multiple negative charges. Unlike Hec1, where multiple Aurora B target sites are grouped in tandem within the small unstructured N-terminal tail, sites within Cdt1 are distributed throughout. Thus we propose that one or both of the following could have elicited destabilization of kMT attachments leading to delayed mitotic progression upon Aurora B phosphorylation of Cdt1: (a) addition of multiple negative charges and (b) a conformational shift. The behavioral parameters and regulatory aspects of the key MT-binding proteins that participate in stabilizing kMT attachments (including Cdt1) at the vertebrate kinetochores are summarized in Fig. 6 C. Based on our observations, we propose a model in which kinetochore-localized Cdt1 provides another MT-binding site for the Ndc80 complex at its loop domain, thus shedding light on the importance of Cdt1 and the loop in kMT attachments (Fig. 6 D). We envisage that the loop domain of the Ndc80 complex, either alone or in conjunction with bound Cdt1, might also coordinate with other kinetochore MAPs to confer stronger attachments. Thus, future studies will be required to delineate the details of hierarchical recruitment of kinetochore components by the Ndc80 loop. Our work, while providing an essential missing link in the series of events that take place during the stabilization of kMT attachments in mitosis (Fig. 6 D), also opens up several stimulating questions such as (a) whether the Ndc80 complex has any synergistic influence on Cdt1 binding to MTs; (b) whether Ska and Cdt1 coordinate for robust kMT attachments, and if so, whether these proteins dock on to the loop as a preformed complex or interact after they are independently recruited at the loop; (c) whether Ska, Ndc80, and Cdt1 can generate a ternary complex for efficient MT binding; (d) whether chTOG, another Ndc80-recruited kinetochore MAP, influences localization of Cdt1 and its interaction with MTs; (e) whether there is cross-talk and/or interplay of several other kinases besides Aurora B kinase to fine-tune the phosphoregulation of Cdt1 in mitosis; and (f) how phosphorylation of Cdt1 with one or many mitotic kinases influences its localization to kinetochores and whether it affects recruitment of other kinetochore-associated proteins directly or indirectly. Answers to these questions will have major impact on the mitosis field in the quest for novel mechanisms that link mitotic kinetochores to spindle MTs and drive accurate chromosome segregation. Materials and methods Chemicals, reagents, cell lines, and bacterial strains Bacterial strains for cloning and expression were grown in Luria–Bertani (LB) broth or agar supplemented with 100 µg/ml ampicillin or 50 µg/ml kanamycin as appropriate. Enzymes for cloning were purchased from New England Biolabs. Custom oligonucleotides and G-blocks were from Integrated DNA Technologies. Sanger DNA sequencing was conducted at the Northwestern University Genomics core facility. DNA manipulations: Cloning and mutagenesis Cdt1 gene sequence was codon optimized and synthesized as a linear DNA strand (G-block, IDT) for subsequent cloning procedures. The largest Cdt1 fragment (92–546 aa), deletion variants (92–232, 92–410, and 410–546 aa), Aurora B phosphomimetic mutants (Ser101Asp, Thr102Asp, Ser143Asp, Ser265Asp; Thr270Asp, Thr287Asp, Thr329Asp, Thr358Asp abbreviated as 8D; and Ser101Asp, Ser143Asp, Thr358Asp abbreviated as 3D), and nonphosphorylatable mutants (Ser101Ala, Thr102Ala, Ser143Ala, Ser265Ala; Thr270Ala, Thr287Ala, Thr329Ala, Thr358Ala abbreviated as 8A) were cloned in fusion with the N-terminal GFP and 6x-His tags using SspI and Gibson cloning in a commercial plasmid procured from Addgene (pET His6 GFP TEV LIC cloning vector [1GFP], plasmid 29663). For stable transfections, the genes encoding Cdt1 with mutations that represent Aurora B phosphomimetic (Thr7Asp, Ser73Asp, Ser101Asp, Thr102Asp, Ser143Asp, Ser265Asp; Thr270Asp, Thr287Asp, Thr329Asp, Thr358Asp) and nonphosphorylatable (Thr7Ala, Ser73Ala, Ser101Ala, Thr102Ala, Ser143Ala, Ser265Ala; Thr270Ala, Thr287Ala, Thr329Ala, Thr358Ala) versions, abbreviated as 10D and 10A, respectively, harboring C-terminal 2x-HA, Strep-tag, 12x-His tag were cloned in pQCXIP plasmid using custom gene synthesis services (Thermo Fisher Scientific). In addition, Aurora B phosphomimetic (Thr7Asp, Ser101Asp, Ser143Asp, Thr358Asp) and nonphosphorylatable (Thr7Ala, Ser101Ala, Ser143Ala, Thr358Ala) versions, abbreviated as 4D and 4A, with the same tags as above were custom synthesized by Gene Universal. In silico analyses Secondary structure prediction The protocol was based on the work of Seeger et al. (2012). The full sequence of human Cdt1 (Uniprot ID: Q9H211) was input into 10 different secondary structure prediction servers to reduce the bias from any one program. The prediction of disordered regions is based on amino acid composition, charge/hydropathy index, known disordered sequences from crystal/NMR structures, and other empirical methods. The servers included were PSIPRED, IUPRED, PONDR, FoldIndex, s2D, ESpritz, PROFsec, MFD2p, DISCoP, and YASPIN. Relatively stringent cutoff of >75% within each algorithm was used to classify a residue as disordered. The results of each server were converted to a binary output of “folded/unknown” or “disordered” (0 or 1, respectively). The scores were summed and plotted as a function of amino acid number. Homology modeling The full sequence of human Cdt1 (Uniprot ID: Q9H211) was input into the Phyre2 (Kelley et al., 2015), I-TASSER (Roy et al., 2010), and SWISS-MODEL (Schwede et al., 2003) homology modeling servers. Models were aligned using the “cealign” plugin on Pymol. The C-terminal W was chosen to align all models. Crystal structures from human WHM domain (PDB ID: 2WVR; De Marco et al., 2009) and mouse WHC domain (PDB ID: 2KLO; Khayrutdinov et al., 2009) were used as templates for the WHDs. Pymol was used to generate protein structure images, and the final figures were generated in Adobe Illustrator. Recombinant protein expression and purification For MT copelleting assays, the fusion proteins along with the GFP protein (as a control) were purified from the soluble fraction after overexpression in BL21(λDE3) at 23°C for 18 h in LB medium, using nickel-affinity chromatography. The proteins purified to 85% purity were dialyzed in 20 mM Tris-HCl, pH 8.0, with 300 mM NaCl and stored at −20°C until further use. Full-length Cdt1 (1–546 aa) was purified as a GST-tagged protein using pGEX-5X-1 followed by tag removal per manufacturer’s instructions. Native tubulin was purified from porcine brains as previously described (Williams and Lee, 1982), flash frozen, stored in aliquots in liquid nitrogen, and clarified of aggregates by centrifugation at 100,000 g before use. Full-length human Cdt1 (1–546 aa) purified from HEK293 cells with a C-terminal Myc/DDK tag dialyzed in 25 mM Tris-HCl, pH 7.3, 100 mM glycine, and 10% glycerol was purchased from Origene (TP301657). For TIR-FM assays, GFP-tagged Cdt1, abbreviated as Cdt192–546, was induced in BL21(λDE3) cells at 30°C for 5 h in enriched terrific broth. Cell pellets were flash frozen in liquid nitrogen and stored at −80°C. Cell pastes were lysed via sonication and spun at 35,000 rpm for 45 min. The soluble fraction was incubated with 6 ml of 50% cobalt IMAC resin (TALON; Clontech) for 1 h at 4°C. The beads were washed with 100 ml wash buffer (25 mM Hepes, pH 7.4, 500 mM NaCl, 2.5 mM imidazole, 5% sucrose wt/vol, and 0.5 mM tris(2-carboxyethyl)phosphine [TCEP]) and eluted with 15 ml elution buffer (wash buffer plus 500 mM imidazole). The eluted protein was concentrated to ∼2.5 ml and injected over a Superdex 200 16/60 SEC column (GE Healthcare) in storage buffer (25 mM Hepes, pH 7.4, 500 mM NaCl, 0.5 mM TCEP, and 5% sucrose wt/vol). Purified, folded Cdt1 eluted at ∼65 ml and was monodisperse. The final protein was ∼95% pure via SDS-PAGE. Protein was flash frozen in liquid nitrogen and stored at −80°C until use. MT copelleting assay For MT-binding assay, tubulin was polymerized and stabilized in general tubulin (GT) buffer, BRB80 (80 mM Pipes, pH 6.8, 1 mM MgCl2, and 1 mM EGTA) with 100 mM GTP, 20 µM paclitaxel/taxol, and 66% glycerol (assay buffer). The proteins (diluted in BRB80 buffer containing 1 mg/ml BSA) were precleared of any aggregates or debris by ultracentrifugation at 70,000 rpm at room temperature for 30 min. The taxol-stabilized MTs (in varying micromolar concentrations) were incubated with indicated concentrations of precleared proteins or 2 µM GFP/GST alone to a final reaction volume of 40 µl. Reactions were incubated at room temperature for 20 min and ultracentrifuged for 30 min at 60,000 rpm in a Beckman TLA100 rotor at 25°C. The binding was also performed in the presence of salt by adding 300 mM NaCl to the assay buffer. Pellets (P) and supernatants (S) were separated carefully, equal volumes of SDS sample buffer were added, and samples were run on SDS-PAGE and analyzed by Western blotting with anti-His monoclonal antibody (1:5,000; Abcam) or Cdt1 polyclonal IgG raised in rabbit (1:2,000; H-300, sc-28262; Santa Cruz Biotech). Immunoblots were scanned into Adobe Photoshop, and any manipulations to the brightness were applied to the entire image. Densitometry quantification of the bands representing the arbitrary amount of protein in supernatants and pellets was done using ImageJ software (National Institutes of Health). The percentage binding was measured by calculating the ratio of protein in the pellet (P) and the total protein (S + P). The fraction of protein bound was then plotted against the concentration of MTs, and the data were fitted to the single-site saturation nonlinear regression curve (with no nonspecific binding but background and Kd > 0 constrain) using GraphPad Prism 6.0. The Kd values generated represent the average and propagated error from at least three independent experiments. For cosedimentation assays with subtilisin-treated MTs, 20 µM taxol-stabilized MTs were incubated for 30 min at 30°C with 200 µg/ml subtilisin A (Calbiochem), a serine protease that cleaves the C-terminal 10–20 aa from α- and β-tubulin. The reaction was stopped with 10 mM PMSF, and digested MTs were pelleted (90,000 rpm, TLA100, 10 min, 25°C) and resuspended in the original volume of GT buffer. The assay was then performed as described above. Curved tubulin oligomers or rings were generated by incubating tubulin with dolastatin (Tocris Bioscience) dissolved in DMSO in a 2:1 molar ratio for 1 h at room temperature in GT buffer supplemented with 6% DMSO. The rings were purified by ultracentrifugation through a cushion of GT buffer with 40% glycerol. The rings were resuspended in the original volume of GT buffer, and assay was performed as described above. Sheared MTs were prepared by passing taxol-stabilized MTs through a 26.5-gauge needle multiple times immediately before the addition to the protein sample. The generation of increased number of ends was confirmed by transmission electron microscopy. TIR-FM Flow chambers containing immobilized MTs were assembled as described previously (McKenney et al., 2014). Imaging was performed on a Nikon Eclipse Ti-E microscope equipped with an Andor iXon EM CCD camera; a 100×, 1.49-NA objective and 1.5× tube lens (yielding a pixel size of 106 nm), four laser lines (405, 491, 568, and 647 nm); and using Micro-Manager software (Edelstein et al., 2014). All assays were performed in assay buffer (30 mM Hepes, pH 7.4, 50 nM K-acetate, 2 mM Mg-acetate, 1 mM EGTA, and 10% glycerol), supplemented with 0.1 mg/ml biotin-BSA, 0.5% Pluronic F-168, and 0.2 mg/ml κ-casein. A final concentration of 1 and 5 nM Cdt192–546 was used. Protein samples from two independent purification batches were used to ensure biological reproducibility. Photobleaching tests (Helenius et al., 2006) revealed that the photobleaching rate of this construct under our conditions exceeded the characteristic dwell time by approximately threefold, and thus our dwell-time calculations do not take photobleaching into account. Dwell times were calculated manually from kymographs of individual MTs and plotted in GraphPad Prism 7.0. For measuring the diffusion coefficient, the TIR-FM imaging time series of Cdt192–546 were first bleach-corrected by histogram matching using ImageJ and saved as 8-bit multi-page TIFF files before being transferred to the DiaTrack 3.04 Pro particle tracking software (http://www.diatrack.org/), which was then used to identify and track GFP particles localized on MTs. The readout for diffusion coefficient obtained from the tracking software was in (pixel)2/s, and the time interval between individual time frames of the TIR-FM time series was ∼65 ms. Thus, each readout in (pixel)2/s was first multiplied by the pixel size (0.106 µm), and again by a factor of ∼15.4 (milliseconds to seconds conversion) to obtain final values in (µm)2/s. The particles that showed high diffusion were excluded for the calculation of the diffusion coefficient (D), along with the stationary particles. This was done because in the time frame used to conduct the analysis, we were unable to ascertain with accuracy whether these particles were indeed bound to and diffusing along MTs. In vitro phosphorylation assay In vitro kinase assay was performed on 1 µg each of hekCdt1 and human vimentin (ab73843, Abcam) at 30°C for 1 h in assay buffer containing 25 mM Hepes, 50 mM NaCl, 1 mM DTT, 2 mM EGTA, 5 mM MgSO4, 10 µM ATP, and 5 µCi γ-[32P]ATP (Perkin Elmer) in the presence of 0.5 µg Aurora B kinase (AURKB-231H; Creative BioMart) alone or with its inhibitor ZM447439 (Sigma; 5 or 10 µM as indicated). Samples (30 µl) were then electrophoresed on SDS-PAGE gel and visualized by autoradiography. The same gel was stained with Coomassie to ascertain migration of the proteins. Sample preparation for MS, liquid chromatography MS/MS, and data analysis Both nonradioactive kinase-treated and control samples were precipitated overnight at −20°C with 8 volumes of acetone and 1 volume of TCA. The protein was pelleted by centrifugation at 15,000 g for 15 min at 4°C. The pellets were washed twice with 200 µl cold acetone, dried, and resuspended in 8 M urea/0.4 M ammonium bicarbonate, then reduced with 4 mM DTT for 30 min at 50°C. Alkylation with 18 mM iodoacetamide was performed at room temperature for 30 min in the dark. The urea was diluted to less than 2 M with ultrapure water followed by addition of sequencing-grade trypsin (Promega) overnight at a ratio of 1:18 (enzyme:substrate). The peptides were desalted using C18 spin columns according to the manufacturer’s instructions. Samples were resuspended in 5% acetonitrile (ACN)/0.1% formic acid (FA) to a final concentration of 0.15 µg/µl. Nano-liquid chromatography MS/MS analyses were performed with a 75 µm × 10.5 cm PicoChip column packed with 3 µm Reprosil C18 beads. A 150 µm × 3 cm trap packed with 3 µm beads was installed in-line. Solvent A consisted of 0.1% FA in water, and solvent B was 0.1% FA in ACN. Peptides were trapped at 5 µl/min for 5 min, then separated at a flow rate of 300 nl/min with a gradient from 5% to 30% B in 95 min. After a 5-min ramp to 60% B, the column was washed at 95% B and reequilibrated to 5% B with a total analysis time of 120 min. Desalted phosphopeptide-enriched fractions were resuspended in 5% ACN/0.1% FA. 6 µl (0.9 µg) was injected in duplicate for each sample. The LC was coupled by electrospray to a Q-Exactive HF mass spectrometer operating in data-dependent MS/MS mode with a top-15 method. Dynamic exclusion was set to 60 s, and charge 1+ ions were excluded as well. MS1 scans were collected from 300–2,000 m/z with resolving power equal to 60,000 at 400 m/z. The MS1 AGC was set to 3 × 106. Precursors were isolated with a 2.0 m/z isolation width, and the HCD normalized collision energy was set to 35%. The MS2 AGC was set to 105 with a maximum ion accumulation time of 120 ms, and the resolving power was 30,000. MS raw files were searched against the human SwissProt database (version downloaded 8/2016, 20,191 sequences) using the Mascot search engine. Cysteine carbamidomethylation was a fixed modification, and the variable modifications were oxidized methionine, acetylation of protein N terminus, deamidation of Asn/Gln, and phosphorylation of Ser, Thr, and Tyr residues. Two missed cleavages were allowed. Statistical cutoffs and data visualization were accomplished using Scaffold Q+ software (Proteome Sciences). Proteins were selected with a 1% false discovery rate cutoff, and peptides with a 90% identification probability or better were considered. All phosphopeptide spectra were inspected manually to determine whether the phosphate group had been assigned to the correct amino acid. Cell culture methods Transient or stable transfections, immunofluorescence, and confocal microscopy HeLa or HEK 293 cells were cultured at 37°C with 5% CO2 in DMEM (Life Technologies) containing 10% FBS (Seradigm, VWR Life Science), 100 U/ml penicillin, and 1 µg/ml streptomycin. Before transfection, cells were seeded overnight onto 18-mm circular coverslips to achieve ∼60–70% confluence. The cells were then transfected with 0.5 µg pJC13-Cdt1-HASH plasmid (carrying C-terminal tags: 2x-HA, Strep-tag, 12x-His tag followed by a stop codon; gift from J. Cook, University of North Carolina, Chapel Hill, NC) using Effectene transfection reagent (Qiagen) according to the manufacturer’s protocol. After 36 h, the cells were fixed in PBS, pH 7.2, containing 4% formaldehyde followed by staining with antibody against the HA tag (1:100; H6908; Sigma), tubulin monoclonal antibody (1:500; T9026, clone DM1A; Sigma), and DAPI dihydrochloride (1:10,000; Life Technologies). Alexa Fluor 488–, Rhodamine Red-X–, or Cy5-labeled donkey secondary antibodies used at 1:250 dilution each were obtained from Jackson ImmunoResearch Laboratories. To generate stable cell lines, pQCXIP, a bicistronic retroviral expression vector, was used (Julius et al., 2000). Upon transfection into HEK-293T packaging cell line, this vector integrates and stably expresses a viral genomic transcript containing the CMV immediate early promoter, gene of interest, internal ribosome entry site (IRES), and the puromycin resistance gene (PurR). The gene of interest and the puromycin resistance gene are cotranscribed, via the IRES, as a bicistronic message (Adam et al., 1991; Goshima and Yanagida, 2000). HEK-293T cells were seeded at a density of 0.75 × 106 in a 10-cm dish. A three-plasmid transfection system was used that contains carrier pQCXIP plasmid (5 µg) expressing the gene of interest (cdt1-WT or 10D or 10A) along with the two helper plasmids pVSV-G, 1.25 µg (Clontech) and pCL-eco, 3.75 µg (Imagenex) encoding the Env and Gag/Pol proteins required for virus production. The HEK cells were transfected using calcium phosphate for 48 h, and the culture supernatant containing viral particles was harvested and passed through a 0.45-µm syringe filter. Fresh medium was added to the dish for a second round of collection at 72 h. Next, polybrene (8 µg/ml) was mixed with the filtered virus followed by addition of the virus to the target HeLa cells (seeded at a density of 6.5 × 104 in a 10-cm dish) for 6 h. Selection for the stably integrated cells was performed for 2 wk by addition of 1 µg/ml puromycin, a concentration determined to kill 100% of control untransduced HeLa cells. For image acquisition, the coverslips were mounted using ProLong Gold Antifade reagent (Invitrogen), and 3D stacks were obtained sequentially at 200-nm steps along the z-axis through the cell using a high-resolution inverted microscope (Eclipse TiE; Nikon) equipped with a spinning disk (CSU-X1; Yokogawa Corporation of America), an Andor iXon Ultra888 EMCCD camera, and a 60× or 100× 1.4-NA Plan-Apochromatic DIC oil immersion objective (Nikon). The images were acquired and processed using NIS Elements software from Nikon. siRNA transfections, double-thymidine synchronization, and cold stability assay HeLa cells were synchronized by treatment with 2 mM thymidine for 18 h followed by release for 9 h and then retreatment with 2 mM thymidine for 18 h. Synthetic duplexed RNA oligonucleotide (siRNA) against Cdt1 was transfected into HeLa cells according to the manufacturer’s instructions as described previously (Varma et al., 2012) during the second thymidine release and before fixing the cells at designated time points, as indicated in Fig. 4 its legend. Western blot was performed using anti-Cdt1 antibody (H-300; Santa Cruz Biotech) to evaluate the efficiency of Cdt1 knockdown. GAPDH was probed for after stripping the same Western blot as loading control. Other cell manipulations included cold treatment for 15 min with ice-cold PBS before fixation. STLC washout assay HeLa cells were double-thymidine synchronized, and siRNA against Cdt1 was added as described above. STLC, an analogue of monastrol (7 µM; Sigma) was added to the cells. After 2 h, cells were either fixed (t = 0) or washed twice with and released into fresh medium containing MG132 for 1 h (t = 60), before fixation and immunofluorescence staining. Live-cell imaging HeLa cell vector controls or those expressing either wt or Aurora B mutant versions of Cdt1 cultured on 35-mm glass-bottomed dishes (MatTek Corporation) were subjected to the same thymidine and Cdt1 siRNA treatment regimen described above and imaged live starting at 9 h after release from second thymidine treatment. The chromosomes were labeled with the live cell DNA dye, Hoechst (45 min, 2.5 µg/ml), before the initiation of live imaging. Just before imaging, the Hoechst-containing medium was replaced with prewarmed L-15 medium (Gibco) supplemented with 10% FBS and fresh Cdt1 siRNA mix. Live imaging was carried out using an incubation chamber for microscopes (Tokai Hit Co.) at 37°C and 5% CO2. Images were recorded using a high-resolution inverted microscope (Eclipse TiE; Nikon) equipped with a spinning disk (CSU-X1; Yokogawa Corporation of America), an Andor iXon Ultra888 EMCCD camera, and a 60× 1.4-NA Plan-Apochromatic DIC oil-immersion objective (Nikon) fitted with an objective heater. Four to six 1.5-µm-separated Z-sections covering the entire volume of the cell were collected every 10 min for up to 12 h. Image processing was performed using NIS Elements. Ni2-NTA agarose-mediated pulldown HEK-293T cells were transfected with the plasmids (Cdt1-HASH or cy-Cdt1 mutant, RRL [68–70] AAA, which prevents Cyclin/Cdk binding, in pQCXIP and pEGFP-N1 as control) for 4 h at 20–30% confluence using PEI max transfection reagent. M-phase synchronization was done by treating the cells with 2 mM thymidine for 18 h, followed by addition of 100 ng/ml (0.33 µM) nocodazole for 10 h. The cells were collected by mitotic shake-off and lysed using lysis buffer (50 mM Hepes, pH 8.0, 33 mM KAc, 117 mM NaCl, 20 mM imidazole, 0.1% Triton X-100, 10% glycerol, 0.1 mM 4-benzenesulfonyl fluoride hydrochloride, 10 µg/ml pepstatin A, 10 µg/ml aprotinin, 10 µg/ml leupeptin, 1 mM ATP, 1 mM MgCl2, 5 µg/ml phosvitin, 1 mM β-glycerol-phosphate, and 1 mM orthovanadate). The whole-cell lysates expressing the His-tagged proteins were incubated with Ni+2-NTA beads for 3 h with end-on rotation at 4°C. Beads were washed three times with the lysis buffer, and bound proteins were eluted by boiling for 5 min in 40 µl of 2× SDS sample buffer. For immunoblotting, proteins were electrophoresed on SDS-PAGE gels and transferred to PVDF membranes. Immunoblots were developed using chemiluminescence and exposed on x-ray films. Antibodies against Aurora B kinase and HA-tag were procured from Abcam (ab2254) and Sigma (H6908), respectively. Statistical analysis GraphPad Prism software (version 7.03) was used for determining statistical significance of the data obtained. A standard nonparametric unpaired two-tailed/sided Student’s t test was used. This test assumes that both groups of data are sampled from Gaussian populations with the same standard deviation. Data distribution was assumed to be normal, but this was not formally tested. Online supplemental material Fig. S1 A shows the topology of a conventional WTH domain; Fig. S1 (B and C) shows that control GST and GFP tags do not bind to MTs; Fig. S1 (D–F) shows Cdt1 binding to straight, curved, and sheared MTs; Fig. S1 G is the TIR-FM image of Cdt192–546 decorating MTs; and Fig. S1 H demonstrates Cdt1 colocalization with MTs in HeLa cells. In Fig. S2 are the chromatograms from phosphoproteomics showing phosphorylation of full-length Cdt1 (1–546 aa) by Aurora B kinase at the indicated residues. Fig. S3 (A and B) demonstrates normal mitotic spindle structure in cells rescued with Aurora B kinase phospho mutants of Cdt1 at 37°C and the localization of Hec1 in these cells. Video 1 shows diffusion of Cdt1 on MTs in TIR-FM. Videos 2 and 3 provide Z-series data of Cdt1-HA colocalization on MTs and kinetochores. Videos 4–7 are live imaging analysis of HeLa cells rescued with Aurora B kinase phospho mutants of Cdt1 after labeling the chromosomes with DNA dye, Hoechst. Supplementary Material Supplemental Materials (PDF) Video 1 Video 2 Video 3 Video 4 Video 5 Video 6 Video 7 Acknowledgments We thank Dr. Jeanette Cook for the HA-tagged Cdt1 expression plasmid and for valuable intellectual input; Dr. Sarah Rice for constructive suggestions and discussions; Drs. Laimonis A Laimins and Kavi Mehta for providing help with the radioactive experiments; Drs. Arshad Desai and Dhanya Cheerambathur for providing Hec1::Nuf2-His-tagged dimeric complex; Northwestern Genomics Core for DNA sequencing services; and Suchithra Seshadrinathan and Dr. Anita Varma for technical support. This work was supported by National Cancer Institute grant R00CA178188 to D. Varma and National Institute of Neurological Disorders and Stroke grant R00NS089428 and National Institute of General Medical Sciences grant R35GM124889 to R.J. McKenney. The authors declare no competing financial interests. Author contributions: S. Agarwal and D. Varma designed and performed majority of the experiments. K.P. Smith assisted in Cdt1 purification, carried out structural and computational studies, and provided intellectual input. Y. Zhou performed the Cdt1-Aurora B kinase pull down experiment. A. Suzuki performed endogenous Cdt1 antibody staining. R.J. McKenney performed single-molecule TIR-FM experiments. S. Agarwal, R.J. McKenney, and D. Varma analyzed the data. S. Agarwal and D. Varma wrote the manuscript with input from all the authors. ==== Refs Abad, M.A., B. Medina, A. Santamaria, J. Zou, C. Plasberg-Hill, A. Madhumalar, U. Jayachandran, P.M. Redli, J. Rappsilber, E.A. Nigg, and A.A. Jeyaprakash. 2014. Structural basis for microtubule recognition by the human kinetochore Ska complex. Nat. Commun. 5 :2964. 10.1038/ncomms3964 24413531 Adam, M.A., N. Ramesh, A.D. Miller, and W.R. Osborne. 1991. Internal initiation of translation in retroviral vectors carrying picornavirus 5′ nontranslated regions. J. Virol. 65 :4985–4990.1651417 Afreen, S., and D. Varma. 2015. Cell Division: Molecular Pathways for KMN Kinetochore Recruitment. Curr. Biol. 25 :R332–R335. 10.1016/j.cub.2015.02.041 25898103 Chan, Y.W., A.A. Jeyaprakash, E.A. Nigg, and A. Santamaria. 2012. Aurora B controls kinetochore-microtubule attachments by inhibiting Ska complex-KMN network interaction. J. Cell Biol. 196 :563–571. 10.1083/jcb.201109001 22371557 Chandrasekaran, S., T.X. Tan, J.R. Hall, and J.G. Cook. 2011. Stress-stimulated mitogen-activated protein kinases control the stability and activity of the Cdt1 DNA replication licensing factor. Mol. Cell. Biol. 31 :4405–4416. 10.1128/MCB.06163-11 21930785 Cheeseman, I.M., and A. Desai. 2008. Molecular architecture of the kinetochore-microtubule interface. Nat. Rev. Mol. Cell Biol. 9 :33–46. 10.1038/nrm2310 18097444 Cheeseman, I.M., J.S. Chappie, E.M. Wilson-Kubalek, and A. Desai. 2006. The conserved KMN network constitutes the core microtubule-binding site of the kinetochore. Cell. 127 :983–997. 10.1016/j.cell.2006.09.039 17129783 Ciferri, C., S. Pasqualato, E. Screpanti, G. Varetti, S. Santaguida, G. Dos Reis, A. Maiolica, J. Polka, J.G. De Luca, P. De Wulf, 2008. Implications for kinetochore-microtubule attachment from the structure of an engineered Ndc80 complex. Cell. 133 :427–439. 10.1016/j.cell.2008.03.020 18455984 DeLuca, J.G., and A. Musacchio. 2012. Structural organization of the kinetochore-microtubule interface. Curr. Opin. Cell Biol. 24 :48–56. 10.1016/j.ceb.2011.11.003 22154944 DeLuca, K.F., S.M. Lens, and J.G. DeLuca. 2011. Temporal changes in Hec1 phosphorylation control kinetochore-microtubule attachment stability during mitosis. J. Cell Sci. 124 :622–634. 10.1242/jcs.072629 21266467 DeLuca, K.F., A. Meppelink, A.J. Broad, J.E. Mick, O.B. Peersen, S. Pektas, S.M.A. Lens, and J.G. DeLuca. 2018. Aurora A kinase phosphorylates Hec1 to regulate metaphase kinetochore-microtubule dynamics. J. Cell Biol. 217 :163–177. 10.1083/jcb.201707160 29187526 De Marco, V., P.J. Gillespie, A. Li, N. Karantzelis, E. Christodoulou, R. Klompmaker, S. van Gerwen, A. Fish, M.V. Petoukhov, M.S. Iliou, 2009. Quaternary structure of the human Cdt1-Geminin complex regulates DNA replication licensing. Proc. Natl. Acad. Sci. USA. 106 :19807–19812. 10.1073/pnas.0905281106 19906994 Edelstein, A.D., M.A. Tsuchida, N. Amodaj, H. Pinkard, R.D. Vale, and N. Stuurman. 2014. Advanced methods of microscope control using μManager software. J. Biol. Methods. 1 :e10. 10.14440/jbm.2014.36 25606571 Espeut, J., D.K. Cheerambathur, L. Krenning, K. Oegema, and A. Desai. 2012. Microtubule binding by KNL-1 contributes to spindle checkpoint silencing at the kinetochore. J. Cell Biol. 196 :469–482. 10.1083/jcb.201111107 22331849 Fujita, M. 2006. Cdt1 revisited: complex and tight regulation during the cell cycle and consequences of deregulation in mammalian cells. Cell Div. 1 :22. 10.1186/1747-1028-1-22 17042960 Goshima, G., and M. Yanagida. 2000. Establishing biorientation occurs with precocious separation of the sister kinetochores, but not the arms, in the early spindle of budding yeast. Cell. 100 :619–633. 10.1016/S0092-8674(00)80699-6 10761928 Goto, H., Y. Yasui, A. Kawajiri, E.A. Nigg, Y. Terada, M. Tatsuka, K. Nagata, and M. Inagaki. 2003. Aurora-B regulates the cleavage furrow-specific vimentin phosphorylation in the cytokinetic process. J. Biol. Chem. 278 :8526–8530. 10.1074/jbc.M210892200 12458200 Guimaraes, G.J., Y. Dong, B.F. McEwen, and J.G. Deluca. 2008. Kinetochore-microtubule attachment relies on the disordered N-terminal tail domain of Hec1. Curr. Biol. 18 :1778–1784. 10.1016/j.cub.2008.08.012 19026543 Helenius, J., G. Brouhard, Y. Kalaidzidis, S. Diez, and J. Howard. 2006. The depolymerizing kinesin MCAK uses lattice diffusion to rapidly target microtubule ends. Nature. 441 :115–119. 10.1038/nature04736 16672973 Hsu, K.S., and T. Toda. 2011. Ndc80 internal loop interacts with Dis1/TOG to ensure proper kinetochore-spindle attachment in fission yeast. Curr. Biol. 21 :214–220. 10.1016/j.cub.2010.12.048 21256022 Jeyaprakash, A.A., A. Santamaria, U. Jayachandran, Y.W. Chan, C. Benda, E.A. Nigg, and E. Conti. 2012. Structural and functional organization of the Ska complex, a key component of the kinetochore-microtubule interface. Mol. Cell. 46 :274–286. 10.1016/j.molcel.2012.03.005 22483620 Joglekar, A.P., K.S. Bloom, and E.D. Salmon. 2010. Mechanisms of force generation by end-on kinetochore-microtubule attachments. Curr. Opin. Cell Biol. 22 :57–67. 10.1016/j.ceb.2009.12.010 20061128 Julius, M.A., Q. Yan, Z. Zheng, and J. Kitajewski. 2000. Q vectors, bicistronic retroviral vectors for gene transfer. Biotechniques. 28 :702–708.10769748 Kapoor, T.M., T.U. Mayer, M.L. Coughlin, and T.J. Mitchison. 2000. Probing spindle assembly mechanisms with monastrol, a small molecule inhibitor of the mitotic kinesin, Eg5. J. Cell Biol. 150 :975–988. 10.1083/jcb.150.5.975 10973989 Kelley, L.A., S. Mezulis, C.M. Yates, M.N. Wass, and M.J. Sternberg. 2015. The Phyre2 web portal for protein modeling, prediction and analysis. Nat. Protoc. 10 :845–858. 10.1038/nprot.2015.053 25950237 Kemmler, S., M. Stach, M. Knapp, J. Ortiz, J. Pfannstiel, T. Ruppert, and J. Lechner. 2009. Mimicking Ndc80 phosphorylation triggers spindle assembly checkpoint signalling. EMBO J. 28 :1099–1110. 10.1038/emboj.2009.62 19300438 Khayrutdinov, B.I., W.J. Bae, Y.M. Yun, J.H. Lee, T. Tsuyama, J.J. Kim, E. Hwang, K.S. Ryu, H.K. Cheong, C. Cheong, 2009. Structure of the Cdt1 C-terminal domain: conservation of the winged helix fold in replication licensing factors. Protein Sci. 18 :2252–2264. 10.1002/pro.236 19722278 Lee, C., B. Hong, J.M. Choi, Y. Kim, S. Watanabe, Y. Ishimi, T. Enomoto, S. Tada, Y. Kim, and Y. Cho. 2004. Structural basis for inhibition of the replication licensing factor Cdt1 by geminin. Nature. 430 :913–917. 10.1038/nature02813 15286659 Maciejowski, J., H. Drechsler, K. Grundner-Culemann, E.R. Ballister, J.A. Rodriguez-Rodriguez, V. Rodriguez-Bravo, M.J.K. Jones, E. Foley, M.A. Lampson, H. Daub, 2017. Mps1 Regulates Kinetochore-Microtubule Attachment Stability via the Ska Complex to Ensure Error-Free Chromosome Segregation. Dev. Cell. 41 :143–156.e6. 10.1016/j.devcel.2017.03.025 28441529 Maure, J.F., S. Komoto, Y. Oku, A. Mino, S. Pasqualato, K. Natsume, L. Clayton, A. Musacchio, and T.U. Tanaka. 2011. The Ndc80 loop region facilitates formation of kinetochore attachment to the dynamic microtubule plus end. Curr. Biol. 21 :207–213. 10.1016/j.cub.2010.12.050 21256019 McKenney, R.J., W. Huynh, M.E. Tanenbaum, G. Bhabha, and R.D. Vale. 2014. Activation of cytoplasmic dynein motility by dynactin-cargo adapter complexes. Science. 345 :337–341. 10.1126/science.1254198 25035494 Meraldi, P., R. Honda, and E.A. Nigg. 2004. Aurora kinases link chromosome segregation and cell division to cancer susceptibility. Curr. Opin. Genet. Dev. 14 :29–36. 10.1016/j.gde.2003.11.006 15108802 Nilsson, J. 2015. Mps1-Ndc80: one interaction to rule them all. Oncotarget. 6 :16822–16823. 10.18632/oncotarget.4837 26219336 Pozo, P.N., and J.G. Cook. 2016. Regulation and Function of Cdt1; A Key Factor in Cell Proliferation and Genome Stability. Genes (Basel). 8 :2. 10.3390/genes8010002 28025526 Ramey, V.H., H.W. Wang, Y. Nakajima, A. Wong, J. Liu, D. Drubin, G. Barnes, and E. Nogales. 2011. The Dam1 ring binds to the E-hook of tubulin and diffuses along the microtubule. Mol. Biol. Cell. 22 :457–466. 10.1091/mbc.e10-10-0841 21169562 Roy, A., A. Kucukural, and Y. Zhang. 2010. I-TASSER: a unified platform for automated protein structure and function prediction. Nat. Protoc. 5 :725–738. 10.1038/nprot.2010.5 20360767 Schmidt, J.C., and I.M. Cheeseman. 2011. Chromosome segregation: keeping kinetochores in the loop. Curr. Biol. 21 :R110–R112. 10.1016/j.cub.2010.12.030 21300272 Schmidt, J.C., H. Arthanari, A. Boeszoermenyi, N.M. Dashkevich, E.M. Wilson-Kubalek, N. Monnier, M. Markus, M. Oberer, R.A. Milligan, M. Bathe, 2012. The kinetochore-bound Ska1 complex tracks depolymerizing microtubules and binds to curved protofilaments. Dev. Cell. 23 :968–980. 10.1016/j.devcel.2012.09.012 23085020 Schwede, T., J. Kopp, N. Guex, and M.C. Peitsch. 2003. SWISS-MODEL: An automated protein homology-modeling server. Nucleic Acids Res. 31 :3381–3385. 10.1093/nar/gkg520 12824332 Seeger, M.A., Y. Zhang, and S.E. Rice. 2012. Kinesin tail domains are intrinsically disordered. Proteins. 80 :2437–2446. 10.1002/prot.24128 22674872 Spittle, C., S. Charrasse, C. Larroque, and L. Cassimeris. 2000. The interaction of TOGp with microtubules and tubulin. J. Biol. Chem. 275 :20748–20753. 10.1074/jbc.M002597200 10770946 Tien, J.F., N.T. Umbreit, D.R. Gestaut, A.D. Franck, J. Cooper, L. Wordeman, T. Gonen, C.L. Asbury, and T.N. Davis. 2010. Cooperation of the Dam1 and Ndc80 kinetochore complexes enhances microtubule coupling and is regulated by aurora B. J. Cell Biol. 189 :713–723. 10.1083/jcb.200910142 20479468 Umbreit, N.T., D.R. Gestaut, J.F. Tien, B.S. Vollmar, T. Gonen, C.L. Asbury, and T.N. Davis. 2012. The Ndc80 kinetochore complex directly modulates microtubule dynamics. Proc. Natl. Acad. Sci. USA. 109 :16113–16118. 10.1073/pnas.1209615109 22908300 Varma, D., and E.D. Salmon. 2012. The KMN protein network--chief conductors of the kinetochore orchestra. J. Cell Sci. 125 :5927–5936. 10.1242/jcs.093724 23418356 Varma, D., S. Chandrasekaran, L.J. Sundin, K.T. Reidy, X. Wan, D.A. Chasse, K.R. Nevis, J.G. DeLuca, E.D. Salmon, and J.G. Cook. 2012. Recruitment of the human Cdt1 replication licensing protein by the loop domain of Hec1 is required for stable kinetochore-microtubule attachment. Nat. Cell Biol. 14 :593–603. 10.1038/ncb2489 22581055 Vavouri, T., J.I. Semple, R. Garcia-Verdugo, and B. Lehner. 2009. Intrinsic protein disorder and interaction promiscuity are widely associated with dosage sensitivity. Cell. 138 :198–208. 10.1016/j.cell.2009.04.029 19596244 Wang, H.W., S. Long, C. Ciferri, S. Westermann, D. Drubin, G. Barnes, and E. Nogales. 2008. Architecture and flexibility of the yeast Ndc80 kinetochore complex. J. Mol. Biol. 383 :894–903. 10.1016/j.jmb.2008.08.077 18793650 Welburn, J.P., E.L. Grishchuk, C.B. Backer, E.M. Wilson-Kubalek, J.R. Yates III, and I.M. Cheeseman. 2009. The human kinetochore Ska1 complex facilitates microtubule depolymerization-coupled motility. Dev. Cell. 16 :374–385. 10.1016/j.devcel.2009.01.011 19289083 Welburn, J.P., M. Vleugel, D. Liu, J.R. Yates III, M.A. Lampson, T. Fukagawa, and I.M. Cheeseman. 2010. Aurora B phosphorylates spatially distinct targets to differentially regulate the kinetochore-microtubule interface. Mol. Cell. 38 :383–392. 10.1016/j.molcel.2010.02.034 20471944 Westhorpe, F.G., and A.F. Straight. 2013. Functions of the centromere and kinetochore in chromosome segregation. Curr. Opin. Cell Biol. 25 :334–340. 10.1016/j.ceb.2013.02.001 23490282 Williams, R.C. Jr., and J.C. Lee. 1982. Preparation of tubulin from brain. Methods Enzymol. 85 (Pt B ):376–385. 10.1016/0076-6879(82)85038-6 7121276 Zaytsev, A.V., J.E. Mick, E. Maslennikov, B. Nikashin, J.G. DeLuca, and E.L. Grishchuk. 2015. Multisite phosphorylation of the NDC80 complex gradually tunes its microtubule-binding affinity. Mol. Biol. Cell. 26 :1829–1844. 10.1091/mbc.e14-11-1539 25808492 Zhang, G., C.D. Kelstrup, X.W. Hu, M.J. Kaas Hansen, M.R. Singleton, J.V. Olsen, and J. Nilsson. 2012. The Ndc80 internal loop is required for recruitment of the Ska complex to establish end-on microtubule attachment to kinetochores. J. Cell Sci. 125 :3243–3253. 10.1242/jcs.104208 22454517 Zhang, Q., S. Sivakumar, Y. Chen, H. Gao, L. Yang, Z. Yuan, H. Yu, and H. Liu. 2017. Ska3 Phosphorylated by Cdk1 Binds Ndc80 and Recruits Ska to Kinetochores to Promote Mitotic Progression. Curr. Biol. 27 :1477–1484.e4. 10.1016/j.cub.2017.03.060 28479321

PMC006xxxxxx/PMC6219728.txt

==== Front J Cell Biol J Cell Biol jcb jcb The Journal of Cell Biology 0021-9525 1540-8140 Rockefeller University Press 30185604 201803105 10.1083/jcb.201803105 Research Articles Article 38 10 43 Retrograde regulation of mossy fiber axon targeting and terminal maturation via postsynaptic Lnx1 Lnx1 stabilizes EphB receptors for axon targeting Liu Xian-Dong 12 Zhu Xiao-Na 1 Halford Michael M. 3 Xu Tian-Le 12 Henkemeyer Mark 3 http://orcid.org/0000-0003-2147-3352 Xu Nan-Jie 124 1 Collaborative Innovation Center for Brain Science, Department of Anatomy and Physiology, Shanghai Jiao Tong University School of Medicine, Shanghai, China 2 Department of Biochemistry and Molecular Cell Biology, Shanghai Key Laboratory for Tumor Microenvironment and Inflammation, Institute of Medical Sciences, Shanghai Jiao Tong University School of Medicine, Shanghai, China 3 Department of Neuroscience, Kent Waldrep Center for Basic Research on Nerve Growth and Regeneration, University of Texas Southwestern Medical Center, Dallas, TX 4 Key Laboratory of Cell Differentiation and Apoptosis of the Chinese Ministry of Education, Shanghai Jiao Tong University School of Medicine, Shanghai China Correspondence to Nan-Jie Xu: xunanjie@sjtu.edu.cn Michael M. Halford’s present address is Peter MacCallum Cancer Centre, Victorian Comprehensive Cancer Centre, Melbourne, Australia. 05 11 2018 217 11 40074024 20 3 2018 25 6 2018 14 8 2018 © 2018 Liu et al. 2018 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). Synapse formation relies on the coordination of dynamic pre- and postsynaptic structures during brain development. Liu et al. reveal that presynaptic terminal maturation of mossy fiber axons is retrogradely regulated by postsynaptic scaffold protein Lnx1 via stabilizing EphB receptor kinases. Neuronal connections are initiated by axon targeting to form synapses. However, how the maturation of axon terminals is modulated through interacting with postsynaptic elements remains elusive. In this study, we find that ligand of Numb protein X 1 (Lnx1), a postsynaptic PDZ protein expressed in hippocampal CA3 pyramidal neurons, is essential for mossy fiber (MF) axon targeting during the postnatal period. Lnx1 deletion causes defective synaptic arrangement that leads to aberrant presynaptic terminals. We further identify EphB receptors as novel Lnx1-binding proteins to form a multiprotein complex that is stabilized on the CA3 neuron membrane through preventing proteasome activity. EphB1 and EphB2 are independently required to transduce distinct signals controlling MF pruning and targeting for precise DG-CA3 synapse formation. Furthermore, constitutively active EphB2 kinase rescues structure of the wired MF terminals in Lnx1 mutant mice. Our data thus define a retrograde trans-synaptic regulation required for integration of post- and presynaptic structure that participates in building hippocampal neural circuits during the adolescence period. Graphical Abstract National Basic Research Program of China2014CB965002 National Natural Science Foundation of China https://doi.org/10.13039/501100001809 91232704 31671062 Shanghai Brain-Intelligence Project 16JC1420500 Shanghai Science and Technology Committee https://doi.org/10.13039/501100003399 16JC1420500 Shanghai Institutions of Higher Learning2013-25 Shanghai Science and Technology Committee https://doi.org/10.13039/501100003399 11DZ2260200 NIH https://doi.org/10.13039/100000002 MH066332 ==== Body pmcIntroduction Proper wiring of the developing brain relies on the dynamic formation of synapses (Cohen-Cory, 2002; Turrigiano and Nelson, 2004; Kolodkin and Tessier-Lavigne, 2011). The development of these specific synapses requires the accurate coordination of multiple developmental events, including axon targeting and pruning, dendritic growth, spinogenesis, and synapse formation (Ackley and Jin, 2004; Jüttner and Rathjen, 2005; Waites et al., 2005; Low and Cheng, 2006). Accumulating evidence has indicated that synapse formation and stabilization are dynamically modulated through the pre- and postsynaptic compartments in an anterograde, retrograde, or bidirectional way (Yuste and Bonhoeffer, 2004; Alvarez and Sabatini, 2007; McAllister, 2007; Bhatt et al., 2009; Shen and Scheiffele, 2010; Siddiqui and Craig, 2011; Sala and Segal, 2014). Extensive studies over the past decades both in vivo and in vitro have demonstrated that the presynaptic compartment plays a dominant role in initializing these processes, especially in the activity- or experience-dependent neuronal connection (Zucker, 1999; Hensch, 2005; Holtmaat and Svoboda, 2009; Kerschensteiner et al., 2009; Kozorovitskiy et al., 2012; Li et al., 2017). The basic connectivity during development is likely through an integration of cell-intrinsic genetic programs and extrinsic influences of guidance cues, neurotrophic factors, and neuronal and synaptic adhesion systems (McAllister, 2007; Chen et al., 2008; Giagtzoglou et al., 2009; Shen and Cowan, 2010; Shen and Scheiffele, 2010; Siddiqui and Craig, 2011; Südhof, 2012; Bennett and Lagopoulos, 2014; Sala and Segal, 2014). Numerous molecules or molecular families, including receptors and adhesion proteins, kinases and small GTPases, and cytoskeletal regulators, interact with various scaffold proteins containing PDZ domains during the formation of functional synapses (Garner et al., 2000, 2002; Sheng and Sala, 2001; Kim and Sheng, 2004; Feng and Zhang, 2009; Sheng and Kim, 2011; Sala and Segal, 2014). Although a large number of PDZ proteins have been identified as participating in postsynaptic morphogenesis, including dendritic development and spinogenesis (El-Husseini et al., 2000; Penzes et al., 2001; Hoogenraad et al., 2005; Nakamura et al., 2011; Geiger et al., 2014; Heisler et al., 2014), these studies are largely restricted in postsynaptic compartments. Previous studies have shown that presynaptic structure and function are also regulated in a retrograde way (Contractor et al., 2002; Jüngling et al., 2006; Regalado et al., 2006; Futai et al., 2007; Hu et al., 2012; Orr et al., 2017), whereas the precise mechanism of how postsynaptic PDZ scaffolds participate in the maturation of presynaptic structure remains relatively less investigated. The dentate mossy fiber (MF)-CA3 synapse in the hippocampus is an excellent model to study the dynamic formation of synaptic structures and neural circuits. The MF axons are composed of two distinct bundles, suprapyramidal bundle (SPB) and infrapyramidal bundle (IPB), which target CA3 neurons. The IPB undergoes a pruning process during the postnatal developing period (Bagri et al., 2003; Xu and Henkemeyer, 2009; Riccomagno et al., 2012) that make it easy to observe the coordinative change with postsynaptic remodeling on a large scale. The MF-CA3 synapses are represented as a large multiheaded morphology composed of highly plastic MF presynaptic terminals with massive separate vesicle release sites and thorny postsynaptic structures that are different from typical glutamatergic asymmetric synapses (Amaral and Dent, 1981; Chicurel and Harris, 1992; Nicoll and Schmitz, 2005; Rollenhagen et al., 2007). This specific axon structure is advantageous for the examination of terminal targeting and maturation with postsynaptic dynamics during postnatal development. In this study, we identify a PDZ scaffold protein, ligand of Numb protein X (Lnx1), which is expressed specifically in the hippocampal CA3 neurons. Through gene targeting in mice, we demonstrate that Lnx1 is required for targeting and remodeling of presynaptic MF axon terminals that wire with postsynaptic spines to form efficient synapses. We further demonstrate that CA3-expressed EphB receptors serve as novel Lnx1-interacting proteins responsible for MF terminal refinement and maturation during MF-CA3 synapse formation. Constitutively active EphB2 receptor kinase in Lnx1−/− mice is sufficient to rescue the presynaptic structure of MF. Thus, our data indicate that presynaptic axon targeting and terminal maturation can be controlled by postsynaptic elements through a trans-synaptic regulation in hippocampus. Results Lnx1 is expressed in CA3 neurons and required for MF axon pruning Lnx1 mRNA has been identified in hippocampal CA3 neurons (Rice et al., 2001), as confirmed by the Allen Brain Atlas (Fig. S1 A). To examine whether PDZ scaffold protein Lnx1 is important for the development of hippocampus in vivo, we generated a protein-null mutant in which the Lnx1 gene was knocked out through homologous recombination in embryonic stem cells. This results in deletion of coding exons needed for both P70 and P80 isoforms of Lnx1 and insertion of LacZ sequences to express β-galactosidase (β-gal), allowing for detection of Lnx1 expression (Fig. 1 A). The Lnx1 knockout was validated by Southern blot analysis using external 5′ and 3′ probes, PCR genotyping using gene-specific oligonucleotides, and protein detection using anti-Lnx1 antibodies (Fig. S1, B–D). Embryos containing the Lnx1 mutation were stained for β-gal expression using X-gal staining and showed strong expression in many tissues such as eyes, ears, and limbs (Fig. S1 E). The Lnx1 null homozygotes (Lnx1−/−) were viable at expected Mendelian ratios, appeared to be healthy, were fertile, and lived until adulthood. In view of the relatively low level of Lnx1 in brain compared with periphery organs (Lenihan et al., 2014), we precisely checked the Lnx1 expression from postnatal week 1 (PW1) until adulthood by immunoprecipitation of β-gal protein from hippocampal lysates of Lnx1 mutant mice (Fig. S2 A). Immunofluorescence with anti–β-gal antibodies in Lnx1 mutant mice indicated specific expression of Lnx1 in the hippocampal CA3 pyramidal neurons (Fig. 1 B). To examine the subcellular localization of Lnx1, we purified postsynaptic density (PSD) fractions from hippocampal tissues and immunoprecipitated with anti-Lnx1 to detect Lnx1 protein, and found that Lnx1 is expressed only in the postsynaptic fraction (Fig. S2 B). To validate the postsynaptic expression of Lnx1, we overexpressed Flag-Lnx1 into cultured hippocampal neurons and observed postsynaptic localization of Lnx1 in dendritic spines (Fig. S2 C). Our analysis thus identifies Lnx1 as a hippocampal CA3-specific postsynaptic protein in the adolescent brain. Figure 1. Defective MF axon pruning in Lnx1 null mice. (A) The schematic indicates the strategy for generating Lnx1 knockout by substituting Lnx1 with LacZ gene. 5′ untranslated sequences and exon coding sequences are shown as gray and dark boxes, respectively, and introns and 5′ and 3′ nontranscribed regions are shown as lines. EcoRI (E) and NcoI (N) restriction sites are indicated. The Lnx1-LacZ targeting vector, including a PGK-Neo cassette, was designed to replace Lnx1 exons including the p80 and p70 promoters after homologous recombination. The 5′ and 3′ external probes used to confirm the results by Southern blot are shown with the expected sizes indicated (see Fig. S1 B). (B) Immunofluorescence of β-gal in Lnx1 mutant mice indicated specific expression of Lnx1 in hippocampal CA3 pyramidal cell layer. Bars: 200 µm (left); 50 µm (right). (C) NeuroTrace dye–labeled CA3 pyramidal cells in 3-wk-old Lnx1−/− mice are loosely packed compared with WT littermates. Calbindin staining showed that the IPB axon–penetrated CA3 pyramidal neurons layer are much longer in Lnx1−/− mice compared with WT littermates. White brackets delineate IPB length, and distance between arrowheads delineates the IPB length in CA3 pyramidal neurons. Bars: 200 µm (left); 100 µm (right). (D) Quantification of the length ratio of IPB to CA3 area (from the hilus to the curvature) in Lnx1−/− mice and WT littermates during postnatal development. n = 5–7 mice. (E) Comparison of IPB length within CA3 pyramidal neurons in Lnx1−/− mice and WT littermates during postnatal development. n = 5–7 mice. (F) Diagram of coculture assay. (G) TdT+ primary hippocampal axons (arrowheads) cocultured with Lnx1−/− neurons showed fewer pruned axons than with WT neurons. Bar, 100 µm. (H) Comparison for neurite lengths in neuronal coculture with WT or Lnx1−/− neurons. (I) Comparison of primary neurite numbers in neuronal coculture with WT or Lnx1−/− hippocampal neurons. n = 72–403 neurons per group in H and I. (J) Comparison of longest neurite length (arrowheads) of calbindin-negative neurons and calbindin-positive neurons dissected from TdT+ hippocampi cocultured with WT or Lnx1−/− hippocampal neurons. n = 16–19 neurons per group. Bar, 100 µm. Means ± SEM; two-way ANOVA with Tukey’s post hoc test (J) or Student’s t test (D, E, and I); *, P < 0.05; **, P < 0.01; ***, P < 0.001. The specific localization of Lnx1 prompted us to determine whether hippocampal morphogenesis is altered in juvenile Lnx1−/− mice. The absence of Lnx1 did not cause an obvious change in the overall structure of the hippocampus, as viewed by staining with the immunofluorescent dye NeuroTrace, to label neurons (Fig. 1 C). According to calbindin immunofluorescence in WT mice, most of the MF axons from the dentate gyrus were seen in the SPB above the CA3 pyramidal cell bodies, whereas a smaller group of IPB axons grew initially underneath the CA3 pyramidal cells but then were pruned back to their mature length and joined the SPB axons as the hippocampus developed in WT mice as reported previously (Bagri et al., 2003; Xu and Henkemeyer, 2009; Riccomagno et al., 2012). Strikingly, we observed that IPB axons in the Lnx1−/− mice grew into split CA3 pyramidal layers that were not shortened in a timely manner during development and instead maintained an inappropriate long and stable length comparable with the SPB until adulthood (Fig. 1, C–E), suggesting defective axon pruning and targeting during a critical development period. Because Lnx1 was expressed restrictively in the CA3 pyramidal neurons but not in dentate granule (DG) cell neurons, the defective pruning and targeting of MF originating from granule cells in Lnx1−/− mice indicates a non–cell-autonomous mechanism. Two possibilities that account for the abnormalities could be considered: one could be that Lnx1 affects postsynaptic structure to mediate axon development upon axon–cell/dendrite contact; the other is that Lnx1 alters extracellular environmental factors that could affect axonal growth, targeting, and retraction in a secreted gradient manner. To test these possibilities, an in vitro coculture assay was developed. After a 12–14-d preculture of hippocampal neurons from Lnx1−/− or WT mice to allow for neuron distribution and contact on the dish, tdTomato–positive (TdT+) primary hippocampal neurons from TdT+ knock-in mice were plated on the culture by either direct addition or loading with coverslips onto the dish to prevent direct axon–cell contact (Fig. 1 F). The primary neurites with TdT fluorescence were analyzed for axon length at the indicated time points. We found that in the cultures of WT or Lnx1−/− hippocampal neurons, the plated TdT+ neurites underwent an initial fast growth by the first 3 d, resembling the early growth of MF axons, and then reached a relatively stable period for elongation and branching, which was followed by a later trimming back of most primary neurites (Fig. 1, G–I). However, neurons cocultured with Lnx1−/− neurons maintained obviously longer neurites than WT neurons from day 4 of coculture (Fig. 1, G–I), while neurons plated on coverslips showed no difference between the two groups (Fig. S3). We further compared the pruning of calbindin-negative and calbindin-positive neurons in the cocultured system (Fig. S4 A) and found that calbindin-positive neurons had even shorter neurites than the calbindin-negative neurons when cocultured with WT neurons, while both remained unchanged when cocultured with Lnx1−/− neurons (Fig. 1 J), indicating that the majority of DG cells become pruned in the coculture. These results suggest that the non–cell-autonomous regulation by Lnx1 on MF axon development is likely in a manner of axon–cell/dendrite contact in developing hippocampus. Retrograde coordination of pre- and postsynaptic arrangement by Lnx1 We then investigated the development of MF axons projected from DG neurons to clarify whether the presynaptic structure was altered in Lnx1−/− mutants. To see with more detail about the morphological changes of axonal terminals in Lnx1−/− mice, brain slices from 3-wk-old WT mice or Lnx1−/− mice were immunostained with anti-ZnT3, a protein marker for the MF axon terminals. In WT brains, robust axon terminals formed along the SPB, while little was found in IPB adjacent to the CA3 pyramidal cell layer. In contrast, the MF axons in Lnx1−/− mice showed a dramatic increase in IPB axon terminals that had contact neurons within the CA3 pyramidal cell layer (Fig. 2 A). We next asked whether the increased MF axon terminals as we observed in Lnx1−/− mice are anatomically matched with the spine morphogenesis that led to functional synapses. By using a Thy1-GFP-M transgenic reporter (Feng et al., 2000), we observed increased spine density but reduced mushroom-shaped spines of CA3 neurons in PW3 Lnx1−/− mice compared with control littermates (Fig. S4 B). We then examined the MF axon terminals that contact CA3 neurons and analyzed the percentage of these terminals on spines (terminal-spine) and dendritic shaft (terminal-shaft; Fig. 2, B and C). We observed a decreased ratio of terminal-spine/terminal-shaft in Lnx1−/− mice compared with control littermates (Fig. 2 D). We then classified the spines in CA3 neurons into two categories: spines overlapped with ZnT3-labeled terminals (ZnT3+ synapse) and spines separated from the ZnT3+ terminals, and we observed a decreased ratio of ZnT3+ synapses/total synapses in Lnx1−/− mice (Fig. 2, B, C, and E). Figure 2. Defective MF terminal targeting in Lnx1-null mice. (A) ZnT3 and NeuN staining showed robust expanded IPB axon terminal field and more terminal-contacted CA3 pyramidal cells in Lnx1−/− mice compared with WT littermates. Bars: 200 µm (top); 30 µm (bottom). PCL, pyramidal cell layer. n = 7 mice per group. (B) Representative images showing spines of CA3 neuron (green) contacting terminals of MF axon (red) in PW3 Lnx1−/− mice and control littermates. The higher magnification deconvolved 3D-rendered views (right) are as shown from regions of white box (left arrowhead). The overlapped spines and ZnT3+ terminals are defined as ZnT3+ synapses. Arrows indicate terminals on shaft. Bars: 100 µm (left); 2.5 µm (right). (C) Schematic showing terminals on spine (Terminal-Spine) or shaft (Terminal-Shaft). (D and E) Quantification of ratio for terminals on spine or shaft and ZnT3+ synapses determined from 3D-rendered views. n = 14–16 neurons from three to four mice for per group. (F and G) Representative and percentage of four types of pre- and postsynaptic dynamics in coculture system. Red labels, presynaptic axons; green labels, postsynaptic dendritic branches or spines. Arrowheads indicate contacted boutons or spines. Dotted line marks the area of bouton. *, AI ≥ 0.2; ns, AI < 0.2. n = 13–17 neurons. (H) The areas of newborn boutons are smaller when cocultured with Lnx1−/− neurons (n = 17 boutons) than with WT neurons (n = 23 boutons). (I) Quantification of the dynamic areas of refined boutons in coculture system. n = 33–47 boutons per group. Means ± SEM; two-way ANOVA with Tukey’s post hoc test (I) or Student’s t test (A, D, E, G, and H); *, P < 0.05; **, P < 0.01; ***, P < 0.001. To further characterize the role of Lnx1 for axon targeting and maturation, we performed live imaging in a dual color–labeled coculture system to observe the temporal dynamics of the axon terminals and matched spines. We quantitated the morphological changes and calculated an area index (AI) to classify these changes into four types of alteration: presynaptic, postsynaptic, both, and none (Fig. 2 F and Videos 1, 2, 3, 4, and 5). Compared with WT neurons, we observed an increased rate in none type but a decreased rate in presynaptic type when cocultured with Lnx1−/− neurons, while the rates of postsynaptic and both types remained unchanged (Fig. 2, F and G). We further classified the presynaptic type into two subtypes: newborn bouton and refined bouton. We found that newborn boutons showed greater expanding terminals in contacted with WT spines than with Lnx1−/− spines (Fig. 2 H). In contrast, the refined boutons underwent rapid terminal shrinkage and split to match with the contacting spines when cocultured with WT neurons, and this effect was attenuated when cocultured with Lnx1−/− neurons (Fig. 2 I). These results indicate that axon targeting and maturation are coordinated with the spine structure by Lnx1 in a retrograde manner. Lnx1 is essential for MF terminal maturation and release probability To determine the targeting specificity of MF axon terminals to spines of CA3 pyramidal neurons, we further validated the synapse formation by transmission EM to quantify the axon terminals and their matched spines. The unique morphological characteristics of the MF synapses consist of giant presynaptic boutons and multi-invaginating spines that are distinct from typical synapses (Fig. 3 A), the massive asymmetric synapses frequently observed in the adjacent area of projected MF axons and cell layer of CA3 pyramidal neurons, particularly the SPB region (∼90% of total synapses) and IPB region (∼20% of total synapses). We saw numerous large and complex presynaptic boutons containing massive vesicles and multiple vesicle release sites embracing the contacted spines in both strains of mice (Fig. 3 A). To evaluate the presynaptic terminal maturation that is associated with vesicle amount, we measured the numbers of vesicle release sites (indicated by PSDs) and docked vesicles in the presynaptic active zone, respectively. We found that both numbers were significantly reduced in Lnx1−/− mice compared with WT littermates (Fig. 3, A and B). Figure 3. Defective presynaptic maturation and function in Lnx1-null mice. (A) Schematic diagram shows synapse with giant bouton (left). Red, presynaptic terminal; green, postsynaptic spine. The docked vesicles are indicated as vesicle-contacted presynaptic membrane. Transmission electron micrographs of SPB regions from PW3 mice show MF terminals (red) and postsynaptic spines (green) in synapses with giant bouton (right). Asterisks mark PSD. The higher-magnification views show docked vesicles. Bars, 0.5 µm (left); 0.1 µm (right). (B) The PSD number per spine and docked vesicles per vesicle release site in MF synapses from SPB (301–312 synapses) or IPB (38–43 synapses) region of WT mice were more than in Lnx1−/− mice. n = 4 mice per group. (C) Schematic diagram shows typical synapse with regular bouton (left). Transmission electron micrographs from PW3 mice show axon terminals (red) and postsynaptic spines (green) in typical synapses (right). Asterisks mark PSD. Red dotted line indicates the presynaptic boutons without observed postsynaptic spine (nonsynaptic). Bars, 0.5 µm. (D) The number of vesicles per terminal (r2 = 0.62; P < 0.001) or areas of vesicles per terminal (r2 = 0.66; P < 0.001) in typical synapses were significantly correlated with PSD length in PW3 WT mice but not in Lnx1−/− mice. A dot presents a synapse. (E) The vesicle density in terminal that formed typical synapse in WT mice was more than that in Lnx1−/− mice. (F) A cumulative frequency plot of average vesicle distance from WT (n = 133 synapses) and Lnx1−/− (n = 165 synapses) mice with histogram distribution fit for the inset. n = 4 mice per group in D–F. (G) Schematic diagram in the upper panel shows EPSC recordings of the MF-CA3 pathway. Stimulating electrode was placed in the SPB layer, and recording pipette was placed in the CA3 area (between dotted line). Stim, stimulating electrode; Rec, recording pipette. Representative average traces (left) and summary graph (right) show PPRs at interstimulus intervals of 25, 50, 100, 200, and 400 ms in PW3 mice. Bar, 50 ms. n = 24–27 neurons from four mice per group. Means ± SEM; two-way ANOVA with Tukey’s post hoc test (E) or Student’s t test (B, F, and G); *, P < 0.05; **, P < 0.01; ***, P < 0.001. In addition to synapses with giant boutons, we observed a number of typical synapses with regular boutons in the region (Fig. 3 C). We examined the maturation of these presynaptic compartments with the modification of postsynaptic structures in Lnx1−/− mice by measuring the number and distributed areas of vesicles in axon terminals as well as their connecting postsynaptic profile area, indicated by PSD length, with EM. We plotted the two presynaptic factors, respectively, versus PSD length and observed a strong positive correlation between the vesicle number/distribution area and PSD length in WT mice, while the correlation was diminished in Lnx1−/− mice (Fig. 3 D). We also examined the vesicle density in presynaptic terminals with (synaptic) or without (nonsynaptic) postsynaptic compartments and the distribution of these vesicles by calculating the average distance of vesicles to the presynaptic membrane, and observed a decreased vesicle density in synaptic terminals and an increased distance of the vesicles to membrane in Lnx1−/− mice (Fig. 3, C, E, and F). These results indicate that loss of Lnx1 causes abnormal development of presynaptic axon terminals projected to CA3 region. We then assayed paired-pulse ratios (PPRs) of MF-CA3 synapses by stimulating SPB or IPB fibers that were measured as the amplitude ratio of the second to the first evoked excitatory postsynaptic current (EPSC) in CA3 neurons to evaluate function of presynaptic release. Compared with control mice, we observed an increased PPR upon either SPB or IPB stimulation in PW3 Lnx1−/− mice (Figs. 3 G and S4 C), indicating an impaired glutamate release probability. These results suggest that Lnx1 integrates postsynaptic structure dynamics and presynaptic terminal maturation for precise connection of functional synapses. EphB receptors are stabilized on membrane through interaction with Lnx1 We then screened transmembrane molecules that might be involved upon MF-CA3 neuron contact and analyzed expression of numerous proteins that have been predicted to interact with Lnx1 (Wolting et al., 2011; Guo et al., 2012). These proteins involve families of EphBs, connexins, claudins, cadherins, and striatin, which were reported to mediate cell–cell interactions and spinogenesis (Söhl et al., 2005; Benoist et al., 2006; Giagtzoglou et al., 2009). We detected members of these proteins for each family that are highly expressed in the hippocampus and found a significant reduction of EphB receptor proteins, a family of tyrosine receptor kinases, in both total cell lysates and membrane components extracted from Lnx1−/− hippocampus (Figs. 4 A and S5 A). Both EphB1 and EphB2 have been demonstrated to express in CA3 neurons, but not in MF, and serve as binding receptors to their membrane-expressed ligand ephrin-B3, which is expressed specifically in MF axons (Xu and Henkemeyer, 2009). In contrast with EphBs, the amount of other transmembrane proteins was not affected by loss of Lnx1 (Figs. 4 A and S5 A). Figure 4. Lnx1 interacts with postsynaptic EphB receptors. (A) The expression of membrane proteins in hippocampus from PW3 Lnx1−/− and WT mice was detected by Western blot and quantified. n = 3 mice per group. (B) Coimmunoprecipitation of Lnx1 and EphB1/2 in synaptosomes of hippocampus. Lnx1 was pulled down by EphB2 antibody from PSD fraction of either WT or EphB2K661R/K661R mice but neither EphB2−/− nor EphB2ΔVEV/ΔVEV mice. ns, nonspecific band; IB, immunoblot. (C) Schematic representation of p70-Lnx1 and its mutant constructs. (D) Immunoprecipitation (IP) of Lnx1 mutants and EphB1 or EphB2 coexpressed in NG108 cells. Lnx1 interacted with EphB1 through the N terminus and with EphB2 through the second PDZ domain. (E) Colocalization of Lnx1 and EphB2 (arrowheads in the right panels) in CA3 pyramidal cell layer (arrow in the left panel) were determined by immunostaining in PW3 WT mice. The higher-magnification views in white boxes indicate superresolution imaging of Lnx1 and EphB2. White dotted lines indicate neuronal boundary. Bars: 250 µm (left); 5 µm (bottom right); 0.5 µm (top right). Means ± SEM. Student’s t test (A); ***, P < 0.001. To determine whether Lnx1 directly interacts with EphB receptors, we pulled down Lnx1 with EphB1 or EphB2 antibody from protein lysates of WT hippocampal tissues compared with EphB1 or EphB2 mutant tissues. We found that Lnx1 could be pulled down by EphB1 or EphB2 antibodies from lysates of WT hippocampal tissues or by EphB2 antibody from EphB2K661R/K661R tissues, a kinase-dead EphB2 mutant (Genander et al., 2009), but not from tissues of EphB2−/−; EphB2LacZ/LacZ, a truncated mutant in which the intracellular domain was replaced with β-gal (Henkemeyer et al., 1996), or EphB2ΔVEV/ΔVEV, a mutant with PDZ domain-binding motif disrupted (Fig. S5 B; Genander et al., 2009). These results indicate that the C-terminal PDZ domain–binding motif is necessary for binding with Lnx1. We further purified the synaptosomes of hippocampal tissues and found that Lnx1 could be pulled down by EphB2 antibodies from PSD but not non-PSD fractions in WT and EphB2K661R/K661R tissues, indicating that Lnx1 interacts with EphB2 in postsynaptic compartments (Fig. 4 B). To clarify the binding sites of Lnx1, we generated Flag-tagged individual PDZ domain mutations of the P70-Lnx1 isoform, which was found to be expressed specifically in the brain (Dho et al., 1998; Xie et al., 2001), and cotransfected them with plasmids encoding HA-tagged EphB1/2 for immunoprecipitation. We found that different domains of Lnx1 were required for binding with the different subtypes of EphB receptors. Lnx1 interacted with EphB1 through the N terminus and with EphB2 through the second PDZ domain (Fig. 4, C and D). To further test colocalization of Lnx1 and EphB2 in CA3 pyramidal neurons, immunostaining was used in hippocampal sections and superresolution imaging with stimulated emission depletion (STED) techniques were performed to visualize the localization of these proteins. The STED imaging resolved discrete puncta containing both Lnx1 and EphB2 around the membrane of CA3 neurons (Fig. 4 E). To figure out whether the decreased level of EphB proteins in Lnx1−/− mice was the result of mRNA reduction in the hippocampus, quantitative RT-PCR was performed, and no change was observed in Lnx1−/− mice (Fig. S5 C). To further determine whether Lnx1 influences the degradation of EphB receptors, we treated cultured Lnx1−/− hippocampal neurons with proteasome inhibitor or lysosomal inhibitors and observed that the decreased EphB1 and EphB2 in Lnx1−/− mutants were restored to the normal level after treatment of proteasome inhibitor MG132, while lysosomal inhibitor leupeptin or NH4Cl had no effect (Fig. 5 A), suggesting an involvement of the proteasome pathway in EphB degradation. Figure 5. Lnx1 sustains stability of EphB receptors. (A) Analysis of degradation of EphB1 and EphB2 after addition of inhibitors of proteasome or lysosome in primary hippocampal neurons from WT and Lnx1−/− mice. *, EphB1; #, EphB2. Quantitative results of three biological replicates are shown. (B) Analysis of endogenous EphB2 protein level in MEF cells infected with lentivirus containing mock, Lnx1, or Lnx1-△PDZ2 in the presence or absence of proteasome inhibitor MG132. Quantitative results of three biological replicates are shown. (C) Analysis of EphB protein levels in primary hippocampal neurons from Lnx1−/− mice infected with lentivirus containing mock, Lnx1, Lnx1-△NT, Lnx1-△PDZ1, or Lnx1-△PDZ2. Quantitative results of four biological replicates are shown. Means ± SEM; two-way ANOVA with Tukey’s post hoc test (A and C) or one-way ANOVA with Tukey’s post hoc test (B); *, P < 0.05; **, P < 0.01; ***, P < 0.001; ###, P < 0.001. IB, immunoblot. To determine whether the binding with Lnx1 is essential for the stable expression of EphB receptors, we infected MEF cells that express endogenous EphB2 protein with lentivirus expressing Lnx1 or Lnx1 mutant lacking the second PDZ domain, which was necessary for Lnx1 binding with EphB2. We observed an obvious increased EphB2 protein level infected with Lnx1 compared with control virus, while deleting the second PDZ domain led to a decreased level of EphB2 protein, and this could be reversed by addition of MG132 (Fig. 5 B). Furthermore, the reduced EphB protein levels in Lnx1−/− primary neurons were also restored to normal levels by infecting with Lnx1 virus, while the protein levels of EphB1 or EphB2 were not rescued by the Lnx1 virus lacking the N terminus or the second PDZ domain, respectively (Fig. 5 C). Thus, these data suggest that Lnx1 stabilizes EphB receptors through specific binding sites to prevent their degradation in proteasome. In contrast with the P70-Lnx1 isoform, the other P80-Lnx1 isoform expressed in periphery tissues has identical PDZ domains but an additional RING domain for E3 ligase activity (Dho et al., 1998; Xie et al., 2001). To investigate whether P80-Lnx1 has a similar role in the stability of the EphB receptor, we overexpressed P70-Lnx1 and P80-Lnx1 into MEF cells, respectively, and observed an increased level of EphB2 protein with P70-Lnx1 but a decreased level with P80-Lnx1. These results indicate that the two Lnx1 isoforms play opposite roles in the expression level of EphB2, owing to their difference in the RING domain (Fig. S5 D). Activating EphB2 kinase promotes MF terminal maturation in Lnx1−/− mice EphB receptors have been reported to be required for spine morphogenesis and synapse formation in CA3 pyramidal neurons in vivo (Henkemeyer et al., 2003). We thus analyzed EphB1−/− and EphB2−/− protein–null mutant mice for MF phenotype compared with Lnx1-null. We found that longer MF axons penetrated CA3 pyramidal cell layers in both mutants (Fig. 6, A–D), which resembled the phenotype observed in Lnx1−/− mice. To further validate the requirement of EphB2 forward signaling that was involved in spinogenesis (Henkemeyer et al., 2003), we examined the IPB morphology of EphB1LacZ/LacZ and EphB2LacZ/LacZ knock-in mice, in which the EphB intracellular segment is substituted with β-gal and the transmembrane and extracellular domains are left intact on the cell surface to activate ephrin-B3–mediated reverse signaling in MF axons (Henkemeyer et al., 1996; Chenaux and Henkemeyer, 2011). Interestingly, we found an opposite phenotype between EphB1LacZ/LacZ and EphB2LacZ/LacZ mice. Unlike the EphB1−/− mutant, EphB1LacZ/LacZ mutant mice showed no obvious change in IPB length compared with controls (Fig. 6, A and B), whereas EphB2LacZ/LacZ knock-in mice showed more serious defective CA3 cell patterning and axon shortening, including both IPB and SPB axons. We further checked the phenotype in EphB2K661R/K661R mice and observed a similar MF defect comparable with EphB2−/− or EphB2LacZ/LacZ mice (Fig. 6, C and D), suggesting that tyrosine kinase activity is required for MF axon pruning. This indicates that EphB1 and EphB2 transduce distinct signals in which the extracellular domain of EphB1 and the EphB2 intracellular kinase-dependent signaling are independently required for MF pruning. Figure 6. EphB receptors differ in signals required for MF pruning. (A) Neurotrace dye–labeled pattern of CA3 pyramidal cells in PW3 WT, EphB1−/−, and EphB1LacZ/LacZ mice. Calbindin staining showed that the IPB axon–penetrated CA3 pyramidal neurons layer are much longer in EphB1−/− mice compared with WT or EphB1LacZ/LacZ mice. White brackets delineate IPB length, and distance between white arrowheads delineates the IPB length in CA3 pyramidal neurons. (B) The ratio of IPB length to the length from hilus to curvature of CA3 area and IPB length within CA3 pyramidal neurons in WT, EphB1−/−, and EphB1LacZ/LacZ mice were quantified. n = 5–6 mice per group. (C and D) Calbindin staining for PW3 WT, EphB2−/−, EphB2LacZ/LacZ, and EphB2K661R/K661R mice. The length ratio of IPB and IPB length in CA3 pyramidal neurons are much longer in EphB2−/−, EphB2LacZ/LacZ, and EphB2K661R/K661R mice compared with WT mice. n = 4–5 mice per group. (E and F) Calbindin staining for PW3 Lnx1−/− and Lnx1−/−; EphB1LacZ/LacZ mice. No difference was observed in IPB axons between Lnx1−/− and Lnx1−/−; EphB1 LacZ/LacZ mice. n = 5 mice per group. (G and H) Calbindin staining for PW3 EphB2 F620D/F620D, Lnx1−/−, and Lnx1−/−; EphB2 F620D/F620D mice. The aberrant IPB axons in Lnx1−/− mice were rescued in Lnx1−/−; EphB2 F620D/F620D mice. n = 5–6 mice per group. Bars, 100 µm. Means ± SEM; one-way ANOVA with Tukey’s post hoc test (B, D, F, and H); *, P < 0.05; **, P < 0.01; ***, P < 0.001. We then crossed EphB1LacZ/LacZ, in which the intracellular segment is substituted with β-gal to prevent binding with intracellular partners for EphB1 protein degradation (Fig. S5 E), or EphB2F620D/F620D, a constitutively active form of EphB tyrosine kinase (Holmberg et al., 2006), with Lnx1−/−, to see whether the extracellular domain of EphB1 or constitutive catalytic activation of EphB2 is able to reverse the morphological defects caused by Lnx1 ablation. We saw that in Lnx1−/−; EphB1LacZ/LacZ mice, the morphological abnormalities remained unchanged compared with Lnx1−/− mice (Fig. 6, E and F). However, an obvious rescue in axon terminal targeting was observed in Lnx1−/−; EphB2F620D/F620D mice (Fig. 6, G and H; and Fig. 7, A and B). To test whether presynaptic signal transduction is involved, we first measured the expression level of ephrin-B3, the EphB binding ligand specifically expressed in MF axons (Xu and Henkemeyer, 2009), in Lnx1 mutants, and we did not see an obvious change (Fig. S5 F). We then crossed Efnb3−/− mice with Lnx1−/−; EphB2F620D/F620D mice and found that MF terminal targeting was disrupted in the Lnx1−/−; Efnb3−/−; EphB2F620D/F620D mice (Fig. 7, A and B), indicating that the presynaptic ephrin-B3 is also required for MF axon targeting. To figure out whether the postsynaptic or cell-autonomous changes of CA3 neurons secondarily contributed to the observed MF phenotypes, we carefully counted the CA3 cell number and cell density in each genetic condition. We did not observe an obvious change in cell number but found a decreased cell density in Lnx1−/− mice, which was exhibited as a loose pattern in CA3 cell layers (Fig. 7, A and C). As the loose cell layer was also observed in EphB mutants as shown previously (Bouché et al., 2013) and in our study (Fig. 6, A and C), the abnormal cell patterning observed in Lnx1−/− mice was likely attributable to the low level of EphBs comparable with that of EphB mutants. Figure 7. EphB kinase activation promotes MF terminal targeting. (A and B) ZnT3 and NeuN staining show the IPB axon terminal–wired CA3 pyramidal cells in different mice. The robust expanded IPB axon terminals in Lnx1−/− mice were rescued in Lnx1−/−; EphB2 F620D/F620D mice but were disrupted again in Lnx1−/−; Efnb3−/−; EphB2 F620D/F620D mice. Bars, 250 µm (top); 25 µm (bottom). n = 5–6 mice per group. (C) Quantification of number of NeuN-positive cells in CA3 region in PW3 different mice. The total number of NeuN+ cells of CA3 area per slice shows no difference among these mice, while the cell density of CA3 area in Lnx1−/− mice and Lnx1−/−; Efnb3−/−; EphB2 F620D/F620D mice showed a reduction compared with other groups. n = 18–19 slices from three mice per group. (D) Immunoprecipitation (IP) of EphB2 protein from hippocampal lysates of PW3 Lnx1−/− and WT mice. The EphB2 tyrosine phosphorylation level was slightly decreased in Lnx1−/− mice compared with WT mice (P = 0.095). n = 3 mice per group. (E) The expression of EphB proteins in hippocampus from PW3 mice was detected by Western blot and quantified. The decreased EphB2 protein level in Lnx1−/− mice was partly restored in Lnx1−/−; EphB2 F620D/F620D mice, while EphB1 remained unchanged. n = 3 mice per group. Means ± SEM; one-way ANOVA with Tukey’s post hoc test (B, C, and E) or Student’s t test (D); *, P < 0.05; **, P < 0.01, ***, P < 0.001. IB, immunoblot. To clarify why constitutively activating EphB2 can reverse morphological defects caused by Lnx1 ablation, we studied the possible changes in tyrosine kinase activity of EphB2 in Lnx1−/− mice. By immunoprecipitation with EphB2 antibody from hippocampal lysates of WT or Lnx1−/− mice, a comparable total amount of EphB2 was extracted and loaded for detection of the tyrosine phosphorylation of EphB2. We observed a slight reduction in phosphorylated EphB2 level in Lnx1−/− mice compared with WT control mice, which suggests that Lnx1 plays a mild role in promoting or sustaining activation of EphB2 (Fig. 7 D). To further determine whether active EphB2 may be more resistant to the degradation, we detected EphB2 expression in Lnx1−/−; EphB2F620D/F620D mice and observed a partial restoration of EphB2 protein level compared with that in Lnx1−/− mice (Fig. 7 E). Finally, we examined the terminal morphology visualized with EM in Lnx1−/−; EphB2F620D/F620D mice. In support of the diminished axon terminals in Lnx1−/−; EphB2F620D/F620D mice observed by ZnT3 staining, the number of release sites and docked vesicles in synapses with giant boutons, the vesicle properties, and their distance to the membrane in typical synapses were also restored to normal levels comparable with WT mice (Fig. 8, A–F). We further assayed the PPRs of MF-CA3 synapses and found that EphB2F620D/F620D mice per se showed no difference compared with the WT mice, while the increased PPR in Lnx1−/− mice was restored to a normal level in Lnx1−/−; EphB2F620D/F620D mice (Fig. 8 G). Figure 8. EphB kinase activation promotes MF terminal maturation. (A) Transmission electron micrographs from PW3 mice show MF terminals (red) and postsynaptic spines (green) in synapses with giant bouton. Asterisks mark PSD. Bars, 0.5 µm. (B) The decreased PSD number per spine and docked vesicles of MF synapses from SPB (281–312 synapses) or IPB (37–46 synapses) regions in Lnx1−/− mice were restored to normal level in Lnx1−/−; EphB2 F620D/F620D mice. n = 3–4 mice per group. (C) Transmission electron micrographs from PW3 mice show axon terminals (red) and postsynaptic spines (green) in typical synapses with regular bouton. Asterisks mark PSD. Bars, 0.5 µm. (D) The number of vesicles per terminal in typical synapses was significantly correlated with the PSD length, determined by EM analysis in PW3 WT (r2 = 0.62; P < 0.001), EphB2 F620D/F620D (r2 = 0.42; P < 0.001), and Lnx1−/−; EphB2 F620D/F620D (r2 = 0.40; P < 0.001) mice but not in Lnx1−/− mice (left). The area of vesicles per terminal in typical synapses was significantly correlated with PSD length in PW3 WT (r2 = 0.66; P < 0.001), EphB2 F620D/F620D (r2 = 0.54; P < 0.001), and Lnx1−/−; EphB2 F620D/F620D (r2 = 0.59; P < 0.001) mice but not in Lnx1−/− mice (right). A dot presents a synapse. (E) EM analysis showed that the decreased vesicle density in terminals that formed synapses in Lnx1−/− mice was restored to normal level in Lnx1−/−; EphB2 F620D/F620D mice. (F) The increased average distance of vesicles to the presynaptic membrane in Lnx1−/− mice was restored to normal level in Lnx1−/−; EphB2 F620D/F620D mice. n = 72–165 synapses per group. n = 3–4 mice per group in D–F. (G) Representative average traces (left) and summary graph (right) showed that increased PPRs upon SPB stimulation in PW3 Lnx1−/− mice were restored to normal level in Lnx1−/−; EphB2 F620D/F620D mice. Bar, 50 ms. n = 22–27 neurons from three to four mice per group. (H) Proposed model for postsynaptic Lnx1–EphB complex–mediated retrograde regulation of MF axon terminal maturation. Lnx1 binds and stabilizes postsynaptic EphB receptors in hippocampal CA3 pyramidal neurons to sculpt postsynaptic structure, which helps to guide MF axon targeting and promote terminal maturation through a trans-synaptic regulation. Means ± SEM; one-way ANOVA with Tukey’s post hoc test (B and E–G); *, P < 0.05; **, P < 0.01; ***, P < 0.001. Taken together, our data suggest a model in which Lnx1 serves as a specific protein stabilizer for postsynaptic EphB receptor kinases to form a protein complex on the membrane of hippocampal CA3 pyramidal neurons to sculpt postsynaptic structure, which helps to guide MF axon targeting and promote terminal maturation through trans-synaptic regulation in a retrograde, non–cell-autonomous way (Fig. 8 H). Discussion In this study, we reveal a postsynaptically driven mechanism for the formation of functional synapses via PDZ scaffold protein Lnx1, which controls the axon targeting and maturation of MF terminals in the developing hippocampus. In contrast with the previous research on Lnx1, which functions as an intrinsic determinant of cell fate by its interaction with the protein Numb during development (Cayouette and Raff, 2002) or as an E3 ubiquitin ligase to cause proteasome-dependent degradation for Notch signaling (Nie et al., 2002), the hippocampal-specific Lnx1 serves as a membrane stabilizer to sustain receptor proteins at postsynaptic compartments in brain to refine the hippocampal presynaptic structure in a non–cell-autonomous manner. Two variants of Lnx1 have been found in vivo, P80-Lnx1 and P70-Lnx1, to share an identical PDZ domain (Dho et al., 1998). In view of the specific expression of P70-Lnx1 isoform in the brain as shown previously (Dho et al., 1998; Xie et al., 2001) and its function on stabilizing EphB receptors, the abnormalities in the Lnx1-null mice observed in this study are attributable to the ablation of P70-Lnx1 protein that does not contain a ring-finger domain for E3 ligase activity. Through a genetic targeting approach, we showed that ablation of postsynaptic expressed protein Lnx1 led to untrimmed presynaptic MF axon terminals during hippocampal development, which resulted in abnormalities of axon targeting and terminal maturation during dynamic coordination of post- and presynaptic compartments. Through live imaging of cocultured neurons, we studied the relevance of pre- and postsynaptic dynamics. Although a high probability of the postsynaptic altered type was observed, which has been revealed by numerous studies focused on how presynaptic inputs and signaling regulate postsynaptic structure and function (Scheiffele, 2003; Zuo et al., 2005; Alvarez and Sabatini, 2007; Südhof, 2008; Kwon and Sabatini, 2011; Li et al., 2017), we also saw a proportion of presynaptic dynamics. We found that deletion of Lnx1 decreased the rate of presynaptic altered type, including the newborn boutons and refined boutons, but did not affect the rate of postsynaptic altered type. Furthermore, we observed more abnormal MF giant boutons and thorny excrescences, with fewer release sites and docked vesicles as well as fewer mature axon terminals in typical synapses at the CA3 area formed in juvenile Lnx1−/− mice. These abnormities in synapse formation might not be limited to the MF boutons as CA3 neurons also receive axon projections from other brain regions (Witter, 2007) where EphBs are expressed (Liebl et al., 2003; Migani et al., 2007). These results suggest that Lnx1 is essential for maturation and stabilization of presynaptic terminals for precise synaptic connection. This study thus uncovers a retrograde modulation of presynaptic structure during synapse formation. As novel interacting partners, EphB1 and EphB2 receptors were identified to bind with different PDZ domains of Lnx1, which serves as a protein stabilizer to prevent degradation in proteasome, through their PDZ-binding motif. Interestingly, we found that EphB1 and EphB2 play distinct roles in MF axon pruning and targeting, and the extracellular segment of EphB1 and intracellular domains of EphB2 are independently required. This raises a presumption that EphB receptors may be integrated differently in a heterogeneous molecular complex upon MF-CA3 neuron contact. We further revealed that EphB2 kinase dead mutant EphB2K661R/K661R showed defective MF axon pruning, while the EphB2F620D/F620D mice with constitutive kinase activity showed normal axon morphology. As the EphB2 protein was not expressed in MF axons (Xu and Henkemeyer, 2009), the defective MF axon targeting observed in EphB2K661R/K661R mice suggests a non–cell-autonomous regulation. Furthermore, the abnormal presynaptic targeting and terminal maturation observed in Lnx1-null mice were restored in the Lnx1−/−; EphB2F620D/F620D compound mutant. This indicates that not only is the EphB2 kinase activity required for normal MF axon pruning during development, but it is also sufficient to rescue the defective axon terminal morphogenesis in the absence of Lnx1, which may be attributed to the resistance of the active EphB2 to the degradation. As the kinase activation of membrane EphB receptors within CA3 pyramidal neurons promotes remodeling of postsynaptic structure (Henkemeyer et al., 2003), this may help to form a precise anchor to receive connection and control refinement of the projected axon terminals in a non–cell-autonomous manner. Mechanistically, this study is distinct from previous studies on regulation of presynaptic function through trans-synaptic molecular signaling upon high neuronal activity (Chavis and Westbrook, 2001; Contractor et al., 2002; Jüngling et al., 2006; Regalado et al., 2006; Futai et al., 2007; Orr et al., 2017). In the specific DG-CA3 synapse, the CA3-expressed EphBs themselves can initiate pruning of MF axons in a retrograde manner through ephrin-B3 reverse signaling (Xu and Henkemeyer, 2009), which is further confirmed in this study. During postnatal development, EphB receptor signaling can be induced upon binding with secreted glycoprotein Reelin via their extracellular domain to form a massive protein complex, and this works together with ApoER2 and VLDL receptor cascade, two members of the LDL receptor family, to regulate neuron cytoskeleton in the CA3 cell layer (Bouché et al., 2013). The stabilization of the postsynaptic multiprotein complex would be critical for CA3-MF trans-synaptic regulation during the long-term developmental process of synaptogenesis. EphB and ephrin-B receptors have been studied in sensory integration and cognitive function through mediating trans-synaptic bidirectional signals (Pasquale, 2008; Sheffler-Collins and Dalva, 2012; Sloniowski and Ethell, 2012). The integration of pre- and postsynaptic remodeling occurs either in an internucleus manner, as shown in our previous research (Zhu et al., 2016a,b), or in an inner-nucleus regulation, as presented in this study. Dysfunction of these early circuits may lead to neurodevelopmental disorders, as supported by accumulating research about the critical role of EphB receptor in brain development and function (Sheffler-Collins and Dalva, 2012; Klein and Kania, 2014; Kania and Klein, 2016). Therefore, our analysis clarifies the mechanisms underlying functional DG-CA3 circuit assembly, leading to a greater understanding of the molecular basis for brain wiring and cognitive functions. Materials and methods Mice and sample preparation EphB1−/− (Williams et al., 2003), EphB1LacZ (Chenaux and Henkemeyer, 2011), EphB2−/− (Henkemeyer et al., 1996), EphB2LacZ (Henkemeyer et al., 1996), EphB2K661R (Genander et al., 2009), EphB2ΔVEV (Genander et al., 2009), EphB2F620D (Holmberg et al., 2006), Efnb3−/− (Xu et al., 2011), TdT (Ai9; Madisen et al., 2010), and Thy1-GFP-M (Feng et al., 2000) knockout and knock-in mice and genotyping methods have been described previously. TdT (Ai9) mice were crossed with a ubiquitous Cre transgene mice to allow TdT expression in brain. The Cre-activated TdT+ mice have been crossed for multiple generations and were used for primary neuronal culture. Mice were anesthetized (chloral hydrate, 350 mg/kg) and perfused with 0.1 M PBS followed by 4% PFA in phosphate buffer. The brains were then removed, postfixed, and sectioned at 30 µm using a vibratome. All experiments involving mice were performed in accordance with the National Institutes of Health (NIH) Guide for the Care and Use of Animals under an Institutional Animal Care and Use Committee–approved protocol at an Association for Assessment and Accreditation of Laboratory Animal Care–approved facility at the Shanghai Jiao Tong University School of Medicine. Parents and pups (10–11 pups per litter) were raised in animal facilities with a constant temperature (22°C) and on a 12-h light-dark cycle. Access to food and water was unlimited. The day of birth was defined as postnatal day 0 (P0). All efforts were made to minimize the number of animals used and their suffering. Generation of Lnx1 mutant mice To construct the Lnx1 targeting vector, a 1.8-kb fragment of cloned mouse genomic DNA upstream of exon 3 (ATG P80-Lnx1) was used for the 5′ arm, and a 4.4-kb fragment downstream of exon4 (ATG P70-Lnx1) was used for the 3′ arm. The two arms were cloned into pPNT vector to provide the neo and TK cassettes used for positive and negative selections. To generate a LacZ marker for the normal expression of Lnx1, the targeting vector was modified to include a tau-β-gal reporter gene. Targeting vectors were electroporated into the ES cell line RI; colonies were isolated following selection in G418 and ganciclovir and expanded; and genomic DNA was screened by Southern blotting. The frequency of homologous recombination was 2 of 623 cell lines screened. Germline transmission was obtained by generating aggregation chimeras with targeted ES cells. The animals used in this study have been backcrossed to CD1 mice for multiple generations. Animals were genotyped by PCR with forward primer 5′-CCAGTAGACAGGCCCCAAGTGATTATTT-3′ and reverse primer 5′-TCGATCCACAGGGCAGAAGTCC-3′ for WT (565 bp), and forward primer 5′-CCAGTAGACAGGCCCCAAGTGATTATTT-3′ and reverse primer 5′-ACTCTTTCAGGCCGGGGTCCAT-3′ for mutant (421 bp). Immunofluorescence For immunofluorescence, vibratome sections were blocked with permeable buffer (0.3% Triton X-100 in PBS) containing 10% donkey serum for half an hour at room temperature and incubated with primary antibodies in permeable buffer containing 2% donkey serum overnight at 4°C. The slices were then washed three times with PBS-T (0.1% Tween-20 in PBS) for 10 min each and incubated with Alexa Fluor secondary antibodies (1:200; Molecular Probes) and NeuroTrace 633 (1:500; Molecular Probes) in the PBS buffer for 2 h at room temperature. Slices were washed in PBS-T three times, mounted on glass slides using Aqua poly/mount (Polysciences), and photographed using a confocal microscope (Leica Application Suite X) or STED imaging (Leica TCS SP8). Fluorescence microscopic images obtained were imported into ImageJ (NIH) for analysis, and all the parameters used were kept consistent during capture. For primary antibodies, we used mouse anti-calbindin 28K (1:1,000; 300; Swant), rabbit anti–β-gal (1:200; MP Biomedicals), mouse anti-lnx1 (1:200; ab22157; Abcam), goat anti-EphB2 (1:500; P54763; R&D), mouse anti–zinc-transporter-3 (ZnT3; 1:500; 197011; Synaptic Systems, rabbit anti-NeuN (1:1,000; D3S3I; Cell Signaling Technology), mouse anti-Flag (1:1,000; F3165; Sigma-Aldrich), rabbit anti-synapsin1 (1:1,000, gift from Ilya Bezprozvanny, University of Texas Southwestern Medical Center, Dallas, TX), and rabbit anti-PSD95 (1:1,000; 3450; Cell Signaling Technology). To accurately analyze the terminals on spines (terminal-spine) and dendritic shaft (terminal-shaft), the fluorescent images were deconvolved according to the instructions of Leica TCS SP8, and 3D rendering was achieved in ImageJ using these deconvolved images. The antibody signal threshold was defined as three times brightness to the background, and brightness/contrast adjustment within linear ranges was made using ImageJ when necessary. To quantify the shape of spine, a procedure was adapted from our previous study (Xu et al., 2011). The shape of neuronal spines in slices was classified by NeuronStudio software package and an algorithm from Rodriguez et al. (2008) with the following cutoff values: AR_thin(crit) = 2.5, head-to-neck ratio (HNR(crit)) = 1.3, and head diameter (HD(crit)) = 0.3 µm. The type of these spines was determined based on the following criteria: (a) spines with HNR greater than HNR(crit) are considered to have a neck and could be either thin or mushroom types; (b) spines with HD greater than HD(crit) are classified as mushroom, otherwise thin; (c) spines lacking significant necks and less than AR_thin(crit) are considered as stubby, otherwise thin. Protrusions with length 0.2–3.0 µm and maximum width 3 µm were counted. Spine density was calculated by dividing the total spine number by the dendritic branch length. The localizations of terminals and spines were carefully identified in the 3D rendering views. The ZnT3+ synapses were defined as the overlapped connections of spines and ZnT3+ terminals with any identical red/green pixels (as shown in Fig. 2 B), otherwise ZnT3− (separated from terminals). Control and experiment conditions were adjusted with the same parameters. Acquisition of the images as well as morphometric quantification was performed under blinded conditions. For quantification of CA3 cell number, serial coronal sections (30 µm) containing hippocampus (from bregma, −1.06 to −2.30 mm) were collected using a vibratome. NeuN immunostaining was performed on every sixth section encompassing the anterior to posterior of the CA3 area. NeuN-positive cells of CA3 area in each section of different mice were counted under blinded conditions. For X-gal staining of embryos, E13–E15 embryos were washed with cold PBS and fixed in cold PBS + 4% PFA for 20 min with gentle agitation. After three 5-min washes with gentle agitation in cold PBS, embryos were transferred to histochemical staining solution (5 mM K4Fe(CN)6, 5 mM K3Fe(CN)6, 2 mM MgCl2, 0.02% [vol/vol] NP-40, 0.01% [wt/vol] sodium deoxycholate, 1 mg/ml 5-bromo-3-indolyl-β-d-galactopyranoside [X-Gal; Amresco], and 20 mM Tris-HCl in PBS, pH 7.3) in a 24-well plate and incubated overnight at 30°C with gentle agitation. Whole mounts were washed with PBS + 0.1% Tween-20 at room temperature for several hours before image capture. Transmission EM Mice were perfused with 2% PFA/2.5% glutaraldehyde in phosphate buffer, pH 7.2, for 30 min, and dissected brains were then postfixed in the same buffer overnight at 4°C. After PBS buffer rinse, samples were postfixed in 1% osmium tetroxide buffer (2 h) on ice in the dark. After a double-distilled water rinse, tissue was stained with 3% aqueous uranyl acetate (0.22-µm filtered; 1 h in the dark), dehydrated in a graded series of ethanol and propylene oxide, and embedded in Epoxy 618 resin. Samples were polymerized at 60°C for 48 h. Thin sections (60–90 nm) were cut with a diamond knife on the LKB V ultramicrotome and picked up with formvar-coated copper slot grids. Grids were stained with lead citrate and observed with transmission microscopy (PHILIP CM-120). Images from the SPB or IPB regions were captured, and the PSDs, terminals, vesicle numbers, and distance of vesicles to presynaptic membrane were counted or analyzed by ImageJ under blinded conditions (n = 3–4 animals per genotype). Western blotting, immunoprecipitation, and isolation of synaptosome and cell-surface protein Western blotting was performed as in a previous study (Sun et al., 2014). Briefly, hippocampal regions from WT, knockout, and knock-in mice at different ages were dissected, homogenized, and solubilized at 4°C for 1 h in lysis buffer (1% CHAPS, 137 mM NaCl, 2.7 mM KCl, 4.3 mM Na2HPO4, 1.4 mM KH2PO4, pH 7.2, 5 mM EDTA, 5 mM EGTA, 1 mM PMSF, 50 mM NaF, 1 mM Na3VO4, and protease inhibitors). Primary hippocampal neurons were collected, homogenized, and solubilized in the same buffer on DIV14. For immunoprecipitation, hippocampal tissues were lysed at 4°C for 1 h in lysis buffer (50 mM Tris-HCl, pH 7.5, 200 mM NaCl, 5 mM MgCl2, 1% NP-40, 10% glycerol, 1 mM DTT, 1 mM PMSF, 50 mM NaF, 1 mM Na3VO4, and protease inhibitors) and then immunoprecipitated with indicated antibodies for 2 h and incubated with protein G beads overnight at 4°C. Bound proteins were separated by SDS-PAGE, transferred to nitrocellulose membranes, and immunoblotted with indicated antibodies. The subcellular fractions of the hippocampus from WT, knockout, and knock-in mice were purified as described previously (Pacchioni et al., 2009). Cell-surface protein of the cultured primary hippocampal neuron samples was isolated using a Pierce cell surface protein isolation kit (Thermo Fisher Scientific) per the manufacturer’s protocol. Analysis of the data was performed using ImageJ, and the mean density of each band was normalized to β-actin or GAPDH signals in the same sample and averaged. For primary antibodies, we used mouse anti-Lnx1 (1:1,000; ab22157; Abcam), rabbit anti–β-gal (1:1,000; MP Biomedicals), rabbit anti-PSD95 (1:1,000; 3450; Cell Signaling Technology), mouse antisynaptophysin (1:3,000; ab8049; Abcam), mouse anti-GAPDH (1:3,000; G8795; Sigma-Aldrich), mouse anti–β-actin (1:3,000; MA5-15739; Thermo Fisher Scientific), goat anti-EphB1 (1:200; M-19; Santa Cruz Biotechnology), goat anti-EphB2 (1:1,000; P54763; R&D), rabbit anti–N-cadherin (1:1,000; ab12221; Abcam), rabbit anti–connexin 43 (1:1,000; ab11370; Abcam), rabbit anti-claudin1 (1:1,000; ab15098; Abcam), mouse anti-striatin (1:1,000; 610838; BD), mouse anti-Flag (F3165; Sigma-Aldrich), anti–Flag-HRP and anti–HA-HRP (1:5,000; Sigma-Aldrich), rabbit anti–ephrin-B3 (1:500; 34-3600; Invitrogen), and mouse anti–phosphotyrosine 4G10 (1:500; 05-321; EMD Millipore). DNA constructs P70-Lnx1 and P80-Lnx1 genes were amplified from hippocampal and renal cDNA, respectively, by PCR and ligated in the EcoRI and XbaI sites of p3XFLAG-CMV-10. All the PDZ mutants of P70-Lnx1 were generated from full-length P70-Lnx1 by PCR and ligated to p3XFLAG-CMV-10. Primary cell culture, coculture, and biochemistry Primary cell culture of hippocampal neurons was performed as described (Xu and Henkemeyer, 2009). Briefly, hippocampal neurons were dissociated from P0 pups. The triturated cells (1 × 105 cells per well) were grown on either six-well dishes or glass coverslips coated with 10 µM polylysine overnight in 24-well dishes. Then the culture was grown in a medium of Neurobasal A media (Gibco) supplemented with B27 and 2 mM glutamine for the indicated number of days. For coculture assays, primary hippocampal neurons from Lnx1+/+ and Lnx1−/− pups were precultured for 12–14 d (transfected with GFP plasmid at DIV7 using the calcium phosphate method), and TdT+ primary hippocampal neurons from P1 TdT+ knock-in pups were dissociated as described (Baranes et al., 1996) and plated on the culture by either direct addition into the dishes or loading with coverslips for another 6 d. During the 6 d, neurons were imaged every day to measure the TdT+ axon length. On the sixth day, neurons were fixed (4% PFA and 4% sucrose in PBS) and imaged to observe the spines and boutons. For the pharmacological treatment assay, primary hippocampal neurons from WT or Lnx1−/− pups were dissociated and cultured for 14 d, and neurons or MEF cells were incubated for 12 h in the presence or absence of proteasome inhibitor (10 µM MG132; Gene Operation) or lysosomal inhibitor (100 µg/ml leupeptin and 50 mM NH4Cl; Sigma-Aldrich). After treatment, cells were lysed and subjected to Western blot analysis. Live imaging in dual color–labeled coculture system Live-imaging experiments were performed on a Nikon A1R confocal microscope. Primary hippocampal neurons from Lnx1+/+ and Lnx1−/− pups were precultured in 35-mm glass-bottom dishes for 12–14 d (transfected with GFP plasmid at DIV7 using the calcium phosphate method), and TdT+ primary hippocampal neurons were plated by direct addition into the dishes for another 6–7 d. Neurons were maintained at room temperature, and images were acquired every 10 min. For the presynaptic or postsynaptic alteration, we calculated the AI: |A60 − A0|/(A60 + A0), where A is the pre- or postsynaptic area and 60 or 0 indicates the time in minutes. When AI ≥ 0.2, we indicate the pre- or postsynaptic alteration as *; otherwise as ns. Quantitative real-time PCR Total RNA was prepared from hippocampus tissue of PW3 Lnx1 mutant and WT littermates using TRI Reagent (Sigma-Aldrich). RNAs were reverse transcribed with high-capacity cDNA reverse transcription kits (Applied Biosystems) according to the manufacturer’s instructions. The PCR mixture contained 1 µl diluted cDNA, 5 µl 2× SYBR Green PCR Master Mix (Applied Biosystems), and 200 nM of each gene-specific primer in a final volume of 10 µl. The real-time PCRs were performed using a Fast 96-Well System (Applied Biosystems). Three biological replicates for each sample were used for real-time PCR analysis, and three technical replicates were analyzed for each biological replicate. The relative copy number of β-actin RNA was quantified and used for normalization. The primer sequences are given in Table S1. Electrophysiology Brain coronal slices were prepared from 3-wk-old naive Lnx1 +/+ and Lnx1−/− mice. Brains were dissected quickly and chilled in ice-cold artificial cerebrospinal fluid (ACSF) containing (in mM): 125 NaCl, 2.5 KCl, 2 CaCl2, 1 MgCl2, 25 NaHCO3, 1.25 NaH2PO4, and 12.5 glucose. Coronal brain slices (300 µm thick) were prepared with a vibratome and recovered in ACSF bubbled with 95% O2 and 5% CO2 at 31°C for 1 h and then maintained at room temperature (22–25°C). For EPSC recording, borosilicate glass pipettes (3–5 MΩ) were filled with an internal solution containing (in mM) 115 CsMeSO3, 10 Hepes, 2.5 MgCl2 ⋅ 6H2O, 20 CsCl2, 0.6 EGTA, 10 Na2 phosphocreatine, 0.4 Na-GTP, and 4 Mg ATP. EPSCs were recorded at −70 mV in the presence of 100 µM picrotoxin. Slices were stimulated using a bipolar concentric electrode (FHC) that was placed in the MF and connected with a stimulator (AMPI) to evoke EPSCs in CA3 pyramidal neurons. PPRs were calculated as a ratio of EPSC2 to EPSC1, separated by interstimulus intervals of 25, 50, 100, 200, and 400 ms. Data were analyzed in pClamp 10.6 (Molecular Devices), and recordings were made from an average of three cells per slice and two to three slices per mouse. Statistical analysis The results are presented as mean ± SEM. Statistical differences were determined by Student’s t test for two-group comparisons or ANOVA followed by Tukey test for multiple comparisons among more than two groups. Online supplemental material Fig. S1 shows mRNA localization of Lnx1 in brain from the Allen Brain Atlas and biochemical characterization of Lnx1 mutant. Fig. S2 shows expression and subcellular localization of Lnx1 in hippocampus. Fig. S3 shows Lnx1 effects on axon growth and pruning when cocultured with coverslips to prevent axon–cell contact. Fig. S4 shows Lnx1 effects on morphology of calbindin-positive/-negative neurons in coculture assay, spine density and mushroom ratio of CA3 neurons, and PPRs with IPB stimulation. Fig. S5 shows interactions between Lnx1 and EphB1/2 and effects of Lnx1 on membrane level of EphB1/2, mRNA level of receptors, and protein level of EphB2/EphB1-β-gal/ephrin-B3. Table S1 shows primer sequences for quantitative real-time PCR used in this study. Videos 1, 2, 3, 4, and 5 show different morphological change types including presynaptic alteration (Videos 1 and 2), postsynaptic alteration (Video 3), both alteration (Video 4), and no alteration (Video 5) in a dual color–labeled coculture system. Supplementary Material Supplemental Materials (PDF) Video 1 Video 2 Video 3 Video 4 Video 5 Acknowledgments We thank Drs. Xiang Yu, Zhiping Pan, and Lan Bao for helpful comments on the manuscript and Liang Zhu, Guang-Ni Xu, Zi-Jun Deng, and Si Chen for laboratory technique support. This research was supported by National Basic Research Program of China (973 Program; 2014CB965002) to N.-J. Xu, the National Natural Science Foundation of China (91232704 and 31671062) to N.-J. Xu, grants from the Shanghai Brain-Intelligence Project from the Shanghai Science and Technology Committee (16JC1420500), the Program for Professor of Special Appointment (Eastern Scholar) at Shanghai Institutions of Higher Learning (2013-25) to N.-J. Xu, the Shanghai Science and Technology Committee (11DZ2260200) to N.-J. Xu, and the NIH (MH066332) to M. Henkemeyer. The authors declare no competing financial interests. Author contributions: X.-D. Liu performed the experiments of morphology and histology. X.-N. Zhu and T.-L. Xu assisted with electrophysiology, M.M. Halford and M. Henkemeyer initiated the Lnx1 project and generated the mutant mouse, and M. Henkemeyer provided the various EphB1 and EphB2 mutants used in the study. X.-D. Liu and N.-J. Xu designed experiments and wrote the manuscript. ==== Refs Ackley, B.D., and Y. Jin. 2004. Genetic analysis of synaptic target recognition and assembly. Trends Neurosci. 27 :540–547. 10.1016/j.tins.2004.07.003 15331236 Alvarez, V.A., and B.L. Sabatini. 2007. Anatomical and physiological plasticity of dendritic spines. Annu. Rev. Neurosci. 30 :79–97. 10.1146/annurev.neuro.30.051606.094222 17280523 Amaral, D.G., and J.A. Dent. 1981. Development of the mossy fibers of the dentate gyrus: I. A light and electron microscopic study of the mossy fibers and their expansions. J. Comp. Neurol. 195 :51–86. 10.1002/cne.901950106 7204652 Bagri, A., H.J. Cheng, A. Yaron, S.J. Pleasure, and M. Tessier-Lavigne. 2003. Stereotyped pruning of long hippocampal axon branches triggered by retraction inducers of the semaphorin family. Cell. 113 :285–299. 10.1016/S0092-8674(03)00267-8 12732138 Baranes, D., J.C. López-García, M. Chen, C.H. Bailey, and E.R. Kandel. 1996. Reconstitution of the hippocampal mossy fiber and associational-commissural pathways in a novel dissociated cell culture system. Proc. Natl. Acad. Sci. USA. 93 :4706–4711. 10.1073/pnas.93.10.4706 8643467 Bennett, M.R., and J. Lagopoulos. 2014. Stress and trauma: BDNF control of dendritic-spine formation and regression. Prog. Neurobiol. 112 :80–99. 10.1016/j.pneurobio.2013.10.005 24211850 Benoist, M., S. Gaillard, and F. Castets. 2006. The striatin family: a new signaling platform in dendritic spines. J. Physiol. Paris. 99 :146–153. 10.1016/j.jphysparis.2005.12.006 16460920 Bhatt, D.H., S. Zhang, and W.B. Gan. 2009. Dendritic spine dynamics. Annu. Rev. Physiol. 71 :261–282. 10.1146/annurev.physiol.010908.163140 19575680 Bouché, E., M.I. Romero-Ortega, M. Henkemeyer, T. Catchpole, J. Leemhuis, M. Frotscher, P. May, J. Herz, and H.H. Bock. 2013. Reelin induces EphB activation. Cell Res. 23 :473–490. 10.1038/cr.2013.7 23318582 Cayouette, M., and M. Raff. 2002. Asymmetric segregation of Numb: a mechanism for neural specification from Drosophila to mammals. Nat. Neurosci. 5 :1265–1269. 10.1038/nn1202-1265 12447381 Chavis, P., and G. Westbrook. 2001. Integrins mediate functional pre- and postsynaptic maturation at a hippocampal synapse. Nature. 411 :317–321. 10.1038/35077101 11357135 Chen, Y., A.K. Fu, and N.Y. Ip. 2008. Bidirectional signaling of ErbB and Eph receptors at synapses. Neuron Glia Biol. 4 :211–221. 10.1017/S1740925X09990287 19785921 Chenaux, G., and M. Henkemeyer. 2011. Forward signaling by EphB1/EphB2 interacting with ephrin-B ligands at the optic chiasm is required to form the ipsilateral projection. Eur. J. Neurosci. 34 :1620–1633. 10.1111/j.1460-9568.2011.07845.x 22103419 Chicurel, M.E., and K.M. Harris. 1992. Three-dimensional analysis of the structure and composition of CA3 branched dendritic spines and their synaptic relationships with mossy fiber boutons in the rat hippocampus. J. Comp. Neurol. 325 :169–182. 10.1002/cne.903250204 1460112 Cohen-Cory, S. 2002. The developing synapse: construction and modulation of synaptic structures and circuits. Science. 298 :770–776. 10.1126/science.1075510 12399577 Contractor, A., C. Rogers, C. Maron, M. Henkemeyer, G.T. Swanson, and S.F. Heinemann. 2002. Trans-synaptic Eph receptor-ephrin signaling in hippocampal mossy fiber LTP. Science. 296 :1864–1869. 10.1126/science.1069081 12052960 Dho, S.E., S. Jacob, C.D. Wolting, M.B. French, L.R. Rohrschneider, and C.J. McGlade. 1998. The mammalian numb phosphotyrosine-binding domain. Characterization of binding specificity and identification of a novel PDZ domain-containing numb binding protein, LNX. J. Biol. Chem. 273 :9179–9187. 10.1074/jbc.273.15.9179 9535908 El-Husseini, A.E., E. Schnell, D.M. Chetkovich, R.A. Nicoll, and D.S. Bredt. 2000. PSD-95 involvement in maturation of excitatory synapses. Science. 290 :1364–1368.11082065 Feng, W., and M. Zhang. 2009. Organization and dynamics of PDZ-domain-related supramodules in the postsynaptic density. Nat. Rev. Neurosci. 10 :87–99. 10.1038/nrn2540 19153575 Feng, G., R.H. Mellor, M. Bernstein, C. Keller-Peck, Q.T. Nguyen, M. Wallace, J.M. Nerbonne, J.W. Lichtman, and J.R. Sanes. 2000. Imaging neuronal subsets in transgenic mice expressing multiple spectral variants of GFP. Neuron. 28 :41–51. 10.1016/S0896-6273(00)00084-2 11086982 Futai, K., M.J. Kim, T. Hashikawa, P. Scheiffele, M. Sheng, and Y. Hayashi. 2007. Retrograde modulation of presynaptic release probability through signaling mediated by PSD-95-neuroligin. Nat. Neurosci. 10 :186–195. 10.1038/nn1837 17237775 Garner, C.C., J. Nash, and R.L. Huganir. 2000. PDZ domains in synapse assembly and signalling. Trends Cell Biol. 10 :274–280. 10.1016/S0962-8924(00)01783-9 10856930 Garner, C.C., R.G. Zhai, E.D. Gundelfinger, and N.E. Ziv. 2002. Molecular mechanisms of CNS synaptogenesis. Trends Neurosci. 25 :243–251. 10.1016/S0166-2236(02)02152-5 11972960 Geiger, J.C., J. Lipka, I. Segura, S. Hoyer, M.A. Schlager, P.S. Wulf, S. Weinges, J. Demmers, C.C. Hoogenraad, and A. Acker-Palmer. 2014. The GRIP1/14-3-3 pathway coordinates cargo trafficking and dendrite development. Dev. Cell. 28 :381–393. 10.1016/j.devcel.2014.01.018 24576423 Genander, M., M.M. Halford, N.J. Xu, M. Eriksson, Z. Yu, Z. Qiu, A. Martling, G. Greicius, S. Thakar, T. Catchpole, 2009. Dissociation of EphB2 signaling pathways mediating progenitor cell proliferation and tumor suppression. Cell. 139 :679–692. 10.1016/j.cell.2009.08.048 19914164 Giagtzoglou, N., C.V. Ly, and H.J. Bellen. 2009. Cell adhesion, the backbone of the synapse: “vertebrate” and “invertebrate” perspectives. Cold Spring Harb. Perspect. Biol. 1 :a003079. 10.1101/cshperspect.a003079 20066100 Guo, Z., E. Song, S. Ma, X. Wang, S. Gao, C. Shao, S. Hu, L. Jia, R. Tian, T. Xu, and Y. Gao. 2012. Proteomics strategy to identify substrates of LNX, a PDZ domain-containing E3 ubiquitin ligase. J. Proteome Res. 11 :4847–4862. 10.1021/pr300674c 22889411 Heisler, F.F., H.K. Lee, K.V. Gromova, Y. Pechmann, B. Schurek, L. Ruschkies, M. Schroeder, M. Schweizer, and M. Kneussel. 2014. GRIP1 interlinks N-cadherin and AMPA receptors at vesicles to promote combined cargo transport into dendrites. Proc. Natl. Acad. Sci. USA. 111 :5030–5035. 10.1073/pnas.1304301111 24639525 Henkemeyer, M., D. Orioli, J.T. Henderson, T.M. Saxton, J. Roder, T. Pawson, and R. Klein. 1996. Nuk controls pathfinding of commissural axons in the mammalian central nervous system. Cell. 86 :35–46. 10.1016/S0092-8674(00)80075-6 8689685 Henkemeyer, M., O.S. Itkis, M. Ngo, P.W. Hickmott, and I.M. Ethell. 2003. Multiple EphB receptor tyrosine kinases shape dendritic spines in the hippocampus. J. Cell Biol. 163 :1313–1326. 10.1083/jcb.200306033 14691139 Hensch, T.K. 2005. Critical period plasticity in local cortical circuits. Nat. Rev. Neurosci. 6 :877–888. 10.1038/nrn1787 16261181 Holmberg, J., M. Genander, M.M. Halford, C. Annerén, M. Sondell, M.J. Chumley, R.E. Silvany, M. Henkemeyer, and J. Frisén. 2006. EphB receptors coordinate migration and proliferation in the intestinal stem cell niche. Cell. 125 :1151–1163. 10.1016/j.cell.2006.04.030 16777604 Holtmaat, A., and K. Svoboda. 2009. Experience-dependent structural synaptic plasticity in the mammalian brain. Nat. Rev. Neurosci. 10 :647–658. 10.1038/nrn2699 19693029 Hoogenraad, C.C., A.D. Milstein, I.M. Ethell, M. Henkemeyer, and M. Sheng. 2005. GRIP1 controls dendrite morphogenesis by regulating EphB receptor trafficking. Nat. Neurosci. 8 :906–915. 10.1038/nn1487 15965473 Hu, Z., S. Hom, T. Kudze, X.J. Tong, S. Choi, G. Aramuni, W. Zhang, and J.M. Kaplan. 2012. Neurexin and neuroligin mediate retrograde synaptic inhibition in C. elegans. Science. 337 :980–984. 10.1126/science.1224896 22859820 Jüngling, K., V. Eulenburg, R. Moore, R. Kemler, V. Lessmann, and K. Gottmann. 2006. N-cadherin transsynaptically regulates short-term plasticity at glutamatergic synapses in embryonic stem cell-derived neurons. J. Neurosci. 26 :6968–6978. 10.1523/JNEUROSCI.1013-06.2006 16807326 Jüttner, R., and F.G. Rathjen. 2005. Molecular analysis of axonal target specificity and synapse formation. Cell. Mol. Life Sci. 62 :2811–2827. 10.1007/s00018-005-5299-5 16237499 Kania, A., and R. Klein. 2016. Mechanisms of ephrin-Eph signalling in development, physiology and disease. Nat. Rev. Mol. Cell Biol. 17 :240–256. 10.1038/nrm.2015.16 26790531 Kerschensteiner, D., J.L. Morgan, E.D. Parker, R.M. Lewis, and R.O. Wong. 2009. Neurotransmission selectively regulates synapse formation in parallel circuits in vivo. Nature. 460 :1016–1020. 10.1038/nature08236 19693082 Kim, E., and M. Sheng. 2004. PDZ domain proteins of synapses. Nat. Rev. Neurosci. 5 :771–781. 10.1038/nrn1517 15378037 Klein, R., and A. Kania. 2014. Ephrin signalling in the developing nervous system. Curr. Opin. Neurobiol. 27 :16–24. 10.1016/j.conb.2014.02.006 24608162 Kolodkin, A.L., and M. Tessier-Lavigne. 2011. Mechanisms and molecules of neuronal wiring: a primer. Cold Spring Harb. Perspect. Biol. 3 :a001727. 10.1101/cshperspect.a001727 21123392 Kozorovitskiy, Y., A. Saunders, C.A. Johnson, B.B. Lowell, and B.L. Sabatini. 2012. Recurrent network activity drives striatal synaptogenesis. Nature. 485 :646–650. 10.1038/nature11052 22660328 Kwon, H.B., and B.L. Sabatini. 2011. Glutamate induces de novo growth of functional spines in developing cortex. Nature. 474 :100–104. 10.1038/nature09986 21552280 Lenihan, J.A., O. Saha, L.M. Mansfield, and P.W. Young. 2014. Tight, cell type-specific control of LNX expression in the nervous system, at the level of transcription, translation and protein stability. Gene. 552 :39–50. 10.1016/j.gene.2014.09.011 25200495 Li, M.Y., W.Y. Miao, Q.Z. Wu, S.J. He, G. Yan, Y. Yang, J.J. Liu, M.M. Taketo, and X. Yu. 2017. A critical role of presynaptic cadherin/catenin/p140Cap complexes in stabilizing spines and functional synapses in the neocortex. Neuron. 94 :1155–1172.28641114 Liebl, D.J., C.J. Morris, M. Henkemeyer, and L.F. Parada. 2003. mRNA expression of ephrins and Eph receptor tyrosine kinases in the neonatal and adult mouse central nervous system. J. Neurosci. Res. 71 :7–22. 10.1002/jnr.10457 12478610 Low, L.K., and H.J. Cheng. 2006. Axon pruning: an essential step underlying the developmental plasticity of neuronal connections. Philos. Trans. R. Soc. Lond. B Biol. Sci. 361 :1531–1544. 10.1098/rstb.2006.1883 16939973 Madisen, L., T.A. Zwingman, S.M. Sunkin, S.W. Oh, H.A. Zariwala, H. Gu, L.L. Ng, R.D. Palmiter, M.J. Hawrylycz, A.R. Jones, 2010. A robust and high-throughput Cre reporting and characterization system for the whole mouse brain. Nat. Neurosci. 13 :133–140. 10.1038/nn.2467 20023653 McAllister, A.K. 2007. Dynamic aspects of CNS synapse formation. Annu. Rev. Neurosci. 30 :425–450. 10.1146/annurev.neuro.29.051605.112830 17417940 Migani, P., C. Bartlett, S. Dunlop, L. Beazley, and J. Rodger. 2007. Ephrin-B2 immunoreactivity distribution in adult mouse brain. Brain Res. 1182 :60–72. 10.1016/j.brainres.2007.08.065 17945206 Nakamura, Y., C.L. Wood, A.P. Patton, N. Jaafari, J.M. Henley, J.R. Mellor, and J.G. Hanley. 2011. PICK1 inhibition of the Arp2/3 complex controls dendritic spine size and synaptic plasticity. EMBO J. 30 :719–730. 10.1038/emboj.2010.357 21252856 Nicoll, R.A., and D. Schmitz. 2005. Synaptic plasticity at hippocampal mossy fibre synapses. Nat. Rev. Neurosci. 6 :863–876. 10.1038/nrn1786 16261180 Nie, J., M.A. McGill, M. Dermer, S.E. Dho, C.D. Wolting, and C.J. McGlade. 2002. LNX functions as a RING type E3 ubiquitin ligase that targets the cell fate determinant Numb for ubiquitin-dependent degradation. EMBO J. 21 :93–102. 10.1093/emboj/21.1.93 11782429 Orr, B.O., R.D. Fetter, and G.W. Davis. 2017. Retrograde semaphorin-plexin signalling drives homeostatic synaptic plasticity. Nature. 550 :109–113.28953869 Pacchioni, A.M., J. Vallone, P.F. Worley, and P.W. Kalivas. 2009. Neuronal pentraxins modulate cocaine-induced neuroadaptations. J. Pharmacol. Exp. Ther. 328 :183–192. 10.1124/jpet.108.143115 18840757 Pasquale, E.B. 2008. Eph-ephrin bidirectional signaling in physiology and disease. Cell. 133 :38–52. 10.1016/j.cell.2008.03.011 18394988 Penzes, P., R.C. Johnson, R. Sattler, X. Zhang, R.L. Huganir, V. Kambampati, R.E. Mains, and B.A. Eipper. 2001. The neuronal Rho-GEF Kalirin-7 interacts with PDZ domain-containing proteins and regulates dendritic morphogenesis. Neuron. 29 :229–242. 10.1016/S0896-6273(01)00193-3 11182094 Regalado, M.P., R.T. Terry-Lorenzo, C.L. Waites, C.C. Garner, and R.C. Malenka. 2006. Transsynaptic signaling by postsynaptic synapse-associated protein 97. J. Neurosci. 26 :2343–2357. 10.1523/JNEUROSCI.5247-05.2006 16495462 Riccomagno, M.M., A. Hurtado, H. Wang, J.G. Macopson, E.M. Griner, A. Betz, N. Brose, M.G. Kazanietz, and A.L. Kolodkin. 2012. The RacGAP β2-Chimaerin selectively mediates axonal pruning in the hippocampus. Cell. 149 :1594–1606. 10.1016/j.cell.2012.05.018 22726444 Rice, D.S., G.M. Northcutt, and C. Kurschner. 2001. The Lnx family proteins function as molecular scaffolds for Numb family proteins. Mol. Cell. Neurosci. 18 :525–540. 10.1006/mcne.2001.1024 11922143 Rodriguez, A., D.B. Ehlenberger, D.L. Dickstein, P.R. Hof, and S.L. Wearne. 2008. Automated three-dimensional detection and shape classification of dendritic spines from fluorescence microscopy images. PLoS One. 3 :e1997. 10.1371/journal.pone.0001997 18431482 Rollenhagen, A., K. Sätzler, E.P. Rodríguez, P. Jonas, M. Frotscher, and J.H. Lübke. 2007. Structural determinants of transmission at large hippocampal mossy fiber synapses. J. Neurosci. 27 :10434–10444. 10.1523/JNEUROSCI.1946-07.2007 17898215 Sala, C., and M. Segal. 2014. Dendritic spines: the locus of structural and functional plasticity. Physiol. Rev. 94 :141–188. 10.1152/physrev.00012.2013 24382885 Scheiffele, P. 2003. Cell-cell signaling during synapse formation in the CNS. Annu. Rev. Neurosci. 26 :485–508. 10.1146/annurev.neuro.26.043002.094940 12626697 Sheffler-Collins, S.I., and M.B. Dalva. 2012. EphBs: an integral link between synaptic function and synaptopathies. Trends Neurosci. 35 :293–304. 10.1016/j.tins.2012.03.003 22516618 Shen, K., and C.W. Cowan. 2010. Guidance molecules in synapse formation and plasticity. Cold Spring Harb. Perspect. Biol. 2 :a001842. 10.1101/cshperspect.a001842 20452946 Shen, K., and P. Scheiffele. 2010. Genetics and cell biology of building specific synaptic connectivity. Annu. Rev. Neurosci. 33 :473–507. 10.1146/annurev.neuro.051508.135302 20367446 Sheng, M., and E. Kim. 2011. The postsynaptic organization of synapses. Cold Spring Harb. Perspect. Biol. 3 :a005678. 10.1101/cshperspect.a005678 22046028 Sheng, M., and C. Sala. 2001. PDZ domains and the organization of supramolecular complexes. Annu. Rev. Neurosci. 24 :1–29. 10.1146/annurev.neuro.24.1.1 11283303 Siddiqui, T.J., and A.M. Craig. 2011. Synaptic organizing complexes. Curr. Opin. Neurobiol. 21 :132–143. 10.1016/j.conb.2010.08.016 20832286 Sloniowski, S., and I.M. Ethell. 2012. Looking forward to EphB signaling in synapses. Semin. Cell Dev. Biol. 23 :75–82. 10.1016/j.semcdb.2011.10.020 22040917 Söhl, G., S. Maxeiner, and K. Willecke. 2005. Expression and functions of neuronal gap junctions. Nat. Rev. Neurosci. 6 :191–200. 10.1038/nrn1627 15738956 Südhof, T.C. 2008. Neuroligins and neurexins link synaptic function to cognitive disease. Nature. 455 :903–911. 10.1038/nature07456 18923512 Südhof, T.C. 2012. The presynaptic active zone. Neuron. 75 :11–25. 10.1016/j.neuron.2012.06.012 22794257 Sun, S., H. Zhang, J. Liu, E. Popugaeva, N.J. Xu, S. Feske, C.L. White III, and I. Bezprozvanny. 2014. Reduced synaptic STIM2 expression and impaired store-operated calcium entry cause destabilization of mature spines in mutant presenilin mice. Neuron. 82 :79–93. 10.1016/j.neuron.2014.02.019 24698269 Turrigiano, G.G., and S.B. Nelson. 2004. Homeostatic plasticity in the developing nervous system. Nat. Rev. Neurosci. 5 :97–107. 10.1038/nrn1327 14735113 Waites, C.L., A.M. Craig, and C.C. Garner. 2005. Mechanisms of vertebrate synaptogenesis. Annu. Rev. Neurosci. 28 :251–274. 10.1146/annurev.neuro.27.070203.144336 16022596 Williams, S.E., F. Mann, L. Erskine, T. Sakurai, S. Wei, D.J. Rossi, N.W. Gale, C.E. Holt, C.A. Mason, and M. Henkemeyer. 2003. Ephrin-B2 and EphB1 mediate retinal axon divergence at the optic chiasm. Neuron. 39 :919–935. 10.1016/j.neuron.2003.08.017 12971893 Witter, M.P. 2007. Intrinsic and extrinsic wiring of CA3: indications for connectional heterogeneity. Learn. Mem. 14 :705–713. 10.1101/lm.725207 18007015 Wolting, C.D., E.K. Griffiths, R. Sarao, B.C. Prevost, L.E. Wybenga-Groot, and C.J. McGlade. 2011. Biochemical and computational analysis of LNX1 interacting proteins. PLoS One. 6 :e26248. 10.1371/journal.pone.0026248 22087225 Xie, Y., W. Zhao, W. Wang, S. Zhao, R. Tang, K. Ying, Z. Zhou, and Y. Mao. 2001. Identification of a human LNX protein containing multiple PDZ domains. Biochem. Genet. 39 :117–126. 10.1023/A:1010269908398 11521506 Xu, N.-J., and M. Henkemeyer. 2009. Ephrin-B3 reverse signaling through Grb4 and cytoskeletal regulators mediates axon pruning. Nat. Neurosci. 12 :268–276. 10.1038/nn.2254 19182796 Xu, N.J., S. Sun, J.R. Gibson, and M. Henkemeyer. 2011. A dual shaping mechanism for postsynaptic ephrin-B3 as a receptor that sculpts dendrites and synapses. Nat. Neurosci. 14 :1421–1429. 10.1038/nn.2931 21964490 Yuste, R., and T. Bonhoeffer. 2004. Genesis of dendritic spines: insights from ultrastructural and imaging studies. Nat. Rev. Neurosci. 5 :24–34. 10.1038/nrn1300 14708001 Zhu, X.N., X.D. Liu, S. Sun, H. Zhuang, J.Y. Yang, M. Henkemeyer, and N.J. Xu. 2016 a. Ephrin-B3 coordinates timed axon targeting and amygdala spinogenesis for innate fear behaviour. Nat. Commun. 7 :11096. 10.1038/ncomms11096 27008987 Zhu, X.N., X.D. Liu, H. Zhuang, M. Henkemeyer, J.Y. Yang, and N.J. Xu. 2016 b. Amygdala EphB2 Signaling Regulates Glutamatergic Neuron Maturation and Innate Fear. J. Neurosci. 36 :10151–10162. 10.1523/JNEUROSCI.0845-16.2016 27683910 Zucker, R.S. 1999. Calcium- and activity-dependent synaptic plasticity. Curr. Opin. Neurobiol. 9 :305–313. 10.1016/S0959-4388(99)80045-2 10395573 Zuo, Y., G. Yang, E. Kwon, and W.B. Gan. 2005. Long-term sensory deprivation prevents dendritic spine loss in primary somatosensory cortex. Nature. 436 :261–265. 10.1038/nature03715 16015331

PMC006xxxxxx/PMC6314548.txt

==== Front J Cell Biol J Cell Biol jcb jcb The Journal of Cell Biology 0021-9525 1540-8140 Rockefeller University Press 30373907 201805016 10.1083/jcb.201805016 Research Articles Article 34 43 Dynein activator Hook1 is required for trafficking of BDNF-signaling endosomes in neurons Hook1 activates BDNF-endosome motility http://orcid.org/0000-0001-5772-1667 Olenick Mara A. 234 Dominguez Roberto 123 http://orcid.org/0000-0001-5389-4114 Holzbaur Erika L.F. 123 1 Department of Physiology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA 2 The Pennsylvania Muscle Institute, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA 3 Biochemistry and Molecular Biophysics Graduate Group, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA 4 Center for Engineering Mechanobiology, University of Pennsylvania, Philadelphia, PA Correspondence to Erika L.F. Holzbaur: holzbaur@pennmedicine.upenn.edu 07 1 2019 218 1 220233 03 5 2018 18 9 2018 15 10 2018 © 2018 Olenick et al. 2018 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). Olenick et al. find that in hippocampal neurons, the cytoplasmic dynein activator Hook1 specifically activates retrograde transport of BDNF–TrkB-signaling endosomes but is not required for motility of other retrograde cargo, supporting a model of cargo-specific effectors for efficient regulation of dynein. Axonal transport is required for neuronal development and survival. Transport from the axon to the soma is driven by the molecular motor cytoplasmic dynein, yet it remains unclear how dynein is spatially and temporally regulated. We find that the dynein effector Hook1 mediates transport of TrkB–BDNF-signaling endosomes in primary hippocampal neurons. Hook1 comigrates with a subpopulation of Rab5 endosomes positive for TrkB and BDNF, which exhibit processive retrograde motility with faster velocities than the overall Rab5 population. Knockdown of Hook1 significantly reduced the motility of BDNF-signaling endosomes without affecting the motility of other organelles. In microfluidic chambers, Hook1 depletion resulted in a significant decrease in the flux and processivity of BDNF-Qdots along the mid-axon, an effect specific for Hook1 but not Hook3. Hook1 depletion inhibited BDNF trafficking to the soma and blocked downstream BDNF- and TrkB-dependent signaling to the nucleus. Together, these studies support a model in which differential association with cargo-specific effectors efficiently regulates dynein in neurons. National Institutes of Health https://doi.org/10.13039/100000002 R35 GM126950 P01 GM087253 T32 GM07229 National Science Foundation https://doi.org/10.13039/100000001 CMMI: 15-48571 ==== Body pmcIntroduction Axonal transport is vital for the maintenance and survival of neurons. Axons have a uniformly polarized microtubule array in which the faster-growing plus ends of microtubules are oriented toward the distal terminal. These microtubules serve as a highway for fast organelle trafficking mediated by molecular motors. While multiple plus end–directed kinesins are responsible for delivery of cargo to the distal end of the axon, the minus end–directed motor cytoplasmic dynein is solely responsible for trafficking a wide variety of cargo back to the soma including autophagosomes, endosomes, and mitochondria (Maday et al., 2014). These organelles not only differ in their lipid and protein compositions, but they also display distinct motility properties. It remains unclear how cytoplasmic dynein attaches to each of its cargos and how the motor is regulated to facilitate the precise trafficking of organelles to the soma. Cytoplasmic dynein 1 (referred to in this study as dynein) is a 1.4-MD AAA+ motor complex that drives the majority of minus end–directed motility in the cell. Alone, dynein is a flexible dimer with low processivity, taking many sideways or backward steps along the microtubule lattice (Reck-Peterson et al., 2006; Ross et al., 2006). Processive motility is enhanced when dynein binds to dynactin, a 1-MD multisubunit complex that reorients the dynein dimer for proper recruitment and motility along microtubules (Ayloo et al., 2014; Zhang et al., 2017). While dynactin has been suggested to play a role in cargo interaction (Zhang et al., 2011; Yeh et al., 2012), adaptor and scaffolding proteins are required to link cargo to the dynein–dynactin motor complex (Kardon and Vale, 2009; Fu and Holzbaur, 2014). Recently, a set of coiled coil effector proteins including BICD2, Hook1, Hook3, Spindly, and NINL have been shown to enhance the dynein–dynactin interaction and induce superprocessive motility (McKenney et al., 2014; Schlager et al., 2014; Olenick et al., 2016; Schroeder and Vale, 2016; Redwine et al., 2017). BICD2, the best characterized of these dynein effectors, has been shown to increase the affinity of dynein–dynactin interaction through coiled-coil contacts along the Arp1 filament that forms the core of dynactin (Chowdhury et al., 2015; Urnavicius et al., 2015). BICD2 also interacts with the N-terminal tail of the dynein heavy chain (Chowdhury et al., 2015; Urnavicius et al., 2015) and the dynein light intermediate chain 1 (LIC1; Schroeder et al., 2014; Lee et al., 2018), leading to a stabilization of the dynein–dynactin-effector complex. Some dynein effectors can recruit two dynein dimers to a single dynactin, which further enhances the force and velocity of the motor complex (Grotjahn et al., 2018; Urnavicius et al., 2018). Hook proteins (HookA or Hok1) are dynein effectors first characterized in filamentous fungi and shown to link dynein to early endosomes (Bielska et al., 2014; Zhang et al., 2014). In mammalian cells, three highly conserved Hook proteins are expressed: Hook1, Hook2, and Hook3. These proteins are characterized by an N-terminal Hook domain, which binds LIC1 of dynein (Schroeder and Vale, 2016; Lee et al., 2018). The Hook domain is followed by a central coiled-coil region and a less well-conserved C-terminal cargo-binding domain (Bielska et al., 2014; Zhang et al., 2014). In vitro studies show that the binding of either Hook1 or Hook3 enhances the dynein–dynactin interaction, leading to significant increases in velocity and run lengths (McKenney et al., 2014; Olenick et al., 2016; Schroeder and Vale, 2016). While Hook1 and Hook3 have been identified as dynein activators in vitro, the role of these proteins in dynein-mediated cargo transport in mammalian cells is less clear. Hook2 has been linked to centrosomal function and homeostasis (Szebenyi et al., 2007; Guthrie et al., 2009; Moynihan et al., 2009), while Hook1 and Hook3 have been implicated in a variety of endosomal trafficking pathways, although there is still no clear consensus on the specific roles of each isoform (Luiro et al., 2004; Xu et al., 2008; Maldonado-Báez et al., 2013). The highly polarized nature and spatial compartmentalization of neurons provide an excellent system to study the role of Hook proteins in endosomal transport. Initial work from Guo et al. (2016) suggested that Hook1 and Hook3 colocalize with retrograde Rab5a vesicles in hippocampal neurons and that knockdown (KD) of Hook1 and Hook3 reduced the retrieval of transferrin receptor from the axon (Guo et al., 2016). These data support a potential role for Hook proteins in dynein-mediated trafficking in axons, prompting us to investigate this question in more detail. In this study, we investigated the role of Hook1 in the dynein-driven transport of endosomes along the axons of hippocampal neurons. We found that Hook1 comigrates with subpopulations of Rab5- and Rab7-positive endosomes. While loss of Hook1 did not significantly change the overall motility of Rab5- or Rab7-positive endosomes, Hook1 siRNA depletion significantly reduced the motility of a specific endosomal compartment that we identified as TrkB–brain-derived neurotrophic factor (BDNF)-signaling endosomes. The motility of TrkB–BDNF-signaling endosomes is also lost if the interaction of Hook1 with dynein is disrupted by targeted mutations at the Hook1-LIC1 interface. In addition, Hook1 is enriched in the distal axon, distinct from the cellular distribution of other dynein effectors like Rab-interacting lysosomal protein (RILP) or BICD2, suggesting a specific function in trafficking from the distal axon. Using microfluidic chambers to model the distal axonal transport of BDNF-signaling endosomes, we found that KD of Hook1 significantly reduced the flux and the processivity of BDNF transport from the distal axon to the soma. In contrast, KD of Hook3 did not affect BDNF uptake or transport. Loss of Hook1 also produced a functional block in downstream BDNF-dependent signaling to the nucleus, which is vital for neuronal survival and maintenance. Overall, this work supports a model in which Hook1 acts as a specific dynein effector for BDNF-signaling endosomes trafficking from the distal axon to induce downstream signaling in the soma of primary hippocampal neurons. Results Hook1 comigrates with the endosomal markers Rab5 and Rab7 To investigate the role of Hook1 in endosomal trafficking, we first expressed the early endosome marker Rab5-GFP or the late endosome marker Rab7-GFP in primary rat hippocampal neurons to observe endosomal motility. Neurons were imaged 7–8 d in vitro (DIV) at 48 h after transfection using live-cell confocal microscopy. Focusing on the axon, Rab5-GFP endosomes were found to be enriched in the distal axon, while Rab7-GFP endosomes were found throughout the distal and mid-axon. In the axon, Rab5-GFP endosomes were mainly stationary or bidirectional, with ∼80% moving <10 μm in any net direction (Fig. S1 B). In contrast, 50% of Rab7-GFP endosomes displayed net retrograde motility (Fig. S1 E). Next, Hook1-Halo was coexpressed with Rab5-GFP or Rab7-GFP to assess whether Hook1 comigrates with a specific population of endosomes since previous studies had found Hook proteins on Rab5-positive endosomes (Bielska et al., 2014; Zhang et al., 2014; Guo et al., 2016). Surprisingly, we observed comigration of Hook1 with both a subpopulation of Rab5 endosomes and a subpopulation of Rab7 endosomes (Fig. 1, A and B). Previous in vitro research indicates that Hook1 enhances the processivity of dynein (Olenick et al., 2016), so we asked whether Hook1-positive endosomes displayed a distinctive retrograde bias in their motility or were significantly more processive than the overall population of axonal endosomes. We found that Hook1-positive Rab5 endosomes had an increased retrograde bias as seen by an increased retrograde:anterograde ratio as compared with neurons expressing Rab5-GFP only (Fig. 1, C and D). Hook1-positive Rab5 endosomes also showed an increase in retrograde-directed instantaneous velocity as compared with the overall population of Rab5-GFP–positive endosomes (Fig. 1 E). In comparison, Hook1-positive Rab7 endosomes had the same retrograde bias and retrograde-directed instantaneous velocities seen for the overall population of Rab7-GFP endosomes, where ∼50% of organelles displayed fast retrograde motility (Fig. 1, F–H). Overall, this analysis indicates that motile endosomes positive for Hook1 were primarily fast, retrogradely moving organelles, indicative of processive dynein-mediated motility. Figure 1. Hook1 comigrates with Rab5 and Rab7 endosomes. (A) Mid-axons of hippocampal neurons expressing Hook1-Halo and Rab5- or Rab7-GFP. Arrows show colocalized Hook1 with the indicated Rab endosome. Bars, 4 µm. (B) The percentage of endosomes with Hook1-colocalization. Scatter plot shows mean ± SEM. Rab5: n = 17 neurons; Rab7: n = 10 neurons. (C) Kymograph of Rab5-GFP and Hook1-Halo motility in axon of hippocampal neuron. Arrows show comigrating Hook1-Rab5 organelles. Bars: 4 µm (horizontal); 20 s (vertical). (D) The ratio of retrograde to anterograde motility events in neurons expressing Rab5-GFP + Hook1-Halo or Rab5-GFP only. Unpaired t test (*, P = 0.0177). Rab5: n = 16 neurons; Rab5 + Hk1: n = 10 neurons. (E) Cumulative histogram of retrograde instantaneous velocities of events in neurons expressing Rab5-GFP + Hook1-Halo or Rab5-GFP only. Dashed line represents one phase decay regression line; half-life (λ) and 95% CI values are shown. (F) Kymograph of Rab7-GFP and Hook1-Halo motility in axon of hippocampal neuron. Arrows show comigrating Hook1-Rab7 organelles. Bars: 4 µm (horizontal); 20 s (vertical). (G) The ratio of retrograde to anterograde motility events in neurons expressing Rab7-GFP + Hook1-Halo or Rab7-GFP only. Unpaired t test (ns, P = 0.8223). Rab7: n = 13 neurons; Rab7 + Hk1: n = 7 neurons. (H) Cumulative histogram of retrograde instantaneous velocities of events in neurons expressing Rab7-GFP + Hook1-Halo or Rab7-GFP only. Dashed line represents one phase decay regression line; half-life(λ) and 95% CI values are shown. Hook1 KD reduces the motility of BDNF-positive signaling endosomes As Hook1 comigrates with a subpopulation of Rab5- and Rab7-positive endosomes, we next asked whether Hook1 depletion would alter the motility of these endosomes. We used a rat siRNA pool to KD Hook1 in cultured neurons; 60% depletion of Hook1 was observed in PC12 cells using this approach (Fig. S1, G and H). Rab5-GFP or Rab7-GFP was transfected along with Hook1 siRNA and imaged 48 h after transfection at DIV 7–8 along the mid- or distal regions of the axon. Hook1 KD did not induce significant differences in the motile fraction or the directionality of either Rab5- or Rab7-positive endosomes at a population level (Fig. S1, A–F). Hook1 KD did induce a trend toward decreased flux of Rab5 endosomes, but the effect was not significantly different than control (Fig. S1 C). We hypothesized that Hook1 might play a more specific role in the transport of a subpopulation of neuronal endosomes, so we focused on signaling endosomes, which initiate distally and mature through the Rab5 and Rab7 endosomal pathway (Deinhardt et al., 2006; Ye et al., 2018). In neurons, neurotrophic factors such as BDNF bind to their transmembrane kinase receptors and are endocytosed to form signaling endosomes that undergo retrograde transport toward the nucleus, leading to changes in gene expression (Cosker and Segal, 2014; Scott-Solomon and Kuruvilla, 2018). To focus on a signaling endosomal population in neurons, we expressed the neurotrophic receptor TrkB-RFP and GFP-Hook1 and found that the labeled proteins comigrated in the axons of hippocampal neurons (Fig. 2 A). Consistent with this observation, immunoisolation of TrkB vesicles from mouse brain lysates demonstrated the coprecipitation of Hook1 along with subunits of both dynein and dynactin (Fig. S2 A). Figure 2. Hook1 KD reduces TrkB–BDNF-signaling endosome motility. (A) Time series of the mid-axon of hippocampal neurons expressing GFP-Hook1 and TrkB-RFP. Arrows show retrograde comigration of Hook1 with TrkB vesicles. Bars, 5 µm. (B) Kymographs of TrkB-RFP in control or Hk1 KD neurons. Traced events shown below are color-coded for ease of interpretation. Arrows show pausing in retrograde events. Bars, 10 µm; 1 min total. (C) Colocalization images of BDNF-Qdots with GFP-Hook1 in the axon of a hippocampal neuron. Arrows show Hook1 puncta colocalized with BDNF. Bar, 2 µm. (D) Line scan through axon in C. Arrow points to colocalized Hook1 and BDNF. (E) Quantification of BDNF-Qdots motility. Bar graph shows mean ± SEM; Kruskal–Wallis one-way ANOVA (***, P < 0.0001; *, P = 0.030; ns, P = 0.7405). Mock: n = 103 neurons; Hk1 KD: n = 63 neurons; Hk1 KD + Hk1-Halo: n = 55 neurons. (F) Kymographs of BDNF-Qdots in control, Hk1 KD, and Hk1 KD + Hk1-Halo neurons. Traced events shown below are color-coded for ease of interpretation. Green arrows indicate retrograde events. Bars: 5 µm (horizontal); 5 s (vertical). Next, we expressed TrkB-RFP in control and Hook1 KD neurons. Upon Hook1 depletion, retrograde TrkB endosomes displayed less processive motility (as indicated by more pausing [66.6% retrograde events displayed pausing per neuron]) than in control cells (32.2% retrograde events with pausing per neuron; Fig. 2 B). To more directly measure the motility of signaling endosomes, we monitored the uptake and motility of the TrkB ligand BDNF. First, we looked for colocalization of BDNF with Hook1-Halo in neurons. Neurons were serum starved for 1 h, and BDNF-biotin was then added to neuronal cultures for at least 1 h before fixation. Approximately 48% of Hook1-Halo puncta colocalized with BDNF–Alexa Fluor 633 in fixed neurons (Fig. S2, B and C). In live neurons, we also found that GFP-Hook1 colocalized with BDNF-conjugated quantum dots (BDNF-Qdots; Fig. 2, C and D). Next, we investigated the motility of BDNF-Qdots in control or Hook1 KD neurons. Neurons with Hook1 KD displayed significantly reduced retrograde BDNF-Qdot motility compared with control neurons (Fig. 2, E and F). This motility defect could be rescued by expression of siRNA-resistant human Hook1-Halo (Fig. 2, E and F). To assess whether the observed effects of Hook1 depletion were specific for signaling endosomes or instead represented a more generalized inhibition of organelle motility, we imaged mitochondrial motility as well as another highly processive retrograde cargo, LC3B-positive autophagosomes. Comparisons of the motility of mitochondria and autophagosomes in Hook1 KD and control neurons demonstrate that no differences were induced by Hook1 depletion (Fig. S1, G–L). Together with the observations that Hook1 depletion does not induce changes in the motility of the overall population of either Rab5-positive or Rab7-positive endosomes, these results indicate that Hook1 plays an essential and specific role in TrkB–BDNF–signaling endosome motility in axons. A direct interaction of Hook1 with dynein is important for signaling endosome motility Dynein subunit LIC1 interacts with several dynein effectors including BICD2, RILP, and FIP3 (Schroeder et al., 2014). Recent studies have found that the Hook domain conserved in both Hook1 and Hook3 also mediates a direct interaction with LIC1 and is important for Hook-mediated dynein processivity (Schroeder and Vale, 2016; Lee et al., 2018). To determine whether the Hook1–LIC1 interaction is important for signaling endosome motility in neurons, we analyzed two constructs: Hook1(Q149A, I156A), based on previous mutations in Hook3 shown to diminish the interaction with LIC1 (Schroeder and Vale, 2016), and Hook1(M146D, I156D), based on our recent structure of a Hook3–LIC1 complex (Fig. 3, A and B; Lee et al., 2018). We used total internal reflection fluorescence (TIRF) microscopy to perform motility assays with single-molecule resolution to test these mutant constructs and found that both Hook1(Q149A, I156A) and Hook1(M146D, I156D) significantly inhibited dynein-driven motility along microtubules (Fig. 3, C and D). Figure 3. Hook1 requires interaction with LIC1 for signaling endosome motility. (A) Diagram of Hook1 domain structure with mutated constructs below. Arrowheads indicate point mutations. (B) Hook3 (yellow)–LIC1 helix (green) structure (PDB: 6B9H) with residues that were mutated highlighted in magenta (Hook1 residue numbering). (C) Kymographs of Halo-Hook1 constructs from a single-molecule TIRF motility assay. Arrows indicate motile events. Bars, 5 µm; total length, 1 min. (D) Quantification of Hook1 motility in a TIRF motility assay. Scatter plot shows mean ± SEM; Kruskal–Wallis one-way ANOVA (***, P < 0.0001; **, P = 0.0012). n = 14–15 videos from three individual experiments. MT, microtubule. (E) Kymographs of BDNF-Qdots in axons of hippocampal neurons. Traced events below are color-coded for ease of interpretation. Green arrows indicate retrograde events. (F) Quantification of BDNF motility. Bar graph shows mean ± SEM; Mann–Whitney t test (*, P = 0.0262). Mock: n = 31 neurons; Hk1(Q149A, I156A): n = 46 neurons. (G) Kymographs of BDNF-Qdots in axons of hippocampal neurons in Hk1 KD-rescue experiments with C-terminal mutated Hook1 constructs. Traced events below are color-coded for ease of interpretation. (E and G) Bars: 5 µm (horizontal); 5 s (vertical). (H) Quantification of BDNF motility in KD-rescue experiments. Bar graph shows mean ± SEM; one-way ANOVA (***, P < 0.0001; ns, P = 0.58–0.98). Mock: n = 29 neurons; Hk1 KD: n = 22 neurons; Hk1 KD + Hk1(K672A,S673A): n = 13 neurons; Hk1 KD + Hk1(S693A,D694A): n = 24 neurons. Next, we tested whether mutating the binding interface of Hook1 and LIC1 would disrupt signaling endosome motility in hippocampal neurons. Similar to our results with Hook1 siRNA KD, BDNF-Qdot motility was significantly reduced in neurons expressing GFP-Hook1(Q149A, I156A) (Fig. 3, E and F). These results indicate that the interaction of Hook1 with LIC1 is important not only for in vitro motility but also during cargo transport of BDNF-signaling endosomes in neurons. The C-terminal domain of the Hook proteins is thought to specify cargo binding. Comparisons of C-terminal Hook1 sequences from several species identify a high degree of sequence conservation (Fig. S3). There is more limited sequence conservation between Hook1 and Hook3, but we noted that immediately following a conserved sequence there was a pair of charged/polar residues in Hook1 not found within Hook3. We engineered a double mutation to generate Halo-tagged Hook1(K672A,S672A) and asked whether this construct could rescue BDNF-Qdot motility in Hook1-depleted neurons. As shown in Fig. 3 G, this construct could not rescue signaling endosome motility. Next, we focused on a 10-aa residue insertion in the C-terminal domain of Hook1 relative to Hook3 and again mutated a pair of polar/charged residues to generate Hook1(S693A, K694A)-Halo. We tested this construct in KD-rescue experiments assaying for BDNF-Qdot motility and found no rescue (Fig. 3 H), further supporting the importance of the C-terminal domain of Hook1 in mediating organelle-specific dynein-driven motility. Hook1 is enriched in the distal axon and is distinctly localized from other dynein effectors There are now several dynein effectors implicated in the regulation of dynein-driven motility in cells (Kardon and Vale, 2009; Fu and Holzbaur, 2014; Reck-Peterson et al., 2018). We hypothesized that dynein effectors might display differential localizations in neurons, reflecting distinct roles in intracellular transport. To address this question, we individually expressed Hook1-Halo, RILP-GFP, and BICD2-GFP in hippocampal neurons for 48 h and then fixed the cells on DIV 7–8. Confocal z stacks were captured of the distal axon, mid-axon, and soma of neurons. Images were deconvolved, and the somal localization was scored as either punctate or cytoplasmic. Over 80% of BICD2- and RILP-expressing neurons displayed an accumulation of these effectors in the soma, associated with either large puncta or clusters, while the distribution of Hook1 was predominantly cytoplasmic within the soma (Fig. 4, A and B). In axons, dynein effectors were imaged either in the mid- or distal axon and puncta quantified per unit length, comparing levels along the same axon to assess relative enrichment of individual effectors. BICD2 had a generally sparse distribution in axons compared with RILP and Hook1. In contrast, RILP was enriched in the mid-axon, and Hook1 was enriched in the distal axon (Fig. 4, A and C). We also quantified the density of puncta for each dynein effector within the growth cone and found there was significantly more Hook1 puncta in the growth cone as compared with BICD2 and RILP (Fig. 4 D). The differential localization of these effectors is consistent with their proposed roles in organelle transport. BICD2 is linked to Rab6 vesicles (Matanis et al., 2002; Matsuto et al., 2015), which are mainly localized to the soma. RILP is a Rab7 adaptor (Cantalupo et al., 2001; Wu et al., 2005; Johansson et al., 2007; Rocha et al., 2009), and Rab7 vesicles are enriched in the mid-axon and soma. In contrast, the enrichment of Hook1 in the distal axon suggests that this effector is involved early in the pathway of neurotrophic factor uptake and trafficking via signaling endosomes. Figure 4. Hook1 localizes to distal axon, while other dynein effectors are enriched in other compartments. (A) Representative images of neurons expressing indicated dynein effectors. White arrows point to effector puncta. Dashed line represents the line used for line scan plots , but line is shifted below the axon so as to not obscure primary data. Line scan graphs of mid- and distal axons are shown to the right of the image. Black arrows point to corresponding puncta peaks. (B) Quantification of percentage of neurons with puncta in soma. Bar graph shows mean ± SEM; one-way ANOVA (***, P ≤ 0.0001; ns, P = 0.3964). n = 3–4 individual experimental averages, 23–26 cells. (C) Graphs of dynein effector enrichment in mid versus distal axons. Red lines indicate enrichment in the distal region of axon. Black lines indicate enrichment in the mid-axon. BICD2: n = 21 neurons; RILP: n = 27 neurons; Hook1: n = 27 neurons. (D) Graph of dynein effector puncta per growth cone. Scatter plot shows mean ± SEM; one-way ANOVA (***, P < 0.0001; ns, P = 0.8901). n = 16–23 growth cones. (E) Representative images of proximity ligation assay of Hook1-Halo–expressing distal axons with and without anti-DIC. Hook1–DIC complexes are presented in green and GFP in magenta. Arrow indicates Hook1–DIC complexes. (A and E) Bars, 10 µm. (F) Quantification of distal axons with Hook1–DIC PLA puncta present. Bar graph shows mean ± SEM; unpaired t test (*, P = 0.0119). n = 2 individual experimental averages, 19–38 neurons. To further visualize the Hook1–dynein interaction, we used a proximity ligation assay to image dynein–Hook1 colocalization. Neurons expressing Hook1-Halo were stained with antibodies to the Halo-tag and to endogenous dynein (anti–dynein intermediate chain [DIC]) and visualized using secondary antibodies conjugated to complementary oligonucleotides to visualize complex formation. Using this assay, Hook1–dynein complexes were localized to the distal axon, a distribution not observed in control reactions lacking anti-DIC or anti-Halo antibody (Fig. 4, E and F). In contrast, we did not see the same distal axon signal using the proximity ligation assay to visualize dynein complexed with either RILP or BICD2. These results show that Hook1 is enriched in the distal axon and primed for motility as it is in a complex with dynein. Hook1 depletion decreases the flux and processivity of BDNF-signaling endosomes from the distal axon To better model the uptake and transport of BDNF from the distal axon, we used microfluidic devices in which axons grow through microchambers to reach the fluidly isolated axonal chamber, permitting BDNF application only to the distal axons (Fig. 5 A). Hippocampal neurons were electroporated with GFP fill and Hook1 siRNA before plating in the microfluidic devices. Neurons were cultured for 7–8 d, allowing the axons to extend to the distal chamber. Prior to imaging, BDNF-Qdots were added to the axonal chamber of the devices. We compared the motility properties of BDNF-Qdots in Hook1 KD and control neurons (see Videos 1 and 2). While BDNF-Qdots still exhibited a retrograde bias in motility upon Hook1 KD, the flux was greatly reduced compared with control conditions (Fig. 5, B–D). In contrast, neurons transfected with Hook3 siRNA did not show this decrease in BDNF-Qdot flux (Fig. 5 D). In addition to decreased flux, we also noticed a difference in the size of BDNF-Qdot vesicles. Previous research has shown that multiple neurotrophic factor–bound Qdots may be internalized into a single signaling endosome (Cui et al., 2007). Thus, we measured the apparent size of BDNF-Qdot organelles as a measure of the number of internalized BDNF-Qdots. We found a significant reduction in organelle area in Hook1 KD neurons (0.39 µm2, with 0.30–0.48 95% confidence interval [CI]) compared with control neurons (0.62 µm2, with 0.51–0.74 95% CI), suggesting less BDNF-Qdots are being endocytosed per vesicle. Overall, this work suggests that Hook1 depletion reduces not only the number of BDNF organelles trafficked down the axon but also the load of individual organelles. Figure 5. Hook1 KD reduces flux of BDNF from distal axon. (A) Schematic of the microfluidic device and experimental setup. (B) Kymographs of BDNF in MOCK, Hk1 KD, and Hk3 KD neurons grown in the microfluidic device. Arrows point to retrograde events. Bars: 10 µm (horizontal); 10 s (vertical). (C) Motility fractioned into retrograde, anterograde, and nonmotile events per neuron. Bar graph shows mean ± SEM; two-way ANOVA (ns, P > 0.113) Mock: n = 35 neurons; Hk1 KD: n = 34 neurons. (D) Quantification of flux of BDNF-Qdots in mid-axons. Scatter plot shows mean ± SEM; one-way ANOVA (***, P < 0.0001; ns, P = 0.9994). Mock: n = 54 neurons; Hk1 KD: n = 36 neurons; Hk3 KD: n = 29 neurons. (E) Number of switches in BDNF-Qdot events. Scatter plot shows mean ± SEM; Mann–Whitney t test (**, P = 0.001). Mock: n = 110 events; Hk1 KD: n = 84 events. (F) Pause duration of retrograde BDNF-Qdot events. Scatter plot shows mean ± SEM, Mann-Whitney t test (**, P = 0.0062). Mock: n = 57 events; Hk1 KD: n = 33 events. (G) Net velocity of retrograde BDNF-Qdot events. Scatter plot shows mean ± SEM, unpaired t test (**, P = 0.0061). Mock: n = 63 events; Hk1 KD: n = 34 events. While significantly fewer BDNF-Qdots were observed to traffic along the axon in Hook1-depleted neurons, we analyzed the motility properties of these organelles to see whether loss of the dynein activator would reduce the processivity of signaling endosomes from the distal axon. In Hook1 KD cells, BDNF-Qdots showed less directed motility along the axon, with significantly more directional switching within individual runs (Fig. 5 E). In addition, retrograde events also displayed increased pause duration and reduced net velocity (Fig. 5, F and G). These results suggest that Hook1 does act as a dynein activator to increase the processivity of retrograde BDNF-signaling endosomes once they have been endocytosed. Since we observed reduced flux and impaired motility of signaling endosomes in Hook1 KD neurons, we hypothesized there might be a resulting accumulation of BDNF-Qdots at the distal ends of axons due to loss of dynein transport (Fig. 6 A). We imaged the distal regions of axons after several washes to remove excess, surface-associated BDNF-Qdots. Quantification of BDNF-Qdot signal per unit area of axon showed no difference between Hook1 KD and control neurons (Fig. 6, B and C). Similar results were seen when dynein motility was blocked with the dynein inhibitor Ciliobrevin D (Fig. 6, B and C), suggesting there is down-regulation of TrkB endocytosis when motility is impaired, preventing distal accumulation. One way to reduce endocytosis of BDNF is to reduce the amount of its receptor TrkB at the plasma membrane. We measured the amount of plasma membrane–associated TrkB on the surface of Hook1 KD or control axons using an antibody to the TrkB extracellular domain, which is not conserved in other Trk proteins. Neurons were fixed and stained with anti-TrkB (aa 54–67) without permeabilization. There was no significant change in surface TrkB levels at the axon tips with Hook1 depletion, suggesting that a reduction in TrkB levels is not contributing to the reduced flux we observed with Hook1 KD (Fig. 6, D and E). Together, these results suggest that endocytosis of BDNF is down-regulated when dynein motility is impaired, which may constitute a potential mechanism to reduce the distal accumulation of cargos in axons. Figure 6. Impaired signaling endosome motility reduces BDNF endocytosis. (A) Schematic model of two possible effects of Hook1 KD on distal axons. (B) Representative images of distal tips of axons with BDNF-Qdots. Arrows indicate BDNF-Qdots, and the dashed line is the cell outline. (C) Area of Qdots in the distal axons normalized by area of axon quantified. Scatter plot shows mean ± SEM; Kruskal–Wallis one-way ANOVA (ns, P > 0.082). Mock: n = 58 neurons; Hk1 KD: n = 46 neurons; Ciliobrevin D [CilioD]: n = 3–6 neurons. (D) Surface TrkB staining with anti-TrkB in distal axon of fixed neurons. Bottom panel (anti-Rabbit594) shows the secondary antibody only (as a control). (B and D) Bars, 5 µm. (E) Area of surface TrkB per micrometer axonal length. The anti-Rabbit 594 condition is a control for the secondary antibody. Scatter plot shows mean ± SEM; one-way ANOVA (***, P = 0.0004; **, P = 0.0055; ns, P = 0.9741). Mock: n = 40 neurons; Hk1 KD: n = 32 neurons; anti-Rabbit594: n = 10 neurons. Hook1 KD reduces downstream BDNF signaling to the nucleus BDNF binds TrkB, which then recruits signaling kinases to produce transcriptional changes in the nucleus (Cosker and Segal, 2014; Mitre et al., 2017). The transport of signaling endosomes is required to elicit downstream signaling to the nucleus (Ye et al., 2003; Heerssen et al., 2004), so we asked whether loss of Hook1 would affect this signaling pathway due to reduced flux of BDNF. In our Qdot assays, significantly less BDNF-Qdots accumulate in the soma upon Hook1 depletion (Fig. 7, A and B). To measure the downstream signaling, we monitored phosphorylated nuclear cAMP response element–binding protein (CREB), which has been previously shown to increase after treatment with BDNF (Watson et al., 2001). Hook1 KD and control neurons were grown in culture for 7 d in microfluidic devices and then treated with 1 nM BDNF for 1 h before being fixed and stained with anti-pCREB(Ser133) antibody. Using epifluorescence microscopy, we imaged neurons with axons that grew through the microchannels to reach the axonal compartment. In control cells, BDNF-treated neurons had increased nuclear pCREB compared with nontreated (NT) control neurons (Fig. 7, C–E). In Hook1 KD neurons, BDNF-treated cells did not show increased pCREB staining compared with NT cells (Fig. 7, C–E). These results indicate that the reduced flux of BDNF impairs downstream signaling to the nucleus in Hook1-deficient neurons. Figure 7. Loss of Hook1 leads to loss of downstream signaling measured by pCREB levels. (A) Representative images of somas after BDNF-Qdots treatment. (B) Quantification of somas with no BDNF-Qdots present. Bar graph shows mean ± SEM; unpaired t test (*, P = 0.0131). n = 3 individual experimental averages, 36–41 cells. (C) Representative images of soma with pCREB staining. (A and C) Bars, 10 µm. (D) Quantification of nuclear pCREB signaling. Scatter plot shows mean ± SEM; one-way ANOVA (*, P = 0.024; ***, P < 0.0002; ns, P = 0.491). Mock, NT: n = 42 somas; mock + BDNF: n = 48 somas; Hk1 KD, NT: n = 51 somas; Hk1 KD + BDNF: n = 55 somas. (E) Normalized pCREB intensity to the NT mock condition per individual experiment. Bar graph shows mean ± SEM; repeated-measures one-way ANOVA (*, P < 0.047; ns, P = 0.99). n = 3 individual experiments. Discussion In this study, we found that Hook1 comigrates with a subpopulation of Rab5- and Rab7-positive endosomes. Previous studies in fungi have implicated Hook proteins in early endosomal transport marked by Rab5, but these systems only express one Hook isoform (Bielska et al., 2014; Zhang et al., 2014). In mammalian systems, Hook1 has been linked to different aspects of the endosomal pathway. In HeLa cells, Hook proteins were found to interact with members of the HOPS complex and to be important for timely trafficking of EGF through endosomal compartments marked by EEA1, CD63, and LAMP1, but this study simultaneously knocked down all three Hook isoforms, making it difficult to determine their individual roles (Xu et al., 2008). Another study using HeLa cells suggested that Hook1 interacts with CD147 to facilitate sorting into Rab22-positive recycling tubules (Maldonado-Báez et al., 2013). In COS-1 cells, Hook1 was found to interact with Rab7, Rab9, and Rab11 using immunoprecipitation (Luiro et al., 2004). The variety of results seen in these studies is likely due to the fluid nature of endosomal pathways and differential cellular demands on these pathways. In our work, Hook1 comigrates primarily with fast, retrograde Rab5-positive vesicles in primary neurons, which suggests a role for Hook1 in activating the motility of these vesicles. Yet Hook1 depletion produced only subtle effects on the dynamics of the total Rab5-positive endosomal population in axons, significantly affecting only signaling endosome motility, suggesting a higher level of specificity than previously observed for dynein effectors that interact directly with Rab proteins such as the interaction of BICD proteins with Rab6 (Schlager et al., 2010; Matsuto et al., 2015; Terawaki et al., 2015; Huynh and Vale, 2017) or RILP with Rab7 (Cantalupo et al., 2001; Wu et al., 2005; Johansson et al., 2007). Instead of cargo attachment through a Rab protein, Hook1 has been suggested to attach to cargo through C-terminal interactions with Fused Toes (FTS) and FTS-Hook interacting (FHIP) proteins (Xu et al., 2008; Yao et al., 2014; Guo et al., 2016). It remains to be determined whether Hook1 is linked to signaling endosomes by FTS–FHIP or another protein complex in neuronal systems. Of note, mutating pairs of residues conserved within the C-terminal domain of Hook1, but not found in Hook3, was sufficient to disrupt the ability of Hook1 to activate the transport of signaling endosomes (Fig. 3, G and H), supporting a key role for the C-terminal of Hook1 in mediating organelle-specific dynein-driven motility. In this study, we found that Hook1 plays a role in signaling endosome processivity. Our previous work has shown that Hook1 increases dynein-dynactin processivity in vitro, with Hook1-bound dynein displaying higher velocities and longer run lengths than BICD2-associated motors (Olenick et al., 2016). Recent cryo-EM structures have shown Hook3 can recruit two dynein dimers per dynactin complex (Urnavicius et al., 2018). Due to the high sequence similarity in the Hook domain and coiled-coil regions, it is likely that Hook1 functions in a similar manner. It is possible that the relatively high velocities observed for signaling endosome transport (averaging 1.4 µm/s) are due to the incorporation of two dynein dimers into the dynein–dynactin–Hook1 complex. We also found that the interaction of Hook1 with dynein subunit LIC1 is essential for signaling endosome motility, a mechanism that is likely conserved in other dynein effectors regulating cargo transport in the cell. Recent structural work showed that a helix (aa 433–458) within the otherwise unstructured C-terminal region of LIC1 is a conserved interface for the binding of dynein effectors including BICD2, Spindly, and Hook proteins (Lee et al., 2018). Since the LIC1 interaction region is conserved among dynein effectors, it is likely that competition for this binding site plays a role in regulating cargo transport. During our investigation of Hook1, we observed differential localization of the dynein effectors Hook1, BICD2, and RILP in hippocampal neurons. It is possible that the compartmentalization of neurons and the differential localization of dynein effectors locally regulates the competition of dynein effectors for dynein–dynactin binding sites. In this study, we found that Hook1 plays a key role early in the transport of TrkB–BDNF endosomes as reflected by Hook1 enrichment in the distal axon. This enrichment is not seen for RILP or BICD2. In contrast, RILP was enriched in the mid-axon and soma, consistent with its role in mediating Rab7-endosome motility (Cantalupo et al., 2001; Johansson et al., 2007). It remains to be investigated whether there is a transition or handoff of dynein effectors during endosomal maturation and conversion from Rab5- to Rab7-positive organelles (Fig. 8) in line with the key role proposed for Rab7-positive multivesicular bodies in mediating the long-distance trafficking of TrkA in sympathetic neurons (Ye et al., 2018). It is also possible that a given cargo could have a mixture of dynein effectors. While Hook and BICD proteins are dynein effectors that modulate dynein processivity, there is as yet no direct evidence of RILP acting as a dynein activator, but other studies have shown that it acts as an adaptor to recruit dynein to Rab7 cargo (Johansson et al., 2007; Rocha et al., 2009). Similarly, Snapin has been reported to recruit dynein to BDNF–TrkB-positive signaling endosomes (Zhou et al., 2012), although this protein lacks the secondary structure expected for a dynein activator (Reck-Peterson et al., 2018). It is possible that a complement of dynein effectors and activators may be required to effectively recruit motors and activate transport; it remains to be seen whether there is coordination between dynein activators and adaptors to help maintain a constant processive dynein pool on a given organelle. Figure 8. Model of neurotrophin uptake and transport from distal axon. In our model, Hook1 plays a role in early dynein-mediated transport of BDNF-TrkB signaling endosomes. As signaling endosomes are transported to the soma, there might be handoff or exchange of dynein effectors as the vesicles mature from Rab5 to Rab7 positive. Using microfluidic devices, we showed that loss of Hook1 reduces the flux and processivity of BDNF-signaling endosomes. However, this reduced flux to the cell body did not result in an accumulation of BDNF-Qdots in the distal axon. Similarly, inhibiting dynein motility with Ciliobrevin D also did not lead to the accumulation of BDNF-Qdot levels in the distal axon, suggesting that endocytosis is down-regulated when endosomal motility is inhibited. BDNF has been reported to be a self-amplifying autocrine factor, which can signal to promote BDNF expression and increase TrkB membrane levels (Cheng et al., 2011). Therefore, loss of BDNF signaling could reduce TrkB surface levels and regulate endocytosis to prevent a buildup of BDNF endosomes at the distal tip. However, we detected no significant change in plasma membrane–associated TrkB levels at the axon tip upon Hook1 depletion, suggesting that another mechanism might be at work. An alternative possibility is that the internalization of the TrkB–BDNF complex via endocytosis is tightly linked to the formation of a high-speed, highly processive Hook1-dependent transport compartment. Thus, if transport is blocked either by Hook1 depletion or dynein inhibition, internalization may also be down-regulated, preventing the distal accumulation of stalled signaling endosomes. Hook1 binding partners such as FTS and FHIP proteins (Xu et al., 2008; Yao et al., 2014; Guo et al., 2016) may mediate this coordination of uptake and transport, an interesting question for future studies. TrkB–BDNF signaling is important for neuronal survival, and disruption of signaling endosome trafficking has been found in models of neurodegenerative diseases including Huntington’s or Parkinson’s disease (Millecamps and Julien, 2013). In Huntington’s disease, the polyQ-expanded huntingtin protein has been shown to impair BDNF retrograde trafficking, leading to reduced neuronal survival (Gauthier et al., 2004). α-Synuclein has also been shown to impair BDNF transport in a mouse model of Parkinson’s disease (Fang et al., 2017). Currently, it is unclear whether Hook proteins play a role in these neurodegenerative diseases, but Hook1 and Hook3 have been localized to tau aggregates, a pathological hallmark of Alzheimer’s disease, frontotemporal dementia, and other tauopathies (Herrmann et al., 2015). With our new understanding of Hook1 as a dynein effector for BDNF transport in nonpathological states, the role of Hook1 in disease states can now be investigated in more detail in future studies. Materials and methods Plasmids and reagents Halo-tagged Hook1 constructs were generated from the human Hook1 sequence (UniProt code Q9UJC3) and using the HaloTag from the pHTN or pHTC Halo tag CMV-neo vector (Promega). GFP-Hook1 constructs were generated in the pEGFP vector. Rab5-GFP was provided by M. Zerial (Max Planck Institute, Dresden, Germany). Rab7-GFP was purchased from Addgene. TrkB-mRFP was provided by M. Chao (New York University, New York, NY). RILP-GFP was provided by J. Neefjes (Leiden University Medical Center, Leiden, Netherlands). BICD2-GFP was provided by A. Akhmanova (Utrecht University, Utrecht, Netherlands). Empty pEGFP-N1 (Addgene) was used as a cell fill to identify neuronal morphology. Antibodies used for biochemistry and Western blotting included anti-Hook1 (rabbit; 1:250; ab104514; for detection of rat protein; Abcam), anti-Hook1 (rabbit; 1:500; ab150397; for detection of mouse protein; Abcam), anti-Hook3 (rabbit; 1:1,000; ProteinTech), anti-actin (mouse; 1:1,000; EMD Millipore), anti-DIC (mouse; 1:1,000; EMD Millipore), anti-TrkB (rabbit; 1:4,000; ab187041; Abcam), and anti-Halo (rabbit; Promega). For immunofluorescence experiments, anti-pCREB(Ser133) (rabbit; 1:1,000; Cell Signaling) and anti-TrkB (aa 55–67; rabbit; 1:1,000; EMD Millipore) were used. For TIRF assays, the monoclonal antibody used was anti–β-tubulin (1:40; mouse; T5201; Sigma-Aldrich). An ON-TARGETplus siRNA SMARTpool of four siRNAs for rat Hook1 or Hook3 was purchased from GE Healthcare. Neuronal culture E18 Sprague–Dawley rat hippocampal neurons were obtained in suspension from the Neuron Culture Service Center at the University of Pennsylvania and plated on 35-mm glass-bottom dishes (MatTek) or 25-mm coverslips (World Precision). Dishes or coverslips were precoated with 0.5 mg/ml poly-L-lysine (Sigma-Aldrich) 24 h before plating. Neurons were cultured at 37°C with 5% CO2 in maintenance media that consisted of Neurobasal (Gibco) supplemented with 2 mM GlutaMAX, 100 U/ml penicillin, 100 mg/ml streptomycin, and 2% B27 (Thermo Fisher Scientific). Every 3–4 d, 40% of the media was replaced with fresh maintenance media supplemented with 1 µM AraC. Neuronal imaging Imaging was done at DIV 7–8 with neurons transfected 24–48 h before imaging. Neurons were transfected using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions, with 0.3–1 µg of each DNA plasmid and, for siRNA conditions, 45 pmol siRNA. Control siRNA labeled with Cy5 was used to confirm transfection of siRNA with Lipofectamine in cultured neurons. Neurons were imaged in low-fluorescence media (HibernateE; Brain Bits) supplemented with 2% B27 and 1% GlutaMAX. For experiments with Halo-Tag constructs, neurons were labeled with tetramethyl rhodamine (TMR) Halo-Tag ligands according to the manufacturer’s instructions (Promega). For mitochondria imaging, TMR ethyl ester (TMRE) was added according to manufacturer’s protocol. Neurons were imaged in an environmental chamber at 37°C on a spinning-disk confocal UltraView VOX (PerkinElmer) on an inverted Nikon Eclipse Ti microscope with the Prefect Focus system using Apochromat 100× 1.49 NA oil-immersion objective and a C9100-50 electron-multiplying charge-coupled device camera (Hamamatsu Photonics) controlled by Volocity software (PerkinElmer). Axons and dendrites were identified based on morphological criteria as outlined elsewhere (Kaech and Banker, 2006). The distal axon is defined as the area <100 μm from the axon terminal, while the mid-axon is defined as the area >100 μm from the axon terminal and >100 μm from the soma. Microfluidic experiments Round microfluidic devices of 450 µm (Xona microfluidics) were used for axon isolation experiments. Devices were UV sterilized and attached to poly-L-lysine–coated imaging dishes (FluoroDish; World Precision Instruments) before plating. On the day of plating, neurons were nucleofected with an Amaxa Nucleofector machine (Lonza) with DNA/siRNA in similar quantities as described above. Cells were plated to one side of microfluidic device at 4 × 105 cells per dish. Fresh maintenance media was added every 2 d (∼30%). BDNF-Qdot experiments Neurons were serum starved in unsupplemented Neurobasal (Gibco) for 2–4 h before BDNF-Qdot addition. 50 nM hBDNF-biotin (Alomone Labs) was combined with 50 nM Quantum Dot ITK 655 Streptavidin conjugate (Invitrogen) for 1 h on ice to generate BDNF-Qdots. After conjugation, BDNF-Qdots were added to neurons in unsupplemented Neurobasal to a final concentration of 0.25 nM for 1–2 h. In microfluidic experiments, BDNF-Qdots were only added to the axon side. For pCREB experiments, 1 nM unconjugated BDNF-biotin was added for 1 h before fixing. For Ciliobrevin D (EMD Millipore) conditions, 20 µM Ciliobrevin D was added 10 min before BDNF-Qdot addition and was present throughout BDNF-Qdot treatment and imaging, with the same timescale as control conditions. Immunofluorescence Neurons cultured on 25-mm glass coverslips were fixed at DIV 7–8 in PBS containing 4% paraformaldehyde and 4% sucrose for 8 min. Coverslips were washed three times in PBS and blocked with cell block (PBS with 5% normal goat serum and 1% BSA). Primary antibodies were incubated for 2 h at RT in cell block. After removing the primary antibodies, the coverslips were washed with PBS and incubated for 1 h at RT with fluorophore-conjugated secondary antibodies diluted in cell block. Following washes with PBS, the coverslips were mounted in ProLong Gold Antifade Mountant (Thermo Fisher Scientific) on glass slides. For proximity ligation assays, Duolink PLA kit with red detection reagents was used according to manufacturer’s protocol (Sigma-Aldrich). TIRF motility assay Single-molecule TIRF motility assays were performed as previously described in detail (Olenick et al., 2016). In brief, HeLa cells 18–20 h after transfection of Halo-Hook1 constructs were labeled with the Halo ligand TMR (Promega) and lysed in buffer containing 40 mM Hepes, 1 mM EDTA, 120 mM NaCl, 0.1% Triton X-100, and 1 mM magnesium ATP, pH 7.4, supplemented with protease inhibitors. Cell lysates were diluted in assay buffer containing 10 mM magnesium ATP, 0.3 mg/ml BSA, 0.3 mg/ml casein, 10 mM DTT, and an oxygen-scavenging system. Diluted cell lysates were then flowed into imaging chambers with Taxol-stabilized microtubules immobilized to the coverslip with a tubulin antibody. TIRF videos were acquired at RT at four frames per second using the Nikon TIRF system (Perkin Elmer) on an inverted Ti microscope with a 100× objective and an ImageEM C9100–13 camera (Hamamatsu Photonics) controlled by Volocity software. Immunoisolation of TrkB vesicles Using three mouse brains per immunoprecipitation experiment, lysates were homogenated in Hepes buffer (10 mM Hepes, pH 7.4, 1 mM EDTA, and protease inhibitors) and then centrifuged at 800× g for 10 min and 3,000× g for 10 min. The supernatant was then added to Dynabeads protein G (Thermo Fisher Scientific) with anti-TrkB (Abcam) or anti-Halo (Promega) as a rabbit IgG control. Lysates were incubated for 15 min at RT and then washed three times with Hepes buffer. Proteins were then eluted in denaturing buffer and boiled 5 min before running on a SDS-PAGE gel. Western blotting To test siRNA efficiency, PC12 cells were transfected at 70–80% confluency with 45 pmol of a pool of Hook1 or Hook3 siRNAs using Lipofectamine RNAiMAX (Invitrogen) and lysed 48 h later. PC12 cells were lysed (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton-X100, and protease inhibitors) and clarified by centrifugation at 13,200 rpm for 10 min at 4°C. For all Western blot experiments, samples were boiled in denaturing buffer for 5 min and run on a SDS-PAGE gel to separate proteins. Imaging analysis For motility analysis, kymographs were generated using the MultipleKymograph plugin for Fiji (ImageJ; National Institutes of Health) and analyzed using custom MATLAB software (MathWorks) or by measurement tools in Fiji. For effector localization, images were deconvolved using Huygens Professional software. Effector puncta were then counted by hand, and axon length was measured in Fiji. Area of BDNF-Qdot was measured using measurement and analysis functions in Volocity. The pCREB signal was measured by outlining the nucleus with a Hoechst stain and measuring the integrated intensity of pCREB in that area with Fiji. Statistical methods Statistics were performed in GraphPad Prism. A Student’s t test or Mann–Whitney test was used when comparing two datasets as indicated, while a ANOVA was used with multiple datasets. For all experiments, data were analyzed from at least three independent replicates. Statistical significance is noted as follows: NS, P > 0.05; *, P ≤ 0.05; **, P ≤ 0.01; and ***, P ≤ 0.001. Figure legends contain specific P value for that figure. Online supplemental materials Fig. S1 shows that Hook1 KD does not significantly change Rab5, Rab7, mitochondria, or autophagosome motility. Fig. S2 shows that Hook1 is present on TrkB-BDNF vesicles using biochemistry and imaging. Fig. S3 is the sequence analysis of Hook1 and Hook3 C terminus that informed design of mutant constructs. Video 1 shows BDNF-Qdots motility in control neurons grown in microfluidic chambers. Video 2 shows BDNF-Qdots motility in Hk1 KD neurons grown in microfluidic chambers. Supplementary Material Supplemental Materials (PDF) Video 1 Video 2 Acknowledgments We thank Mariko Tokito for her expertise in molecular biology, Pedro Guedes Dias and Sydney Cason for critical review of the manuscript, and Chantell Evans, Alex Böcker, and Andrea Stavoe for helpful insights and discussions. This work was supported by the National Institutes of Health (grant R35 GM126950 to E.L.F. Holzbaur, grant P01 GM087253 to E.L.F. Holzbaur and R. Dominguez, and grant T32 GM07229 to M.A. Olenick). This work was also supported by the Center for Engineering MechanoBiology, a National Science Foundation Science and Technology Center (grant agreement CMMI: 15-48571). The authors declare no competing financial interests. Author contributions: M.A. Olenick, R. Dominguez, and E.L.F. Holzbaur designed the experiments. M.A. Olenick performed the experiments. M.A. Olenick and E.L.F. Holzbaur wrote the manuscript. All authors reviewed the results and approved the final version of the manuscript. ==== Refs Ayloo, S., J.E. Lazarus, A. Dodda, M. Tokito, E.M. Ostap, and E.L.F. Holzbaur. 2014. Dynactin functions as both a dynamic tether and brake during dynein-driven motility. Nat. Commun. 5 :4807. 10.1038/ncomms5807 25185702 Bielska, E., M. Schuster, Y. Roger, A. Berepiki, D.M. Soanes, N.J. Talbot, and G. Steinberg. 2014. Hook is an adapter that coordinates kinesin-3 and dynein cargo attachment on early endosomes. J. Cell Biol. 204 :989–1007. 10.1083/jcb.201309022 24637326 Cantalupo, G., P. Alifano, V. Roberti, C.B. Bruni, and C. Bucci. 2001. Rab-interacting lysosomal protein (RILP): the Rab7 effector required for transport to lysosomes. EMBO J. 20 :683–693. 10.1093/emboj/20.4.683 11179213 Cheng, P.-L., A.-H. Song, Y.-H. Wong, S. Wang, X. Zhang, and M.-M. Poo. 2011. Self-amplifying autocrine actions of BDNF in axon development. Proc. Natl. Acad. Sci. USA. 108 :18430–18435. 10.1073/pnas.1115907108 22025720 Chowdhury, S., S.A. Ketcham, T.A. Schroer, and G.C. Lander. 2015. Structural organization of the dynein-dynactin complex bound to microtubules. Nat. Struct. Mol. Biol. 22 :345–347. 10.1038/nsmb.2996 25751425 Cosker, K.E., and R.A. Segal. 2014. Neuronal signaling through endocytosis. Cold Spring Harb. Perspect. Biol. 6 :a020669. 10.1101/cshperspect.a020669 24492712 Cui, B., C. Wu, L. Chen, A. Ramirez, E.L. Bearer, W.-P. Li, W.C. Mobley, and S. Chu. 2007. One at a time, live tracking of NGF axonal transport using quantum dots. Proc. Natl. Acad. Sci. USA. 104 :13666–13671. 10.1073/pnas.0706192104 17698956 Deinhardt, K., S. Salinas, C. Verastegui, R. Watson, D. Worth, S. Hanrahan, C. Bucci, and G. Schiavo. 2006. Rab5 and Rab7 control endocytic sorting along the axonal retrograde transport pathway. Neuron. 52 :293–305. 10.1016/j.neuron.2006.08.018 17046692 Fang, F., W. Yang, J.B. Florio, E. Rockenstein, B. Spencer, X.M. Orain, S.X. Dong, H. Li, X. Chen, K. Sung, 2017. Synuclein impairs trafficking and signaling of BDNF in a mouse model of Parkinson’s disease. Sci. Rep. 7 :3868. 10.1038/s41598-017-04232-4 28634349 Fu, M.M., and E.L.F. Holzbaur. 2014. Integrated regulation of motor-driven organelle transport by scaffolding proteins. Trends Cell Biol. 24 :564–574. 10.1016/j.tcb.2014.05.002 24953741 Gauthier, L.R., B.C. Charrin, M. Borrell-Pagès, J.P. Dompierre, H. Rangone, F.P. Cordelières, J. De Mey, M.E. MacDonald, V. Lessmann, S. Humbert, and F. Saudou. 2004. Huntingtin controls neurotrophic support and survival of neurons by enhancing BDNF vesicular transport along microtubules. Cell. 118 :127–138. 10.1016/j.cell.2004.06.018 15242649 Grotjahn, D.A., S. Chowdhury, Y. Xu, R.J. McKenney, T.A. Schroer, and G.C. Lander. 2018. Cryo-electron tomography reveals that dynactin recruits a team of dyneins for processive motility. Nat. Struct. Mol. Biol. 25 :203–207. 10.1038/s41594-018-0027-7 29416113 Guo, X., G.G. Farías, R. Mattera, and J.S. Bonifacino. 2016. Rab5 and its effector FHF contribute to neuronal polarity through dynein-dependent retrieval of somatodendritic proteins from the axon. Proc. Natl. Acad. Sci. USA. 113 :E5318–E5327. 10.1073/pnas.1601844113 27559088 Guthrie, C.R., G.D. Schellenberg, and B.C. Kraemer. 2009. SUT-2 potentiates tau-induced neurotoxicity in Caenorhabditis elegans. Hum. Mol. Genet. 18 :1825–1838. 10.1093/hmg/ddp099 19273536 Heerssen, H.M., M.F. Pazyra, and R.A. Segal. 2004. Dynein motors transport activated Trks to promote survival of target-dependent neurons. Nat. Neurosci. 7 :596–604. 10.1038/nn1242 15122257 Herrmann, L., C. Wiegmann, A. Arsalan-Werner, I. Hilbrich, C. Jäger, K. Flach, A. Suttkus, I. Lachmann, T. Arendt, and M. Holzer. 2015. Hook proteins: association with Alzheimer pathology and regulatory role of hook3 in amyloid beta generation. PLoS One. 10 :e0119423. 10.1371/journal.pone.0119423 25799409 Huynh, W., and R.D. Vale. 2017. Disease-associated mutations in human BICD2 hyperactivate motility of dynein-dynactin. J. Cell Biol. 216 :3051–3060. 10.1083/jcb.201703201 28883039 Johansson, M., N. Rocha, W. Zwart, I. Jordens, L. Janssen, C. Kuijl, V.M. Olkkonen, and J. Neefjes. 2007. Activation of endosomal dynein motors by stepwise assembly of Rab7-RILP-p150Glued, ORP1L, and the receptor betalll spectrin. J. Cell Biol. 176 :459–471. 10.1083/jcb.200606077 17283181 Kaech, S., and G. Banker. 2006. Culturing hippocampal neurons. Nat. Protoc. 1 :2406–2415. 10.1038/nprot.2006.356 17406484 Kardon, J.R., and R.D. Vale. 2009. Regulators of the cytoplasmic dynein motor. Nat. Rev. Mol. Cell Biol. 10 :854–865. 10.1038/nrm2804 19935668 Lee, I.-G., M.A. Olenick, M. Boczkowska, C. Franzini-Armstrong, E.L.F. Holzbaur, and R. Dominguez. 2018. A conserved interaction of the dynein light intermediate chain with dynein-dynactin effectors necessary for processivity. Nat. Commun. 9 :986. 10.1038/s41467-018-03412-8 29515126 Luiro, K., K. Yliannala, L. Ahtiainen, H. Maunu, I. Järvelä, A. Kyttälä, and A. Jalanko. 2004. Interconnections of CLN3, Hook1 and Rab proteins link Batten disease to defects in the endocytic pathway. Hum. Mol. Genet. 13 :3017–3027. 10.1093/hmg/ddh321 15471887 Maday, S., A.E. Twelvetrees, A.J. Moughamian, and E.L.F. Holzbaur. 2014. Axonal transport: cargo-specific mechanisms of motility and regulation. Neuron. 84 :292–309. 10.1016/j.neuron.2014.10.019 25374356 Maldonado-Báez, L., N.B. Cole, H. Krämer, and J.G. Donaldson. 2013. Microtubule-dependent endosomal sorting of clathrin-independent cargo by Hook1. J. Cell Biol. 201 :233–247. 10.1083/jcb.201208172 23589492 Matanis, T., A. Akhmanova, P. Wulf, E. Del Nery, T. Weide, T. Stepanova, N. Galjart, F. Grosveld, B. Goud, C.I. De Zeeuw, 2002. Bicaudal-D regulates COPI-independent Golgi-ER transport by recruiting the dynein-dynactin motor complex. Nat. Cell Biol. 4 :986–992. 10.1038/ncb891 12447383 Matsuto, M., F. Kano, and M. Murata. 2015. Reconstitution of the targeting of Rab6A to the Golgi apparatus in semi-intact HeLa cells: A role of BICD2 in stabilizing Rab6A on Golgi membranes and a concerted role of Rab6A/BICD2 interactions in Golgi-to-ER retrograde transport. Biochim. Biophys. Acta. 1853 (10 , 10 Pt A ):2592–2609. 10.1016/j.bbamcr.2015.05.005 25962623 McKenney, R.J., W. Huynh, M.E. Tanenbaum, G. Bhabha, and R.D. Vale. 2014. Activation of cytoplasmic dynein motility by dynactin-cargo adapter complexes. Science. 345 :337–341. 10.1126/science.1254198 25035494 Millecamps, S., and J.-P. Julien. 2013. Axonal transport deficits and neurodegenerative diseases. Nat. Rev. Neurosci. 14 :161–176. 10.1038/nrn3380 23361386 Mitre, M., A. Mariga, and M.V. Chao. 2017. Neurotrophin signalling: novel insights into mechanisms and pathophysiology. Clin. Sci. (Lond.). 131 :13–23. 10.1042/CS20160044 27908981 Moynihan, K.L., R. Pooley, P.M. Miller, I. Kaverina, and D.M. Bader. 2009. Murine CENP-F regulates centrosomal microtubule nucleation and interacts with Hook2 at the centrosome. Mol. Biol. Cell. 20 :4790–4803. 10.1091/mbc.e09-07-0560 19793914 Olenick, M.A., M. Tokito, M. Boczkowska, R. Dominguez, and E.L.F. Holzbaur. 2016. Hook Adaptors Induce Unidirectional Processive Motility by Enhancing the Dynein-Dynactin Interaction. J. Biol. Chem. 291 :18239–18251. 10.1074/jbc.M116.738211 27365401 Reck-Peterson, S.L., A. Yildiz, A.P. Carter, A. Gennerich, N. Zhang, and R.D. Vale. 2006. Single-molecule analysis of dynein processivity and stepping behavior. Cell. 126 :335–348. 10.1016/j.cell.2006.05.046 16873064 Reck-Peterson, S.L., W.B. Redwine, R.D. Vale, and A.P. Carter. 2018. The cytoplasmic dynein transport machinery and its many cargoes. Nat. Rev. Mol. Cell Biol. 1 . 10.1038/s41580-018-0004-3 Redwine, W.B., M.E. DeSantis, I. Hollyer, Z.M. Htet, P.T. Tran, S.K. Swanson, L. Florens, M.P. Washburn, and S.L. Reck-Peterson. 2017. The human cytoplasmic dynein interactome reveals novel activators of motility. eLife. 6 :e28257. 10.7554/eLife.28257 28718761 Rocha, N., C. Kuijl, R. van der Kant, L. Janssen, D. Houben, H. Janssen, W. Zwart, and J. Neefjes. 2009. Cholesterol sensor ORP1L contacts the ER protein VAP to control Rab7-RILP-p150 Glued and late endosome positioning. J. Cell Biol. 185 :1209–1225. 10.1083/jcb.200811005 19564404 Ross, J.L., K. Wallace, H. Shuman, Y.E. Goldman, and E.L.F. Holzbaur. 2006. Processive bidirectional motion of dynein-dynactin complexes in vitro. Nat. Cell Biol. 8 :562–570. 10.1038/ncb1421 16715075 Schlager, M.A., L.C. Kapitein, I. Grigoriev, G.M. Burzynski, P.S. Wulf, N. Keijzer, E. de Graaff, M. Fukuda, I.T. Shepherd, A. Akhmanova, and C.C. Hoogenraad. 2010. Pericentrosomal targeting of Rab6 secretory vesicles by Bicaudal-D-related protein 1 (BICDR-1) regulates neuritogenesis. EMBO J. 29 :1637–1651. 10.1038/emboj.2010.51 20360680 Schlager, M.A., H.T. Hoang, L. Urnavicius, S.L. Bullock, and A.P. Carter. 2014. In vitro reconstitution of a highly processive recombinant human dynein complex. EMBO J. 33 :1855–1868. 10.15252/embj.201488792 24986880 Schroeder, C.M., and R.D. Vale. 2016. Assembly and activation of dynein-dynactin by the cargo adaptor protein Hook3. J. Cell Biol. 214 :309–318. 10.1083/jcb.201604002 27482052 Schroeder, C.M., J.M. Ostrem, N.T. Hertz, and R.D. Vale. 2014. A Ras-like domain in the light intermediate chain bridges the dynein motor to a cargo-binding region. eLife. 3 :e03351. 10.7554/eLife.03351 25272277 Scott-Solomon, E., and R. Kuruvilla. 2018. Mechanisms of neurotrophin trafficking via Trk receptors. Mol. Cell. Neurosci. 91 :25–33. 10.1016/j.mcn.2018.03.013 29596897 Szebenyi, G., B. Hall, R. Yu, A.I. Hashim, and H. Krämer. 2007. Hook2 localizes to the centrosome, binds directly to centriolin/CEP110 and contributes to centrosomal function. Traffic. 8 :32–46. 10.1111/j.1600-0854.2006.00511.x 17140400 Terawaki, S., A. Yoshikane, Y. Higuchi, and K. Wakamatsu. 2015. Structural basis for cargo binding and autoinhibition of Bicaudal-D1 by a parallel coiled-coil with homotypic registry. Biochem. Biophys. Res. Commun. 460 :451–456. 10.1016/j.bbrc.2015.03.054 25796327 Urnavicius, L., K. Zhang, A.G. Diamant, C. Motz, M.A. Schlager, M. Yu, N.A. Patel, C.V. Robinson, and A.P. Carter. 2015. The structure of the dynactin complex and its interaction with dynein. Science. 347 :1441–1446. 10.1126/science.aaa4080 25814576 Urnavicius, L., C.K. Lau, M.M. Elshenawy, E. Morales-Rios, C. Motz, A. Yildiz, and A.P. Carter. 2018. Cryo-EM shows how dynactin recruits two dyneins for faster movement. Nature. 554 :202–206. 10.1038/nature25462 29420470 Watson, F.L., H.M. Heerssen, A. Bhattacharyya, L. Klesse, M.Z. Lin, and R.A. Segal. 2001. Neurotrophins use the Erk5 pathway to mediate a retrograde survival response. Nat. Neurosci. 4 :981–988. 10.1038/nn720 11544482 Wu, M., T. Wang, E. Loh, W. Hong, and H. Song. 2005. Structural basis for recruitment of RILP by small GTPase Rab7. EMBO J. 24 :1491–1501. 10.1038/sj.emboj.7600643 15933719 Xu, L., M.E. Sowa, J. Chen, X. Li, S.P. Gygi, and J.W. Harper. 2008. An FTS/Hook/p107(FHIP) complex interacts with and promotes endosomal clustering by the homotypic vacuolar protein sorting complex. Mol. Biol. Cell. 19 :5059–5071. 10.1091/mbc.e08-05-0473 18799622 Yao, X., X. Wang, and X. Xiang. 2014. FHIP and FTS proteins are critical for dynein-mediated transport of early endosomes in Aspergillus. Mol. Biol. Cell. 25 :2181–2189. 10.1091/mbc.e14-04-0873 24870033 Ye, H., R. Kuruvilla, L.S. Zweifel, and D.D. Ginty. 2003. Evidence in support of signaling endosome-based retrograde survival of sympathetic neurons. Neuron. 39 :57–68. 10.1016/S0896-6273(03)00266-6 12848932 Ye, M., K.M. Lehigh, and D.D. Ginty. 2018. Multivesicular bodies mediate long-range retrograde NGF-TrkA signaling. eLife. 7 :e33012. 10.7554/eLife.33012 29381137 Yeh, T.-Y., N.J. Quintyne, B.R. Scipioni, D.M. Eckley, and T.A. Schroer. 2012. Dynactin’s pointed-end complex is a cargo-targeting module. Mol. Biol. Cell. 23 :3827–3837. 10.1091/mbc.e12-07-0496 22918948 Zhang, J., X. Yao, L. Fischer, J.F. Abenza, M.A. Peñalva, and X. Xiang. 2011. The p25 subunit of the dynactin complex is required for dynein-early endosome interaction. J. Cell Biol. 193 :1245–1255. 10.1083/jcb.201011022 21708978 Zhang, J., R. Qiu, H.N. Arst Jr., M.A. Peñalva, and X. Xiang. 2014. HookA is a novel dynein-early endosome linker critical for cargo movement in vivo. J. Cell Biol. 204 :1009–1026. 10.1083/jcb.201308009 24637327 Zhang, K., H.E. Foster, A. Rondelet, S.E. Lacey, N. Bahi-Buisson, A.W. Bird, and A.P. Carter. 2017. Cryo-EM Reveals How Human Cytoplasmic Dynein Is Auto-inhibited and Activated. Cell. 169 :1303–1314.e18. 10.1016/j.cell.2017.05.025 28602352 Zhou, B., Q. Cai, Y. Xie, and Z.-H. Sheng. 2012. Snapin recruits dynein to BDNF-TrkB signaling endosomes for retrograde axonal transport and is essential for dendrite growth of cortical neurons. Cell Reports. 2 :42–51. 10.1016/j.celrep.2012.06.010 22840395

PMC006xxxxxx/PMC6548120.txt

==== Front J Cell Biol J Cell Biol jcb jcb The Journal of Cell Biology 0021-9525 1540-8140 Rockefeller University Press 31092556 201902017 10.1083/jcb.201902017 Research Articles Tools 1 25 Proteome of the central apparatus of a ciliary axoneme Proteome of a ciliary central apparatus Zhao Lei http://orcid.org/0000-0002-7038-675X Hou Yuqing http://orcid.org/0000-0003-0049-1148 Picariello Tyler http://orcid.org/0000-0003-3180-5947 Craige Branch http://orcid.org/0000-0002-9497-9218 Witman George B. Division of Cell Biology and Imaging, Department of Radiology, University of Massachusetts Medical School, Worcester, MA Correspondence to George B. Witman: george.witman@umassmed.edu B. Craige’s present address is Department of Biochemistry, Virginia Tech, Blacksburg, VA. 28 6 2019 15 5 2019 218 6 20512070 03 2 2019 13 4 2019 17 4 2019 © 2019 Zhao et al. 2019 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). The central apparatus is an essential component of “9+2” cilia. Zhao et al. identify more than 40 new potential components of the central apparatus of Chlamydomonas. Many are conserved and will facilitate genetic screening of patients with a form of primary ciliary dyskinesia that is difficult to diagnose. Nearly all motile cilia have a “9+2” axoneme containing a central apparatus (CA), consisting of two central microtubules with projections, that is essential for motility. To date, only 22 proteins are known to be CA components. To identify new candidate CA proteins, we used mass spectrometry to compare axonemes of wild-type Chlamydomonas and a CA-less mutant. We identified 44 novel candidate CA proteins, of which 13 are conserved in humans. Five of the latter were studied more closely, and all five localized to the CA; therefore, most of the other candidates are likely to also be CA components. Our results reveal that the CA is far more compositionally complex than previously recognized and provide a greatly expanded knowledge base for studies to understand the architecture of the CA and how it functions. The discovery of the new conserved CA proteins will facilitate genetic screening to identify patients with a form of primary ciliary dyskinesia that has been difficult to diagnose. National Institutes of Health https://doi.org/10.13039/100000002 R37 GM030626 R35 GM122574 Robert W. Booth EndowmentUniversity of Massachusetts Medical School ==== Body pmcIntroduction Nearly all motile cilia and flagella (terms here used interchangeably) contain a “9+2” axoneme consisting of nine outer doublet microtubules and two central microtubules. Periodically arranged along the outer doublet microtubules are a number of substructures, including outer and inner dynein arms, radial spokes, and nexin-dynein regulatory complexes (N-DRCs), that work together to generate and control motility. Genetic and biochemical analyses of these substructures in humans and model organisms, especially Chlamydomonas reinhardtii, have resulted in detailed knowledge of their protein composition and function and have revealed a high level of evolutionary conservation of their constituent subunits among motile organisms (Yang et al., 2006; Lin et al., 2011; Bower et al., 2013; King, 2018). Less well characterized are the central microtubules and their projections, here collectively termed the central apparatus (CA; Loreng and Smith, 2017). The two microtubules of the CA, termed C1 and C2, differ in their stability and projections, which repeat with 16- or 32-nm periodicities along the microtubules. The CA interacts with the radial spokes and has been implicated in the control of ciliary waveform, presumably via the CA → radial spoke → N-DRC → inner dynein arm pathway (Wirschell et al., 2009). It also has a role in the regulation of motility by Ca2+. It is essential for the motility of 9+2 cilia in organisms ranging from Chlamydomonas to mammals, and defects in the CA result in infertility, hydrocephalus, and severe respiratory problems in mice and humans (Zhang et al., 2006, 2007; Lechtreck et al., 2008; Olbrich et al., 2012; McKenzie et al., 2015; Edelbusch et al., 2017). Hence, it is imperative to have a detailed knowledge of the CA as well, both to understand how ciliary motility is controlled and to better diagnose human diseases caused by defects in this critical component of the 9+2 axoneme. In an elegant analysis of Chlamydomonas mutants lacking the CA, Adams et al. (1981), using then state-of-the-art 1D and 2D gel electrophoresis, reported that the CA contains 18 different proteins in addition to tubulin. In the decades since then, and in apparently good agreement with the results of Adams et al. (1981), further research on Chlamydomonas has resulted in the characterization, at the level of amino-acid sequence, of 22 non-tubulin proteins that are components of the CA. Eighteen of these are unique to the CA, while four appear also to be present elsewhere in the axoneme (Table 1). All have human homologues. Most have been localized to specific projections of either the C1 or C2 microtubule. Table 1. Known CA proteins Location Protein NCBI accession number Mass (kD) References C1a projection PF6 AAK38270.1 240 Dutcher et al., 1984; Rupp et al., 2001; Wargo et al., 2005 FAP101 AAZ31187.1 86 Wargo et al., 2005 FAP114 AAZ31185.1 32 Wargo et al., 2005 FAP119 XP_001696622.1 34 Wargo et al., 2005 FAP227 AAZ31184.1 18 Wargo et al., 2005 Calmodulina 1206346A 18 Wargo et al., 2005 C1b projection CPC1 XP_001702926.1 265 Mitchell and Sale, 1999; Zhang and Mitchell, 2004; Mitchell et al., 2005 FAP42 XP_001697065.1 350 Mitchell et al., 2005 FAP69 XP_001703508.1 135 Mitchell et al., 2005 Enolasea XP_001702971.1 56 Mitchell et al., 2005 HSP70Aa XP_001701326.1 78 Mitchell et al., 2005; Shapiro et al., 2005 C1d projection FAP46 XP_001702776.1 289 DiPetrillo and Smith, 2010 FAP54 XP_001696950.1 318 DiPetrillo and Smith, 2010 FAP74 ADD85930.1 204 DiPetrillo and Smith, 2010 FAP221 ADD85929.2 100 DiPetrillo and Smith, 2010 FAP297 XP_001690036.1 87 Brown et al., 2012 C1 microtubule PP1ca AAD38856.1 35 Yang et al., 2000 C2b projection Hydin XP_001689997.1 540 Lechtreck and Witman, 2007 C2c projection KLP1 XP_001701617.1 83 Bernstein et al., 1994; Yokoyama et al., 2004 Others PF16 XP_001694680.1 57 Dutcher et al., 1984; Smith and Lefebvre, 1996 PF20 P93107.1 67 Smith and Lefebvre, 1997 FAP174 ACR55627.1 10 Rao et al., 2016 C1-kinesin (Fox et al., 1994) and AKAP240 (Gaillard et al., 2001) are not listed because their sequences are unknown. a Protein is located both in the CA and elsewhere in the flagellum. Despite the good agreement between the number of CA proteins estimated by 1D and 2D gel electrophoresis and the number that have been characterized at the molecular level, there are reasons to believe that the CA may contain many more proteins that have not yet been identified. First, the 2D gels used for the estimate of Adams et al. (1981) could resolve only ∼250 axonemal proteins, whereas a proteomic analysis of the Chlamydomonas flagellum by mass spectrometry (MS) revealed that the axoneme contains approximately twice that many proteins (Pazour et al., 2005), suggesting that the CA also might contain twice as many proteins as previously believed. Second, a recent cryo-electron tomography (cryo-ET) analysis of the CA revealed unexpected structural complexity, including four new projections not previously reported (Carbajal-González et al., 2013). Based on this analysis, the C1 microtubule has a total of six projections, termed C1a through C1f, and the C2 microtubule has a total of five projections, termed C2a through C2e. There also is a complex bridge between the two central microtubules, as well as small microtubule inner proteins that are attached to the inside of the C2 microtubule wall. It is difficult to imagine that all of these structures could be built from just 22 proteins. Indeed, the known CA proteins have been localized to just five of the CA projections. Third, the sum of the masses of all the projections as estimated by cryo-ET is >14 MD (Carbajal-González et al., 2013). However, the sum of the masses of all the proteins that have been localized to these projections is just over 3 MD (Table 1). This also suggests that there are many more CA proteins waiting to be discovered. To search for previously uncharacterized proteins of the CA, we have now compared the proteomes of Chlamydomonas WT and CA-less axonemes by label-free quantitative MS. We identified 44 proteins as candidates for being novel CA proteins; at least 13 of these are highly conserved in humans. Detailed studies of five of the conserved proteins confirmed that all five are associated with the CA and cause impaired flagellar motility when missing or defective. Using a combination of genetic, biochemical, and proteomic approaches, we were able to assign many of these proteins to either the C1 or C2 microtubule, and in some cases have been able to predict the specific projections and/or interacting partners with which they are associated. Mutants defective for the confirmed novel CA proteins have a variety of motility phenotypes, indicating different roles for the different proteins. These findings are an important step toward understanding how the CA performs its functions in motile cilia and will facilitate the identification and diagnosis of human patients with defects in the CA. Results Selection of pf18 for comparative MS analysis to identify novel CA proteins To select the best mutant strain for our studies, we first examined cells of pf15, pf18, pf19, and pf20, which partially or totally lack the CA (Adams et al., 1981). As previously reported (Adams et al., 1981), some pf20 cells had motile flagella, suggesting that some of these cells retained a CA that was at least partially functioning. The flagella of pf19 were shorter than those of WT, pf15, and pf18 (Fig. S1 A), raising the possibility that flagellar components other than the CA were defective in this mutant. Thus, strains pf19 and pf20 were not considered further. Transmission EM (TEM) of isolated pf15 and pf18 axonemes confirmed that all axonemes lacked central microtubules, but revealed that 31% of pf15 axonemes had electron-dense material (Witman et al., 1978) in the lumens of their axonemes, whereas only 7.5% of pf18 axonemes had such material (Fig. 1 and Fig. S1, B and C). Except for the absence of the CA, the axonemes of pf18 appeared structurally normal. Therefore, pf18 was selected for our comparative proteomics analysis. Figure 1. Structural characterization of WT and pf18 axonemes used for MS analysis. (A and B) Representative TEMs of isolated axonemes of WT (A) and pf18 (B). Nearly all WT axonemes contained a CA, whereas all pf18 axonemes lacked the CA (see Fig. S1 for quantitation). The axonemes of both preparations were highly pure, with little apparent cell body contamination. Insets: Higher-magnification images confirming integrity of the 9+2 structure in WT axonemes and the 9+0 structure in pf18 axonemes. The lower inset in B shows that the electron-dense material in the lumens of some pf18 axonemes is not the CA. Bars, 0.5 µm (A and B); 0.1 µm (insets). Quantitative MS accurately reports abundances of known axonemal proteins in WT and pf18 axonemes Preparations of isolated WT and pf18 axonemes were analyzed by label-free quantitative MS. Both intensity-based absolute quantification (IBAQ; Schwanhäusser et al., 2011) and Top3 precursor quantification (Silva et al., 2006) methods were used to estimate protein abundance; in general, similar results were obtained for the two methods (Table S1). The experiment was repeated twice, generating two independent biological replicates. Data from each replicate (WT vs. pf18) were put into a separate dataset. The two datasets together contained a total of 1,364 proteins, of which 75% were detected in both replicates (Table S1). All known proteins of the outer and inner dynein arms, the radial spokes, and the N-DRC were detected in both WT and pf18 in both replicates, and all known CA proteins were detected in WT in both replicates. Therefore, the datasets are likely to contain most axonemal proteins. To evaluate our quantitative analysis methods, we first looked at the known axonemal proteins (Fig. 2 A and Table S2). As expected, all outer dynein arm subunits and associated proteins were found in a 1:1 ratio (pf18:WT) except for Lis1. That the level of Lis1 is much higher in pf18 than WT axonemes was previously shown by Western blot analysis and apparently is a cellular response to reduced or disrupted flagellar beating (Rompolas et al., 2012). Similarly, all 25 subunits of the inner dynein arms, all 19 radial spoke subunits for which sequences are available, and all 11 N-DRC subunits were present in a 1:1 ratio (pf18:WT). Figure 2. pf18:WT ratios for select flagellar proteins, abundances for known CA proteins, and criteria used to screen for candidate novel CA proteins. (A) pf18:WT ratios for axonemal and IFT proteins (in this example, the ratios are based on IBAQ scores from replicate 1; see Table S2). Ratios for subunits such as LC7 and LC8, which are present in more than one axonemal structure, are repeated for each structure with which they are associated. The four CA proteins with the highest ratios (calmodulin, enolase, HSP70, and PP1c, in order of decreasing ratios) are located both in the CA and elsewhere in the flagellum. NAP, an unconventional actin previously shown to be a subunit of some inner arm dyneins in the absence of conventional actin, is grouped separately. Our first criterion for considering an uncharacterized protein as a potentially novel CA component was that it had to have a pf18:WT ratio ≤0.2 (red line), ensuring that it is reduced in pf18 axonemes at least as much as PP1c and the known CA-specific proteins. (B) IBAQ values, representing protein abundance, for known CA proteins (values from replicate 1 used as example here). Based on both IBAQ and Top3 values, the least abundant of the known CA proteins in replicate 1 was FAP297. Therefore, our second criterion for considering an uncharacterized protein as a potentially novel CA component was that it had to have an abundance based on IBAQ or Top3 score ≥0.8 of the FAP297 value (red line). Both datasets identified 21 of 22 known intraflagellar transport (IFT) particle proteins (IFT43 was not detected). All were significantly elevated in the pf18 axonemes (Fig. 2 A and Table S2). This is consistent with the previously reported trapping of IFT particles in the axonemal lumen of CA-less mutants (Lechtreck et al., 2013). BBSome proteins and ODA16, which function in IFT (Ahmed et al., 2008; Lechtreck et al., 2009; Hou and Witman, 2017; Taschner et al., 2017), also were greatly increased in pf18 axonemes; a similar increase for the BBSome protein BBS4 in axonemes of pf18 and pf19 was previously shown by Western blotting (Lechtreck et al., 2013). All 22 of the known CA proteins with reported protein sequences (Table 1) were identified in both datasets (Fig. 2 A and Table S2). As expected, the 18 proteins specific for the CA were greatly decreased in pf18 (ratio of pf18:WT < 0.2). The degree to which each protein was reduced varied, consistent with the observation that residual amounts of some but not other CA proteins could be detected in axonemes of pf18 and pf19 by Western blotting (Lechtreck et al., 2013). Protein phosphatase 1 catalytic subunit (PP1c), heat shock protein 70 (HSP70), enolase, and calmodulin, which are located both in the CA and elsewhere in the flagellum (Yang et al., 2000, 2001; Mitchell et al., 2005; Shapiro et al., 2005), were reduced to a lesser extent. Interestingly, katanin 60/PF19 and katanin 80/PF15, which are required for the assembly of the CA (Dymek et al., 2004; Dymek and Smith, 2012), either were not found or were found at a much lower abundance in WT axonemes than were known CA proteins, and their abundance was not less in pf18 axonemes (Table S2). This indicates that they are not CA structural proteins, which is consistent with evidence that katanin 80 is located in the basal body region (Esparza et al., 2013). Taken together, the above results show that our approach has identified the vast majority of axonemal proteins and is accurately reporting their relative amounts in pf18 versus WT axonemes. 44 novel CA candidate proteins were identified by MS analysis The quantitative data for the known CA proteins provided a good reference for identifying novel candidate CA proteins. For PP1c, which is predominantly but not exclusively located in the CA (Yang et al., 2000), the pf18:WT ratio was 0.2; for all CA-specific proteins, the ratio was <0.2 (Fig. 2 A). Therefore, we screened our datasets for proteins that had pf18:WT ratios ≤0.2 (red line in Fig. 2 A). We also needed to set a lower boundary for abundance to eliminate proteins that might be low-level contaminants in the WT preparations but not the pf18 preparations. Fig. 2 B shows the IBAQ scores, which are a measure of protein abundance, for each of the known CA proteins in the WT axoneme as reported for replicate 1. FAP297, which consistently was the least abundant of these proteins, had an IBAQ score of 9.30 × 108, so we set a boundary at 80% of this value (7.44 × 108) as the lowest IBAQ score that a protein in replicate 1 could have and still be considered a potential candidate CA protein (red line in Fig. 2 B). Proteins in this replicate were similarly screened based on Top3 values and pooled with those selected on the basis of IBAQ scores. Proteins in replicate 2 were screened in the same way using FAP297 abundance values specific to that replicate. From the two replicates, a total of 63 proteins were identified that met both criteria (pf18:WT ratio and protein abundance). By definition, these included PP1c and the 18 previously known CA-specific proteins. The remaining 44 proteins (Tables 2 and S3 and Fig. S2) are candidates for being novel CA proteins. Table 2. Novel candidate CA proteins Protein a NCBI accession number Mass (kD) Phytozome gene number Conserved domains Human homolog b Location FAP7 XP_001693026.1 55 Cre12.g531800 N N ?c FAP39 XP_001694877.1 101 Cre02.g145100 Calcium-transporting ATPase N C1d FAP47 XP_001691575.1 310 Cre17.g704300 ASH, Calponin homolog CFAP47 ?c FAP49 XP_001696011.1 295 Cre08.g362050 PAS domain N C2c FAP65 XP_001702074.1 220 Cre07.g354551e ASH domain CCDC108 C2c FAP70 XP_001692552.1 114 Cre07.g345400 TPR repeat, VMR2 CFAP70/TTC18 C2c FAP72 XP_001696096.1 595 Cre08.g362000 PAS domain N NA FAP75 XP_001696434.1 125 Cre06.g249900 Adenylate kinase N C2c FAP76 XP_001694909.1 162 Cre09.g387689 DUF4455, DUF4456 CCDC180/C9orf174 C1c FAP81 XP_001691318.1 172 Cre06.g296850 ASH domain DLEC1 C1c,d FAP92 XP_001693652.1 150 Cre13.g562250 N N C1c FAP99 XP_001692825.1 90 Cre14.g624400 Neuromodulin family CFAP99 C1c FAP105 XP_001691070.1 31 Cre12.g511750 N N C1c FAP108 XP_001691315.1 22 Cre06.g297200 N N C1c FAP123 XP_001703476.1 34 Cre03.g171800 DUF4045 domain N ?c FAP125 XP_001693959.1 112 Cre12.g546100 Kinesin motor N ?c FAP139 XP_001694913.1 76 Cre09.g387912 DUF390 domain N ?c FAP147 XP_001692304.1 97 Cre04.g224250 MYCBP-associated protein family, ASH domain MYCBP-associated protein C2c FAP154 XP_001696012.1 467 Cre08.g362100 PAS domain N ?c FAP171 XP_001692822.1 81 Cre14.g624900 N N C2c FAP178 XP_001702799.1 20 Cre10.g418150 Calponin homolog N C2c FAP194 XP_001697045.1 52 Cre12.g522150 Armadillo repeat Spag6 ?c FAP216 XP_001702365.1 79 Cre12.g497200 N N C1c FAP225 XP_001689770.1 81 Cre01.g051050 EF-hand N ?c FAP239 XP_001695620.1 23 Cre03.g145387 Mechanosensitive channel N C2c FAP246 XP_001689839.1 32 Cre14.g618750 LRR, TGC, EF-hand LRGUK C1c,d FAP266f XP_001699180.1 22 Cre16.g690450 MORN repeat RSP10B ?c XP_001699181.1 FAP275 XP_001697713.1 18 Cre05.g239200 N N C1c FAP286 XP_001690861.1 17 Cre12.g509800 SMC family N C2c FAP289 XP_001700259.1 46 Cre01.g009800 N N C1c FAP312 XP_001692669.1 34 Cre14.g630200 Cdk activating kinase N C2c FAP345 XP_001701140.1 13 Cre15.g640000 N N C1c FAP348 XP_001699354.1 96 Cre16.g693204 N N C1c FAP380 XP_001689865.1 20 Cre01.g010400 N N ?c FAP411 XP_001696779.1 13 Cre09.g409600 N N C1c FAP412 XP_001702388.1 57 Cre12.g497450 LRR domain N C1c FAP413 XP_001696110.1 37 Cre08.g364000 N N C1c FAP414 XP_001694676.1 17 Cre09.g394065 N N C1c FAP415 XP_001698357.1 23 Cre10.g454600 N N NA FAP416 XP_001691634.1 17 Cre17.g697750 N N NA FAP417 XP_001702692.1 53 Cre10.g429750 N N ?b DIP13 XP_001697145.1 13 Cre17.g724550 DUF3552 domain SSNA1 NA DPY30 XP_001695420.1 11 Cre06.g279100 Dpy-30 domain Dpy-30 C1c,d MOT17 XP_001697337.1 28 Cre11.g482300 Myosin head family Spata17 C1c,d NAPg XP_001703266.1 42 Cre03.g176833 Actin superfamily N C1c,d N, no conserved domain or no human homolog; NA, data on location incomplete or not reproducible, see “Other” cluster in Fig. 3. a FAP proteins with numbers >410 are given the FAP designation for the first time in this study. b E value ≤10−10 and reciprocal best match (except for Spag6, whose reciprocal best match is PF16). c Cluster analysis in Fig. 3. d Coimmunoprecipitation. e The correct Phytozome gene number for FAP65 is Cre07.g354551, but Phytozome has incorrectly associated the name “FAP65” with a different gene. f The sequences corresponding to NCBI XP_001699180.1 and XP_001699181.1 (the former annotated as FAP266) are merged into a single sequence (Cre16.g690450) in Phytozome. Our search identified numerous peptides expressed from both halves of the Phytozome sequence, suggesting that the conjoined model is the correct one for FAP266. g NAP was below the abundance threshold used to identify candidate CA proteins but is included here for reasons discussed in text. The criteria used to select these 63 proteins were relatively stringent; as a result, Table S1 likely includes additional candidates for novel CA proteins that were excluded from our list of 63 proteins because they fell above our cutoff for pf18:WT ratio or below our cutoff for abundance. One such protein may be NAP (novel actin-like protein), an enigmatic protein that replaces conventional actin in some inner arm dyneins when conventional actin is missing (Kato-Minoura et al., 1997) but whose function in the WT cell has been a mystery. Our MS analysis showed that NAP was present in WT axonemes at levels consistently 0.1–0.2 that of FAP297, so it was excluded from our initial list of candidate novel CA proteins. However, it is greatly reduced in pf18 axonemes (Fig. 2 A), a result that was confirmed by Western blot analysis (Fig. S3 A). Previously, NAP either was not detected in WT axonemes by Western blotting or was detected only in very small amounts (Kato-Minoura et al., 1998; Hirono et al., 2003). Our results suggest that, in the WT axoneme, NAP has some role associated with the CA. Consequently, we have added it to our list of candidate novel CA proteins (Table 2). Assignment of candidate novel CA proteins to the C1 or C2 microtubule We used two approaches to assign the candidate novel CA proteins to either the C1 or C2 microtubule. In the first approach, we took advantage of the fact that the C2 microtubule can be selectively solubilized from isolated axonemes by high-salt treatment, thus leaving most C2-associated proteins in the soluble fraction and most C1-associated proteins in the insoluble fraction (Dutcher et al., 1984; Mitchell and Sale, 1999; Lechtreck and Witman, 2007). We isolated WT axonemes, fractionated them by high-salt treatment and centrifugation, and then analyzed the fractions by quantitative MS to determine the supernatant-to-pellet ratio for each candidate CA protein. A ratio substantially lower than 1 predicts that the protein is a component of the C1 microtubule, whereas a ratio substantially higher than 1 predicts that the protein is a component of the C2 microtubule. In the second approach, we took advantage of the fact that the C1 microtubule is unstable and thus missing from isolated axonemes of the paralyzed flagella mutant pf16 when the axonemes are demembranated by treatment with detergent in buffer containing 25 mM NaCl (Dutcher et al., 1984; Smith and Lefebvre, 1996; Mitchell and Sale, 1999). We isolated axonemes of WT and pf16 in this way and then used quantitative MS to estimate the pf16:WT ratio for each candidate CA protein. A ratio substantially lower than 1 predicts that the protein is a C1 protein, whereas a ratio of ∼1 predicts that the protein is a C2 protein. The ratios from the two approaches were then used to manually group the known and candidate CA proteins into clusters (Fig. 3 and Table S4). Figure 3. Assignment of novel candidate CA proteins to the C1 or C2 microtubule. The results of the two different approaches described in the text were combined to predict association with either the C1 or the C2 microtubule. Each approach was repeated twice to provide two biological replicates (Rep. 1 and Rep. 2), and each replicate was analyzed using IBAQ and Top3 methods. Proteins with similar ratios were then grouped manually into the clusters shown. The soluble:insoluble ratios for replicates 1 and 2 are indicated by green and blue bars, respectively; the pf16:WT ratios for replicates 1 and 2 are indicated by purple and red bars, respectively. For each replicate, the ratios as reported by IBAQ and Top3 are shown in that order using the same color. To the left of each bar graph, known C1 proteins are indicated by a light-red background; known C2 proteins have a light-green background. FAP174, indicated by a light-blue background, previously was assigned to C2 (Rao et al., 2016), but this assignment does not fully agree with our other results (see Some candidate proteins can be assigned to specific CA projections). Candidate CA proteins in Cluster 1 are predicted to be C1 proteins, whereas those in Cluster 3 are predicted to be C2 proteins. The locations of candidate CA proteins in the other clusters are less certain. For the proteins in cluster 1, the soluble:insoluble ratios were <1 and the pf16:WT ratios were <1, which means that both approaches predicted that these proteins are associated with the C1 microtubule. Indeed, all 13 known CA proteins in this cluster were previously assigned to C1. Thus, the 18 candidate CA proteins in this cluster are predicted with high confidence to be associated with C1. NAP also is in this cluster and is predicted to be a C1 protein. Cluster 2 contains hydin, a C2 protein that remains associated with C1 when the C2 microtubule is solubilized (Lechtreck and Witman, 2007). Like hydin, candidate CA proteins FAP47, FAP194, and FAP380 had ratios relatively close to 1 in both experimental approaches, meaning that they are not readily lost from the axoneme when either C1 or C2 is destabilized. They could bridge C1 and C2 directly, or like hydin, they could be part of a projection on one central microtubule and have a stable attachment to a projection on the other central microtubule. Cluster 3 contains proteins that had soluble:insoluble ratios much greater than 1 and had pf16:WT ratios ∼1, meaning that both approaches predicted that these are C2 proteins. Indeed, the two known CA proteins in this cluster previously were assigned to C2. Therefore, the 10 candidate CA proteins in this cluster are predicted with high confidence to be C2 proteins. The location of the candidate CA proteins in cluster 4 is less certain. These proteins had soluble:insoluble ratios >1 (indicating association with C2) but pf16:WT ratios <1 (indicating association with C1). These proteins could be C1 proteins that are extracted from WT axonemes by high salt or C2 proteins that are unstable in pf16 axonemes. PP1c and FAP297, both previously assigned to C1, are in this cluster. There is evidence that FAP297 is a subunit of the C1d projection of the C1 microtubule (Brown et al., 2012), but it was not identified in an earlier biochemical study of this projection (DiPetrillo and Smith, 2010), suggesting that it may indeed be prone to dissociation from the rest of the projection. In any case, the candidate CA proteins in this cluster cannot be assigned to a specific central microtubule based on these results alone. For proteins in the “other” cluster, the ratios were not reproducible in the biological replicates for one or both approaches (e.g., FAP7 and FAP174), the ratios were different from those of any other protein (FAP225), or the data were incomplete, i.e., the proteins were not detected in one or more of the samples. Selected candidate CA proteins were confirmed to localize in the CA To determine how reliable our identification of potentially novel CA proteins was, six of these proteins (DPY30, FAP47, FAP70, FAP76, FAP99, and FAP246) that are highly conserved in humans were selected for further analysis. We also selected two proteins, FAP196 and WNK1, that were greatly reduced in pf18 axonemes but fell just below our cutoff for abundance. Chlamydomonas insertional mutants reported to be defective in the genes encoding these proteins were obtained from the Chlamydomonas Library Project (https://www.chlamylibrary.org/; Li et al., 2016). All the mutants were reported to have insertions in the coding regions of the affected genes. We checked the insertion sites by PCR and confirmed the sites for all but fap70-1 and wnk1-1, which were not investigated further (Fig. S4, A–F). The remaining mutants were backcrossed to WT, and progenies containing the insertions (Fig. S4 G) were then transformed with constructs designed to express HA-tagged versions of the WT proteins. Western blot analysis of isolated axonemes from the transformed strains confirmed the presence of these HA-tagged proteins in the axonemes (Fig. S3 B). However, the abundance of FAP196-HA in isolated axonemes was very low; most of the protein was in the cell body (Fig. S3 C). It is possible that the HA tag impeded assembly of FAP196 into the axoneme. We examined these six mutants for defects in flagella length and swimming speed (Fig. 4, A and B). Among them, fap76-1 has the shortest flagella and swims the slowest; it also turns more frequently than WT (Fig. 4 C). These fap76-1 phenotypes were rescued by the HA-tagged FAP76, showing that the insertion in the FAP76 gene is the underlying cause of the phenotypes. fap47-1 has flagella that may be slightly longer than normal but swims significantly slower than WT; its slow-swimming phenotype was partially rescued by FAP47-HA. The strain also has a phototaxis defect, which was not rescued by FAP47-HA (Fig. 4 D). However, a second insertional mutant, fap47-2, has a similar phototaxis defect (unpublished data), suggesting that FAP47 is important for phototactic steering. Cells of fap99-1, fap246-1, and dpy30-1 have normal- or nearly normal-length flagella and swim slightly slower than WT; their slow swimming phenotypes were rescued by the HA-tagged constructs. fap196-1 had no obvious defects in flagella length or swimming speed. Figure 4. Phenotypic analysis of insertional mutants and rescued strains. (A) Flagellar length of WT, the fap47-1, fap76-1, fap99-1, fap196-1, fap246-1, and dpy30-1 insertional mutants, and the mutants following transformation with constructs designed to express WT HA-tagged versions of the proteins defective in each. n = number of flagella scored; error bars indicate standard deviation; *, significant difference from WT (Student’s t test, P ≤ 0.05). (B) Swimming speed of the same strains as in A. n = number of cells scored; error bars indicate standard deviation; *, significant difference (Student’s t test, P ≤ 0.05) between WT and mutant or mutant and rescued strains as indicated. (C) Swimming paths of WT, the fap76-1 mutant, and the mutant following transformation with the construct expressing FAP76-HA. The exposure time was 1 s; bar, 50 µm. (D) Swimming paths of WT, the fap47-1 mutant, and the mutant following transformation with the construct expressing FAP47-HA, in the absence of photostimulation (top panels) and in the presence of photostimulation coming from the left (arrow; lower panels). The large dot in each swimming path indicates the end of the track. fap47-1 has a slow-swimming phenotype that was partially rescued by FAP47-HA; the phototaxis defect was not noticeably rescued. The exposure time was 0.5 s; bar, 50 µm. To determine if these proteins are components of the CA, we used superresolution structured illumination microscopy (SIM) to investigate the locations of the HA-tagged proteins in isolated axonemes. Fig. 5 A shows axonemes from control and FAP47-HA cells probed with anti-HA and anti-acetylated tubulin antibodies. The antibody to the HA tag labels a thin structure that is centered in and runs the entire length of the FAP47-HA axoneme (B, B′, and B″ in Fig. 5 A); no HA signal was detected in control axonemes (i.e., the nonrescued mutants; A, A′, and A″). To confirm that the labeled structure is the CA, the CA was partially (C, C′, and C″ in Fig. 5 A) or completely (D, D′, and D″) extruded from the axoneme by addition of ATP (Mitchell and Nakatsugawa, 2004; Lechtreck and Witman, 2007). In this case, the FAP47-HA signal was clearly associated with the CA but not the rest of the axoneme. Therefore, within the axoneme, FAP47 is located exclusively in the CA. Figure 5. Localization of selected candidate CA proteins to the CA. (A–E) SIM images of isolated axonemes from mutant cells rescued with FAP47-HA (A), FAP76-HA (B), FAP99-HA (C), FAP246-HA (D), and DPY30-HA (E). Each set of images shows control intact axonemes from the nonrescued mutant (subpanels A, A′, A″), intact axonemes from the rescued mutant (subpanels B, B′, B″), and axonemes from the rescued mutant after treatment with ATP to induce partial (subpanels C, C′, C″) or complete (subpanels D, D′, D″) extrusion of the CA. Axonemes were double labeled with antibodies to acetylated tubulin (red; A–D) and the HA tag (green; A′–D′); merged images are shown in A″–D″. There was no anti-HA signal from the control axonemes. In the intact rescued axonemes, the CA can be seen as a thin anti-HA–labeled structure centered in and extending the length of the axonemes. In the samples of rescued axonemes with partially or fully extruded CAs, the CAs appear, respectively, as thin microtubular structures projecting (arrows in C″) or completely separated (arrows in D″) from the axonemes. In the merged images, regions of overlap between the CA and outer doublet microtubules appear yellow. (F) (A) Enlargement of B″ from B, and (B) enlargement of C″ from E. Bar, 5 µm. Using the same approach, we determined that FAP76, FAP99, FAP246, and DPY30 also are located exclusively in the CA (Fig. 5, B, C, D, and E, respectively; and see Fig. 5 F for enlargements). No anti-HA signal could be detected in axonemes from FAP196-HA cells (not depicted). Interacting partners of novel CA proteins and assignment to specific projections of the CA Having established that the selected candidate CA proteins were indeed components of the CA, we next used immunoprecipitation coupled with MS to identify possible interacting partners for these proteins. Axonemes from WT (control) and the mutant cells rescued with HA-tagged proteins were isolated and extracted with 0.6 M KCl to solubilize most C2 and some C1 microtubules (Pazour et al., 2005). The solubilized HA-tagged proteins were then immunoprecipitated with anti-HA antibodies, and the immunoprecipitates were analyzed by MS. The specificity of coimmunoprecipitation was then assessed on the basis of both abundance relative to the bait protein and enrichment relative to the WT control. The previously described fap46-1 FAP46-HA strain (Brown et al., 2012) was used as a positive control to make sure that our extraction and immunoprecipitation conditions worked for a CA projection. The mutant fap46-1 is null for FAP46, which is a component of the C1d projection that also contains FAP54, FAP74, FAP221/Pcdp1, and FAP297/C1d-87; fap46-1 lacks the C1d projection but is completely rescued by FAP46-HA (DiPetrillo and Smith, 2010; Brown et al., 2012). The anti-HA antibody specifically coimmunoprecipitated FAP46-HA and all other known components of C1d with the exception of FAP297 (Table 3 and Table S5). Thus, our conditions appear to be appropriate for investigation of CA protein-interacting partners. Table 3. Proteins identified in immunoprecipitations from axonemal extracts Bait Identified protein NCBI accession number Description Location FAP46-HA FAP46 XP_001702776.1 Flagellar associated protein C1d projection FAP54 XP_001696950.1 Flagellar associated protein C1d projection FAP74 ADD85930.1 Flagellar associated protein C1d projection FAP221 ADD85929.2 Flagellar associated protein C1d projection FAP47-HA FAP47 XP_001691575.1 Flagellar associated protein Unknown FAP76-HA FAP76 XP_001694909.1 Flagellar associated protein C1 microtubulea FAP99-HA FAP99 XP_001692825.1 Flagellar associated protein C1 microtubulea FAP246-HA CPC1 XP_001702926.1 Protein associated with central pair microtubule complex C1b projection FAP42 XP_001697065.1 Adenylate/guanylate kinase-like protein C1b projection FAP69 XP_001703508.1 Flagellar associated protein C1b projection HSP70A XP_001701326.1 HSP70A C1b projection Enolaseb XP_001702971.1 Enolase C1b projection FAP39b XP_001694877.1 Calcium-transporting ATPase C1b projection FAP174b ACR55627.1 Flagellar associated protein C1b projection FAP246 XP_001689839.1 Flagellar associated protein C1b projection DPY30-HA PF6 AAK38270.1 PF6 protein C1a projection FAP101 AAZ31187.1 CA-associated protein C1a-86 C1a projection FAP114 AAZ31185.1 CA-associated protein C1a-32 C1a projection FAP119 XP_001696622.1 CA-associated protein C1a-34 C1a projection FAP227 AAZ31184.1 CA-associated protein C1a-18 C1a projection Calmodulinb 1206346A Calmodulin C1a projection DPY30b XP_001695420.1 Subunit of chromatin modifying protein C1a projection FAP81 XP_001691318.1 Flagellar associated protein C1a projection MOT17 XP_001697337.1 Predicted protein C1a projection NAP XP_001703266.1 Actin-related protein C1a projection Previously known CA proteins are indicated by a gray background. a Predicted location based on the cluster analysis of Fig. 3. b Protein was identified or specifically enriched in only one replicate. The anti-HA immunoprecipitates from the DPY30-HA strain specifically contained all six of the known C1a proteins (PF6, FAP101, FAP114, FAP119, FAP227, and calmodulin). DPY30-HA and calmodulin were identified by MS in only one of the two biological replicates, probably because they are small proteins (11 and 18 kD, respectively). However, calmodulin is a known component of the C1a projection (Wargo et al., 2005), and Western blotting confirmed that DPY30-HA was present in the sample where MS had failed to detect it (unpublished data). Therefore, DPY30 likely is associated with the C1a projection. In addition, the immunoprecipitate specifically contained two other candidate CA proteins, MOT17 and FAP81, indicating that they also are likely to be novel C1a components. Finally, NAP was specifically and highly enriched in this immunoprecipitate, suggesting that it also is associated with C1a. The anti-HA immunoprecipitates from the FAP246-HA strain specifically contained the known C1b proteins CPC1, HSP70A, FAP42, and FAP69. In addition, the C1b protein enolase was specifically coimmunoprecipitated in one of the two replicates. Therefore, FAP246 is very likely a C1b protein. The candidate CA proteins FAP39 (101 kD) and FAP174 (10 kD) were specifically enriched in one of the replicates, suggesting, with less confidence, that they also are components of C1b. The immunoprecipitates from the FAP47-HA, FAP76-HA, and FAP99-HA strains contained the expected HA-tagged protein, but no other proteins were specifically present at similar abundance. It is likely that these HA-tagged proteins were dissociated from their interacting partners by our extraction conditions. Proteins missing or reduced in axonemes of insertional mutants To investigate the expression of the novel CA proteins in axonemes of the mutants with defects in the genes encoding these proteins, we used quantitative MS to compare the axonemal proteome of each mutant with that of WT in two biological replicates. In each analysis, we focused on the known and candidate CA proteins that were present in the mutant axonemes in amounts ≤0.2 that of WT (Table 4 and Table S6). Table 4. Comparative proteomic analysis of WT and mutant axonemes Mutant Protein missing or greatly decreased Accession number Description Predicted location fap47-1 FAP47a XP_001691575.1 Flagellar associated protein Unknown FAP49 XP_001696011.1 Flagellar associated protein FAP72 XP_001696096.1 Flagellar associated protein FAP154 XP_001696012.1 Flagellar associated protein FAP414a XP_001694676.1 Predicted protein FAP416a XP_001691634.1 Predicted protein fap76-1 FAP76 XP_001694909.1 Flagellar associated protein C1 microtubuleb FAP413 XP_001696110.1 Predicted protein FAP414 XP_001694676.1 Predicted protein fap99-1 FAP99 XP_001692825.1 Flagellar associated protein C1 microtubuleb fap196-1 FAP196 XP_001697178.1 Flagellar associated protein Unknown FAP414c XP_001694676.1 Predicted protein fap246-1 FAP246 XP_001689839.1 Flagellar associated protein C1b projectiond FAP413 XP_001696110.1 Predicted protein dpy30-1 DPY30 XP_001695420.1 Subunit of chromatin modifying protein C1a projectiond a Repeatable in two biological replicates by IBAQ and one replicate by Top3; in the other replicate, Top3 reported that the protein was reduced in the mutant axonemes to only 0.27–0.37 of the WT level. Our cutoff for specificity was that a protein had to be reduced to ≤0.2 of the WT level. b Predicted location based on cluster analysis of Fig. 3. c IBAQ reported that this protein was reduced to <0.2 of the WT level in both biological replicates, whereas Top3 reported that it was present at 0.38 and 2.52 of the WT level in replicates 1 and 2, respectively. d Predicted location based on cluster analysis and supported by coimmunoprecipitation. MS of fap47-1 axonemes detected some FAP47 peptides. In both fap47-1 samples, all but one of these peptides were encoded by sequence located upstream of the insertion site. The single exception was present in a very low amount compared with the more N-terminal peptides, which themselves were present at a level 0.15–0.27 that in WT. These results indicate that a C-terminally truncated FAP47 is assembled into the fap47-1 axoneme at a level well below that of full-length FAP47 in the WT axoneme. In addition, candidate CA proteins FAP49, FAP72, FAP154, and FAP416 were greatly reduced in fap47-1 axonemes, indicating that they are likely to be in the same CA substructure as FAP47 and dependent on FAP47 for their assembly into the axoneme. FAP414 also was substantially reduced, but it was reduced in axonemes of three other mutants as well (see below). Because none of the previously known CA proteins were noticeably affected in this mutant, FAP47 likely is a subunit of a CA substructure that has not previously been characterized biochemically. FAP47 contains ASH (ASPM, SPD-2, Hydin) domains, which may have a role in microtubule binding (Ponting, 2006; Schou et al., 2014); consequently, it is tempting to speculate that FAP47 anchors the complex to the CA via its ASH domains. Axonemes of the fap76-1 mutant contained some FAP76 peptides that were encoded only by sequence downstream of the insertion site, indicating that an N-terminally truncated protein is expressed. The abundance (Top3 method) of the truncated protein in the mutant axonemes was much less than that of the full-length protein in WT, suggesting that only a small amount of the truncated protein is assembled into the fap76-1 axoneme. The only other candidate CA proteins greatly reduced in fap76-1 axonemes were FAP413, which also was reduced in axonemes from fap246-1 (see below), and FAP414. As none of the known CA proteins were markedly decreased in this mutant, FAP76 also is likely to be a component of a previously uncharacterized CA substructure. In the two samples of fap99-1 axonemes, MS detected one and two exclusive unique FAP99 peptides compared with 22 and 20 exclusive unique peptides in the WT samples, indicating that very little of this protein is incorporated into the mutant axoneme. No other candidate or known CA proteins were comparably reduced in these axonemes. In the two samples of fap246-1 axonemes, MS detected three and five exclusive unique FAP246 peptides compared with nine exclusive unique peptides in each of the WT samples. IBAQ and Top3 scores for these peptides were much lower in the mutant samples than in WT, suggesting that very little of FAP246 is assembled into the mutant axoneme. FAP413 also was reduced. No DPY30 peptides were detected in dpy30-1 axonemes, compared with three and four exclusive unique peptides in the two control samples. No other candidate CA proteins were greatly reduced in the dpy30-1 axonemes. FAP196 was readily detected in WT axonemes, but not in fap196-1 axonemes, indicating that the mutant axoneme is completely devoid of the protein. In addition, FAP414 was considerably reduced. As noted above, FAP413 was decreased in both fap76-1 and fap246-1 axonemes, and FAP414 was decreased in fap47-1, fap76-1, and fap196-1 axonemes. These results suggest one of the following possibilities: (a) FAP413 and FAP414 are present in multiple locations in the CA; (b) FAP413 interacts with FAP76 and FAP246, while FAP414 interacts with FAP47, FAP76, and FAP196; or (c) FAP413 and FAP414 associate with the CA nonspecifically. Discussion The ensemble of CA proteins is likely to be at least three times greater than previously known In an effort to identify more CA proteins, we used label-free MS to carry out a comparative proteomics analysis of Chlamydomonas WT and CA-less axonemes. The analysis generated a list of 45 candidate novel CA proteins (including NAP), of which at least 13 are conserved in humans (BLAST E ≤ 10−10; Table 2). Further study of five of the conserved proteins by SIM showed that all five are indeed located specifically in the CA, and analysis of these proteins' interacting partners provided evidence that additional candidate proteins are in fact located in the CA. Thus, the total number of CA proteins is likely to be at least three times the number that was previously known. Assignment of proteins to the C1 or C2 microtubule By combining data from coimmunoprecipitations and experiments in which either the C1 or C2 microtubule was selectively solubilized, we were able to tentatively assign 20 of the candidate CA proteins to the former and 10 others to the latter (Fig. 6). Eighteen of the previously known CA proteins had been localized to C1, so the total of known and predicted proteins for that microtubule is 38. Only four of the previously known CA proteins were assigned to C2, so the new candidates bring the total for that microtubule to 14 (including FAP174, but see next section). Therefore, the C1 microtubule seems to be associated with about twice as many proteins as the C2 microtubule. This is consistent with the estimate that the total mass of the projections of the C1 microtubule (10.3 MD) is about twice that of the projections of the C2 microtubule (4.3 MD; Carbajal-González et al., 2013). Figure 6. Summary of CA proteins and their predicted locations in the C1 and C2 microtubules. Diagram of cross section of the Chlamydomonas CA (Carbajal-González et al., 2013) showing predicted locations of previously known CA proteins (normal font) and new candidate or confirmed CA proteins (bold font). Earlier biochemical and structural studies assigned many known proteins to the C1a, C1b, and C2c projections, but more recent cryo-ET data indicate that these structures should be divided into C1a and C1e, C1b and C1f, and C2c and C2d projections, respectively (Carbajal-González et al., 2013). Hence these proteins and new candidate CA proteins that interact with them may be located in either member of these pairs of projections, as indicated by the curved brackets. The FAP47 complex (box, upper right) includes FAP49 and so likely is associated with C2, but where in C2 has not been determined. The question mark indicates proteins whose locations in the CA are not yet known. One of the proteins that we predict is associated with the C2 microtubule is FAP70 (114 kD). Cryo-ET comparison of the outer doublet microtubules of WT Chlamydomonas and a fap70 mutant generated by CRISPR/Cas9 found no structural defect in the mutant doublets, although extra density was detected at the base of the outer dynein arm and on the N-DRC of doublet microtubules from the fap70 mutant following its rescue with FAP70 fused to the biotin carboxyl carrier protein tag (Shamoto et al., 2018). This led the authors to conclude that FAP70 is located at the base of the outer dynein arm. However, a structural defect or a more definitive localization of the tagged protein might have been observed if the CA had been examined. Another candidate protein that we predict is associated with the C2 microtubule is FAP49. FAP49, along with FAP47 and three candidate CA proteins (excluding FAP414), was greatly reduced in axonemes of the fap47-1 mutant, indicating that all five proteins are likely to be part of the same complex (Fig. 6). FAP47 had a solubility profile closely resembling that of hydin, which is located in the C2b projection but remains associated with C1 (and the C1b projection) when C2 is solubilized (Lechtreck and Witman, 2007). The other three proteins could not be clearly assigned to C1 or C2. Thus, it is possible that this complex has connections to both C1 and C2, or to projections emanating from both C1 and C2. Of the five proteins in this complex, three (FAP49, FAP72, and FAP154) are predicted to have PAS (Per-ARNT-Sim) domains. This is remarkable because the entire axoneme contains only six or seven PAS proteins (The Chlamydomonas Flagellar Proteome Project: ChlamyFP; http://chlamyfp.org). Intriguingly, PAS domains in many other proteins are involved in responses to environmental factors such as light, oxygen, and redox potential (Taylor and Zhulin, 1999; Vogt and Schippers, 2015). Therefore, it may not be coincidental that the fap47 mutants appear to have defects in phototaxis. In the course of these analyses, we also have determined the likely identities of two proteins previously shown to be associated with the CA, but not yet known at the sequence level. One of these, termed C1 kinesin (Loreng and Smith, 2017), had an estimated mass of 110 kD as determined by SDS-PAGE, was recognized by two different anti-peptide polyclonal antibodies generated against conserved kinesin sequences, and was predicted to be a C1 protein (Fox et al., 1994; Mitchell and Sale, 1999). FAP125 is likely to be this kinesin: it has a kinesin motor domain, has a mass of 110,544 D, contains sequences 100% and 70% identical to those used to generate the two pan-kinesin antibodies, and had a pf16:WT ratio <1, consistent with it being associated with the C1 microtubule. The second protein, AKAP240, had an estimated mass of 240 kD as determined by SDS-PAGE and was predicted to be associated with the C2 microtubule (Gaillard et al., 2001; Rao et al., 2016). Among our candidate CA proteins, only one matches this description: FAP65 has a predicted mass of 233,007 D and has solubility properties consistent with it being a C2 protein. Some candidate proteins can be assigned to specific CA projections When the novel CA protein DPY30-HA was immunoprecipitated, nine other proteins were specifically coimmunoprecipitated, including six previously known CA proteins, two candidate CA proteins (FAP81 and MOT17), and NAP (see next section). The six previously characterized proteins are all known to be subunits of the C1a projection. Therefore, DPY 30 and the other proteins coimmunoprecipitated with it are likely to also be located in the C1a projection, or in an adjacent projection that physically connects to the C1a projection. Similarly, when the novel protein FAP246 was immunoprecipitated, seven other proteins coimmunoprecipitated with it. Six of these previously had been localized to the CA, and five of them had been localized specifically to the C1b projection. This suggests that FAP246 and the candidate CA protein FAP39, which is one of the proteins coimmunoprecipitated with FAP246, also are components of the C1b projection. The other protein that was coimmunoprecipitated with FAP246 was FAP174. FAP174 previously was assigned to the C2 microtubule because it was present in demembranated axonemes of pf16 but was reduced or absent in axonemes of pf18 and other CA-less mutants (Rao et al., 2016). Our findings that FAP174 is present along with several C1b proteins in FAP246-HA immunoprecipitates suggests that FAP174 is a subunit of either C1b or a projection that interacts with C1b. The C1b and C2b projections physically interact (Mitchell and Sale, 1999; Lechtreck and Witman, 2007; Carbajal-González et al., 2013), and C1b remains associated with the C2 microtubule when the rest of the C1 microtubule is solubilized (Mitchell and Sale, 1999), so a location of FAP174 in either C1b or C2b would be consistent with all evidence to date. NAP NAP is a highly divergent member of the actin family that is expressed at very low levels in WT cells but strongly up-regulated in the absence of conventional actin (Kato-Minoura et al., 1998; Hirono et al., 2003). Conventional actin is a subunit of several flagellar inner arm dyneins (Piperno and Luck, 1979), and NAP can substitute for actin in some of these dyneins when conventional actin is absent (Kato-Minoura et al., 1997). However, Western blot analysis using an antibody specific for NAP failed to reproducibly detect NAP in axonemes of WT cells, leading to the conclusion that NAP is “present in WT axonemes in a very small concentration, if present at all” (Kato-Minoura et al., 1998). Further investigation revealed that NAP expression in WT cells is strongly induced by deflagellation, that NAP can be detected in newly formed WT flagella and axonemes by Western blotting, and that the amount of NAP in the axoneme appeared to gradually decrease with time (Hirono et al., 2003), suggesting that NAP may have a role in flagellar assembly. Our MS analysis of WT axonemes provides compelling evidence that NAP is associated with the axoneme even in the presence of conventional actin. Its 1,000-fold reduction in pf18 axonemes relative to WT axonemes even though actin levels are unchanged suggests that NAP is uniquely associated with the CA, although it is present in the WT axoneme in an amount ∼100 times less than any previously known CA protein. This very low abundance likely explains why NAP was difficult to reproducibly detect in isolated WT axonemes in the study of Kato-Minoura et al. (1998). Its enrichment along with several known subunits of the C1a projection in immunoprecipitates of DPY30-HA suggests that it may be associated with that projection or with an interacting structure. In support of this, its solubility profile (Fig. 3) is consistent with it being associated with the C1 microtubule. Therefore, we propose that axonemal NAP has some function related to the CA. It may be a structural component of the CA, or it may have a scaffolding function important for CA assembly. Alternatively, NAP may function as an IFT cargo adaptor that links a preassembled CA structure to the anterograde IFT machinery for transport into the flagellum during flagellar assembly, and then largely disappears from the flagellum once assembly is complete. An analogous function recently was reported for IDA3, an IFT cargo adaptor that couples the inner arm dynein I1 to the anterograde IFT machinery (Hunter et al., 2018). It will be of interest to determine if one of the human actin isoforms has a CA association similar to that of NAP. Mutants with defects in novel CA proteins reveal new roles for CA structures Previous studies have shown that diverse phenotypes result from defects in specific CA proteins and structures. The cpc1 mutant, lacking the C1b projection, may be the least affected, with apparently normal forward swimming at a beat frequency that is about two-thirds that of WT and with apparently normal phototactic and photoshock responses (Mitchell and Sale, 1999); the reduced beat frequency may be due to a reduced intraflagellar ATP level as a result of loss of enolase, a subunit of C1b (Mitchell et al., 2005). Other mutations or knockdowns cause more severe phenotypes. The pf6-2 mutant, which lacks the C1a projection, has flagella that twitch ineffectively so that cells make little or no forward progress (Rupp et al., 2001). Loss of most of projection C2c as a result of RNAi knockdown of the central-pair kinesin Klp1 to ∼20% of its WT level results in a reduction of beat frequency to ∼60% that of WT in most cells, and paralysis in ∼5% of cells (Yokoyama et al., 2004). Loss of the C2b projection as a result of hydin RNAi knockdown causes an unusual form of paralysis in which the flagella of most cells are arrested at the transition points between effective and recovery strokes, indicating a defect in switching dynein arms on and off at specific times during the flagellar beat cycle (Lechtreck and Witman, 2007). Mutations in subunits of the C1d projection cause immotility of most cells and reduced swimming velocity via reduced flagellar beat frequency in those cells that can swim; importantly, phototaxis and photoshock are severely impaired in these mutants, documenting a role for C1d in these behaviors (DiPetrillo and Smith, 2011; Brown et al., 2012). These diverse phenotypes indicate that different CA projections have specific roles, a conclusion that is supported by the unique architecture of each projection (Carbajal-González et al., 2013). The novel mutants described here expand the range of phenotypes associated with CA defects. Cells of fap76-1 exhibit a novel CA phenotype in which nearly all cells are motile but swim erratically at about one-half the speed of WT, suggesting impairment of the mechanism controlling cell steering. In another phenotype not previously reported, nearly all fap47 cells move forward in normal swimming paths at ∼80% the speed of WT and appear to be defective specifically in phototaxis but not photoshock. As mentioned above, this is intriguing because three of the proteins in the FAP47 complex have PAS domains, which have been implicated in transducing environmental signals in many organisms. Estimated mass of CA projections versus predicted mass of CA proteins The sum of the masses of the individual CA projections in Chlamydomonas as estimated from cryo-ET volume measurements is ∼14.6 MD (Carbajal-González et al., 2013). The total mass of the 22 previously known CA proteins is ∼3.5 MD, and the total mass of the candidate CA proteins identified here is 4.2 MD. Therefore, if all 44 of the latter are indeed CA proteins, the total mass of the identified CA proteins would be 7.7 MD, leaving 6.9 MD still unaccounted for. This “missing” mass would of course be less if some of the proteins are present in multiple copies within a projection. Nevertheless, if there is no systematic error in the mass estimates from cryo-ET, there likely are still many more CA proteins to be identified. Some of the proteins that we found to be greatly reduced in pf18 axonemes but which did not meet our criteria for inclusion as candidate CA proteins may very well be components of the CA. However, it can be very difficult to precisely estimate masses of large structures by cryo-ET volume measurements, and estimates for the numbers of several subunits in each Chlamydomonas N-DRC and radial spoke based on cryo-ET are twice the estimates based on quantitative immunofluorescence and immunoblotting of tagged proteins, suggesting that the volumes/masses were overestimated twofold by cryo-ET (Oda, 2017). If that is also the case for the CA projections, then it is possible that our study has identified nearly all proteins of these structures. Concluding remarks Our results reveal that the CA is far more compositionally complex than previously recognized and provide a greatly expanded knowledge base for future studies to understand the architecture of the CA and how it performs its essential functions in motile cilia and flagella. Our results also have significant clinical relevance. In humans, defects in motile cilia cause primary ciliary dyskinesia (PCD; Horani and Ferkol, 2018). Most known PCD genes affect the nodal cilia, which initiate left–right asymmetry in the early embryo. Errors in this process lead to laterality defects such as situs inversus and heterotaxy in about half of PCD patients; these clinical manifestations, when present, are very helpful for screening potential PCD patients (Shapiro et al., 2018). However, nodal cilia lack a CA, so defects in CA proteins do not cause laterality defects, thus eliminating a key indicator of PCD. Moreover, disruption of individual CA projections is not readily detected by TEM, thus potentially leading to false negatives using that historically important diagnostic tool (Shapiro et al., 2018). As a result, patients who have PCD due to mutations in genes encoding CA components currently are very difficult to diagnose (Edelbusch et al., 2017). Our identification of a large number of novel candidate or confirmed CA proteins that are conserved in humans provides many new potential PCD genes that will be very useful for enhanced genetic screening of patients referred for clinical suspicion of PCD. Materials and methods Strains and culture conditions All Chlamydomonas strains used in the work are listed in Table S7. Cells were grown in modified M medium I (Witman, 1986) at 23°C with aeration of 5% CO2 and a light/dark cycle of 14/10 h. Flagella preparation and fractionation Flagella were isolated as described previously (Witman, 1986) and demembranated by resuspension in HMDEK (30 mM Hepes, 5 mM MgSO4, 1 mM DTT, 0.5 mM EGTA, and 25 mM KCl) containing 0.5% NP-40 (Calbiochem) for 10 min at 4°C. The resulting axonemes were collected by centrifugation (30,000 g for 20 min) at 4°C. Axonemes prepared in this way were used for the MS analysis of axonemes of WT vs. pf18, fap47-1, fap76-1, fap99-1, fap196-1, fap246-1, and dyp30-1. For assignment of CA proteins to the C1 or C2 microtubule, WT axonemes were further extracted with 0.6 M NaCl in HMDEK for 30 min on ice to destabilize the C2 microtubule (Mitchell and Sale, 1999). The sample was then separated into soluble and insoluble fractions by centrifugation, the pellet was resuspended in a volume of HMDEK equal to the volume of the supernatant, and the two fractions were analyzed by MS. To compare pf16 and WT axonemes, isolated flagella were demembranated in HMDEN (10 mM Hepes, 5 mM MgSO4, 1 mM DTT, 0.5 mM EDTA, 30 mM NaCl, and 0.5% PEG 20,000) with 0.5% NP-40 (Mitchell and Sale, 1999). The axonemes were then collected by centrifugation as above and analyzed by MS. Genetic crossing Mutant and WT strains were first spread onto Tris-acetate-phosphate plates (Harris, 1989) and grown for 7–10 d. The cells were then transferred from the plates into a sterile 250-ml flask containing 10 ml of M-N medium (M medium without NH4NO3 and with KH2PO4 added only to adjust pH to 7.6) and shaken in full light overnight. On the next day, cells were harvested by centrifugation and resuspended in 10 ml of 1/5 M-N medium for 4 h. Aliquots (2–3 ml) of each cell type were then mixed and incubated under light for 2–3 h to allow mating to proceed, after which the mixed cells were collected by centrifugation and resuspended in 1–2 ml of the supernatant. 200–300 µl of the suspension were spread on 4% M medium plates. The plates were dried in the light, sealed with parafilm, left for 3 d in the light, wrapped in foil, left for 10–12 d at room temperature, and then placed at −20°C for 48 h, after which time the cells were allowed to grow on the plates under normal culture conditions. Single colonies were picked for further analysis. Immunofluorescence and EM In experiments to determine the locations of select candidate CA proteins, the CA complex was extruded spontaneously from reactivated axonemes as described by Mitchell and Nakatsugawa (2004)with the following modifications: 2.5 µl of 10 mM ATP was added to 18 µl of freshly prepared axonemes to initiate reactivation, and the reactivated axonemes were then incubated at room temperature for 30 min. Then 1.4 µl of 37% formaldehyde (final concentration, 2–3%) was added to fix the samples. For immunofluorescence microscopy, slides were first treated with 0.1% poly-l-lysine (Sigma-Aldrich) for 5–8 min and rinsed with deionized H2O, and the excess water was wicked off. An aliquot of intact axonemes or axonemes with extruded CAs was then placed on the slide, and the axonemes allowed to adhere for 10 min, after which excess sample was wicked off. The slide was then placed in −20°C methanol for 8–10 min, the excess methanol was wicked off, and the slide was allowed to dry for 20 min at room temperature. The slide was then wetted with 1× PBS for 1 h and flooded with blocking buffer (5% [wt/vol] BSA, 1% [vol/vol] fish skin gelatin, and 10% [vol/vol] goat serum, in 1× PBS) for at least 1 h. The slides were then treated overnight at 4°C with blocking buffer containing the diluted primary antibody (rat anti-HA antibody [Roche; clone 3F10, 1:150]; mouse anti-acetylated tubulin [Sigma-Aldrich; clone 6-11B-1, 1:1,000]). The next day, the slides were washed four times over 1 h with blocking buffer, and then treated for 1 h with blocking buffer containing the secondary antibodies (goat anti-rat IgG [H+L] cross-adsorbed secondary antibody, Alexa Fluor 488 [Invitrogen; A11006, 1:200]; and F(ab′)2-goat anti-mouse IgG [H+L] cross-adsorbed secondary antibody, Alexa Fluor 568 [Invitrogen; A11019, 1:1,000]). The slides were then given two 15-min washes with blocking buffer and transferred to a large volume of 1× PBS for a final wash. After the slides were dried at room temperature, specimens were mounted with ProLong Gold (Invitrogen). SIM was performed using a DeltaVision OMX system (GE Healthcare) with a 1.42 NA 60× Plan-Apochromat objective (Olympus) and immersion oil with a refraction index of 1.512. SIM images were reconstructed with softWoRx 6.1.3 (GE Healthcare). Capture times and adjustments were the same for images with the same antibody. Image brightness and contrast were adjusted using ImageJ (National Institutes of Health). Figures for publication were assembled using Illustrator 8.0 (Adobe). For TEM, freshly prepared axonemes were fixed in glutaraldehyde and processed as described previously (Hoops and Witman, 1983) with modification. Briefly, the axoneme pellet was fixed at room temperature with 2% glutaraldehyde in 100 mM sodium cacodylate buffer, pH 7.2, for 15 min. The pellet was then teased off from the bottom of the tube, and fixation continued for 1 h. After three 15-min washes with 100 mM sodium cacodylate buffer, the sample was fixed with 1% osmium tetroxide in 75 mM cacodylate buffer at room temperature for 1 h. Next, the pellet was washed twice in 50 mM sodium cacodylate buffer, washed three times in water, and stained en bloc overnight in 1% uranyl acetate at 4°C. On the following day, the sample was given three 10-min washes with water at 4°C, then dehydrated through a graded ethanol series (10%, 30%, 50%, 15 min each; 70%, 85%, 95%, 20 min each; 100%, 30 min), embedded in SPI-Pon 812 Epoxy Resin (SPI), and sectioned at 60–80 nm. Sections were stained with uranyl acetate and lead citrate and viewed with a Phillips CM10 electron microscope. Gene cloning and vector construction The hygromycin cassette (Berthold et al., 2002) was first cloned into the HindIII site in pNEB193 (New England Biolabs) to make pBH, and then the vectors were constructed as described immediately below. All the primers used in vector construction are listed in Table S8. Bacterial artificial chromosome (BAC) clones were obtained from Clemson University Genomics Institute (the clones are now available from the Chlamydomonas Resource Center, https://www.chlamycollection.org/). For the FAP47 rescue construct, BAC 18J10 containing FAP47 was digested with BamHI and NheI to obtain 6.9-kb (BamH1-NheI) and 10.7-kb (NheI-NheI) fragments. The 6.9-kb fragment was cloned into pNEB193, yielding pBC1. The 10.7-kb fragment was ligated into pBC1 digested with NheI to yield pBC2. pBC2 was then digested with NdeI and SbfI and ligated into pBH digested with the same enzymes to yield pBC3. To introduce a triple-HA tag, pBC3 was first digested with pfIFI, and the resulting 3.7- and 18.3-kb fragments were purified. The 3.7-kb fragment was ligated into pNEB193 to make pBC4. Sequence encoding the triple-HA tag was amplified with primers F9/R9 from plasmid p3xHA (Silflow et al., 2001) and inserted into pBC4 at the AscI site to yield pBC5. Finally, the 3.7-kb fragment with sequence encoding the triple-HA tag was cut from pBC5 with pfIFI and ligated with the 18.3-kb fragment to yield pBC6. For the FAP76 rescue construct, BAC 01H08 was digested with AscI and XbalI to obtain an 8.5-kb fragment containing part of the FAP76 gene; it was cloned directly into pBH, yielding pBC7. Separately, BAC 01H08 was digested with XbalI to yield a 9.6-kb fragment that contained the rest of the FAP76 gene. The 9.6-kb fragment was inserted into pBC7 at the XbalI site, yielding pBC8. To introduce a triple-HA tag, sequence encoding the triple-HA tag was amplified from p3xHA with primers F10/R10 and then cloned into pBC8 at the MauB1 site, yielding pBC9. For the FAP99 rescue construct, part of FAP99 was amplified from BAC 32A23 with primers F11/R11 and ligated into pBH between the NdeI and AgeI sites to yield pBC10. A 7.7-kb fragment containing the rest of FAP99 was cut from BAC 32A23 with AgeI and ligated into pBC10, yielding pBC11. Sequence encoding the triple-HA tag was amplified from the plasmid p3xHA by means of primers F12/R12 and then inserted into pBC11 at the MauB1 site, yielding pBC12. For the FAP196 rescue construct, BAC 03D04 containing FAP196 was digested with AscI and SbfI to obtain 5.6-kb (AscI-SbfI) and 9.1-kb (SbfI-SbfI) fragments. The 5.6-kb fragment was cloned into pBH between the AscI and the SbfI sites, yielding pBC13. The 9.1-kb fragment was ligated into pBC13 at the SbfI site, yielding pBC14. To introduce a triple-HA tag, pBC14 was first digested with MauB1 and BglII to yield 2.8- and 16.3-kb fragments. The 2.8-kb fragment was ligated into pNEB193 to make pBC15. Sequence encoding the triple-HA tag was amplified from the plasmid p3xHA by means of primers F13/R13 and inserted into pBC15 at the PfoI site, yielding pBC16. Then, the 2.8-kb fragment tagged with sequence encoding 3HA was cut from pBC16 with MauB1 and BglII and ligated with the 16.3-kb fragment, yielding pBC17. For the FAP246 rescue construct, a 10.2-kb fragment containing part of the FAP246 gene was cut from BAC 38L21 with AgeI and NheI and ligated into pBH, yielding pBC18. The rest of the FAP246 gene was then amplified from BAC 38L21 with primers F14/R14 and ligated into pBC18 between its PacI and AgeI sites to yield pBC19. To insert a sequence encoding the triple-HA tag into the gene, part of FAP246 was amplified from BAC clone 38L21 using primers F15/R15; the resulting fragment was ligated into pNEB193 at the BamHI and HindIII sites to yield pBC20. Sequence encoding the triple-HA tag was amplified from the plasmid p3xHA by means of primers F16/R16 and cloned into pBC20 at the SacII site, yielding pBC21. Then the partial FAP246 gene with 3xHA-encoding sequence was cut from pBC21 with PasI and SfiI and used to replace its untagged counterpart in pBC19 to yield pBC22. For the DPY30 rescue construct, DPY30 was amplified from BAC 23P20 with primers F17/R17 and cloned into pBH with XmaI and PacI, yielding pBC23. To introduce a triple-HA tag, sequence encoding the triple-HA tag was amplified from the plasmid p3xHA by means of primers F18/R18 and then cloned into pBC23 with Asc1, yielding pBC24. All constructs were verified by sequencing. DNA transformation was done by the glass bead method (Kindle, 1990). After transformation, cells were grown on Tris-acetate-phosphate agar supplemented with 10 µg/ml hygromycin (Sigma-Aldrich). Colonies derived from single cells were picked and screened for incorporation of the rescuing construct by PCR with primers listed in Table S8. Incorporation of the construct was confirmed by Western blotting before phenotypic analysis. Immunoprecipitation and identification of interacting partners For immunoprecipitation with the anti-HA peptide antibody, axonemes isolated as above were resuspended in high-salt buffer (30 mM Hepes, pH 7.4, 5 mM MgSO4, 1 mM DTT, 0.5 mM EGTA, and 0.6 M KCl) and incubated on ice for 30 min to solubilize most C2 microtubules and some C1 microtubules (Pazour et al., 2005). The suspension was then centrifuged (30,000 g for 20 min at 4°C), and the supernatant was collected as the KCl extract. 50 µl of anti-HA Affinity Matrix (Roche) and Mini Protease Inhibitor Cocktail (Roche; 11836170001) was mixed with the KCl extract (77 µg of total protein), and the mixture was incubated with rotation overnight at 4°C. After four washes with high-salt buffer, bound proteins were eluted for 20 min at room temperature with 80 µl of 1 mg/ml HA peptide in 167 mM NaCl, 10 mM Tris-HCl, pH 7.5, 0.05% Tween 20, and 0.25× Mini Protease Inhibitor Cocktail. Western blotting was used to confirm that the tagged protein was eluted from the matrix. Immunoprecipitates were then analyzed by MS, and protein abundance was estimated with both IBAQ or Top3 quantification methods. Experiments for DPY30 and FAP246 were repeated twice to generate two biological replicates. Only previously known or newly identified candidate CA proteins were considered as potential interacting partners in each immunoprecipitation. In general, to be designated a specifically interacting partner, a protein had to have an abundance in the experimental immunoprecipitate ≥1/10 that of the bait protein and had to be enriched in the experimental immunoprecipitate ≥10-fold relative to the same protein in the WT control immunoprecipitate, based on either IBAQ or Top3 scores. For those cases in which previously known subunits of a CA projection were identified in the experimental immunoprecipitate, a protein was considered specifically interacting (a) if its abundance in the experimental immunoprecipitate was greater than that of the least abundant subunit from the projection, and (b) if it was enriched in the experimental immunoprecipitate at least as much as the least enriched of the previously known subunits, or enriched ≥10-fold, whichever was larger. In the cases of DPY30 and FAP246, if a protein from either replicate met these criteria, it was designated an interacting partner. Western blotting Protein samples were subjected to SDS-PAGE using 4–20% precast gels (Bio-Rad). Western blots were prepared by transferring proteins from gels to polyvinylidene difluoride membrane (Immobilon P; Millipore). The Western blots were probed with rat monoclonal anti-HA antibody (Roche; clone 3F10) at 1:1,000 dilution or rabbit polyclonal anti-NAP antibody (Kato-Minoura et al., 1998; kindly provided by Dr. S. King, University of Connecticut Health Center, Farmington, CT) at 1:2,000 dilution. For detection of loading controls, blots were probed with mouse monoclonal antibody to outer arm dynein intermediate chain IC2 (King et al., 1985) at 1:250 dilution or mouse monoclonal antibody to α-tubulin (Sigma-Aldrich; clone B-5-1-2) at 1:1,000 dilution. Motility analysis, photobehavioral assays, and flagellar length measurement All observations and recordings were performed at room temperature. To analyze swimming speed, 50 µl of cell culture were transferred to a plastic chamber (0.127-mm-deep Fisherbrand UriSystem DeciSlide; Thermo Fisher Scientific). Cells were imaged with nonactinic (deep-red) light using a Zeiss inverted microscope equipped with a 16×/0.35 NA Plan objective and a Kopp #2408 red long-pass filter (Kopp Glass). Videos were recorded at 30 images/s with a digital charge-coupled camera (UP-610; UNIQ Vision) and Video Savant 3.0 software (IO Industries). Swimming speeds were determined using ImageJ software as previously described (Awata et al., 2015). To assess phototactic behavior, cells in 0.8-mm-deep chambers constructed of two coverslips were illuminated from one side with a stimulus beam of light as previously described (Moss et al., 1995) and recorded using the above microscope, camera, and software. Photoshock was visually assayed by viewing swimming cells with red light using a Zeiss Universal microscope equipped with a NEOFLUAR 16×/0.40 Ph2 objective, and then quickly removing the red filter. For video microscopy of photoshock, swimming cells illuminated by nonactinic red light, in the same setup as for assay of phototaxis, were suddenly exposed to a stimulus beam of white light coming from the side. To record swimming tracks, 1-s exposures were acquired using white light and phase-contrast optics on a Zeiss Axioskop 2 Plus microscope equipped with a 20× Plan-NEOFLUAR 0.5 NA Ph2 objective, a digital charge-coupled camera (AxioCam MRm), and AxioVision 3.1 software (Zeiss). For flagellar length measurement, cells were fixed in 2% glutaraldehyde and then imaged using phase-contrast optics and the Axioskop 2 Plus microscope equipped with a 40× Plan-NEOFLUAR 0.75 NA Ph2 objective. Images were recorded using AxioVision software, and flagellar lengths were then determined with ImageJ. MS For comparison of WT and mutant axonemes, axonemes were isolated as above, and protein concentration was determined using a Bradford Protein Assay Kit (Thermo Fisher Scientific; 23200). Equivalent amounts of WT and mutant axonemal protein were fully resolved by 1D 4–20% gradient SDS-PAGE, and each lane was then cut into five segments, such that each segment had an approximately equal amount of protein. Each segment was then processed separately. For analysis of soluble and insoluble fractions generated by extraction of WT axonemes with 0.6 M NaCl, the samples were treated similarly except that the gel lanes were loaded with equal volumes of the samples, and each lane was cut into four segments. For analysis of immunoprecipitates, proteins in control and experimental samples were electrophoresed a short distance into an SDS-polyacrylamide gel and the segments containing the unresolved proteins then excised. In-gel digestion and liquid chromatography–MS/MS analysis were performed in the Mass Spectrometry Facility of University of Massachusetts Medical School as described before (Kubo et al., 2018) with modification. Briefly, the gel segments were treated with trypsin and then subjected to reduction with DTT and alkylation with iodoacetamide. Peptides eluted from the gel were lyophilized and resuspended in 25 µl of 5% acetonitrile (0.1% [vol/vol] TFA). 3 µl of each sample, in technical singlicate, were loaded by a Waters NanoAcquity UPLC in 5% acetonitrile (0.1% formic acid) at 4.0 µl/min for 4.0 min onto a 100-µm-internal-diameter (ID) fused-silica precolumn packed with 2 cm of 5 µm (200 Å) Magic C18AQ (Bruker-Michrom). Peptides were eluted at 300 nl/min from a 75-µm-ID gravity-pulled analytical column packed with 25 cm of 3-µm (100 Å) Magic C18AQ particles using a linear gradient from 5 to 35% of mobile phase B (acetonitrile + 0.1% formic acid) in mobile phase A (water + 0.1% formic acid) over 60 min. Ions were introduced by positive electrospray ionization via liquid junction at 1.5 kV into a Thermo Fisher Scientific Q Exactive hybrid mass spectrometer. Mass spectra were acquired over m/z 300–1,750 at 70,000 resolution (m/z 200) with an AGC target of 1e6, and data-dependent acquisition selected the top 10 most abundant precursor ions for tandem MS by HCD fragmentation using an isolation width of 1.6 D, maximum fill time of 110 ms, and AGC target of 1e5. Peptides were fragmented by a normalized collisional energy of 27, and fragment spectra acquired at a resolution of 17,500 (m/z 200). Data analysis Raw data files were peak processed with Proteome Discoverer (version 2.1; Thermo Fisher Scientific) followed by identification using Mascot Server (version 2.5 or 2.6; Matrix Science) against the National Center for Biotechnology Information (NCBI) protein databases. Search parameters included Trypsin/P specificity, up to two missed cleavages, a fixed modification of carbamidomethyl cysteine, and variable modifications of oxidized methionine, pyroglutamic acid for Q, and N-terminal acetylation. Assignments were made using a 10-ppm mass tolerance for the precursor and 0.05-D mass tolerance for the fragments. All nonfiltered search results were loaded into Scaffold (version 4.8; Proteome Software), with gel fractions loaded as technical replicates and further processed by Scaffold using the Trans-Proteomic Pipeline (Institute for Systems Biology) with all peptides filtered to a 1% false discovery rate. Only proteins identified by two or more peptides, with a protein threshold of 90% probability and a peptide threshold of 90% probability, were considered “detected” and were included in the subsequent analysis. IBAQ (Schwanhäusser et al., 2011) and Top3 precursor quantification methods (Silva et al., 2006) were used to estimate the abundance of each protein. Domain predictions ASH domains were confirmed using the NCBI Conserved Domain tool (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi; threshold, 100). PAS domains were confirmed by searching proteins on the SMART website (http://smart.embl-heidelberg.de/). Online supplemental material Fig. S1 shows the flagellar length and ultrastructure of axonemes of WT and pf mutants. Fig. S2 shows the Venn diagram of known and candidate CA proteins in replicates 1 and 2. Fig. S3 shows that NAP is present in WT axonemes and absent from pf18 axonemes, and that HA-tagged proteins are expressed in the axonemes of the rescued strains as confirmed by Western blots. Fig. S4 shows confirmation of mutant insertion sites by PCR. Table S1 lists all the proteins quantified in two independent biological replicates comparing WT and pf18 axonemes. Table S2 contains the MS data for the proteins in Fig. 2. Table S3 contains the MS data for the candidate CA proteins. Table S4 contains the MS data for Fig. 3. Table S5 contains the MS data for the immunoprecipitation experiments of Table 3. Table S6 contains the MS data for the comparative proteomics experiments of Table 4. Tables S7 and S8 list the strains and primers used in this study, respectively. Supplementary Material Supplemental Materials (PDF) Tables S1-S6 (.zip) Acknowledgments We are grateful to Dr. S. King for his generous gift of anti-NAP antibody. From the University of Massachusetts Medical School, we thank Dr. G. Pazour and Mr. N. McNeill for help with protein analysis; Ms. D. Cochran for technical assistance; Drs. G. Hendricks and L. Strittmatter and Mr. K. Reddig of the Electron Microscopy Facility for expert assistance with EM; Drs. J. Leszyk and M. Dubuke at the Mass Spectrometry Facility for expert assistance with MS; and Dr. C. Baer of the Sanderson Center for Optical Experimentation (SCOPE) Imaging Facility for expert assistance with SIM. This work was supported by National Institutes of Health grants R37 GM030626 and R35 GM122574 to G.B. Witman and by the Robert W. Booth Endowment at the University of Massachusetts Medical School to G.B. Witman. The authors declare no competing financial interests. Author contributions: L. Zhao, Y. Hou, and G.B. Witman conceived the project and designed the experiments. L. Zhao performed most of the experiments and data analysis. Y. Hou performed the MS analysis of insertional mutants. T. Picariello and B. Craige contributed advice and reagents for the genetic crossing, vector construction, and immunofluorescence microscopy. L. Zhao, Y. Hou, and G.B. Witman wrote the manuscript. ==== Refs Adams, G.M., B. Huang, G. Piperno, and D.J. Luck. 1981. Central-pair microtubular complex of Chlamydomonas flagella: polypeptide composition as revealed by analysis of mutants. J. Cell Biol. 91 :69–76. 10.1083/jcb.91.1.69 7028763 Ahmed, N.T., C. Gao, B.F. Lucker, D.G. Cole, and D.R. Mitchell. 2008. ODA16 aids axonemal outer row dynein assembly through an interaction with the intraflagellar transport machinery. J. Cell Biol. 183 :313–322. 10.1083/jcb.200802025 18852297 Awata, J., K. Song, J. Lin, S.M. King, M.J. Sanderson, D. Nicastro, and G.B. Witman. 2015. DRC3 connects the N-DRC to dynein g to regulate flagellar waveform. Mol. Biol. Cell. 26 :2788–2800. 10.1091/mbc.E15-01-0018 26063732 Bernstein, M., P.L. Beech, S.G. Katz, and J.L. Rosenbaum. 1994. A new kinesin-like protein (Klp1) localized to a single microtubule of the Chlamydomonas flagellum. J. Cell Biol. 125 :1313–1326. 10.1083/jcb.125.6.1313 8207060 Berthold, P., R. Schmitt, and W. Mages. 2002. An engineered Streptomyces hygroscopicus aph 7” gene mediates dominant resistance against hygromycin B in Chlamydomonas reinhardtii. Protist. 153 :401–412. 10.1078/14344610260450136 12627869 Bower, R., D. Tritschler, K. Vanderwaal, C.A. Perrone, J. Mueller, L. Fox, W.S. Sale, and M.E. Porter. 2013. The N-DRC forms a conserved biochemical complex that maintains outer doublet alignment and limits microtubule sliding in motile axonemes. Mol. Biol. Cell. 24 :1134–1152. 10.1091/mbc.e12-11-0801 23427265 Brown, J.M., C.G. Dipetrillo, E.F. Smith, and G.B. Witman. 2012. A FAP46 mutant provides new insights into the function and assembly of the C1d complex of the ciliary central apparatus. J. Cell Sci. 125 :3904–3913. 10.1242/jcs.107151 22573824 Carbajal-González, B.I., T. Heuser, X. Fu, J. Lin, B.W. Smith, D.R. Mitchell, and D. Nicastro. 2013. Conserved structural motifs in the central pair complex of eukaryotic flagella. Cytoskeleton (Hoboken). 70 :101–120. 10.1002/cm.21094 23281266 DiPetrillo, C.G., and E.F. Smith. 2010. Pcdp1 is a central apparatus protein that binds Ca(2+)-calmodulin and regulates ciliary motility. J. Cell Biol. 189 :601–612. 10.1083/jcb.200912009 20421426 DiPetrillo, C.G., and E.F. Smith. 2011. The Pcdp1 complex coordinates the activity of dynein isoforms to produce wild-type ciliary motility. Mol. Biol. Cell. 22 :4527–4538. 10.1091/mbc.e11-08-0739 21998195 Dutcher, S.K., B. Huang, and D.J. Luck. 1984. Genetic dissection of the central pair microtubules of the flagella of Chlamydomonas reinhardtii. J. Cell Biol. 98 :229–236. 10.1083/jcb.98.1.229 6707088 Dymek, E.E., and E.F. Smith. 2012. PF19 encodes the p60 catalytic subunit of katanin and is required for assembly of the flagellar central apparatus in Chlamydomonas. J. Cell Sci. 125 :3357–3366. 10.1242/jcs.096941 22467860 Dymek, E.E., P.A. Lefebvre, and E.F. Smith. 2004. PF15p is the chlamydomonas homologue of the Katanin p80 subunit and is required for assembly of flagellar central microtubules. Eukaryot. Cell. 3 :870–879. 10.1128/EC.3.4.870-879.2004 15302820 Edelbusch, C., S. Cindrić, G.W. Dougherty, N.T. Loges, H. Olbrich, J. Rivlin, J. Wallmeier, P. Pennekamp, I. Amirav, and H. Omran. 2017. Mutation of serine/threonine protein kinase 36 (STK36) causes primary ciliary dyskinesia with a central pair defect. Hum. Mutat. 38 :964–969. 10.1002/humu.23261 28543983 Esparza, J.M., E. O’Toole, L. Li, T.H. Giddings Jr., B. Kozak, A.J. Albee, and S.K. Dutcher. 2013. Katanin localization requires triplet microtubules in Chlamydomonas reinhardtii. PLoS One. 8 :e53940. 10.1371/journal.pone.0053940 23320108 Fox, L.A., K.E. Sawin, and W.S. Sale. 1994. Kinesin-related proteins in eukaryotic flagella. J. Cell Sci. 107 :1545–1550.7962196 Gaillard, A.R., D.R. Diener, J.L. Rosenbaum, and W.S. Sale. 2001. Flagellar radial spoke protein 3 is an A-kinase anchoring protein (AKAP). J. Cell Biol. 153 :443–448. 10.1083/jcb.153.2.443 11309423 Harris, E.H. 1989. The Chlamydomonas Sourcebook. Academic Press, San Diego. 780 pp. Hirono, M., S. Uryu, A. Ohara, T. Kato-Minoura, and R. Kamiya. 2003. Expression of conventional and unconventional actins in Chlamydomonas reinhardtii upon deflagellation and sexual adhesion. Eukaryot. Cell. 2 :486–493. 10.1128/EC.2.3.486-493.2003 12796293 Hoops, H.J., and G.B. Witman. 1983. Outer doublet heterogeneity reveals structural polarity related to beat direction in Chlamydomonas flagella. J. Cell Biol. 97 :902–908. 10.1083/jcb.97.3.902 6224802 Horani, A., and T.W. Ferkol. 2018. Advances in the Genetics of Primary Ciliary Dyskinesia: Clinical Implications. Chest. 154 :645–652. 10.1016/j.chest.2018.05.007 29800551 Hou, Y., and G.B. Witman. 2017. The N-terminus of IFT46 mediates intraflagellar transport of outer arm dynein and its cargo-adaptor ODA16. Mol. Biol. Cell. 28 :2420–2433. 10.1091/mbc.e17-03-0172 28701346 Hunter, E.L., K. Lechtreck, G. Fu, J. Hwang, H. Lin, A. Gokhale, L.M. Alford, B. Lewis, R. Yamamoto, R. Kamiya, . 2018. The IDA3 adapter, required for intraflagellar transport of I1 dynein, is regulated by ciliary length. Mol. Biol. Cell. 29 :886–896. 10.1091/mbc.E17-12-0729 29467251 Kato-Minoura, T., M. Hirono, and R. Kamiya. 1997. Chlamydomonas inner-arm dynein mutant, ida5, has a mutation in an actin-encoding gene. J. Cell Biol. 137 :649–656. 10.1083/jcb.137.3.649 9151671 Kato-Minoura, T., S. Uryu, M. Hirono, and R. Kamiya. 1998. Highly divergent actin expressed in a Chlamydomonas mutant lacking the conventional actin gene. Biochem. Biophys. Res. Commun. 251 :71–76. 10.1006/bbrc.1998.9373 9790909 Kindle, K.L. 1990. High-frequency nuclear transformation of Chlamydomonas reinhardtii. Proc. Natl. Acad. Sci. USA. 87 :1228–1232. 10.1073/pnas.87.3.1228 2105499 King, S.M. 2018. Dyneins: dynein mechanics, dysfunction, and disease. Elsevier, Oxford, UK. 530 pp. King, S.M., T. Otter, and G.B. Witman. 1985. Characterization of monoclonal antibodies against Chlamydomonas flagellar dyneins by high-resolution protein blotting. Proc. Natl. Acad. Sci. USA. 82 :4717–4721. 10.1073/pnas.82.14.4717 3161075 Kubo, T., Y. Hou, D.A. Cochran, G.B. Witman, and T. Oda. 2018. A microtubule-dynein tethering complex regulates the axonemal inner dynein f (I1). Mol. Biol. Cell. 29 :1060–1074. 10.1091/mbc.E17-11-0689 29540525 Lechtreck, K.F., and G.B. Witman. 2007. Chlamydomonas reinhardtii hydin is a central pair protein required for flagellar motility. J. Cell Biol. 176 :473–482. 10.1083/jcb.200611115 17296796 Lechtreck, K.F., P. Delmotte, M.L. Robinson, M.J. Sanderson, and G.B. Witman. 2008. Mutations in Hydin impair ciliary motility in mice. J. Cell Biol. 180 :633–643. 10.1083/jcb.200710162 18250199 Lechtreck, K.F., E.C. Johnson, T. Sakai, D. Cochran, B.A. Ballif, J. Rush, G.J. Pazour, M. Ikebe, and G.B. Witman. 2009. The Chlamydomonas reinhardtii BBSome is an IFT cargo required for export of specific signaling proteins from flagella. J. Cell Biol. 187 :1117–1132. 10.1083/jcb.200909183 20038682 Lechtreck, K.F., T.J. Gould, and G.B. Witman. 2013. Flagellar central pair assembly in Chlamydomonas reinhardtii. Cilia. 2 :15. 10.1186/2046-2530-2-15 24283352 Li, X., R. Zhang, W. Patena, S.S. Gang, S.R. Blum, N. Ivanova, R. Yue, J.M. Robertson, P.A. Lefebvre, S.T. Fitz-Gibbon, . 2016. An Indexed, Mapped Mutant Library Enables Reverse Genetics Studies of Biological Processes in Chlamydomonas reinhardtii. Plant Cell. 28 :367–387. 10.1105/tpc.15.00465 26764374 Lin, J., D. Tritschler, K. Song, C.F. Barber, J.S. Cobb, M.E. Porter, and D. Nicastro. 2011. Building blocks of the nexin-dynein regulatory complex in Chlamydomonas flagella. J. Biol. Chem. 286 :29175–29191. 10.1074/jbc.M111.241760 21700706 Loreng, T.D., and E.F. Smith. 2017. The Central Apparatus of Cilia and Eukaryotic Flagella. Cold Spring Harb. Perspect. Biol. 9 :a028118. 10.1101/cshperspect.a028118 27770014 McKenzie, C.W., B. Craige, T.V. Kroeger, R. Finn, T.A. Wyatt, J.H. Sisson, J.A. Pavlik, L. Strittmatter, G.M. Hendricks, G.B. Witman, and L. Lee. 2015. CFAP54 is required for proper ciliary motility and assembly of the central pair apparatus in mice. Mol. Biol. Cell. 26 :3140–3149. 10.1091/mbc.e15-02-0121 26224312 Mitchell, D.R., and M. Nakatsugawa. 2004. Bend propagation drives central pair rotation in Chlamydomonas reinhardtii flagella. J. Cell Biol. 166 :709–715. 10.1083/jcb.200406148 15337779 Mitchell, D.R., and W.S. Sale. 1999. Characterization of a Chlamydomonas insertional mutant that disrupts flagellar central pair microtubule-associated structures. J. Cell Biol. 144 :293–304. 10.1083/jcb.144.2.293 9922455 Mitchell, B.F., L.B. Pedersen, M. Feely, J.L. Rosenbaum, and D.R. Mitchell. 2005. ATP production in Chlamydomonas reinhardtii flagella by glycolytic enzymes. Mol. Biol. Cell. 16 :4509–4518. 10.1091/mbc.e05-04-0347 16030251 Moss, A.G., G.J. Pazour, and G.B. Witman. 1995. Assay of Chlamydomonas phototaxis. Methods Cell Biol. 47 :281–287. 10.1016/S0091-679X(08)60821-3 7476500 Oda, T. 2017. Three-dimensional structural labeling microscopy of cilia and flagella. Microscopy (Oxf.). 66 :234–244. 10.1093/jmicro/dfx018 28541401 Olbrich, H., M. Schmidts, C. Werner, A. Onoufriadis, N.T. Loges, J. Raidt, N.F. Banki, A. Shoemark, T. Burgoyne, S. Al Turki, UK10K Consortium. 2012. Recessive HYDIN mutations cause primary ciliary dyskinesia without randomization of left-right body asymmetry. Am. J. Hum. Genet. 91 :672–684. 10.1016/j.ajhg.2012.08.016 23022101 Pazour, G.J., N. Agrin, J. Leszyk, and G.B. Witman. 2005. Proteomic analysis of a eukaryotic cilium. J. Cell Biol. 170 :103–113. 10.1083/jcb.200504008 15998802 Piperno, G., and D.J. Luck. 1979. Axonemal adenosine triphosphatases from flagella of Chlamydomonas reinhardtii. Purification of two dyneins. J. Biol. Chem. 254 :3084–3090.155062 Ponting, C.P. 2006. A novel domain suggests a ciliary function for ASPM, a brain size determining gene. Bioinformatics. 22 :1031–1035. 10.1093/bioinformatics/btl022 16443634 Rao, V.G., R.B. Sarafdar, T.S. Chowdhury, P. Sivadas, P. Yang, P.M. Dongre, and J.S. D’Souza. 2016. Myc-binding protein orthologue interacts with AKAP240 in the central pair apparatus of the Chlamydomonas flagella. BMC Cell Biol. 17 :24. 10.1186/s12860-016-0103-y 27287193 Rompolas, P., R.S. Patel-King, and S.M. King. 2012. Association of Lis1 with outer arm dynein is modulated in response to alterations in flagellar motility. Mol. Biol. Cell. 23 :3554–3565. 10.1091/mbc.e12-04-0287 22855525 Rupp, G., E. O’Toole, and M.E. Porter. 2001. The Chlamydomonas PF6 locus encodes a large alanine/proline-rich polypeptide that is required for assembly of a central pair projection and regulates flagellar motility. Mol. Biol. Cell. 12 :739–751. 10.1091/mbc.12.3.739 11251084 Schou, K.B., S.K. Morthorst, S.T. Christensen, and L.B. Pedersen. 2014. Identification of conserved, centrosome-targeting ASH domains in TRAPPII complex subunits and TRAPPC8. Cilia. 3 :6. 10.1186/2046-2530-3-6 25018876 Schwanhäusser, B., D. Busse, N. Li, G. Dittmar, J. Schuchhardt, J. Wolf, W. Chen, and M. Selbach. 2011. Global quantification of mammalian gene expression control. Nature. 473 :337–342. 10.1038/nature10098 21593866 Shamoto, N., K. Narita, T. Kubo, T. Oda, and S. Takeda. 2018. CFAP70 Is a Novel Axoneme-Binding Protein That Localizes at the Base of the Outer Dynein Arm and Regulates Ciliary Motility. Cells. 7 :E124. 10.3390/cells7090124 Shapiro, A.J., S.D. Davis, D. Polineni, M. Manion, M. Rosenfeld, S.D. Dell, M.A. Chilvers, T.W. Ferkol, M.A. Zariwala, S.D. Sagel, American Thoracic Society Assembly on Pediatrics. 2018. Diagnosis of Primary Ciliary Dyskinesia. An Official American Thoracic Society Clinical Practice Guideline. Am. J. Respir. Crit. Care Med. 197 :e24–e39. 10.1164/rccm.201805-0819ST 29905515 Shapiro, J., J. Ingram, and K.A. Johnson. 2005. Characterization of a molecular chaperone present in the eukaryotic flagellum. Eukaryot. Cell. 4 :1591–1594. 10.1128/EC.4.9.1591-1594.2005 16151252 Silflow, C.D., M. LaVoie, L.W. Tam, S. Tousey, M. Sanders, W. Wu, M. Borodovsky, and P.A. Lefebvre. 2001. The Vfl1 Protein in Chlamydomonas localizes in a rotationally asymmetric pattern at the distal ends of the basal bodies. J. Cell Biol. 153 :63–74. 10.1083/jcb.153.1.63 11285274 Silva, J.C., M.V. Gorenstein, G.Z. Li, J.P. Vissers, and S.J. Geromanos. 2006. Absolute quantification of proteins by LCMSE: a virtue of parallel MS acquisition. Mol. Cell. Proteomics. 5 :144–156. 10.1074/mcp.M500230-MCP200 16219938 Smith, E.F., and P.A. Lefebvre. 1996. PF16 encodes a protein with armadillo repeats and localizes to a single microtubule of the central apparatus in Chlamydomonas flagella. J. Cell Biol. 132 :359–370. 10.1083/jcb.132.3.359 8636214 Smith, E.F., and P.A. Lefebvre. 1997. PF20 gene product contains WD repeats and localizes to the intermicrotubule bridges in Chlamydomonas flagella. Mol. Biol. Cell. 8 :455–467. 10.1091/mbc.8.3.455 9188098 Taschner, M., A. Mourão, M. Awasthi, J. Basquin, and E. Lorentzen. 2017. Structural basis of outer dynein arm intraflagellar transport by the transport adaptor protein ODA16 and the intraflagellar transport protein IFT46. J. Biol. Chem. 292 :7462–7473. 10.1074/jbc.M117.780155 28298440 Taylor, B.L., and I.B. Zhulin. 1999. PAS domains: internal sensors of oxygen, redox potential, and light. Microbiol. Mol. Biol. Rev. 63 :479–506.10357859 Vogt, J.H., and J.H. Schippers. 2015. Setting the PAS, the role of circadian PAS domain proteins during environmental adaptation in plants. Front. Plant Sci. 6 :513. 10.3389/fpls.2015.00513 26217364 Wargo, M.J., E.E. Dymek, and E.F. Smith. 2005. Calmodulin and PF6 are components of a complex that localizes to the C1 microtubule of the flagellar central apparatus. J. Cell Sci. 118 :4655–4665. 10.1242/jcs.02585 16188941 Wirschell, M., D. Nicastro, M.E. Porter, and W.S. Sale. 2009. The regulation of axonemal bending. In The Chlamydomonas Sourcebook. Vol. 3: Cell Motility and Behavior. G.B. Witman, editor. Elsevier, New York. 253–282. Witman, G.B. 1986. Isolation of Chlamydomonas flagella and flagellar axonemes. Methods Enzymol. 134 :280–290. 10.1016/0076-6879(86)34096-5 3821567 Witman, G.B., J. Plummer, and G. Sander. 1978. Chlamydomonas flagellar mutants lacking radial spokes and central tubules. Structure, composition, and function of specific axonemal components. J. Cell Biol. 76 :729–747. 10.1083/jcb.76.3.729 632325 Yang, P., L. Fox, R.J. Colbran, and W.S. Sale. 2000. Protein phosphatases PP1 and PP2A are located in distinct positions in the Chlamydomonas flagellar axoneme. J. Cell Sci. 113 :91–102.10591628 Yang, P., D.R. Diener, J.L. Rosenbaum, and W.S. Sale. 2001. Localization of calmodulin and dynein light chain LC8 in flagellar radial spokes. J. Cell Biol. 153 :1315–1326. 10.1083/jcb.153.6.1315 11402073 Yang, P., D.R. Diener, C. Yang, T. Kohno, G.J. Pazour, J.M. Dienes, N.S. Agrin, S.M. King, W.S. Sale, R. Kamiya, . 2006. Radial spoke proteins of Chlamydomonas flagella. J. Cell Sci. 119 :1165–1174. 10.1242/jcs.02811 16507594 Yokoyama, R., E. O’toole, S. Ghosh, and D.R. Mitchell. 2004. Regulation of flagellar dynein activity by a central pair kinesin. Proc. Natl. Acad. Sci. USA. 101 :17398–17403. 10.1073/pnas.0406817101 15572440 Zhang, H., and D.R. Mitchell. 2004. Cpc1, a Chlamydomonas central pair protein with an adenylate kinase domain. J. Cell Sci. 117 :4179–4188. 10.1242/jcs.01297 15292403 Zhang, Z., I. Kostetskii, W. Tang, L. Haig-Ladewig, R. Sapiro, Z. Wei, A.M. Patel, J. Bennett, G.L. Gerton, S.B. Moss, . 2006. Deficiency of SPAG16L causes male infertility associated with impaired sperm motility. Biol. Reprod. 74 :751–759. 10.1095/biolreprod.105.049254 16382026 Zhang, Z., W. Tang, R. Zhou, X. Shen, Z. Wei, A.M. Patel, J.T. Povlishock, J. Bennett, and J.F. Strauss III. 2007. Accelerated mortality from hydrocephalus and pneumonia in mice with a combined deficiency of SPAG6 and SPAG16L reveals a functional interrelationship between the two central apparatus proteins. Cell Motil. Cytoskeleton. 64 :360–376. 10.1002/cm.20189 17323374

PMC006xxxxxx/PMC6605749.txt

==== Front J Exp Med J Exp Med jem jem The Journal of Experimental Medicine 0022-1007 1540-9538 Rockefeller University Press 31123083 20182359 10.1084/jem.20182359 Research Articles Article 314 Tumor suppression of novel anti–PD-1 antibodies mediated through CD28 costimulatory pathway Enhanced anti–PD-1–mediated immune recovery Fenwick Craig 1 Loredo-Varela Juan-Luis 3* Joo Victor 1* http://orcid.org/0000-0003-2090-9572 Pellaton Céline 1 Farina Alex 1 Rajah Navina 1 Esteves-Leuenberger Line 1 http://orcid.org/0000-0003-1036-6148 Decaillon Thibaut 1 Suffiotti Madeleine 1 Noto Alessandra 1 http://orcid.org/0000-0002-0715-6046 Ohmiti Khalid 1 http://orcid.org/0000-0002-3867-0232 Gottardo Raphael 4 Weissenhorn Winfried 3 http://orcid.org/0000-0003-3651-2721 Pantaleo Giuseppe 12 1 Service of Immunology and Allergy, Department of Medicine, Lausanne University Hospital, University of Lausanne, Lausanne, Switzerland 2 Swiss Vaccine Research Institute, Lausanne University Hospital, University of Lausanne, Lausanne, Switzerland 3 University Grenoble Alpes, Commissariat à l'Energie Atomique, Centre National de la Recherche Scientifique, Institut de Biologie Structurale, Grenoble, France 4 Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA Correspondence to Giuseppe Pantaleo: giuseppe.pantaleo@chuv.ch * J.-L. Loredo-Varela and V. Joo contributed equally to this paper. 01 7 2019 23 5 2019 216 7 15251541 19 12 2018 20 3 2019 01 5 2019 © 2019 Fenwick et al. 2019 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). Novel anti–PD-1 antibodies (Abs) not blocking the PD-1–PDL-1 interaction are presented with equivalent antagonistic activity to classical blocking anti–PD-1 Abs and have distinct mechanisms of action that synergize in functional recovery of exhausted CD8 T cells and enhancing tumor suppression in an immunogenic mouse tumor model. Classical antagonistic antibodies (Abs) targeting PD-1, such as pembrolizumab and nivolumab, act through blockade of the PD-1–PDL-1 interaction. Here, we have identified novel antagonistic anti–PD-1 Abs not blocking the PD-1–PDL-1 interaction. The nonblocking Abs recognize epitopes on PD-1 located on the opposing face of the PDL-1 interaction and overlap with a newly identified evolutionarily conserved patch. These nonblocking Abs act predominantly through the CD28 coreceptor. Importantly, a combination of blocking and nonblocking Abs synergize in the functional recovery of antigen-specific exhausted CD8 T cells. Interestingly, nonblocking anti–PD-1 Abs have equivalent antitumor activity compared with blocker Abs in two mouse tumor models, and combination therapy using both classes of Abs enhanced tumor suppression in the mouse immunogenic tumor model. The identification of the novel nonblocker anti–PD-1 Abs and their synergy with classical blocker Abs may be instrumental in potentiating immunotherapy strategies and antitumor activity. Institut Universitaire de France http://doi.org/10.13039/501100004795 Grenoble Instruct-ERIC Center UMS 3518 CNRS-CEA-UJF-EMBL French Infrastructure for Integrated Structural Biology InitiativeANR-10-INSB-05-02 Grenoble Alliance for Integrated Structural Cell Biology ANR-10-LABX-49-01 Grenoble Partnership for Structural Biology European Synchrotron Research Facility European Molecular Biology Laboratory 10.13039/100013060 ==== Body pmcIntroduction The programmed cell death 1 receptor (PD-1) is the master regulator of T cells through the suppression of T cell activation following the binding to its primary ligand, programmed death ligand 1 (PDL-1; Trautmann et al., 2006; Tumeh et al., 2014; Pauken and Wherry, 2015; Wherry and Kurachi, 2015). The proposed mechanism by which PD-1 acts as an immune checkpoint inhibitor includes recruitment of the SHP-2 phosphatase into the vicinity of the TCR complex, resulting in dephosphorylation of membrane proximal signaling proteins, including CD3ζ, ZAP70, and PLC-γ1 (Sheppard et al., 2004; Yokosuka et al., 2012). The PD-1–SHP-2 complex also acts on the CD28 costimulatory receptor and the associated PI3K and AKT signaling pathway needed for optimal T cell activation and survival (Parry et al., 2005; Patsoukis et al., 2012). These dynamics have been observed in fluorescence microscopy imaging studies where PD-1 exists in microclusters on the cell surface and is recruited along with SHP-2 phosphatase into the immunological synapse to suppress phosphorylation events during TCR activation (Chemnitz et al., 2004; Sheppard et al., 2004; Yokosuka et al., 2012; Wherry and Kurachi, 2015). Two recent studies have also indicated that anti–PD-1–mediated tumor-suppressive activity is primarily dependent of the CD28 costimulatory receptor (Hui et al., 2017; Kamphorst et al., 2017). Monoclonal antibodies (Abs) acting through PD-1 blockade represent a major breakthrough in oncology, showing significant clinical success in the treatment of several types of cancer, including advanced melanoma, non–small cell lung cancer, and head and neck squamous cell carcinoma (Baitsch et al., 2011; Mellman et al., 2011; Topalian et al., 2012; Hamid et al., 2013; Rizvi et al., 2015; Sharma and Allison, 2015; Callahan et al., 2016). Despite these successes, only ∼30–40% of patients show a response to anti–PD-1 immunotherapy, and only a fraction of these show a durable clinical response. These limitations highlight the need for a better understanding of the mechanism by which anti–PD-1 Abs act and for the identification of new therapies that synergize to improve the response rate and/or breadth of cancers that can be treated. The objective of the present study was to identify novel antagonist Abs with more potent antitumor activity and/or acting through a mechanism independent of the PD-1–PDL-1 blockade. A panel of Ab clones binding with high affinity to diverse epitopes on PD-1 was generated and profiled for antagonistic activity in recovering antigen (Ag)–specific CD8 T cells from functional exhaustion. A novel class of anti–PD-1 Ab was identified that was not blocking the PD-1–PDL-1 interaction with antagonistic activity comparable to pembrolizumab and nivolumab. Antagonistic activity of the novel anti–PD-1 Abs was determined by evaluating their ability to recover proliferation and/or to potentiate cytokine production of exhausted Ag-specific CD8 T cells. Biochemical and structural studies demonstrated that these Abs bound to the opposite face of the PD-1 protein relative to the PD-1–PDL-1 interaction site. In mechanistic studies, nonblocking anti–PD-1 Abs act predominantly through the CD28 coreceptor that restores signaling through the AKT–NF-κB pathway and leads to T cell proliferation and survival. Consisted with nonblocking anti–PD-1 Abs acting through a distinct mechanism of action, combinations of nonblocking and blocking anti–PD-1 Abs synergize to recover the functional activity of exhausted Ag-specific CD8 T cell in vitro and resulted in significantly enhanced antitumor activity in the MC38 immunogenic in vivo mouse tumor model. Results Characterization of a diverse panel of anti–PD-1 Abs binding different epitopes on PD-1 An immunization campaign with human PD-1 was launched in mice, and over 2,000 hybridoma clones were generated and screened in a Luminex assay for binding to recombinant human PD-1. Forty different Ab families with subnanomolar affinity were selected based on (1) possessing low nanomolar binding affinity for cell-surface PD-1, (2) competitive binding profile with a commercially available anti–PD-1 Ab that acted as a surrogate assay to identify blocking Abs of the PD-1–PDL-1 interaction, and (3) heavy-chain complementarity-determining region (CDR) variable region. Anti–PD-1 Abs were further screening to assess both the ability to block the PD-1–PDL-1 interaction in a Luminex biochemical assay and recover Ag-specific CD8 T cells from functional exhaustion. Therefore, the simultaneous use of a functional assay allowed also for the selection of anti–PD-1 Abs with antagonistic activity independent of PD-1–PDL-1 blockade. The antagonistic, immune-enhancing activity of the novel anti–PD-1 Abs was evaluated in a highly standardized CFSE proliferation assay measuring the recovery of Ag-specific proliferation in blood mononuclear cells from chronically infected viremic HIV patients. HIV-specific CD8 T cells are instrumental to evaluate the ability of anti–PD-1 Abs to recover T cells from functional exhaustion because of the expression of high levels of PD-1 and poor proliferation in response to Ag-specific stimulation (Barber et al., 2006; Day et al., 2006; Trautmann et al., 2006; Zhang et al., 2007). Stimulation of blood mononuclear cells with HIV-derived peptides followed by 6 d in culture led to an increase in CFSE-low CD8 T cells that have undergone proliferation. However, the addition of two classical PDL-1–blocking anti–PD-1 Abs (e.g., pembrolizumab or nivolumab) led to an enhanced level of proliferation relative to the peptide alone control, thus indicating that both anti–PD-1 Abs recover CD8 T cells from exhaustion (Fig. 1, A and B). Validation studies have shown that proliferating CD8 T cells following pembrolizumab treatment were HIV specific, as indicated by pentamer MHC–HIV peptide complex staining. On the basis of the functional activity in this proliferation assay, 10 novel anti–PD-1 Abs showed potency similar to a pembrolizumab, which has been used as a benchmark control (Fig. 1 B). Anti–PD-1 Ab clones of interest had antagonistic activity that was also replicated using blood mononuclear cells from five different HIV-infected donors having CD8 T cells specific to seven different HIV-derived peptides (Fig. 1, A and B; and Figs. S1 and S2). Figure 1. Anti–PD-1 Abs significantly enhance proliferation of Ag-specific exhausted CD8 T cells. (A) Recovery of proliferation in HIV-specific CD8 T cells following stimulation with an HIV-derived peptide. Results from a representative experiment are shown and expressed as the percentage of CFSE-low CD8 T cells. 8–10 replicates were performed for each experimental condition. (B) Cumulative results from multiple CFSE experiments (n = 2–6) are shown for the 10 anti–PD-1 Abs identified with antagonistic properties similar to pembrolizumab. Results have been generated assessing the proliferation of CD8 T cells specific to three HIV-derived peptides (FLGKIWPSYK restricted by A*0201 and RLRPGGKKK or RMRGAHTNDVK restricted by A*0301) in patients B08 and B09. For comparative purposes across multiple assays and the different anti–PD-1 Abs, the level of proliferation in the pembrolizumab-treated samples was set as a 100% reference. Pembrolizumab was used as a positive control, while untreated (Neg) or mouse IgG1 isotype control Ab was used as a negative control in each experiment. Graphs show the mean ± SD. ****, P < 0.0001 for all anti–PD-1 Abs relative to the IgG1 control (unpaired t test with Welch’s correction). We then investigated whether the newly identified antagonistic anti–PD-1 Abs act through PD-1–PDL-1 blockade, as with pembrolizumab and nivolumab. Initial profiling in an Ab competitive binding assay was consistent with two clones binding to a different epitope than the PDL-1 blocking Abs. A Luminex biochemical protein–protein interaction assay was established to test the novel Abs and directly determine PDL-1 binding to a preformed Ab–PD-1 complex. The 135C12 and 136B4 immune-enhancing anti–PD-1 Abs with distinct heavy- and light-chain CDR regions induced only a minor 1.5- to 3-fold reduced affinity of the PDL-1 protein for PD-1 (Fig. 2 A). In contrast, 137F2, which bound competitively to PD-1 with a blocking anti–PD-1 Ab, completely prevented binding of PDL-1 to PD-1–coated beads at all concentrations tested. It is important to underscore that out of the 156 Ab clones with high affinity for PD-1, only 3.2% corresponded to nonblocking Abs with antagonistic activity. The remaining Abs with antagonistic activity (Fig. 1 B) were tested in a PD-1–PDL-1 protein interaction screening assay, and all blocked the PD-1–PDL-1 interaction (Fig. 2 B). Figure 2. Epitope mapping and structural studies of the prioritized anti–PD-1 Ab clones. (A) Ability of anti–PD-1 Abs to block the PD-1–PDL-1 interaction in a Luminex biochemical assay. PD-1–coated beads were incubated in the presence or absence of a competitor anti–PD-1 Ab, and then beads were stained with different concentrations of biotin-labeled PDL-1 protein. Data (n = 2) are mean ± SD. MFI, mean fluorescence intensity. (B) Potency of anti–PD-1 Abs in blocking the PD-1–PDL-1 interaction in a Luminex biochemical assay. PD-1–coated beads were incubated with a fixed concentration of PDL-1 equivalent to the half-maximal inhibitory concentration value for the PD-1–PDL-1 interaction in this assay. The PD-1–PDL-1 complex, bound at 50% in equilibrium, was then treated with increasing concentrations of anti–PD-1 Ab to determine if they were capable of completely disrupting the PD-1–PDL-1 interaction with pembrolizumab used as a positive blocking Ab control. (C) Epitope mapping by site directed mutagenesis of PD-1. Defined epitopes were identified for anti–PD-1 Abs that were either blocking or nonblocking of the PD-1–PDL-1 interaction using HeLa cells transfected with expression vectors encoding PD-1 with substitutions at solvent accessible residues. Amino acid substitutions in PD-1 are indicated in blue lettering above each histogram. Representative data are shown for n = 3 experiments. (D) Ab competitive binding studies for cell-surface PD-1. Jurkat PD-1 cells were incubated with excess of 137F2, 135C12, or 136B4 mouse Abs and then stained with a minimal concentration of the indicated humanized anti–PD-1 Abs (n = 3). (E) hPD-1 and NB01a Fab (humanized version of the mouse 135C12 Ab) complexes were purified by size-exclusion chromatography and crystallized. Crystals diffracted to 2.2 Å resolution, and the structure was solved by molecular replacement. The structure reveals that the binding site of NB01a is adjacent to residues in purple involved in the PD-1 interaction with either PDL-1 or PDL-2. The CC′ loop (residues 70–74) of hPDL-1 is disordered and indicated as a dashed line. Loops connecting β strands BC (57–63), C′D (84–92), and FG (127–133) were also disordered. Strands are named following the canonical designation. Cα superpositioning of the hPD-1 present in the NB01a Fab and hPDL-1 (PDB accession no. 4ZQK) complexes show that NB01a Fab and PDL-1 bind distinct nonoverlapping sites on PD-1. (F) Mapping of variable residues between human PD-1 and monkey, dog, horse, mouse, and rat PD-1 revealed an evolutionarily conserved patch (P1) on the opposite face of PD-1 from the PDL-1 or PDL-2 interaction site (yellow, orange, and light green colored residues). The P1 patch overlaps with the binding epitopes for the 135C12/NB01 (orange residues) and 136B4 (light green residues) antagonistic Abs that are nonblocking of PD-1–PDL-1. Residues N49, N58, N74, and N116 that are predicted N-linked glycosylation sites are show in brown on the PD-1 model to be excluded from the P1 patch, 135C12/NB01, and 136B4 Ab binding epitopes. (G) Cα superpositioning of hPDL-1 coordinates of the NB01a complex with pembrolizumab (PDB accession no. 5GGS) and the nivolumab (PDB accession no. 5GGR) confirms that NB01a Fab binding to PD-1 does not interfere with the binding of either pembrolizumab or nivolumab anti–PD-1 Abs. hPDL-1 is shown as a ribbon diagram in E and F, with the hPDL-1 binding surface (PDB accession no. 4ZQK) colored in purple in G. The PD-1–PDL-1 blockade and the CFSE assay results clearly indicate that 135C12 and 136B4 represent novel anti–PD-1 Abs that may exert antagonistic activity independent of blockade of the PD-1–PDL-1 interaction. For these reasons, the two novel nonblocking anti–PD-1 Abs were advanced for more in-depth profiling along with 137F2 blocking Ab, which displayed an improved immune-enhancing activity in our in vitro functional assay relative to pembrolizumab. The CDR sequences for the 135C12 and 137F2 mouse IgG1 Abs were also transferred into a human IgG4 Ab with a panel of constant region clones screened to identify humanized Abs with subnanomolar affinity for cell-surface PD-1. These humanized clones were renamed NB01 for the nonblocking 135C12 Ab and B01 for the blocking 137F2 Ab. Binding epitopes of antagonistic anti–PD-1 Abs that do not block the PD-1–PDL-1 interaction Binding epitopes for the three prioritized Abs were then mapped by monitoring binding to a panel of PD-1 proteins expressed with amino acid substitutions at solvent accessible residues of the extracellular domain (Fig. S3 A). Transiently transfected HeLa cells expressed mutant or wild-type forms of PD-1 and allowed for comparative Ab binding to cell-surface PD-1 by flow cytometry (Fig. 2 C). Discrete amino acid substitutions in PD-1 resulted in a loss in binding for selected Abs. Pembrolizumab and 137F2 Abs both mapped to regions previously identified as being important for PDL-1 binding to PD-1 (Lázár-Molnár et al., 2008; Lin et al., 2008), but binding of nonblocking antagonistic Abs was unaffected by these substitutions. 135C12 and 136B4 Ab binding was almost completely disrupted with either M18 (L41A/V43T/S137A/L138A/R139T substitutions) or M32 (D105A substitution) PD-1 constructs, respectively, which are situated on the opposite face to the PD-1 interaction with either PDL-1 or PDL-2. As a further evaluation of Ab-binding characteristics, competitive binding studies were performed with a PD-1–expressing Jurkat cell line. Saturating the PD-1 receptor with the 137F2 blocking Ab prevented cell staining with pembrolizumab, nivolumab, and B01 (humanized 137F2) but did not prevent staining with NB01 (humanized 135C12). Conversely, cell-surface PD-1 saturated with either 135C12 or 136B4 nonblocking Abs had no impact on the binding of pembrolizumab, nivolumab, or B01 Abs to PD-1. Both 135C12 and 136B4 Abs prevented the binding of NB01, the humanized version of the mouse 135C12 Ab (Fig. 2 D), indicating that these Abs bind overlapping epitopes on PD-1. These studies show that antagonistic blocking and nonblocking Abs of the PD-1–PDL-1 interaction can bind to cell-surface PD-1 simultaneously. To complete the molecular details of the binding epitope of the newly generated PDL-1 nonblocker Abs, the structure of the NB01a (135C12) Fab in complex with hPD-1 was solved at a resolution of 2.2 Å (Table S1). The NB01a Fab binds at the opposite side of the PDL-1–binding site (Fig. 2 E) with interacting residues on PD-1 confirming our mapping studies performed with the 135C12 Ab. The overall structure of hPD-1 fits previously published structures (Zak et al., 2015). Superposition of the Cα atoms of hPD-1 in complex with the NB01a Fab and hPD-1 in complex with hPDL-1 (Protein Data Bank [PDB] accession no. 4ZQK) showed no overlap of the binding footprints for NB01a and hPDL-1, indicating that they can bind PD-1 simultaneously (Fig. 2 E). One potential clash was noted between the hPDL-1 side chains E60/D61 and VL residues S52/S65 of NB01a; however, these side chains may adopt different rotamers in solution, thereby allowing concomitant binding of hPDL-1 and NB01a to hPD-1. Consequently, the affinity of PDL-1 for hPD-1 in the presence of 135C12/NB01 may be slightly reduced, as observed in the biochemical binding data (Fig. 2 A). It was important to determine whether the four predicted N-glycosylation sites on PD-1 were interfering with NB01 Ab binding. However, the modeling of these sites shows that glycosylation at N49, N58, N74, and N116 does not overlap with the bind epitopes of NB01 or 136B4 Abs and would not interfere with Ab interaction with PD-1 (Fig. 2 F). The epitope mapping and structural studies both revealed that Ab binding to a specific region of PD-1 on the opposing face of the PDL-1–PDL-2 interaction site resulted in an antagonistic functional activity (Fig. 1 and Fig. 2, C and E). Interestingly, sequence alignment across five species revealed that despite a low 52.3% identity with human PD-1, conserved regions throughout the sequence come together to form a highly conserved patch on PD-1 (Fig. S3 B) that is partially overlapping with the binding epitopes of both 135C12 (NB01 hIgG4) and 136B4 Abs (P1 patch in Fig. 2 F). Structural models of hPD-1–NB01a Fab with either pembrolizumab (PDB accession no. 5GGS) or nivolumab (PDB accession no. 5GGR) were also generated and clearly show that NB01a can bind hPD-1 simultaneously with either PDL-1 blocking Ab (Fig. 2 G; Lee et al., 2016). This model is consistent with our Ab competitive binding studies (Fig. 2 D). Nonblocking and blocking anti–PD-1 Abs synergize in recovering Ag-specific CD8 T cells from exhaustion Given that nonblocking anti–PD-1 Abs can bind simultaneously to cell-surface PD-1 with blocking Abs such as pembrolizumab, an important question is whether the simultaneous binding results in a synergistic potentiation of their antagonistic activity. This was addressed by testing combinations of blocking and nonblocking Abs in the in vitro functional recovery exhaustion assay. Studies presented in Fig. 3 show a significant increase in the recovery of proliferation of HIV-specific CD8 T cells when NB01a was used in combination of either pembrolizumab or B01 with 155% and 158% proliferating cells, respectively, relative to single Abs. These increases in proliferation were observed with the 135C12 mouse IgG1 variants of NB01 and several different humanized IgG4 clones (Fig. S2, A–D). It is important to underscore that this enhanced proliferation was beyond levels that could be achieved by higher Ab concentrations or with combinations of two blocking Abs that did not lead to enhancements in the recovery of proliferation (Fig. S2, E and F). Figure 3. Blocking and nonblocking anti–PD-1 Ab combinations synergize in recovering both the proliferation and functional activity of exhausted Ag-specific CD8 T cells. (A) Cumulative results (three to six experiments) of the recovery of the proliferation of HIV-specific CD8 T cells after treatment with single and/or the combination of blocking and/or nonblocking anti–PD-1 Abs. Results are expressed as the percentage of CFSE-low CD8 T cells, and 8–10 replicates were performed for each experimental condition. (B) Recovery of T cell functionality evaluated by measuring cytokine levels of IFNγ, IL-2, TNFα, and IL-10 in the cell medium following an Ag-specific CD8 T cell stimulation. Cumulative results are shown using the PBLs from six to eight different viremic HIV-positive donors. Untreated samples (Neg) were used as a negative control in each experiment. Data represent mean ± SD. *, P < 0.036; **, P < 0.0079; ***, P < 0.0009; ****, P < 0.0001 (unpaired t test with Welch’s correction). To further determine the biological activity of the two classes of anti–PD-1 Abs, the synergistic effect of the combination of blocking and nonblocking anti–PD-1 Abs was assessed on the production of cytokines. For these purposes, blood mononuclear cells from eight chronically infected viremic HIV individuals were stimulated with the specific antigenic peptides. Cells treated with anti–PD-1 Ab combinations released significantly higher levels of IFNγ, TNFα, and IL-10 relative to antigenic peptide alone treatments (Fig. 3 B). Additionally, results across the different donors showed that Ab combination treatment lead to IFNγ production with significantly higher levels than either blocking or nonblocking anti–PD-1 monotherapy treatments alone. Nonblocking anti–PD-1 Abs act predominantly through the CD28 costimulatory receptor to promote the AKT–NF-κB pathway Mechanism-of-action studies with anti–PD-1 Abs were then investigated using phosphoflow intracellular staining of proteins important to the T cell signaling cascade. Blood mononuclear cells from a viremic HIV-positive donor were used for these signaling studies with 60–63% of memory T cells expressing the PD-1 exhaustion marker. T cell stimulation with anti-CD3/CD28 Abs induced a temporal increase in protein phosphorylation that was proximal to (ZAP70, SLP76, and Lck/Src) and downstream of (AKT, PDK1, PLCγ1, NF-κB, ERK1/2, p38, and CREB) the TCR complex. T cell signaling was suppressed with human PDL-1 Fc fusion protein when added before the anti-CD3/CD28 stimulation. Importantly, this protocol using exhausted primary T cells identified the same canonical pathways reported for PDL-1–PD-1 mediated suppression of T cell activation using immunoblotting techniques that were performed with cell lines or in vitro stimulated T cell to augment PD-1 expression (Patsoukis et al., 2012; Yokosuka et al., 2012). T cell stimulation in the presence of PDL-1 resulted in statistically significant reduced phosphorylation of ZAP70 and Lck394/Src 1 min after stimulation and ERK1/2, AKT, and PDK1 5–15 min after stimulation (Figs. 4 A and S4). Figure 4. Nonblocking anti–PD-1 Abs restore Ca2+ flux and AKT signaling to exhausted T cells. (A) Phospho-flow signaling studies performed with PD-1–expressing functionally exhausted T cells. Intracellular staining of phosphoproteins important to T cell signaling showed increased phosphorylation upon stimulation with anti-CD3/CD28 Abs. FACS histogram profiles for memory CD4 T cells show that a PDL-1 Fc fusion protein suppressed phosphorylation of AKT pT308, AKT pS473, and PDK1 pS242. Preincubation of cells with NB01b nonblocking or pembrolizumab blocking anti–PD-1 Abs significantly relieved PDL-1–mediated suppression of these phosphosignaling proteins at 5 and 15 min after stimulation. Data are presented as mean ± SEM for five to eight individual experiments. ns, not significant; *, P < 0.05; **, P < 0.0064; ****, P < 0.0001. (B) Exhausted T cells had reduced Ca2+ mobilization when stimulated with anti-CD3/CD28 in the presence of PDL-1 Fc fusion protein. (C) This suppression was restored with NB01b, pembrolizumab, or nivolumab. Synergistic increase in Ca2+ release was observed in exhausted T cells stimulated with anti-CD3/CD28 + PDL-1 when coincubated with NB01b and either pembrolizumab (B) or nivolumab (C). Representative data are shown from three independent experiments, and clinical Ab preparation of pembrolizumab was used for these studies. FDR, false discovery rate; MFI, mean fluorescence intensity. Pretreatment of cells with NB01b or pembrolizumab partially restored phosphorylation of PDK1-pS242, AKT-pT308, and AKT-pS473 when stimulated with anti-CD3/CD28 in the presence of PDL-1 Fc protein. A trend toward recovery was also observed with ERK1/2, ZAP70, and LCK394/Src phosphorylation levels; however, these levels did not reach statistical significance. Anti–PD-1 Ab–mediated rescue was not evident for AKT at the earliest time point evaluated, but phosphorylation levels gradually built to a significant increase at 5 and 15 min for both PDK1 and AKT (Fig. 4 A). This increased phosphorylation was more statistically significant for the nonblocking anti–PD-1 Ab compared with pembrolizumab-treated cells, which also had a trend toward lower phospho-AKT levels 15 min after stimulation. This pathway is directly downstream of the CD28 costimulatory receptor, which was previous reported to be necessary for the antitumor activity of blocking anti–PD-1 Abs (Hui et al., 2017). These results are consistent with a mechanism whereby anti–PD-1 Ab therapy prevents the PDL-1–mediated recruitment of a phosphatase to the PDK1–AKT complex. Prolonged phosphorylation and activation of the PDK1–AKT complex would lead to increased T cell survival, trafficking and effector function. An important aspect of T cell stimulation is the rapid increase in cytoplasmic levels of calcium (Ca2+) released from intracellular stores and through the opening of the plasma membrane calcium release–activated channels. This Ca2+ release helps to propagate the T cell activation signal, including NFAT activation and cytoskeletal rearrangement. Similar to the phospho-flow experiments, stimulation of exhausted PD-1+ T cells in the presence of a PDL-1 Fc fusion protein led to reduced levels of intracellular Ca2+ relative to an anti-CD3/CD28 Ab stimulation control (Fig. 4, B and C). Pretreatment of cells with either blocking (pembrolizumab or nivolumab) or nonblocking NB01b anti–PD-1 Abs restored Ca2+ mobilization to levels observed in cells stimulated in the absence of PDL-1 Fc protein. Importantly, combinations of NB01b and either pembrolizumab (Fig. 4 B) or nivolumab (Fig. 4 C) before T cell stimulation in the presence of PDL-1 resulted in a synergistic increase in Ca2+ levels. This Ca2+ mobilization was more intense and gave area under the curve values more than twofold higher compared with the anti-CD3/CD28 stimulation condition. To provide insights on the mechanistic differences in the effect of blocking and nonblocking anti–PD-1 Abs, the two classes of Abs were further probed by immunoprecipitation (IP) of PD-1 expressed on a Jurkat cell line stably expressing high levels of PD-1. Cells were unstimulated or stimulated with anti-CD3/CD28 Abs and then the PD-1 receptor and associated cellular complex was coprecipitated with NB01b, pembrolizumab or a combination of NB01b and pembrolizumab Abs covalently coupled to beads. In stimulated cells, pembrolizumab effectively pulled down a PD-1 complex that included CD28, SHP-2, PI3K, and phosphorylated Src (Fig. 5 A). In contrast, although NB01b pulled down equivalent levels of PD-1 and the PD-1–associated SHP-2 compared with pembrolizumab, there were significantly lower levels of CD28 receptor and PI3K in the nonblocking anti–PD-1 Ab IP complex (Fig. 5, A and B). AKT was weakly immunoprecipitated with the PD-1 complex, but there was a trend toward higher levels of AKT pulled down with pembrolizumab compared with the NB01b Ab. IPs performed with pembrolizumab- and NB01b-coated beads resulted in reduced levels of CD28 being coprecipitated with PD-1 relative to the pembrolizumab-alone IP (Fig. 5 B). In the above experiments, recruitment of PD-1 in the immunological synapsis after anti-CD3/CD28 stimulation occurred in the absence of PDL-1 engagement. To explain how the recruitment of PD-1 occurs in the absence of PDL-1 engagement, in control experiments, IPs were performed using biotinylated pembrolizumab and NB01b not coupled to beads. These experiments showed that uncoupled Abs did not pull down the CD28 in either the presence or absence of stimulation (Fig. 5 C). Therefore, these control experiments indicate that aggregation induced by the anti–PD-1 Abs coupled to beads deliver to PD-1 a signal similar to the engagement of PDL-1 and promote recruitment of PD-1 to the CD28 costimulatory receptor following T cell activation as shown in Fig. 5 B. Figure 5. Nonblocking anti–PD-1 Abs act primarily through the CD28 costimulatory receptor–associated pathway. (A) IPs for two representative experiments are shown where the PD-1 receptor and associated protein complex were immunoprecipitated with NB01b, pembrolizumab, or a combination of NB01b and pembrolizumab in nonstimulated and anti-CD3/CD28 Ab–stimulated Jurkat PD-1 cells. In stimulated cells, pembrolizumab coprecipitated high levels of PD-1, CD28, SHP-2, PI3K, and the phosphorylated form of Src (p-Src) with the PD-1 complex. (B) NB01b pulled down equivalent levels of PD-1, SHP-2, and p-Src, but significantly reduced levels of CD28 and PI3K were observed in three or four separate experiments. AKT was only weakly pulled down in the PD-1 complex (A; right blot); however, there was a trend toward higher levels of AKT being immunoprecipitated with pembrolizumab compared with NB01b. (C) Control IPs performed by preincubating cells with either biotinylated pembrolizumab or NB01b before T cell stimulation shows that PD-1 pulled down with streptavidin (S) does not constitutively form a complex with CD28. Similarly, IPs performed with anti-CD3/CD28 coated beads show a strong pull-down of CD28 without detectable levels of associated PD-1. (D) The nonblocking anti–PD-1 Ab NB01b showed weak to no activation of the NFAT promoter in a Jurkat PD-1 luciferase reporter cell line when stimulated with 293T cells coexpressing a TCR activator and the PDL-1 receptor. Pretreatment of Jurkat PD-1 cells with the blocking anti–PD-1 Ab pembrolizumab strongly promoted NFAT activation. In contrast, both NB01b and pembrolizumab relieved PD-1–mediated suppression of NF-κB activation. Graphs show the mean ± SD and are representative of two independent experiments. *, P < 0.045; ***, P < 0.001; ****, P < 0.0001 (unpaired t test with Welch’s correction). RLU, relative luminescence units. Taken together, these results indicate that PD-1 is recruited to the CD28 costimulatory receptor upon T cell activation and that this complex is unaffected by the blocking with pembrolizumab anti–PD-1 Ab. Importantly, nonblocking Abs such as NB01b inhibit the formation of the PD-1 complex that includes CD28. This would effectively reduce the recruitment of the SHP-2 phosphatase to the CD28 costimulatory receptor and its associated intracellular kinases that includes PI3K, the upstream activator of AKT (Parry et al., 2005; Patsoukis et al., 2012). Given that the CD28 costimulatory receptor contributes to T cell activation through enhanced signaling of the AKT–NF-κB pathways and TCR stimulation acts strongly through the NFAT pathway, anti–PD-1 Abs were tested in NF-κB and NFAT reporter assays. A Jurkat–PD-1 T cell line stably expressing luciferase under the control of a NFAT promoter was transiently transfected with the NanoLuc NF-κB reporter plasmid in order to monitor the activation of both pathways in the same cells. Experiments using this cell line were designed to evaluate the potency of blocking anti–PD-1 Abs, and stimulation was achieved through coculture of Jurkat PD-1 cells with 293T cells coexpressing a membrane-associated anti-CD3 TCR activator and the PDL-1 receptor. Pretreating Jurkat PD-1 cells with the blocking pembrolizumab anti–PD-1 Ab before stimulation effectively relieved the PD-1–PDL-1–mediated suppression of the NFAT reporter by 6.4-fold in this assay relative to an IgG4 control Ab, while the NB01b nonblocking anti–PD-1 Ab resulted in low to no NFAT activation (Fig. 5 D). In contrast, parallel evaluation of the NF-κB pathway showed that both blocking and nonblocking anti–PD-1 Abs significantly relived PD-1–mediated suppression of the NF-κB promoter following TCR-mediated activation (1.7-fold and 2.4-fold relative to an IgG4 isotope control, respectively; Fig. 5 D). Consistent with the functional data (Fig. 3, A and B) and Ca2+ flux studies (Fig. 4, B and C), combination of blocking and nonblocking anti–PD-1 Abs result in a significant increase in NFAT activation relative to single anti–PD-1 Ab treatments given alone. Taken together, signaling and IP studies show that classical blocking and antagonistic nonblocking anti–PD-1 Abs act through distinct mechanisms in relieving T cell exhaustion. Both Ab classes are shown to act on pathways linked to PD-1–mediated suppression of T cell activation. However, blocking Abs have a more pronounced effect on the NFAT pathway, while nonblocking Abs act predominantly through the CD28 coreceptor that promotes the AKT–NF-κB pathway and leads to T cell proliferation and survival. In vivo efficacy of nonblocking anti–PD-1 Abs The antagonistic activity of PDL-1 nonblocking Abs and the synergistic effect observed in vitro provided the scientific rationale to determine the ability of nonblocking Abs alone or in combination with blocking Abs in suppressing tumor growth in vivo. The in vivo efficacy of the NB01b antagonistic nonblocking Ab was then evaluated in the PD-1 HuGEMM in vivo tumor model (Lute et al., 2005; Huang et al., 2017). HuGEMM mice were genetically engineered to express a chimeric human/mouse PD-1 protein with the majority of the ectodomain (residues 26–146) encoded by the human PD-1 protein. Mice successfully engrafted with tumors formed from the PDL-1–high MC38 colon adenocarcinoma cell line were randomly ascribed to give the same average tumor volumes for the different arms of the study with the indicated treatments administered twice weekly (Fig. 6 A). The nonblocking NB01b Ab administered at 10 mg/kg effectively suppressed tumor growth similar to pembrolizumab (clinical Ab batch) or nivolumab administered at the same dose in three separate studies (Fig. 6, B–D and F). We then determined whether the antitumor activity was enhanced by combining NB01b and either pembrolizumab or nivolumab. Mice that were coadministered with NB01b and pembrolizumab or nivolumab at a dose of 5 mg/kg of each Ab, for a total dose of 10 mg/kg, had more potent suppression in tumor growth relative to either Ab dosed alone (Fig. 6, E and F). A global analysis of three separate studies was performed where mice were treated with NB01b, pembrolizumab, and nivolumab monotherapies or combination therapies consisting of NB01b with either pembrolizumab or nivolumab. This analysis showed that the combination of blocking and nonblocking anti–PD-1 Abs was associated with a significantly greater suppression of tumor growth throughout the study. In modeling the cubic root–transformed tumor volumes as a function of time and Ab therapy using a mixed-effects statistical framework (Fig. 6 F), it can be seen that Ab combination therapy had a significant reduction in tumor volume over time compared with the anti–PD-1 monotherapy arms in the studies (P = 0.00045). Importantly, the proportion of mice with complete control of tumor growth, having a smaller tumor volume at the end of the study compared with the initial day of anti–PD-1 therapy, was 47% (9 out of 19 mice) in the combination treatment group (NB01b plus pembrolizumab or nivolumab) Abs as compared with 22% (13 out of 59) in the anti–PD-1 Ab monotherapy treatment groups (Fig. 6 G; P = 0.0354). In a study that was extended to 33 d after MC38 inoculation, mouse survival was significantly improved for the combination therapy arm relative to vehicle control–treated mice (Fig. 6 G; P = 0.0005). The significance of this extended survival was greater for the combination therapy relative to either NB01b or nivolumab monotherapies administered alone (Fig. 6 G, 0.0079 and 0.0223, respectively). Figure 6. Enhancement of tumor clearance by the combination of blocking and nonblocking anti–PD-1 Abs in the PD-1 HuGEMM in vivo MC38 tumor model. (A) Experimental scheme. (B–E) Mice successfully engrafted with the MC38 tumor cell line were treated twice weekly with PBS control (B), NB01b (C), pembrolizumab or nivolumab (D), or a combination of NB01b with either pembrolizumab or nivolumab (E), with tumor volumes measured in parallel. A collective analysis of three separate studies with n = 9–10 mice per arm per study is presented in A–D. P values determined by pairwise comparison using a mixed-effect linear model showed that all Ab arms of the study had reduced tumor growth compared with the vehicle control arm. (F) Suppression of tumor volume in mice engrafted with PDL-1–high MC38 cell line with either single or the combination of blocking and nonblocking anti–PD-1 Abs. Blocking (pembrolizumab or nivolumab) and nonblocking (NB01b) anti–PD-1 Ab monotherapies exerted equivalent suppression of tumor growth and mean percent tumor inhibition. Modeling of the cubic root transformed tumor volume in three separate studies as a function of time demonstrated a statistically significant reduction in tumor volume over time for the anti–PD-1 combination therapy relative to anti–PD-1 monotherapies. (G) Combination anti–PD-1 Ab therapy significantly enhanced the proportion of mice that controlled tumor growth (9 out of 19 mice) relative to Ab monotherapies (13 out of 59 mice). Survival analysis for a study performed using NB01b and nivolumab, with statistical differences determined using the log-rank test. Graphs show the mean ± SEM, unless otherwise indicated. *, P = 0.0223; **, P < 0.008; ***, P = 0.0005; ****, P < 0.0001. In a second in vivo study, PD-1 HuGEMM mice were subcutaneously implanted with the highly aggressive and poorly immunogenic wild-type B16F10 cell line (Kuzu et al., 2015; Kokolus et al., 2017). B16F10 tumor, which is engrafted subcutaneously, is characterized by a very fast growth and generation of rapid lung metastasis. As for the MC38 tumor model, mice were dosed twice weekly with blocking pembrolizumab, nonblocking NB01b, and/or the combination of both blocker/nonblocker Abs (Fig. 7 A). Suppression of tumor growth in the active groups versus the vehicle control untreated group was similar (Fig. 7, B–E). Although the suppression of tumor growth was significant (P < 0.008) in the active groups, the effect was transient and associated with rapid escape of the tumor. Furthermore, significant prolonged mouse survival was observed in the active groups as compared with the vehicle control group as determined using a log-rank statistical test (Fig. 7 F). Figure 7. Blocking and nonblocking anti–PD-1 Abs suppress tumor growth and prolong survival of mice implanted with B16F10 cells in the PD-1 HuGEMM in vivo tumor model. (A) Experimental scheme. (B–E) Mice successfully engrafted with the poorly immunogenic B16F10 tumor cell line were treated twice weekly with PBS control (B), NB01b (C), pembrolizumab (D), or a combination of NB01b and pembrolizumab (E), with tumor volumes measured in parallel in the treatment (n = 10 mice) and control (vehicle; n = 10 mice) groups. P values determined by pairwise comparison using a mixed-effect linear model showed that blocking (pembrolizumab), nonblocking (NB01b), and combination anti–PD-1 Ab therapies had reduced tumor volume growth relative to the vehicle control. (F) Survival analysis with statistical difference determined using the log-rank test. Graphs show the mean ± SEM. **, P = 0.008; ****, P < 0.0001. Discussion Here, we report the identification of a novel class of antagonistic anti–PD-1 Abs acting independently of PDL-1 blockade. These Abs were rare among hybridoma clone–producing Abs with high affinity for PD-1 (3.2% of 156 clones), which may partially explain why this novel anti–PD-1 Ab class has eluded discovery until now. Given the prevailing theory that PD-1–PDL-1 blockade is an important element needed for relieving T cell exhaustion, we performed several different lines of investigation to validate that our novel anti–PD-1 Abs were truly nonblocking of the PD-1–PDL-1 interaction. Protein-to-protein (PD-1–PDL-1) interaction studies, together with competitive binding studies using the Jurkat PD-1 stable cell line, provided evidence that 135C12 or 136B4 nonblocking Abs interfere neither with the PD-1–PDL-1 interaction nor with the simultaneous binding of blocking Abs to PD-1 expressed at the cell surface. These results indicate that nonblocking anti–PD-1 Abs do not sterically interfere with PDL-1 binding to PD-1 and do not exert a conformational change to the PD-1 protein that would significantly reduce its affinity for the PDL-1 ligand, and they exclude the possibility that either conformational or posttranslational differences occur between recombinant PD-1 and the cell-surface PD-1 receptor. Epitope mapping studies performed through PD-1 site-directed mutagenesis together with the hPD-1–NB01a Fab crystal structure have shown that the novel two nonblocking anti–PD-1 Abs with antagonistic activity both bind PD-1 on different epitopes located on the opposing face of PD-1 relative to the PDL-1 interaction site. Importantly, we discovered that this novel functional active region of PD-1 recognized by nonblocking Abs overlaps with a patch of PD-1 residues that are highly conserved across six different species. Evolutionary conservation of this region suggests that PD-1 is under a selective pressure to preserve this site, which could be explained by the site being functionally implicated in PD-1 signaling or potentially through interaction with an alternate ligand from PDL-1–PDL-2. The fact that 135C12/NB01 and 136B4 Abs both overlap with an edge of this conserved site on PD-1 indicates that steric blocking of an alternate ligand and/or a complex formed with CD28 may represent the most likely hypothesis. Consistent with blocking and nonblocking Abs both exerting distinct immune-enhancing functional activity through PD-1, combination therapy using the two classes of anti–PD-1 resulted in synergistic functional recovery of Ag-specific CD8 T cells from exhaustion. Mechanism-of-action studies demonstrated that blocking and nonblocking anti–PD-1 Abs relieve PD-1 mediated T cell exhaustion through distinct effects on the TCR and CD28 related signaling cascades. We have provided evidence that the blocking Abs have a more pronounced effect on the NFAT pathway, while nonblocking Abs act predominantly through the CD28 coreceptor that promotes the AKT–NF-κB pathway. Previous studies (Patsoukis et al., 2012, 2013; Yokosuka et al., 2012; Hui et al., 2017) have indicated that PD-1 exerts its suppressive effects on T cell activation through PDL-1–mediated recruitment of PD-1 into the immunological synapse. This recruitment brings the PD-1–associated SHP-2 phosphatase into contact with the TCR-associated kinases, including ZAP70, Lck/Src, and ERK1/2, thus resulting in dephosphorylation of these kinases and dampening the intensity and duration of the T cell–specific activation. Blocking anti–PD-1 and anti–PDL-1 Abs relieve T cell exhaustion by preventing this recruitment of PD-1 into the TCR complex (Chemnitz et al., 2004; Sheppard et al., 2004; Yokosuka et al., 2012; Wherry and Kurachi, 2015; Hui et al., 2017; Kamphorst et al., 2017). Interestingly, our IP studies performed with the blocking anti–PD-1 Ab pembrolizumab showed that a complex formed between PD-1 and CD28 upon T cell activation. This indicates that elevated levels of PD-1 on a T cell may suppress TCR activation by a second mechanism, whereby PD-1 forms a cis complex with the CD28 costimulatory receptor. This colocalization would bring CD28-associated intracellular kinases essential for T cell costimulation into contact with PD-1–associated SHP-2 phosphatase, resulting in a reduced costimulatory effect. In contrast, IPs performed with the nonblocking anti–PD-1 NB01b show significantly lower levels of a PD-1–CD28 complex after T cell activation. Taken together, our results support the model that in binding PD-1 on the opposing face of the PDL-1 interaction site, NB01b disrupts the inhibitory close-contact complex that forms between PD-1 and CD28 on activated T cells. The importance of this PD-1–CD28 interaction may explain the evolutionary conservation of the residues on PD-1 overlapping with the binding site of the nonblocking Abs. Our results are also consistent with Hui et al., who recently demonstrated that CD28 is the primary substrate for dephosphorylation through PD-1/SHP-2, leading to T cell suppression (Hui et al., 2017). The distinct mechanism of nonblocking anti–PD-1 Abs may also help rationalize our in vitro functional data, where there is a trend toward lower cytokine production with NB01b compared with pembrolizumab. Since nonblocking Abs have a more limited effect on restoring PD-1–mediated suppression of the NFAT–Ca2+ flux pathways, one would predict a reduced enhancement in cytokine production. Conversely, the more pronounced effect of nonblocking anti–PD-1 Abs on the AKT–NF-κB pathway is consistent with the enhanced proliferation observed when used in combination with blocking anti–PD-1 Abs. The finding that blocker and nonblocker Abs act through distinct mechanisms explains the synergistic effect observed both in vitro and in vivo studies in combining the two classes of Abs. We have provided three lines of evidence in in vitro studies on the synergistic effect using a combination of blocking and nonblocking anti–PD-1 Abs: (1) increased Ag-specific CD8 T cell proliferation, (2) increased cytokine production in response to Ag-specific stimulation, and (3) enhanced intracellular Ca2+ mobilization following anti-CD3/CD28 stimulation in the presence of PDL-1 suppression. The immunotherapy with anti–PD-1 Abs is a major breakthrough in the fight against cancer, and anti–PD-1 therapy using pembrolizumab or nivolumab achieves a 20–50% objective response rate depending upon the type of cancer and the selection of criteria used for determining the percentage of PDL-1–expressing tumor cells (Carbognin et al., 2015; Gandini et al., 2016). Many patients have only a partial response to treatment, and a variable proportion of patients (depending on the type of cancer) experience recurrence of the tumor. Therefore, it has become clear that the potentiation of the antitumor activity of blocker anti–PD-1 Abs and the development of combinations of immunotherapy strategies are urgently needed. In addition to combination of anti–PD-1 Abs with anti–CTLA-4 (Larkin et al., 2015), combination therapies with Abs targeting other immune checkpoint inhibitors have entered clinical development. In the present study, we have shown in two in vivo mouse tumor models that the novel nonblocking anti–PD-1 Abs have an antitumor activity equal to the conventional blocking Abs such as pembrolizumab. We have validated the results using the immunogenic colon carcinoma MC38 cell line and the highly aggressive and poorly immunogenic melanoma B16F10 cell line. Of note, in the MC38 studies, combination therapy using both classes of anti–PD-1 Abs significantly suppressed tumor growth (74%) and was associated with a 2.4-fold increase in mice with complete tumor control and a trend toward improved survival relative to anti–PD-1 monotherapies. The antitumor activity in the B16F10 model observed with anti–PD-1 monotherapy using either blocking or nonblocking Abs was inferior and transient as compared with MC38 model. The limited antitumor effect observed in our study is consistent with previous studies (Chen et al., 2015; Woods et al., 2015; Zamarin et al., 2018) in the B16F10 tumor model using anti–PD-1 treatment. Along the same line, combination of blocking and nonblocking Abs did not provide an additional therapeutic advantage in the poorly immunogenic B16F10 model. This may be attributed to the exceptionally rapid growth of the B16F10 cells, where tumor volume was >1,000 mm3 in 70% of mice in the vehicle control arm by day 7 after initiation of anti–PD-1 therapy. Because of the rapid tumor escape, there is likely not sufficient time to develop a fully competent immune response and allow for the combined anti–PD-1 treatment to enhance tumor suppression, since only two doses of the combined treatment were administered by day 7. It should be underscored that this high rate of tumor growth in the B16F10 mouse tumor model is not representative of tumor development in humans. In conclusion, we report the identification of a novel class of antagonistic anti–PD-1 Abs acting independently of PDL-1 blockade and acting through the CD28–AKT–NF-κB costimulatory pathway. These nonblocking anti–PD-1 Abs synergize with the classical blocker anti–PD-1 and are associated with an improvement of antitumor activity. These results provide important advances in the biology and function of PD-1 and may open new avenues for enhanced antitumor activity. Materials and methods Abs Novel mouse anti–PD-1 Abs were isolated from the media of hybridoma cell lines and tested for binding to human PD-1 protein in the Luminex assay described below. For hybridomas that produced Abs with affinity < 1 nM, RNA was extracted using the Rneasy micro kit (Qiagen) followed by cDNA synthesis with the Cellsdirect 1-step QRT-PCR kit (Thermo Fisher Scientific) using the mouse Ig-Primer set (EMD Biosciences). The heavy-chain CDR region for each hybridoma was PCR amplified using the mouse Ig-Primer sets and then sequenced using the Applied Biosystems 3500 Genetic Analyzer sequencer (Thermo Fisher Scientific) with sequences described in a patent application (Pantaleo and Fenwick, 2016). Prioritized clones were expressed from 200 ml of hybridoma cells with the monoclonal Abs purified from the cell medium using a protein A column (Thermo Fisher Scientific) with Abs eluted with a 100 mM glycine buffer pH 3.0 into a 1 M Tris-HCl eluate at pH8.0. Abs were dialyzed twice against PBS, concentrated using a JumboSep centrifuge filter with a 3-kD molecular weight cutoff (Pall Laboratories) and sterile filtered with a Millex GP 0.22 µm filter (Millipore). Prior to anti–PD-1 Ab evaluation in the in vitro functional recovery assay, Abs were tested with the Food and Drug Administration–licensed Endosafe-PTS kit (Charles River Laboratory), and all had low endotoxin levels at <5 EU per mg of protein. Humanized Abs of the 135C12 and 137F2 mouse clones were produced by GenScript with the Deluxe Antibody Humanization Service and using a framework assembly method. Briefly, frameworks of different human germlines with high sequence identities for each of the mouse anti–PD-1 Abs were selected and assembled using overlapping PCR. The resulting libraries of humanized Fabs with different framework sequences and encoding the mouse Ab CDR sequences for either 135C12 or 137F2 were then used to generate a phage display library encoding panels of humanized Fabs. Panning the library against the human PD-1 Fc fusion protein (R&D Systems) allowed for the enrichment of binding constructs and individual clones were screened for affinity using a phage ELISA assay. The FASEBA screen (fast screening for expression, biophysical properties, and affinity) was employed to identify Fab fragments that expressed well and had good biophysical properties, with candidate clones ranked based on their dissociation rate after binding to the PD-1 Fc fusion protein in a Biacore T200 assay (GE Healthcare). Candidate sequences with low-binding off rates were transferred into a proprietary vector at GenScript to produce the humanized anti–PD-1 Abs with an IgG4 Fc region that encoded a serine 228 to proline substitution in the hinge region to help prevent in vivo Fab arm exchange. High-affinity humanized clones NB01a, NB01b, and NB01c were identified for the 135C12 Ab and B01 for the 137F2 Ab. Pembrolizumab and nivolumab used in the in vitro functional studies were produced by GenScript using codon-optimized genes to express IgG4 Abs. A clinical lot of pembrolizumab was used in the in vivo mouse studies and was obtained through the Hospitalier Universitaire Vaudois. The following Abs were used in flow cytometry and Luminex binding studies: allophycocyanin (APC)–H7–conjugated anti-CD3 (clone SK7), Pacific blue–, FITC-, or PE-CF594–conjugated anti-CD4 (clone RPA-T4), anti-CD8 Pacific blue (clone RPA-T8) Abs (Becton Dickinson); PE-labeled goat anti-mouse IgG secondary, PE-labeled F(ab′)2-goat anti-mouse IgG secondary Ab, and mouse IgG1 isotype control (P3.6.2.8.1; eBioscience); PE-labeled goat anti-human IgG secondary Ab (Invitrogen); human IgG4 Isotype control Ab (ET904; Eureka Therapeutics); anti–PD-1 Ab (clone EH12.2H7) and PE-labeled streptavidin (BioLegend) and energy-coupled dye (phycoerythrin–Texas red conjugate)–conjugated anti-CD45RA (clone 2H4; Beckman Coulter). Phospho-flow and Ca2+ signaling studies were performed with the following Abs: CD3 Alexa Fluor 700 (UCHT1; BD), CD4 Qdot 605 (S3.5; Thermo Fisher Scientific), CD8 APC-eFluor780 (RPA-T8; eBioscience), CD8 PerCP-Cy5.5 (SK1; BD), and CD45RA Brilliant violet 650 (HI100; BioLegend) was used for surface T cell characterization. For PD-1 staining of humanized mAbs, goat anti-human PE secondary (Thermo Fisher Scientific) was used. For intracellular phosphoprotein staining, ZAP70 pY319 PE-Cy7 (17A/P-ZAP70; BD), Erk1/2 pT202/pY204 PE-CF594 (20A; BD), SLP76 pY128 Alexa Fluor 488 (J141-668.36.58; BD), p38 pT180/pY182 PE-CF594 (36/p38; BD), PLCγ1 pY783 (27/PLC; BD), AKT pS473 Brilliant violet 421(M89-61; BD), AKT pT308 PE (J1-223.371, BD), PDK1 pS241 Alexa Fluor 647 (J66-653.44.17; BD), NF-κB p65 pS529 Alexa Fluor 488 (K10-895.12.50; BD), CREB pS133 PE (J151-21; BD), Src pY418 Alexa Fluor 488 (K98-37; BD), Lck505 pY505 Alexa Fluor 647 (4/LCK-Y505; BD), and S6 pS235/236 PE-Cy7 (D57.2.2E; Cell Signaling Technology) were used. For stimulation conditions, anti-CD3 (OKT3; BD), anti-CD28 (CD28.2; BD), and goat anti-mouse secondary Ab (Jackson ImmunoResearch) was used. Recombinant Fc chimeric PDL-1 protein (R&D Systems) was resuspended in PBS and used for suppression experiments. Fluo-4 AM (Thermo Fisher Scientific) was used for calcium flux experiments. Immunoblot studies were performed with the Abs against PD-1 (D4W23), CD28 (D2Z4E), PI3K (19H8), and pSRC (D49G4) from Cell Signaling Technology and against SHP-2 (79) and AKT (M89-61) from BD Biosciences. Detection of primary Abs used the anti-mouse IgG or the anti-rabbit IgG HRP-linked Abs (catalog numbers 58802 and 7074, respectively; Cell Signaling Technology). Cell culture Peripheral blood mononuclear cells were cultured in Roswell Park Memorial Institute (RPMI) medium and HeLa cells in DMEM (GIBCO BRL Life Technologies), each containing 10% heat-inactivated FBS or 6% human serum as indicated (both from Institut de Biotechnologies Jacques Boy), 100 IU/ml penicillin, and 100 µg/ml streptomycin (Bio Concept). Incubation of cells was performed at 37°C with 5% CO2. HIV-positive donors, ethics statement, and cell isolation The present study was approved by the Institutional Review Board of the Centre Hospitalier Universitaire Vaudois, and all individuals gave written informed consent. Blood mononuclear cells used in the in vitro functional assay were obtained following leukapheresis performed on eight HIV-positive donors that were chronically infected based on virological and clinical profiles (Vajpayee et al., 2005). Blood mononuclear cells were isolated as previously described (Perreau and Kremer, 2005) and cryopreserved in liquid nitrogen. Luminex binding assay The human PD-1 Fc fusion protein (R&D Systems) was conjugated to Bio-Plex magnetic beads (Bio-Rad) according to the manufacturer’s protocol and used in a primary screen for anti–PD-1 Ab binding affinity. Ab serial dilutions were incubated with PD-1 Fc-coated beads for 2 h with bound mouse Ab detected using a PE-labeled anti-mouse IgG secondary in a Luminex binding assay. In the competitive binding assay performed with a commercial anti–PD-1 Ab, PD-1 Fc-conjugated Bio-Plex beads were preincubated with one of the novel anti–PD-1 Ab clones. A biotinylated clone EH12.2H7 anti–PD-1 Ab was then added to the beads and incubated for 2 h before staining for the bound commercial Abs with PE-labeled streptavidin. Competitive Ab binding to PD-1 results in a reduction in mean fluorescence intensity relative to biotinylated EH12.2H7 Ab binding in the absence of a competitor Ab. Beads were analyzed on a FLEXMAP 3D or Luminex 100 instrument (Luminex Corporation). The PD-1–PDL-1 biochemical protein–protein interaction assay was performed by preincubating PD-1 Fc coated Bio-Plex beads in the presence of 20 nM of anti–PD-1 Ab (clones 137F2, 135C12, and 136B4) or a mouse IgG1 isotype control for 1 h. Duplicate samples of biotinylated PD-L1 Fc protein (R&D Systems) were added in a concentration range from 0.08 to 10 nM to the preformed Ab–PD-1 complex and incubated for a further 4 h before removing unbound PDL-1 and staining beads with PE-labeled streptavidin. In a variation of this assay to identify blocking Abs, PD-1 Fc–coated Bio-Plex beads were preincubated for 1 h with up to 20 nM anti–PD-1 Ab followed by a 4-h incubation with 1.25 nM biotinylated PDL-1 protein, a concentration that gave half-maximal binding to the PD-1–coated beads in the absence of Ab competitor. Bound PDL-1 was stained with PE-labeled streptavidin and measured with the Luminex 100 instrument. Ab epitope mapping studies The pReceiver-M67 vector encoding the open reading frame the PDCD1 gene (GeneCopoeia) was used to express human PD-1, and vectors encoding the amino acid substitutions in PD-1 listed in Fig. S4 were generated using the Q5 Site-Directed Mutagenesis Kit (New England Biolabs) according to the manufacturer’s protocol. These PD-1 expression vectors together with an enhanced GFP pcDNA3 construct (Addgene) were used to transiently transfect HeLa cells using the FuGENE 6 transfection reagent (Promega). Following a 2-d incubation at 37°C, the adherent HeLa cells were detached from the plates with a stream of PBS and labeled with the Aqua LIVE/DEAD staining kit (Thermo Fisher Scientific). Ab binding to wild-type or mutant forms of PD-1 was performed in parallel by incubating the different transfected cell samples with 2 µg/ml of the indicated mouse anti–PD-1 Ab followed by a PE-labeled F(ab′)2-goat anti-mouse IgG secondary Ab. Binding of pembrolizumab to the transiently transfected HeLa cells expressing the different PD-1 constructs was detected using PE-labeled goat anti-human IgG secondary Ab. Flow cytometry analysis of the samples was performed on an LSR II (Becton Dickinson) with Ab binding for the different wild-type and mutant PD-1 constructs evaluated for the live, enhanced GFP–expressing HeLa cells. Flow cytometry histogram results presented are representative of two to four independent experiments. Ab binding to cell-surface PD-1 and a competitive binding assay A Jurkat cell line stably transfected to express high levels of human PD-1 (BPS Biosciences) was used to evaluate Ab binding affinity to cell-surface PD-1. These cells were incubated with different Ab concentrations, washed, stained with a PE–labeled anti-mouse or anti-human secondary Ab, and analyzed by flow cytometry. Percent detection of PD-1 positive cells was determined for each Ab concentration and binding affinities were determined with the GraphPad Prism7 software using a nonlinear curve-fitting analysis. Competitive binding between different anti–PD-1 Ab clones to Jurkat PD-1 cells was performed by adding saturating amounts (40 µg/ml) mouse IgG1 competitor Ab (137F2, 135C12, or 136B4) with cells incubated for 30 min followed by 1 µg/ml of the indicated human IgG4 anti–PD-1 Abs. The PE-labeled goat anti-human IgG secondary Ab was used to detect the levels of human anti–PD-1 Abs binding to cell-surface PD-1 in the presence of competitor. Maximal binding to the Jurkat PD-1 cells was equivalent for all human anti–PD-1 Abs, and the positive control in each graph is shown for pembrolizumab staining. Functional exhaustion CFSE proliferation assay Cryopreserved blood mononuclear cells from a patient with chronic HIV infection were thawed and resuspended in RPMI medium containing 20% FBS, washed with 37°C PBS, and rested in RPMI medium with 10% FBS overnight. The following morning, cells were washed twice with 37°C PBS and then incubated with 20 µM CFSE (Invitrogen) at 37°C for 7 min in the dark. The staining was quenched with the addition of FBS at a final concentration of 10% and then washed twice with RPMI medium. An aliquot of CFSE-stained blood mononuclear cells was set aside for the negative control samples, and the remaining peripheral blood mononuclear cells were batch stimulated with the addition of the indicated HIV antigenic peptide (GPT) to give a final concentration of 1.11 µg/ml. Peptide-stimulated and nonstimulated cells were then distributed into 48-well plates at 106 cells per well in 900 µl. Replicates of 8–10 wells were prepared for each condition tested. Abs tested were prepared in a 10× stock in RPMI + 6% human serum, then 100 µl was added to the appropriate wells to have a final peptide concentration of 1 µg/ml and final Ab concentration of 5 µg/ml unless otherwise indicated. As additional controls, a mouse IgG1 or human IgG4 isotype control Ab at a final concentration of 5 µg/ml was included in most experiments, and one well of cells was stimulated with staphylococcal enterotoxin B as a positive control for the induction of cellular proliferation. All samples within the 48-well plates were then returned to the 37°C, 5% CO2 incubator for 6 d. On the sixth day, the cells were harvested from wells and washed with warm PBS, as before. The cells were then labeled with Aqua viability stain and incubated with an Ab cocktail containing anti-CD3 APC Cy7, anti-CD4 PE CF 594, and anti-CD8 Pacific blue. The samples were then analyzed by flow cytometry on the LSR II SORP four-laser (405, 488, 532, and 633 nm) instrument. The five different chronic HIV-infected patients used for the CFSE proliferation assay studies were B08, B09, C10, M34, and M114. Assessment of cytokine production Cryopreserved blood mononuclear cells from eight different HIV-infected donors (B08, B09, C10, M34, M114, B02, M24, and M46) were used in studies monitoring cytokine production following Ag-specific stimulations in the presence and absence of anti–PD-1 Abs. The protocol for HIV antigenic peptide stimulations is the same as described above for the proliferation assay with the omission of the CFSE cell staining. Following a 3-d incubation in culture, cell medium supernatants were assessed for levels of INF-γ, TNF-α, IL-2, and IL-10 by Luminex assay using a ProcartaPlex custom human 4-plex kit (Thermo Fisher Scientific) according the manufacturer’s protocol with measurements performed on the Luminex 100 instrument. Purification of hPD1 and the hPD-1–Fab NB01a complex The expression of hPD-1 33–150 (UniProt Q15116), cloned into pET-24d (kindly provided by Krzysztof M. Zak and Tad A. Holak; Faculty of Biochemistry, Biophysics, and Biotechnology, Jagiellonian University, Krakow, Poland), was induced with 1 mM isopropyl β-D-1-thiogalactopyranoside for 5 h at 37°C in Escherichia coli BL21 (DE3) cells, and the protein purification protocol was adapted from Zak et al. (2015). Briefly, induced cells were lysed by sonication with recovery of the inclusion bodies by centrifugation at 20,000 rpm. The inclusion bodies were resuspended in 50 mM Tris, pH 8.0, 6 M guanidine hydrochloride, 200 mM NaCl, and 20 mM β-mercaptoethanol before drop-wise dilution into ice-cold refolding buffer containing 0.1 M Tris-HCl, pH 8.0, 0.4 M L-arginine, 2 mM EDTA, 2 mM reduced glutathione, and 0.2 mM oxidized glutathione. Overnight refolding at 4°C was followed by dialysis and sample loading onto HiTrap Q HP and HiTrap SP FF columns (GE Healthcare). The flow through was concentrated and run in a Superdex 200 10/300 size-exclusion chromatography (GE Healthcare) column. (Fab)2 and Fab species were generated by digestion of the NB01a IgG4 with the FabRICATOR protease at 4°C overnight according to the manufacturer’s protocol with the Fc domain removed using a Protein A affinity column (Thermo Fisher Scientific). (Fab)2 were reduced to Fab with the addition of 2 mM dithiothreitol. Complexes were assembled by mixing hPD-1 with the Fab or reduced (Fab)2 NB01a species in a 1.5:1 molar ratio. After 45 min of incubation at room temperature, the complexes were loaded onto a size-exclusion chromatography Superdex 200 10/300 column equilibrated with 10 mM Tris HCl, 70 mM NaCl, and 2 mM dithiothreitol. Fractions in which the complex eluted were concentrated down to 4.3 mg/ml. Data collection, processing, and refinement of the x-ray structure X-ray data were collected at the European Synchrotron Radiation Facility (ESRF), Grenoble, France, at the ID29 beamline (Table S1). Data were collected at a wavelength of 0.97625 Å with a crystal to detector distance of 398.81 mm and a 0.15° rotation of the crystal to complete a total rotation of 150°. The data were processed with XDS (Kabsch, 2010) using the resolution range 65.49–2.2 Å. The structure of the complex was solved by molecular replacement in Phaser MR (McCoy et al., 2007; Read and McCoy, 2011; Z-score 14.9) using hPD-1 in complex with hPD-L1 (PDB accession no. 4ZQK) and a model built in Phenix using coordinates from Fab BL3-6 (PDB accession no. 4Q9Q; Huang et al., 2014), which exhibits high sequence homology with Fab NB01a (90% and 73% with the heavy and light chains, respectively) as search models. The model of the complex was refined in Phenix (Adams et al., 2010) and Refmac5 (Murshudov et al., 2011) and adjusted manually in Coot (Emsley and Cowtan, 2004) to values of Rfactor = 0.2120 and Rfree= 0.2760. The software Contact, part of the CCP4 suite like Phaser MR and REFMAC5 (Winn et al., 2011) and PDBePISA (Krissinel and Henrick, 2007), were used to identify contacts in the binding interface. Structure representations were prepared using CCP4mg (McNicholas et al., 2011). The data for this study are available under PDB accession no. 6HIG. Modeling of the hPD-1–136B4Fv interaction The model of 136B4 was built using Phenix software and the aligned sequences of 136B4 and the Fv from PDB file 1DZB (Aÿ et al., 2000), which shares 63% identity with 136B4. In addition to the high sequence identity, this PDB file was selected because it comprises an Fv, which is expected to adopt a similar fold to 136B4. Subsequently, models for the light and heavy chains were searched on the PDB to model the CDRs. The heavy and light chains from PDB files 1MVU and 4M61 (Stanfield and Eilat, 2014) share 79% and 96% of identity with those from 136B4, respectively, and the length of the CDR is comparable. The CDR segments were built into the model provided by Phenix software using Coot. Phosphoprotein signaling by flow cytometry Blood mononuclear cells from a viremic HIV-positive donor were incubated with 20 µg/ml of pembrolizumab, NB01b, or IgG4 isotype control for 30 min at 37°C. PDL-1 Fc fusion protein was then added to the indicated test conditions at 20 µg/ml for an additional 30 min. For all conditions except the IgG4 control, anti-CD3/CD28 Ab stimulation was performed at 5 µg/ml of each Ab and cross-linked with an anti-mouse secondary Ab at 50 µg/ml to initiate T cell activation. Cells were incubated at 37°C for various time points (1, 5, and 15 min) and fixed immediately with 2% formaldehyde in PBS at 37°C for 10 min to stop the reaction. Blood mononuclear cells were washed and stained for CD3, CD4, CD8, and CD45RA T cell surface markers at room temperature for 20 min. Following cell permeabilization in chilled 70% methanol for 30 min, intracellular staining with various phosphoproteins was performed. The stained cells were washed and fixed once more in 2% formaldehyde in PBS before FACS acquisition. The percentage of positive phosphoproteins was determined at the different time points following stimulation by applying a global positive gate using the time 0 sample for each experiment as reference. Calcium flux Blood mononuclear cells were washed in PBS and incubated with 1 µM of Fluo-4 AM for 15 min at room temperature. Cells were then washed again and incubated with an anti-CD4/CD8 Ab cocktail for T cell markers. Blood mononuclear cells were washed once more and resuspended in RPMI medium. NB01b, pembrolizumab, and/or nivolumab anti–PD-1 Abs were incubated at 10 µg/ml for 30 min at 37°C and PD-L1 protein was incubated at 5 µg/ml for an additional 30 min before the samples were stimulated with the same anti-CD3/CD28 Ab protocol described above and directly analyzed by FACS acquisition. Calcium flux from TCR stimulation was measured in real time for 10 min before ending the reaction with the addition of 2 mM EDTA, which was acquired for a further 1 min. Immunoprecipitation of the cellular PD-1 protein complex Immunoprecipitation studies used Dynabeads M-450 Epoxy beads (Thermo Fisher Scientific) covalently coupled with either anti–PD-1 or IgG4 control Abs according to the manufacturer’s protocol. For each condition, 10 million Jurkat cells stably expressing PD-1 (BPS Bioscience) were either left unstimulated or stimulated with biotinylated anti-CD3 (5 µg/ml; OKT3; eBioscience) and anti-CD28 (5 µg/ml; CD28.2; eBioscience) then aggregated with the addition of 50 µg/ml of avidin (Invitrogen). Ab-coated beads were added directly and cells were incubated at 37°C for 10 min. Cell membranes where then lysed with ice cold PBS containing 1% Triton X-100, protease inhibitor cocktail (Complete; Roche) and a phosphatase inhibitor cocktail followed by a 15-min incubation on ice with frequent mixing. Magnetic beads were attracted to the side of each tube with a magnet and washed three times with cell lysis buffer and then once with lysis buffer without detergent. Immune complexes analyzed by Western blot had proteins separated on a NuPAGE 10% Bis-Tris gel (Invitrogen) and then transferred onto nitrocellulose (Bio-Rad). Immunoblotting was performed by staining with the primary Ab overnight at 4°C followed by washing steps and then incubation with the HRP-conjugated secondary Ab and detection with the Pierce ECL Western blotting substrate (Thermo Fisher Scientific) in accordance with the manufacturer’s protocol. Blots were imaged with the Fusion FX Vilber Lurmat camera (Witec), and the density of individual bands was evaluated using ImageJ software. Jurkat PD-1 NFAT and NF-κB reporter assays The Jurkat PD-1 NFAT reporter assay was performed according the manufactures protocol with minor modifications (BPS Bioscience). 2 d before performing the assay, Jurkat PD-1 NFAT cells were transiently transfected with the NanoLuc reporter vector with an NF-κB response element (Promega) using the Fugene 6 transfection reagent (Promega). Preparation of the activator cells involved transient cotransfected of 293T cells with the TCR activator and PDL-1 expression vectors using the Fugene 6 transfection reagent. The following day, cells were resuspended and seeded at 60–70% confluence in wells of a 96-well plate and then incubated for 5 h to allow cell adherence. The NanoLuc transfected Jurkat PD-1 NFAT reporter cells were added at 100,000 cells per well in the absence of presence of the indicated Abs. The NFAT and NF-κB activation was measured 18 h later using the Nano-Glo Dual-luciferase Reporter assay system (Promega) on the Synergy H1 Hybrid Multi-Mode Microplate Reader (BioTek Instruments). Mouse tumor model The in vivo tumor model studies using PD-1 HuGEMM mice (background: C57BL/6) were performed at CrownBio (Taicang, China) in three separate MC38 studies and one B16F10 tumor study. Each mouse was inoculated subcutaneously at the right hind flank with 106 MC38 cells or 105 B16F10 cells. Mice with successfully engrafted tumors of the desired sized were enrolled in the studies with 10 mice per arm with mean tumor volumes of 82, 84, and 85 mm3 for the first, second, and third MC38 tumor study, respectively. Mice enrolled for the B16F10 study had mean tumor volumes of 65.1 mm3 for the 10 mice per arm of the study. Mice were administered the indicated Abs at a concentration of 0.9 to 1.0 mg/ml in PBS or a sterile PBS vehicle control at twice weekly intervals over the course of the study. All in vivo therapies used a total dose of 10 mg/kg Ab apart from the combination therapy arm of the B16F10 study, where 10 mg/kg of pembrolizumab and NB01b were administered. Tumor volumes were measured twice weekly in two dimensions using a caliper, and the volume was expressed in cubic millimeters using the formula V = 0.5 a × b2, where a and b are the long and short diameters of the tumor, respectively. Body weight was also measured twice weekly. Mice that controlled MC38 tumor growth had a decrease in tumor volume at the end of the study relative to the start, just before initiating Ab therapy. The protocol and any amendments or procedures involving the care and use of animals in this study were reviewed and approved by the Institutional Animal Care and Use Committee of CrownBio before conduct. During the study, the care and use of animals was conducted in accordance with the regulations of the Association for Assessment and Accreditation of Laboratory Animal Care. Statistical analyses Statistical significance (P values) between two sample conditions in the in vitro functional assay was obtained using two-tailed unpaired t tests with a Welch correction to account for samples with nonequivalent standard deviations. In the mouse study, a mixed-effects linear model, modeling the volume (after cubic-root transformation) as a function of group, time, and their interaction, was used to test whether the combination therapy had lower mean volume over time than any single therapy. Our model allows for a group-specific slope, taking into account the correlation within mice across time (due to the repeated measurements) and using a mice-specific random effect, which is common for such studies (Galecki and Burzykowski, 2013). The cubic-root transformation was used to stabilize the variance and make the data more normally distributed. It is also a natural transformation for volume measurements as the scale is easily interpretable. Since the original unit is a volume expressed in cubic millimeters, the transformed unit is a length expressed in millimeters. The cube-root transformation is often used to model volume data (Matthews et al., 1990; Gallegos et al., 1996; Mandonnet et al., 2003). Using our model, we tested the null hypothesis that the slope of the combination therapy is equal to the slope of the anti–PD-1 monotherapies alone versus the alternative that the slope is smaller using a two-sided test. In our model, the slope can be interpreted as the rate of increase of the mean tumor diameter over time. Online supplemental material Fig. S1 shows the enhanced proliferation of Ag-specific exhausted CD8 T cells in the presence of a panel of prioritized anti–PD-1 Abs. Fig. S2 shows four separate experiments where blocking and nonblocking anti–PD-1 Ab combinations synergize in recovering the proliferation of exhausted Ag-specific CD8 T cells. Fig. S3 shows a PD-1 sequence map of the mutations used for Ab epitope mapping and the sequence alignment of PD-1 from different species. Fig. S4 shows the PD-1–PDL-1–mediated suppression in phosphosignaling following stimulation of exhausted T cells and the relief of this suppression mediated by anti–PD-1 Abs. Table S1 shows the x-ray data statistics for the hPD-1–NB01a Fab complex. Supplementary Material Supplemental Materials (PDF) Acknowledgments We are grateful to N. Grandchamp, P. Pochon, X. Bron, M. Graff, A. Crétignier, R. Mamin, and C. André for technical assistance. We are also grateful to D. Hacker at the Ecole Polytechnique Fédérale de Lausanne for the production and purification of the nonblocking anti–PD-1 Abs used in this study. W. Weissenhorn acknowledges support from the Institut Universitaire de France, the platforms of the Grenoble Instruct-ERIC Center (Integrated Structural Biology Grenoble; UMS 3518 CNRS-CEA-UJF-EMBL) supported by the French Infrastructure for Integrated Structural Biology Initiative (ANR-10-INSB-05-02) and Grenoble Alliance for Integrated Structural Cell Biology (ANR-10-LABX-49-01) within the Grenoble Partnership for Structural Biology, the European Synchrotron Research Facility and European Molecular Biology Laboratory Joint Structural Biology Group for access and support at the ESRF beam lines, and J. Marquez (European Molecular Biology Laboratory) from the crystallization platform. G. Pantaleo and C. Fenwick are cofounders of MabQuest SA, which owns the patent rights to the novel Abs described in this report (WO 2016/020856 A2, US patent number 9,982,052 B2 and WO 2017/125815A2). The remaining authors declare no competing financial interests. Author contributions: V. Joo performed the signaling studies; C. Pellaton, T. Decaillon, and A. Noto performed the CFSE proliferation and cytokine assays; and A. Farina, L. Esteves-Leuenberger, and N. Rajah performed the remaining experiments. J.-L. Loredo-Varela and W. Weissenhorn designed and performed the crystallography and structural modeling studies. M. Suffiotti, K. Ohmiti, and R. Gottardo performed statistical analyses. G. Pantaleo and C. Fenwick conceived the study, designed the experiments, and wrote the manuscript. ==== Refs Adams, P.D., P.V. Afonine, G. Bunkóczi, V.B. Chen, I.W. Davis, N. Echols, J.J. Headd, L.W. Hung, G.J. Kapral, R.W. Grosse-Kunstleve, . 2010. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D Biol. Crystallogr. 66 :213–221. 10.1107/S0907444909052925 20124702 Aÿ, J., T. Keitel, G. Küttner, H. Wessner, C. Scholz, M. Hahn, and W. Höhne. 2000. Crystal structure of a phage library-derived single-chain Fv fragment complexed with turkey egg-white lysozyme at 2.0 A resolution. J. Mol. Biol. 301 :239–246. 10.1006/jmbi.2000.3971 10926506 Baitsch, L., P. Baumgaertner, E. Devêvre, S.K. Raghav, A. Legat, L. Barba, S. Wieckowski, H. Bouzourene, B. Deplancke, P. Romero, . 2011. Exhaustion of tumor-specific CD8+ T cells in metastases from melanoma patients. J. Clin. Invest. 121 :2350–2360. 10.1172/JCI46102 21555851 Barber, D.L., E.J. Wherry, D. Masopust, B. Zhu, J.P. Allison, A.H. Sharpe, G.J. Freeman, and R. Ahmed. 2006. Restoring function in exhausted CD8 T cells during chronic viral infection. Nature. 439 :682–687. 10.1038/nature04444 16382236 Callahan, M.K., M.A. Postow, and J.D. Wolchok. 2016. Targeting T Cell Co-receptors for Cancer Therapy. Immunity. 44 :1069–1078. 10.1016/j.immuni.2016.04.023 27192570 Carbognin, L., S. Pilotto, M. Milella, V. Vaccaro, M. Brunelli, A. Caliò, F. Cuppone, I. Sperduti, D. Giannarelli, M. Chilosi, . 2015. Differential Activity of Nivolumab, Pembrolizumab and MPDL3280A according to the Tumor Expression of Programmed Death-Ligand-1 (PD-L1): Sensitivity Analysis of Trials in Melanoma, Lung and Genitourinary Cancers. PLoS One. 10 :e0130142. 10.1371/journal.pone.0130142 26086854 Chemnitz, J.M., R.V. Parry, K.E. Nichols, C.H. June, and J.L. Riley. 2004. SHP-1 and SHP-2 associate with immunoreceptor tyrosine-based switch motif of programmed death 1 upon primary human T cell stimulation, but only receptor ligation prevents T cell activation. J. Immunol. 173 :945–954. 10.4049/jimmunol.173.2.945 15240681 Chen, S., L.F. Lee, T.S. Fisher, B. Jessen, M. Elliott, W. Evering, K. Logronio, G.H. Tu, K. Tsaparikos, X. Li, . 2015. Combination of 4-1BB agonist and PD-1 antagonist promotes antitumor effector/memory CD8 T cells in a poorly immunogenic tumor model. Cancer Immunol. Res. 3 :149–160. 10.1158/2326-6066.CIR-14-0118 25387892 Day, C.L., D.E. Kaufmann, P. Kiepiela, J.A. Brown, E.S. Moodley, S. Reddy, E.W. Mackey, J.D. Miller, A.J. Leslie, C. DePierres, . 2006. PD-1 expression on HIV-specific T cells is associated with T-cell exhaustion and disease progression. Nature. 443 :350–354. 10.1038/nature05115 16921384 Emsley, P., and K. Cowtan. 2004. Coot: model-building tools for molecular graphics. Acta Crystallogr. D Biol. Crystallogr. 60 :2126–2132. 10.1107/S0907444904019158 15572765 Galecki, A., and T. Burzykowski. 2013. Linear Mixed-Effects Models Using R: A Step-by-Step Approach. Springer, New York. 542 pp. 10.1007/978-1-4614-3900-4 Gallegos, A., J.R. Gasdaska, C.W. Taylor, G.D. Paine-Murrieta, D. Goodman, P.Y. Gasdaska, M. Berggren, M.M. Briehl, and G. Powis. 1996. Transfection with human thioredoxin increases cell proliferation and a dominant-negative mutant thioredoxin reverses the transformed phenotype of human breast cancer cells. Cancer Res. 56 :5765–5770.8971189 Gandini, S., D. Massi, and M. Mandalà. 2016. PD-L1 expression in cancer patients receiving anti PD-1/PD-L1 antibodies: A systematic review and meta-analysis. Crit. Rev. Oncol. Hematol. 100 :88–98. 10.1016/j.critrevonc.2016.02.001 26895815 Hamid, O., C. Robert, A. Daud, F.S. Hodi, W.J. Hwu, R. Kefford, J.D. Wolchok, P. Hersey, R.W. Joseph, J.S. Weber, . 2013. Safety and tumor responses with lambrolizumab (anti-PD-1) in melanoma. N. Engl. J. Med. 369 :134–144. 10.1056/NEJMoa1305133 23724846 Huang, A., D. Peng, H. Guo, Y. Ben, X. Zuo, F. Wu, X. Yang, F. Teng, Z. Li, X. Qian, and F.X. Qin. 2017. A human programmed death-ligand 1-expressing mouse tumor model for evaluating the therapeutic efficacy of anti-human PD-L1 antibodies. Sci. Rep. 7 :42687. 10.1038/srep42687 28202921 Huang, H., N.B. Suslov, N.S. Li, S.A. Shelke, M.E. Evans, Y. Koldobskaya, P.A. Rice, and J.A. Piccirilli. 2014. A G-quadruplex-containing RNA activates fluorescence in a GFP-like fluorophore. Nat. Chem. Biol. 10 :686–691. 10.1038/nchembio.1561 24952597 Hui, E., J. Cheung, J. Zhu, X. Su, M.J. Taylor, H.A. Wallweber, D.K. Sasmal, J. Huang, J.M. Kim, I. Mellman, and R.D. Vale. 2017. T cell costimulatory receptor CD28 is a primary target for PD-1-mediated inhibition. Science. 355 :1428–1433. 10.1126/science.aaf1292 28280247 Kabsch, W. 2010. Xds. Acta Crystallogr. D Biol. Crystallogr. 66 :125–132. 10.1107/S0907444909047337 20124692 Kamphorst, A.O., A. Wieland, T. Nasti, S. Yang, R. Zhang, D.L. Barber, B.T. Konieczny, C.Z. Daugherty, L. Koenig, K. Yu, . 2017. Rescue of exhausted CD8 T cells by PD-1-targeted therapies is CD28-dependent. Science. 355 :1423–1427. 10.1126/science.aaf0683 28280249 Kokolus, K.M., Y. Zhang, J.M. Sivik, C. Schmeck, J. Zhu, E.A. Repasky, J.J. Drabick, and T.D. Schell. 2017. Beta blocker use correlates with better overall survival in metastatic melanoma patients and improves the efficacy of immunotherapies in mice. OncoImmunology. 7 :e1405205. 10.1080/2162402X.2017.1405205 29399407 Krissinel, E., and K. Henrick. 2007. Inference of macromolecular assemblies from crystalline state. J. Mol. Biol. 372 :774–797. 10.1016/j.jmb.2007.05.022 17681537 Kuzu, O.F., F.D. Nguyen, M.A. Noory, and A. Sharma. 2015. Current State of Animal (Mouse) Modeling in Melanoma Research. Cancer Growth Metastasis. 8 (Suppl 1 ):81–94.26483610 Larkin, J., V. Chiarion-Sileni, R. Gonzalez, J.J. Grob, C.L. Cowey, C.D. Lao, D. Schadendorf, R. Dummer, M. Smylie, P. Rutkowski, . 2015. Combined Nivolumab and Ipilimumab or Monotherapy in Untreated Melanoma. N. Engl. J. Med. 373 :23–34. 10.1056/NEJMoa1504030 26027431 Lázár-Molnár, E., Q. Yan, E. Cao, U. Ramagopal, S.G. Nathenson, and S.C. Almo. 2008. Crystal structure of the complex between programmed death-1 (PD-1) and its ligand PD-L2. Proc. Natl. Acad. Sci. USA. 105 :10483–10488. 10.1073/pnas.0804453105 18641123 Lee, J.Y., H.T. Lee, W. Shin, J. Chae, J. Choi, S.H. Kim, H. Lim, T. Won Heo, K.Y. Park, Y.J. Lee, . 2016. Structural basis of checkpoint blockade by monoclonal antibodies in cancer immunotherapy. Nat. Commun. 7 :13354. 10.1038/ncomms13354 27796306 Lin, D.Y., Y. Tanaka, M. Iwasaki, A.G. Gittis, H.P. Su, B. Mikami, T. Okazaki, T. Honjo, N. Minato, and D.N. Garboczi. 2008. The PD-1/PD-L1 complex resembles the antigen-binding Fv domains of antibodies and T cell receptors. Proc. Natl. Acad. Sci. USA. 105 :3011–3016. 10.1073/pnas.0712278105 18287011 Lute, K.D., K.F. May Jr., P. Lu, H. Zhang, E. Kocak, B. Mosinger, C. Wolford, G. Phillips, M.A. Caligiuri, P. Zheng, and Y. Liu. 2005. Human CTLA4 knock-in mice unravel the quantitative link between tumor immunity and autoimmunity induced by anti-CTLA-4 antibodies. Blood. 106 :3127–3133. 10.1182/blood-2005-06-2298 16037385 Mandonnet, E., J.Y. Delattre, M.L. Tanguy, K.R. Swanson, A.F. Carpentier, H. Duffau, P. Cornu, R. Van Effenterre, E.C. Alvord Jr., and L. Capelle. 2003. Continuous growth of mean tumor diameter in a subset of grade II gliomas. Ann. Neurol. 53 :524–528. 10.1002/ana.10528 12666121 Matthews, J.N., D.G. Altman, M.J. Campbell, and P. Royston. 1990. Analysis of serial measurements in medical research. BMJ. 300 :230–235. 10.1136/bmj.300.6719.230 2106931 McCoy, A.J., R.W. Grosse-Kunstleve, P.D. Adams, M.D. Winn, L.C. Storoni, and R.J. Read. 2007. Phaser crystallographic software. J. Appl. Cryst. 40 :658–674. 10.1107/S0021889807021206 19461840 McNicholas, S., E. Potterton, K.S. Wilson, and M.E. Noble. 2011. Presenting your structures: the CCP4mg molecular-graphics software. Acta Crystallogr. D Biol. Crystallogr. 67 :386–394. 10.1107/S0907444911007281 21460457 Mellman, I., G. Coukos, and G. Dranoff. 2011. Cancer immunotherapy comes of age. Nature. 480 :480–489. 10.1038/nature10673 22193102 Murshudov, G.N., P. Skubák, A.A. Lebedev, N.S. Pannu, R.A. Steiner, R.A. Nicholls, M.D. Winn, F. Long, and A.A. Vagin. 2011. REFMAC5 for the refinement of macromolecular crystal structures. Acta Crystallogr. D Biol. Crystallogr. 67 :355–367. 10.1107/S0907444911001314 21460454 Pantaleo, G., and C. Fenwick. 2016. Immunological reagents. MabQuest SA patent application WO/2016/020856 A2, filed August 5, 2015, and published February 11, 2016. Parry, R.V., J.M. Chemnitz, K.A. Frauwirth, A.R. Lanfranco, I. Braunstein, S.V. Kobayashi, P.S. Linsley, C.B. Thompson, and J.L. Riley. 2005. CTLA-4 and PD-1 receptors inhibit T-cell activation by distinct mechanisms. Mol. Cell. Biol. 25 :9543–9553. 10.1128/MCB.25.21.9543-9553.2005 16227604 Patsoukis, N., J. Brown, V. Petkova, F. Liu, L. Li, and V.A. Boussiotis. 2012. Selective effects of PD-1 on Akt and Ras pathways regulate molecular components of the cell cycle and inhibit T cell proliferation. Sci. Signal. 5 :ra46. 10.1126/scisignal.2002796 22740686 Patsoukis, N., L. Li, D. Sari, V. Petkova, and V.A. Boussiotis. 2013. PD-1 increases PTEN phosphatase activity while decreasing PTEN protein stability by inhibiting casein kinase 2. Mol. Cell. Biol. 33 :3091–3098. 10.1128/MCB.00319-13 23732914 Pauken, K.E., and E.J. Wherry. 2015. Overcoming T cell exhaustion in infection and cancer. Trends Immunol. 36 :265–276. 10.1016/j.it.2015.02.008 25797516 Perreau, M., and E.J. Kremer. 2005. Frequency, proliferation, and activation of human memory T cells induced by a nonhuman adenovirus. J. Virol. 79 :14595–14605. 10.1128/JVI.79.23.14595-14605.2005 16282459 Read, R.J., and A.J. McCoy. 2011. Using SAD data in Phaser. Acta Crystallogr. D Biol. Crystallogr. 67 :338–344. 10.1107/S0907444910051371 21460452 Rizvi, N.A., J. Mazières, D. Planchard, T.E. Stinchcombe, G.K. Dy, S.J. Antonia, L. Horn, H. Lena, E. Minenza, B. Mennecier, . 2015. Activity and safety of nivolumab, an anti-PD-1 immune checkpoint inhibitor, for patients with advanced, refractory squamous non-small-cell lung cancer (CheckMate 063): a phase 2, single-arm trial. Lancet Oncol. 16 :257–265. 10.1016/S1470-2045(15)70054-9 25704439 Sharma, P., and J.P. Allison. 2015. The future of immune checkpoint therapy. Science. 348 :56–61. 10.1126/science.aaa8172 25838373 Sheppard, K.A., L.J. Fitz, J.M. Lee, C. Benander, J.A. George, J. Wooters, Y. Qiu, J.M. Jussif, L.L. Carter, C.R. Wood, and D. Chaudhary. 2004. PD-1 inhibits T-cell receptor induced phosphorylation of the ZAP70/CD3zeta signalosome and downstream signaling to PKCtheta. FEBS Lett. 574 :37–41. 10.1016/j.febslet.2004.07.083 15358536 Stanfield, R.L., and D. Eilat. 2014. Crystal structure determination of anti-DNA Fab A52. Proteins. 82 :1674–1678. 10.1002/prot.24514 24449198 Topalian, S.L., F.S. Hodi, J.R. Brahmer, S.N. Gettinger, D.C. Smith, D.F. McDermott, J.D. Powderly, R.D. Carvajal, J.A. Sosman, M.B. Atkins, . 2012. Safety, activity, and immune correlates of anti-PD-1 antibody in cancer. N. Engl. J. Med. 366 :2443–2454. 10.1056/NEJMoa1200690 22658127 Trautmann, L., L. Janbazian, N. Chomont, E.A. Said, S. Gimmig, B. Bessette, M.R. Boulassel, E. Delwart, H. Sepulveda, R.S. Balderas, . 2006. Upregulation of PD-1 expression on HIV-specific CD8+ T cells leads to reversible immune dysfunction. Nat. Med. 12 :1198–1202. 10.1038/nm1482 16917489 Tumeh, P.C., C.L. Harview, J.H. Yearley, I.P. Shintaku, E.J. Taylor, L. Robert, B. Chmielowski, M. Spasic, G. Henry, V. Ciobanu, . 2014. PD-1 blockade induces responses by inhibiting adaptive immune resistance. Nature. 515 :568–571. 10.1038/nature13954 25428505 Vajpayee, M., S. Kaushik, V. Sreenivas, N. Wig, and P. Seth. 2005. CDC staging based on absolute CD4 count and CD4 percentage in an HIV-1-infected Indian population: treatment implications. Clin. Exp. Immunol. 141 :485–490. 10.1111/j.1365-2249.2005.02857.x 16045738 Wherry, E.J., and M. Kurachi. 2015. Molecular and cellular insights into T cell exhaustion. Nat. Rev. Immunol. 15 :486–499. 10.1038/nri3862 26205583 Winn, M.D., C.C. Ballard, K.D. Cowtan, E.J. Dodson, P. Emsley, P.R. Evans, R.M. Keegan, E.B. Krissinel, A.G. Leslie, A. McCoy, . 2011. Overview of the CCP4 suite and current developments. Acta Crystallogr. D Biol. Crystallogr. 67 :235–242. 10.1107/S0907444910045749 21460441 Woods, D.M., A.L. Sodré, A. Villagra, A. Sarnaik, E.M. Sotomayor, and J. Weber. 2015. HDAC Inhibition Upregulates PD-1 Ligands in Melanoma and Augments Immunotherapy with PD-1 Blockade. Cancer Immunol. Res. 3 :1375–1385. 10.1158/2326-6066.CIR-15-0077-T 26297712 Yokosuka, T., M. Takamatsu, W. Kobayashi-Imanishi, A. Hashimoto-Tane, M. Azuma, and T. Saito. 2012. Programmed cell death 1 forms negative costimulatory microclusters that directly inhibit T cell receptor signaling by recruiting phosphatase SHP2. J. Exp. Med. 209 :1201–1217. 10.1084/jem.20112741 22641383 Zak, K.M., R. Kitel, S. Przetocka, P. Golik, K. Guzik, B. Musielak, A. Dömling, G. Dubin, and T.A. Holak. 2015. Structure of the Complex of Human Programmed Death 1, PD-1, and Its Ligand PD-L1. Structure. 23 :2341–2348. 10.1016/j.str.2015.09.010 26602187 Zamarin, D., J.M. Ricca, S. Sadekova, A. Oseledchyk, Y. Yu, W.M. Blumenschein, J. Wong, M. Gigoux, T. Merghoub, and J.D. Wolchok. 2018. PD-L1 in tumor microenvironment mediates resistance to oncolytic immunotherapy. J. Clin. Invest. 128 :5184. 10.1172/JCI125039 Zhang, J.Y., Z. Zhang, X. Wang, J.L. Fu, J. Yao, Y. Jiao, L. Chen, H. Zhang, J. Wei, L. Jin, . 2007. PD-1 up-regulation is correlated with HIV-specific memory CD8+ T-cell exhaustion in typical progressors but not in long-term nonprogressors. Blood. 109 :4671–4678. 10.1182/blood-2006-09-044826 17272504

PMC006xxxxxx/PMC6719409.txt

==== Front J Gen Physiol J Gen Physiol jgp jgp The Journal of General Physiology 0022-1295 1540-7748 Rockefeller University Press 31409663 201912421 10.1085/jgp.201912421 Research Articles Communications 511 508 501 Triple arginines as molecular determinants for pentameric assembly of the intracellular domain of 5-HT3A receptors Triple arginines determine pentamerization and conductance Pandhare Akash * https://orcid.org/0000-0001-8098-9549 Pirayesh Elham * Stuebler Antonia G. https://orcid.org/0000-0003-1325-9700 Jansen Michaela Department of Cell Physiology and Molecular Biophysics and Center for Membrane Protein Research, School of Medicine, Texas Tech University Health Sciences Center, Lubbock, TX Correspondence to Michaela Jansen: michaela.jansen@ttuhsc.edu * A. Pandhare and E. Pirayesh contributed equally to this paper. 02 9 2019 13 8 2019 151 9 11351145 13 6 2019 15 7 2019 © 2019 Pandhare et al. 2019 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). Serotonin type 3A receptors are homopentameric ligand-gated ion channels that are thought to assemble via interactions involving the subunits’ extracellular and transmembrane domains. Pandhare et al. reveal that channel assembly is also determined by three arginine residues in the receptor’s intracellular domain. Serotonin type 3 receptors (5-HT3Rs) are cation-conducting pentameric ligand-gated ion channels and members of the Cys-loop superfamily in eukaryotes. 5-HT3Rs are found in the peripheral and central nervous system, and they are targets for drugs used to treat anxiety, drug dependence, and schizophrenia, as well as chemotherapy-induced and postoperative nausea and emesis. Decades of research of Cys-loop receptors have identified motifs in both the extracellular and transmembrane domains that mediate pentameric assembly. Those efforts have largely ignored the most diverse domain of these channels, the intracellular domain (ICD). Here we identify molecular determinants within the ICD of serotonin type 3A (5-HT3A) subunits for pentameric assembly by first identifying the segments contributing to pentamerization using deletion constructs of, and finally by making defined amino acid substitutions within, an isolated soluble ICD. Our work provides direct experimental evidence for the contribution of three intracellular arginines, previously implicated in governing the low conductance of 5-HT3ARs, in structural features such as pentameric assembly. National Institute of Neurological Disorders and Stroke https://doi.org/10.13039/100000065 National Institutes of Health https://doi.org/10.13039/100000002,%22National%20Institutes%20of%20Health%22 R01NS077114 ==== Body pmcIntroduction Serotonin type 3 receptors (5-HT3Rs) are cation-conducting pentameric ligand-gated ion channels (pLGICs), also known as the Cys-loop superfamily in eukaryotes. 5-HT3Rs were first discovered in the peripheral nervous system in the gut (Gaddum and Picarelli, 1957). Only much later was it identified that agents developed for the treatment of chemotherapy-induced emesis elicited behavioral effects in rodents indicative of a central nervous system effect. These receptors play a key role in the process of rapid excitatory neurotransmission in the human brain (Yang, 1990; Hargreaves et al., 1994). 5-HT3Rs are targeted by many therapeutic drugs currently prescribed for the management of cancer chemotherapy–induced vomiting (Aapro, 1991) as well as depression (Pandhare et al., 2017). They are also implicated as potential targets for the treatment of some neurological and psychiatric diseases and disorders (Lummis, 2012). The Cys-loop superfamily additionally includes neuronal- and muscle-type nicotinic acetylcholine receptors (nAChRs), glycine receptors, and γ–aminobutyric acid type A receptors (Changeux and Edelstein, 1998; Miller and Smart, 2010). Cys-loop receptors assemble from five homologous subunits to form homo- or heteropentamers. Each subunit contains three domains: (1) an N-terminal extracellular domain (ECD) mainly structured into two antiparallel β-sheets that harbors the ligand-binding site at subunit interfaces, (2) a transmembrane domain (TMD) consisting of four α-helical segments that shape the ion channel, and (3) a variable intracellular domain (ICD; Forman et al., 2015). Recently, there has been remarkable progress in the determination of atomic resolution 3-D structures of a number of eukaryotic anion- (Huang et al., 2015; Phulera et al., 2018; Zhu et al., 2018; Laverty et al., 2019) and cation-conducting (Hassaine et al., 2014; Morales-Perez et al., 2016) members of the Cys-loop superfamily. As a result, our structural understanding of subunit assembly, allosteric regulation, and conformational transitions associated with gating has improved significantly (Nury et al., 2011; Spurny et al., 2013; Du et al., 2015; Huang et al., 2015, 2017; Polovinkin et al., 2018; Zhu et al., 2018; Masiulis et al., 2019). While significant advances with regard to the structure and function of the ECD and the TMD have been achieved, structure determination of the ICD has remained elusive. The ICD of 5-HT3ARs (5-HT3A–ICD) begins at the C-terminus of M3 as a loop (L1; Fig. 1) and a short α-helix (MX) followed by a stretch of ∼60 amino acids forming an intrinsically disordered region (L2) and a membrane-associated α-helix (MA). The MA helix forms a continuous helix with M4 at the cytoplasmic leaflet of the membrane bilayer. Pioneering EM studies on the cation-conducting Torpedo nAChR reported the presence of the MA helix in these receptors (Unwin, 2005). Similar helical content pre-M4 is predicted for most other cation-conducting Cys-loop receptors but absent from anion-conducting superfamily members (Fig. S1). On the contrary, α-helical content corresponding to the MX helix is predicted for most anion- and cation-conducting channels. Overall, the lengths and sequences of ICDs are remarkably dissimilar between different subunits. The discovery that replacement of the ICD with a short linker for both anion- and cation-conducting Cys-loop receptors led to functional channels (Jansen et al., 2008) inspired a series of studies using the same or similar modifications for functional (Bar-Lev et al., 2011; McKinnon et al., 2011, 2012; Moroni et al., 2011; Unterer et al., 2012) and structural (Hibbs and Gouaux, 2011; Bondarenko et al., 2012, 2014; Mowrey et al., 2013; Hassaine et al., 2014; Miller and Aricescu, 2014; Morales-Perez et al., 2016) studies. Recent cryo-EM studies using full-length 5-HT3ARs resolved L1-MX and the MA helix (Basak et al., 2018; Polovinkin et al., 2018), similar to a study where the ICD had been proteolyzed before crystallization and x-ray structure determination (Hassaine et al., 2014). The inability to resolve L2 likely stems from its inherently “very dynamic” nature (Polovinkin et al., 2018). Therefore, structural, albeit incomplete, insights into the ICD are limited to neuronal and Torpedo nAChRs and 5-HT3ARs, and entirely lacking for anion-conducting Cys-loop receptors (Unwin, 2005; Hassaine et al., 2014; Morales-Perez et al., 2016). Even though full-length γ–aminobutyric acid type A constructs were used for recent cryo-EM structure determination, only a few amino acids of the ICD without defined secondary structure could be resolved (Phulera et al., 2018; Laverty et al., 2019; Masiulis et al., 2019). This provides the first experimental indication for the ICD of anion-conducting pLGICs being significantly different from cation-conducting channels. Figure 1. Structural representation and alignments of constructs. (A) Cartoon representation of the pentameric mouse 5-HT3A receptor x-ray structure (PDB ID: 4PIR; Hassaine et al., 2014) viewed parallel to the plane of the membrane. ECD and TMD in green, MA and MX helices in blue and cyan. (B) A single subunit of the 5-HT3A receptor viewed parallel to the membrane. The 62 residues not resolved in L2 are depicted as a dashed line. (C) Cartoon representation of MBP–5-HT3A–ICD–WT. All structural models of the fusion proteins are hypothetical models. The cytoplasmic domain of 5-HT3AR is fused to the C-terminus of MBP and serves as the template for the chimeras. (D) The chimera MBP-ΔMA has the entire MA helix removed. (E) MBP-MA has the MA helix attached to the C-terminus of MBP. (F) MBP-Δ44 contains a 44-amino acid deletion in L2. (G) MBP-QDA contains a triple mutation in its MA helix and no deletions. (H) Multiple sequence alignment of 5-HT3AR and the ICD chimeras highlighting deletions (dashed lines) or mutations (red). The ICD is involved in determining ion conductance (Kelley et al., 2003; Jansen et al., 2008) and modulation by drugs (Moraga-Cid et al., 2011). Additionally, it interacts with various entities of the intracellular apparatus, which facilitate receptor sorting, assembly, trafficking, and anchorage (Thompson et al., 2010; Nishtala et al., 2016). Studies of subunit assembly and oligomerization have investigated the individual contributions of the ECD (Verrall and Hall, 1992; Kouvatsos et al., 2016) as well as the TMD (Bondarenko et al., 2012; Mowrey et al., 2013) at the molecular level. While both the ECD and the TMD can assemble individually into pentamers (Wells et al., 1998; Liu et al., 2008; Bondarenko et al., 2012, 2014), formation of hexamers of the ECD alone for GLIC (Nury et al., 2010) and nonnative stoichiometries for α2β4 nAChR (Bondarenko et al., 2012) exemplify that additional drivers outside these domains may mediate precise subunit stoichiometries and arrangements observed in nature. Intriguingly, this points toward the highly diverse ICD as a mediator for defined oligomerization and pentamerization, in general. Indeed, we showed for the first time that the 5-HT3A-ICD alone can assemble into highly stable pentamers in the absence of the ECD and the TMD (Pandhare et al., 2016). This remarkable experimental observation led us to investigate and uncover the molecular determinants for pentameric assembly of the 5-HT3A-ICD. In the present study, we introduced defined deletions over the entire span of the ICD, and finally a set of amino acid substitutions, to determine the motif necessary for its pentameric assembly. Materials and methods The following materials were used: BL21-CodonPlus-(DE3)-RIPL cells (Agilent Technologies), ampicillin (Fisher Scientific), chloramphenicol (Fisher Scientific), isopropyl β-D-thiogalactoside (Fisher Scientific), leupeptin (AdipoGen Life Sciences), pepstatin (AdipoGen Life Sciences), phenylmethylsulfonyl fluoride (PMSF; Research Products International), tris(2-carboxyethyl)phosphine hydrochloride (TCEP; Oakwood Chemical), lysozyme (MP Biomedicals), protease inhibitor cocktail III (Research Products International), and DNase I (Alfa Aesar). Molecular biology The intracellular domain of the mouse 5-HT3AR (NCBI Protein database accession no. Q8K1F4) was generated as a fusion construct with N-terminal maltose-binding protein (MBP) as we described previously (Pandhare et al., 2019) using the pMAL-c2x (New England Biolabs) variant pMALX (Moon et al., 2010). Based on this MBP–5-HT3A–ICD template containing the entire WT ICD, deletions and substitutions were generated using the QuikChange II Site-Directed Mutagenesis kit (Agilent Technologies) and confirmed by DNA sequencing (GENEWIZ). The resulting amino acid sequences for all constructs are displayed in Fig. 1. Expression and purification of the MBP–5-HT3A–ICD constructs All plasmids for expression of MBP–5-HT3A–ICD constructs were transformed into Escherichia coli BL21-CodonPlus-(DE3)-RIPL cells (E. coli; Agilent Technologies). Cells were grown in Terrific Broth medium (1.2% [wt/vol] tryptone, 2.4% [wt/vol] yeast extract, and 0.4% [vol/vol] glycerol) supplemented with ampicillin (100 µg/ml), chloramphenicol (34 µg/ml), and 0.2% (wt/vol) sterile glucose, at 37°C and 250 rpm, in a shaking incubator. All concentrations indicated are final concentrations unless otherwise stated. At OD600 of 0.4–0.5, the cultures were induced by adding 0.4 mM isopropyl β-D-thiogalactoside and allowed to continue growth at 18°C for an additional 8 h. The cells were harvested by centrifugation at 4,600 g and 4°C for 15 min and then resuspended (10 ml buffer/g of the cell pellet) in buffer A (20 mM Tris, pH 7.4; 200 mM NaCl, 1 mM TCEP, and 2 mM EDTA) enriched with a freshly prepared protease inhibitor cocktail containing leupeptin (10 µg/ml), pepstatin (10 µg/ml), and 1 mM PMSF. After the cells were disrupted by treatment with lysozyme (100 µg/ml) and a freeze–thaw cycle, the lysate was clarified by ultracentrifugation at 100,000 g and 4°C for 1 h. The supernatant was passed through a 0.2-µm pore-size filter and loaded onto a pre-equilibrated amylose resin column. After a 30–bed volume wash of buffer A, proteins were eluted with 20 mM maltose. Fractions were analyzed after separation in stain-free precast SDS-PAGE gels (4–20% Mini-PROTEIN TGX Stain-Free; Biorad). These stain-free gels contain trihalo compounds that undergo UV-induced covalent modification of tryptophan residues for subsequent fluorescent detection. Fractions containing purified protein were pooled and concentrated (up to ∼5 mg/ml) using a 50- or 100-kD molecular weight cut-off centrifugal filter (Amicon Ultra-15; Merck Millipore Ltd.) for an additional purification step by size exclusion chromatography (SEC). SEC The amylose column–purified and concentrated protein samples were passed through an ENrich SEC 650 10 × 300 high-resolution column (Biorad) pre-equilibrated with buffer B (20 mM HEPES, 150 mM NaCl, 1 mM TCEP, 5 mM maltose, and 0.01% NaN3, pH 7.4) for additional purification as well as molecular mass determination. For the apparent molecular mass estimation of each 5-HT3A-ICD construct (WT, deletion, and substitution constructs), the SEC column was calibrated using thyroglobulin, 669 kD; ferritin, 440 kD; aldolase, 158 kD; conalbumin, 75 kD; and ovalbumin, 44 kD according to the instruction manual (GE Healthcare). The column void volume (Vo) was established with Blue Dextran 2000. The molecular mass of each construct was determined based on the calibration curve of the gel-phase distribution coefficient (Kav) versus log molecular weight (log Mr). The gel-phase distribution coefficient is calculated by the equation Kav = (Ve − Vo)/(Vc − Vo), where Ve = elution volume, Vc = geometric column volume, and Vo = column void volume. The purified protein samples obtained after the final SEC step were analyzed by stain-free precast SDS-PAGE gels. SEC coupled with multi-angle light scattering (SEC-MALS) We configured a Biorad Biological DuoFlow 10 system coupled with a miniDAWN-TREOS static 3-angle laser light-scattering detector and an Optilab-rEX refractive index detector (Wyatt Technology) for conducting the SEC-MALS experiments. 250 µl of purified protein sample (0.1–0.5 mg) was passed, at a constant flow rate of 0.5 ml/min, through an ENrich SEC 650 10 × 300 high-resolution column (Biorad) thoroughly pre-equilibrated with buffer B at room temperature. UV absorbance was measured with the detector at 280 nm. Light scattering and refractive index were monitored at a wavelength of 658 nm. BioLogic DuoFlow software version 5.3 (Biorad) was used to control the chromatography system, and Astra 5.3.4 software (Wyatt Technology) was used for data collection and analysis. The light-scattering (LS) detectors were normalized with monomeric BSA (Sigma). Baseline settings for all laser detectors as well as peak alignment and band-broadening correction between the UV, LS, and refractive index detectors were performed using Astra software algorithms. Data were processed to determine the weight-average molar mass and polydispersity of the protein sample using the Debye model as per the manufacturer’s instructions. A minimum of three repeat runs was conducted for each construct under identical experimental conditions. Online supplemental material Fig. S1 provides sequence alignments for cation- and anion-conducting pLGICs with predicted α-helical segments highlighted to illustrate predicted structural diversity of these two groups of ICDs. Fig. S2 presents an illustration of charged amino acids of the MA-helix and post-M3 loop using the x-ray structure of 5-HT3A (PDB ID: 4PIR; Hassaine et al., 2014). Results Overview of approach We have shown previously that the intracellular domain of 5-HT3ARs in fusion with MBP self-associates as a pentamer in solution (Pandhare et al., 2016, 2019). To gain further insights into the molecular mechanisms governing pentameric assembly of the 5-HT3A–ICD, we adopted a systematic approach wherein a series of MBP–5-HT3A–ICD constructs, bearing either large deletions or a set of amino acid substitutions within the ICD, were engineered. Our nomenclature for the different constructs is illustrated in detail in Fig. 1. In brief, five constructs were investigated in this study: the initial full-length ICD construct containing the 5-HT3A–ICD consisting of all 115 amino acids of the ICD; three deletion constructs, and one construct based on the full-length ICD with three arginine residues in the MA helix mutated to glutamine, aspartic acid, and alanine (R432Q/R436D/R440A). All constructs were expressed in E. coli and purified to homogeneity. We then used a combination of biophysical and biochemical methods, including denaturing gel electrophoresis, SEC, and SEC-MALS to investigate the oligomeric state of each construct. Expression and purification of WT and engineered 5-HT3A–ICD constructs Cytosolic expression in E. coli cells of the 5-HT3A–ICD as a chimeric protein containing an N-terminal MBP was under the control of a Ptac promoter. After careful optimization, ∼20–30 mg of ICD could be purified per liter of E. coli cell culture (Pandhare et al., 2019). Modifications (point or deletion) of this original construct (Fig. 1) were similarly produced with yields of 8–30 mg/liter of E. coli culture. Two-step purification was accomplished on an amylose resin affinity column followed by SEC with SDS-PAGE indicating a purity of >90% and >95%, respectively (Figs. 2 and 3). Figure 2. Affinity column purification of the ICD chimeras. The chimeras were purified by amylose resin column chromatography. The soluble protein was loaded onto columns (L, loading material; FT, flow-through), followed by washes (W), and eluted. SDS-PAGE indicates the quality of purified protein and identifies peak fractions. Figure 3. Weight determination of ICD chimeras by SDS-PAGE and SEC. (A) The weight of the ICD chimeras in monomeric state was determined by resolving SEC-purified protein on SDS-gel electrophoresis. Molecular weight standards are shown on top, and ICD (53 kD), ΔMA (49 kD), MA (45 kD), Δ44 (48 kD), and QDA (52 kD) are shown below. (B) The complete SDS-gel electrophoresis image from which the analysis in section A is drawn. A standard protein ladder is shown next to each protein for increased accuracy of the analysis. Lane compositions are as follows: lane 2, ICD; lane 4, ΔMA; lane 6, MA; lane 8, Δ44; and lane 10, QDA. (C) SEC yielded a weight of 267 kD for ICD, 261 kD for Δ44, 180 kD for ΔMA, 141 kD for QDA, and 73 kD for MA. Chromatogram of standard proteins (T, thyroglobulin; F, ferritin; A, aldolase; C, conalbumin; O, ovalbumin) in gray, and chimeras in their respective colors are shown above. A280, UV absorbance at 280 nm. Oligomeric structure of full-length 5-HT3A–ICD Affinity- or SEC-purified full-length ICD, when analyzed by SDS-PAGE under reducing conditions, migrated as a major band with a relative molecular weight of 53 kD (Figs. 2 and 3), consistent with its theoretical mass of 53.4 kD. Notably, during SEC ICD eluted much earlier than would be anticipated from its relative molecular weight determined by SDS-PAGE. The elution profile of the protein was similar over the entire range of concentrations examined (0.1–5 mg/ml; data not shown). The hydrodynamic volume (13.01 ml) of a well-resolved peak of ICD predicted an apparent molecular mass of 267 kD when compared with the elution profiles of proteins with known molecular masses (Fig. 3). The elution volume indicates a mass that is five times the monomer mass and therefore a potential pentameric assembly. Asymmetrically shaped molecules containing multiple helical regions with an overall extended conformation, such as human mannose-binding protein, are known to display anomalous behavior on SEC (Lipscombe et al., 1995). As shape-dependent factors can influence the elution volume of macromolecules when analyzed on SEC, the results from this hydrodynamic technique are possibly vulnerable to misinterpretation (Wen et al., 1996). Therefore, we used the shape-independent technique of SEC-MALS to study the self-assembly properties of 5-HT3A–ICD constructs further (Table 1). SEC-MALS determines the absolute molecular mass of macromolecules independent of molecular shape, and the high-resolution separation by SEC facilitates analysis of each resolved protein species in a tandem manner. Examination of the SEC-MALS results revealed that WT 5-HT3A–ICD exists predominantly as pentamers with an absolute molecular mass of 259 kD (Pandhare et al., 2016, 2019). Samples consisting of MBP alone were analyzed by SEC-MALS, indicating that MBP was monomeric in solution (data not shown), consistent with a previous small-angle x-ray scattering study (Rodgers et al., 1996), and that it likely did not influence oligomeric assembly of the ICD. Table 1. Monomer and oligomer masses Construct Theoretical molar mass of the monomer (Mw, 103 g/mol) Calculated molar mass from SEC-MALS (Mw, 103 g/mol) Dispersity (Ð = Mw/Mn) Number of monomers (calculated Mw/theoretical Mw) (Mean ± SEM) (Mean ± SEM) ICD 53.4 259 ± 8 1.073 ± 0.015 4.8 Δ44 49.3 245 ± 9 1.002 ± 0.001 4.9 ΔMA 49.8 149 ± 6 1.001 ± 0.001 2.9 MA 44.3 89 ± 2 1.000 ± 0.003 2.0 QDA 53.0 106 ± 1 1.004 ± 0.003 2.0 Effect of deletions on ICD self-assembly Our data so far demonstrate that the pentameric assembly is an intrinsic property of the ICD. Based on a recent cryo-EM structure of the mouse 5-HT3AR with only a partially resolved ICD, the MX and MA helices form a “putative contact” with the MA helix of the neighboring subunit (Polovinkin et al., 2018). Since the involvement of unresolved regions in domain–domain interactions was not apparent, we first sought to understand the effects of systematic deletions within the ICD on their oligomeric states. We generated three ICD deletion constructs: (1) ICD lacking 44 amino acids from the disordered flexible region between the MX and MA helices (Δ44), (2) ICD lacking the MA helix (ΔMA), and (3) MA helix alone (MA; Fig. 1). The theoretical molecular weights, calculated from their amino acid composition of 49.3 kD, 49.8 kD, and 44.3 kD were comparable to the experimentally determined relative molecular weights of 48 kD, 49 kD, and 45 kD, respectively, when purified protein samples were analyzed by SDS-PAGE (Fig. 3). The faint higher molecular weight bands observed in SDS-PAGE (Fig. 2) are likely comprised of oligomeric assemblies resistant to SDS under the given electrophoresis condition. However, as observed later on for each construct, the purified protein analyzed after SEC (Fig. 3) as well as by multi-angle light scattering (Fig. 4) was homogenous and monodisperse, respectively. Figure 4. SEC-MALS determination of oligomeric assembly. (A) Both the ICD (dotted black) and the Δ44 construct (purple line) maintain pentameric assembly, whereas deletions in the ΔMA (pink line) and the MA (orange line) constructs abolish pentamerization of the ICD. (B) The QDA substitutions (red line) disrupt the pentameric assembly of the ICD. A representative SEC-MALS profile showing the Rayleigh ratio. Note that different flow rates were used during the monitoring period for A (0.5 ml/min) and B (0.3 ml/min). The three deletion constructs showed strikingly different elution profiles in SEC (Fig. 3). Importantly, all constructs eluted in single monodisperse peaks. Based on the elution volumes of 13.04 ml, 13.57 ml, and 14.85 ml for Δ44, ΔMA, and MA, respectively, apparent molecular weights of 261 kD, 171 kD, and 73 kD, respectively, were estimated. These results were inconsistent with a monomeric state for all deletion constructs. While SEC data for Δ44 may indicate a pentamer (49.3 kD × 5 = 247 kD vs. 261 kD), data for ΔMA and MA did not correspond to a defined multimeric state. Therefore, at this juncture, it was imperative to investigate all constructs with SEC-MALS to determine the absolute molecular weight (Table 1). The predominant peak for the Δ44 construct yielded a weight-average molar mass of 245 ± 9 kD (mean ± SEM, n = 5), which corresponds well to a pentameric assembly (theoretical molecular mass: 49.3 kD). Moreover, the pentameric protein is monodisperse because the experimentally determined molar masses were equivalent across the peak, as indicated by the measurements of polydispersity: Mw/Mn of 1.002 ± 0.001, where Mw is the weight-average molar mass, and Mn is the number-average molar mass (Fig. 4; Table 1). Similarly, for the other two deletion constructs, ΔMA and MA, analysis yielded the weight-average molar masses of 149 ± 6 kD (mean ± SEM, n = 3) and 89 ± 2 kD (mean ± SEM, n = 3), respectively, and polydispersity of 1.001 ± 0.001 (mean ± SEM, n = 3) and 1.000 ± 0.003 (mean ± SEM, n = 3), respectively. Therefore, SEC-MALS results established that ΔMA is a trimer (theoretical molecular mass: 49.8 kD) and MA is a dimer (theoretical molecular mass: 44.3 kD) in solution. It is also important to note that we did not observe a plateau in the molar mass profiles or a peak in the elution profiles that corresponded to major quantities of the monomeric form of either of the deletion constructs. Together, the results suggested that we were able to disrupt pentameric assembly of the ICD. Since Δ44 assembled into pentamers, but both ΔMA and MA did not do so, we inferred at this point that the segment removed in Δ44 is not involved in pentamerization, whereas motifs contained in both MA and ΔMA may interact to mediate pentamerization. Effect of specific amino acid substitutions in the MA helix on the oligomeric structure of the ICD chimera The observed involvement of both the proximal and distal segments of the ICD in pentamerization was reminiscent of the salt-bridge networks identified in some of the structural studies (Fig. S2). Three arginine residues of the MA helix (R432, R436, and R440), which have been identified as determinants for the 5-HT3A subunit’s low single-channel conductance, participate in salt bridges with acidic amino acids from neighboring ICDs (Kelley et al., 2003; Hassaine et al., 2014). We replaced these three arginine residues with the aligned residues glutamine, aspartic acid, and alanine of serotonin type 3B (5-HT3B) subunits to obtain the QDA construct (R432Q/R436D/R440A) to investigate their impact on the pentamerization of the ICD. The QDA construct eluted, at a volume of 13.92 ml, as a major symmetrical peak after SEC, which predicted its apparent molecular weight as 137 kD (Fig. 3). The size of the monomer, calculated from its amino acid sequence as well as determined by SDS-PAGE, is 53 kD (Fig. 3). Analogous to the experimental strategy employed for the deletion constructs, we determined an absolute molar mass for the QDA construct by SEC-MALS. The result of the weight-average molar mass of 106 ± 1 kD (mean ± SEM, n = 3) and polydispersity of 1.004 ± 0.003 (mean ± SEM, n = 3) suggested that the QDA construct exists as a dimer in solution (Fig. 4). This result indicates that mutation of the RRR motif and the likely resulting disruption of salt bridges is sufficient to abolish pentameric assembly of the ICD. Discussion The ICD is the most diverse and least understood portion of pLGICs across the Cys-loop receptor superfamily. Several studies show that the ICD plays a pivotal role in receptor assembly, trafficking, and targeting (Connolly, 2008; Kracun et al., 2008), functional interactions with cytoplasmic proteins (Lansdell et al., 2005; Nishtala et al., 2016), gating and desensitization (Papke and Grosman, 2014), inward rectification (Baptista-Hon et al., 2013), Ca2+ permeability (Livesey et al., 2008), and single-channel conductance (Kelley et al., 2003; Hales et al., 2006). In the Torpedo nAChR structure, the five MA helices were viewed as an inverted cone protruding into the cytosol and enclosing the intracellular vestibule of the transmembrane pore (Unwin, 2005). The MA helices were thought to frame lateral portals that function as cytosolic ion pathways. In the case of the 5-HT3A–ICD, it has become apparent that several of the experimentally determined functions are, perhaps most critically, contingent on the integrity of its secondary and/or quaternary structure. Mutagenesis studies have demonstrated that the variation in the length and amino acid composition of the ICD can influence the desensitization rate of the receptor channel (McKinnon et al., 2012; Baptista-Hon et al., 2013). Reducing the length of the loop connecting the M3 to the MA helix has been shown to result in loss of inward rectification of macroscopic currents, which has been inferred to be an indicator for the disruption of the portal architecture (Baptista-Hon et al., 2013). Key residues within the MA helix not only profoundly affect the permeability of Ca2+ but also the single-channel conductance uniquely observed in experiments with homomeric 5-HT3ARs. The homomeric 5-HT3AR displays a very low single-channel conductance in the sub-pS range (Brown et al., 1998). This is in stark contrast to the larger currents (16–30 pS) measured from heteromeric serotonin type 3AB receptors as well as other Cys-loop receptors (Davies et al., 1999; Krzywkowski et al., 2008). A seminal study attributed the low conductance to a group of three arginine residues, which are found in the MA helix of the 3A subunit but are absent in the 3B subunit (Kelley et al., 2003). The x-ray structure of mouse 5-HT3AR revealed that some of the conductance-restricting arginine residues as well as other basic residues lining the MA helix are within salt-bridge distance to several acidic residues on the neighboring MA helices and post-M3 loops (Hassaine et al., 2014). Notably, the post-M3 loops thread through the MA portals, leaving no paths for ions to access the channel. In the x-ray structure, the side chain of D312, at the center of the short post-M3 loop, is within salt-bridging distance and straddled by arginine residues R436 and R440 on the adjacent subunit. Interestingly, it has been postulated that inter- and intra-subunit salt bridges confer structural rigidity to the inverted pentacone formed by the five MA helices, and that a controlled removal of these salt bridges increases flexibility in this area (Kozuska et al., 2014). The same study proposed that it was widening of the lateral portals between the MA helices, due to the increased flexibility in the absence of certain salt bridges, and not the abolition of electrostatic repulsion, that relieved the limited conductance of the 5-HT3AR channel. Importantly, the authors based their prediction on a cautious assumption that the mutations they introduced would not destabilize global interhelical interactions that might further widen the portals (Kozuska et al., 2014). Taken together, these findings suggest that the extensive interhelical interactions between the MA helices may be central to holding the intact quaternary structure of the ICD, and thus contribute to the functional landscape of full-length 5-HT3ARs. In this regard, it is intriguing to find that the 5-HT3A–ICD is structured as pentamers in the absence of both the ECD and the TMD (Pandhare et al., 2016). Multiple studies underscore the importance of salt-bridge formation in the stabilization of a protein oligomeric assembly (Binter et al., 2009; Skinner et al., 2017; Pacheco et al., 2018). Therefore, we investigated whether some of the aforementioned function-governing salt bridges additionally play a fundamental role in pentameric assembly. We previously developed a chimeric strategy, where the 5-HT3A–ICD was linked N-terminally to MBP (Pandhare et al., 2016, 2019). In the present work, we systematically introduced deletions or defined amino acid substitutions within this construct. First, in order to obtain insights into the role of the flexible unresolved region of the ICD, we removed 44 amino acids from this region (Δ44). A similar construct was generated in full-length channels in a study that investigated the significance of the lateral portals within MA helices for inward rectification observed in 5-HT3ARs (Baptista-Hon et al., 2013). Using a series of truncation mutants, the authors determined the minimum number of residues necessary to link M3 with the MA helix without abolishing function. While truncations of 44 and 24 amino acids from the ICD produced functional receptors, the construct with 44 amino acids deleted exhibited loss of rectification. This indicated reduced impediment to ion conduction that was inferred to be caused by deconstruction of the lateral portals. Interestingly, our results showed that truncation of the exact 44 amino acids from the soluble ICD maintained pentamers in solution. The study additionally reported a truncation of 55 amino acids in L2, or removal of 10 amino acids from the proximal part of the ICD that did not produce functional homomeric 5-HT3ARs (Baptista-Hon et al., 2013). We infer that the large 55-residue truncation, as well as truncations made in the proximal ICD, can herald a global disruption and lead to the formation of a global nonnative ICD conformation, which may further impose unfavorable constraints for channel function. Our assumption is supported by the previously reported data in which an engineered 5-HT3AR containing a heptapeptide replacing the entire ICD was not only fully functional but also exhibited enhanced single-channel conductance (Jansen et al., 2008). Within this context, given our experimental evidence of intact pentamers formed by Δ44, we inferred that this shortened ICD plausibly underwent merely a local perturbation likely limited only to the lateral portals without disturbing the key interactions holding together the oligomeric assembly. This might have resulted in a minimally aberrant native-like ICD architecture responsible for the observed retention of function in correspondingly engineered 5-HT3ARs (Baptista-Hon et al., 2013). In the structures there is an ∼54-Å gap between the end of MX and the start of the MA α-helices (M3-WILC to VRG-M4). At least 18 amino acids are needed for a linker to span this distance. Δ44 and Δ55 contain 14 and 3 amino acids in our constructs, respectively. Since Baptista-Hon et al. (2013) used the long variant with 6 additional amino acids in this segment, their Δ44 and Δ55 contain 20 and 9 amino acids, respectively. Consequently, it would be expected that Δ44 has a sufficient remaining L2 loop length to not disrupt the relative arrangement of MX and MA, and indeed it pentamerizes. Δ55, on the other hand, has a too-short L2-linker, and a drastic distortion is expected. In preliminary experiments using dynamic light scattering, we observed disruption of pentamers for Δ55. Collectively, we postulated that the underlying molecular interactions driving the ICD pentamerization must lie within the proximal (L1-MX) as well as the distal (MA) regions of the ICD. To test the above stated hypothesis, we decided to explore the effect of deletion of a proximal (MA construct) or a distal (ΔMA construct) portion of the ICD on its oligomeric state. SEC-MALS indicated disruption of pentameric assembly in both constructs, and MA was determined as a dimer and ΔMA as a trimer. The fact that the MA construct did not pentamerize was somewhat surprising since the MA helices are tightly associated in the crystal structure, where rings of hydrophobic residues L402, L406, I409, and L413 shape a 17-Å-long narrow channel (Hassaine et al., 2014). Given the observation that ΔMA also disrupts pentamerization, while Δ44 retains the pentameric state, we hypothesized that the elements of both MA and ΔMA are required for driving the pentamerization. In particular, the results pointed toward both the proximal and distal segments of the ICD as being determinants for pentameric assembly. Interestingly, when we previously made chimeras by using the prokaryotic pLGIC from GLIC that naturally lacks an ICD and inserted ICDs from different eukaryotic anion- or cation-conducting channels, we found that the linking amino acids, on both the proximal and distal insertion sites, were crucial for functional 5-HT3A–ICD insertion (Goyal et al., 2011). Remarkably, for other ICDs that we investigated (nAChα7, Glyα1, GABAρ1), the linking amino acids for the ICDs had little impact on function, indicating a unique characteristic for the 5-HT3A–ICD (Mnatsakanyan et al., 2015). We next sought to examine if the conductance-limiting arginines are additionally involved in maintaining the quaternary pentameric structure of the ICD. Interestingly, the location of the RRR coincides with a kink in the paths of the long continuous MA-M4 helices that collectively leads to a significantly tighter packing of MA helices as they traverse away from the membrane. The bottom aperture of the MA bundle has a diameter of 4.2 Å, too small for conduction of hydrated cations, but wider than the remaining opening of the lateral windows through which the post-M3 loop is threaded (Hassaine et al., 2014). Combined replacement of RRR with QDA substantially augments single-channel conductance (Kelley et al., 2003), a phenomenon originally attributed to the removal of repulsive forces of the positive charges of the arginines upon conducted cations. On the other hand, constrained geometric simulations, upon abolishing certain salt bridges, have predicted enhanced structural flexibility within the MA helices and led to postulation that a rigid structural framework of the inverted cone formed by the MA helices might be crucial for the limited conductance of the 5-HT3AR channel (Kozuska et al., 2014). Interestingly, the QDA substitutions in our soluble construct disrupted the pentameric state of the full-length ICD and reduced it to dimers. Clearly, this finding does not appear to support the “charge-repulsion theory,” but rather provides the first experimental evidence suggesting a structural role of the triple arginines that are critical to sustaining tight packing of the MA bundle as well as pentameric assembly of the ICD. We infer that the disjunction of interactions due to the QDA substitutions is related to breakage of salt bridges observed in the x-ray structure. Sequence alignment between all available 5-HT3A sequences indicates that the triple arginines as well as the aspartate in the post-M3 loop are absolutely conserved. Among 5-HT3B subunit sequences, neither motif is observed. Many 5-HT3B sequences have negatively charged side chains one or two residues before 5-HT3AD312, with or without a positive side chain at the D312-equivalent position. Only very few 5-HT3B sequences have one of the three MA helix arginines. However, the 5-HT3B sequences, similar to the 5-HT3A sequences, have many other charged residues within the MA helix that may contribute to salt-bridge networks (Hassaine et al., 2014; Kozuska et al., 2014). However, as our MA construct shows, those are not sufficient to drive pentamerization of the ICD alone, and therefore we do not expect homopentameric 5-HT3B or heteropentameric serotonin type 3AB ICDs to be as tightly interdigitated as homopentameric 5-HT3A subunits alone. We infer that the heteropentameric ICDs are therefore more loosely packed and that this contributes to their higher single-channel conductance. We do not know exactly why the ΔMA construct forms trimers under the given experimental conditions; nor do we know the exact driving forces behind dimerization of the MA and QDA constructs in solution. It is a possibility, as documented elsewhere (Skinner et al., 2017), that the breakage of preexisting salt bridges can induce protein remodeling via initiation of partial unfolding and reorganization of electrostatic interactions leading to the formation of nonnative oligomers. Nevertheless, it is important to emphasize that we did not observe pentamers formed by any of the engineered proteins during our investigation, with the notable exceptions of the ICD and Δ44. Finally, to the best of our knowledge, this is the first experimental evidence that extends our understanding beyond the known functional contributions of the three arginine residues to limiting conductance, and posits that their existence within the MA helix of the 5-HT3AR is critical for preserving a pentameric state of the ICD. Supplementary Material Supplemental Materials (PDF) Acknowledgments Merritt C. Maduke served as editor. We thank the Texas Tech University Health Sciences Center Core Facilities; some of the images and or data were generated in the Image Analysis Core Facility & Molecular Biology Core Facility supported by the Texas Tech University Health Sciences Center, Lubbock, TX. Research reported in this publication was supported by the National Institute of Neurological Disorders and Stroke of the National Institutes of Health under award number R01NS077114 (to M. Jansen). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The authors declare no competing financial interests. Author contributions: M. Jansen designed the research. A. Pandhare, E. Pirayesh, A.G. Stuebler and M. Jansen conceived the WT and engineered constructs, performed the experiments, analyzed the data, and wrote the paper. ==== Refs Aapro, M.S. 1991. 5-HT3 receptor antagonists. An overview of their present status and future potential in cancer therapy-induced emesis. Drugs. 42 :551–568. 10.2165/00003495-199142040-00002 1723361 Baptista-Hon, D.T., T.Z. Deeb, J.J. Lambert, J.A. Peters, and T.G. Hales. 2013. The minimum M3-M4 loop length of neurotransmitter-activated pentameric receptors is critical for the structural integrity of cytoplasmic portals. J. Biol. Chem. 288 :21558–21568. 10.1074/jbc.M113.481689 23740249 Bar-Lev, D.D., N. Degani-Katzav, A. Perelman, and Y. Paas. 2011. Molecular dissection of Cl−-selective Cys-loop receptor points to components that are dispensable or essential for channel activity. J. Biol. Chem. 286 :43830–43841. 10.1074/jbc.M111.282715 21987577 Basak, S., Y. Gicheru, A. Samanta, S.K. Molugu, W. Huang, M. Fuente, T. Hughes, D.J. Taylor, M.T. Nieman, V. Moiseenkova-Bell, et al. 2018. Cryo-EM structure of 5-HT3A receptor in its resting conformation. Nat. Commun. 9 :514. 10.1038/s41467-018-02997-4 29410406 Binter, A., N. Staunig, I. Jelesarov, K. Lohner, B.A. Palfey, S. Deller, K. Gruber, and P. Macheroux. 2009. A single intersubunit salt bridge affects oligomerization and catalytic activity in a bacterial quinone reductase. FEBS J. 276 :5263–5274. 10.1111/j.1742-4658.2009.07222.x 19682074 Bondarenko, V., D. Mowrey, T. Tillman, T. Cui, L.T. Liu, Y. Xu, and P. Tang. 2012. NMR structures of the transmembrane domains of the α4β2 nAChR. Biochim. Biophys. Acta. 1818 :1261–1268. 10.1016/j.bbamem.2012.02.008 22361591 Bondarenko, V., D.D. Mowrey, T.S. Tillman, E. Seyoum, Y. Xu, and P. Tang. 2014. NMR structures of the human α7 nAChR transmembrane domain and associated anesthetic binding sites. Biochim. Biophys. Acta. 1838 :1389–1395. 10.1016/j.bbamem.2013.12.018 24384062 Brown, A.M., A.G. Hope, J.J. Lambert, and J.A. Peters. 1998. Ion permeation and conduction in a human recombinant 5-HT3 receptor subunit (h5-HT3A). J. Physiol. 507 :653–665. 10.1111/j.1469-7793.1998.653bs.x 9508827 Changeux, J.P., and S.J. Edelstein. 1998. Allosteric receptors after 30 years. Neuron. 21 :959–980. 10.1016/S0896-6273(00)80616-9 9856454 Connolly, C.N. 2008. Trafficking of 5-HT3 and GABAA receptors (Review). Mol. Membr. Biol. 25 :293–301. 10.1080/09687680801898503 18446615 Davies, P.A., M. Pistis, M.C. Hanna, J.A. Peters, J.J. Lambert, T.G. Hales, and E.F. Kirkness. 1999. The 5-HT3B subunit is a major determinant of serotonin-receptor function. Nature. 397 :359–363. 10.1038/16941 9950429 Du, J., W. Lü, S. Wu, Y. Cheng, and E. Gouaux. 2015. Glycine receptor mechanism elucidated by electron cryo-microscopy. Nature. 526 :224–229. 10.1038/nature14853 26344198 Forman, S.A., D.C. Chiara, and K.W. Miller. 2015. Anesthetics target interfacial transmembrane sites in nicotinic acetylcholine receptors. Neuropharmacology. 96 (Pt B ):169–177. 10.1016/j.neuropharm.2014.10.002 25316107 Gaddum, J.H., and Z.P. Picarelli. 1957. Two kinds of tryptamine receptor. Br. J. Pharmacol. Chemother. 12 :323–328. 10.1111/j.1476-5381.1957.tb00142.x 13460238 Goyal, R., A.A. Salahudeen, and M. Jansen. 2011. Engineering a prokaryotic Cys-loop receptor with a third functional domain. J. Biol. Chem. 286 :34635–34642. 10.1074/jbc.M111.269647 21844195 Hales, T.G., J.I. Dunlop, T.Z. Deeb, J.E. Carland, S.P. Kelley, J.J. Lambert, and J.A. Peters. 2006. Common determinants of single channel conductance within the large cytoplasmic loop of 5-hydroxytryptamine type 3 and α4β2 nicotinic acetylcholine receptors. J. Biol. Chem. 281 :8062–8071. 10.1074/jbc.M513222200 16407231 Hargreaves, A.C., S.C. Lummis, and C.W. Taylor. 1994. Ca2+ permeability of cloned and native 5-hydroxytryptamine type 3 receptors. Mol. Pharmacol. 46 :1120–1128.7808432 Hassaine, G., C. Deluz, L. Grasso, R. Wyss, M.B. Tol, R. Hovius, A. Graff, H. Stahlberg, T. Tomizaki, A. Desmyter, . 2014. X-ray structure of the mouse serotonin 5-HT3 receptor. Nature. 512 :276–281. 10.1038/nature13552 25119048 Hibbs, R.E., and E. Gouaux. 2011. Principles of activation and permeation in an anion-selective Cys-loop receptor. Nature. 474 :54–60. 10.1038/nature10139 21572436 Huang, X., H. Chen, K. Michelsen, S. Schneider, and P.L. Shaffer. 2015. Crystal structure of human glycine receptor-α3 bound to antagonist strychnine. Nature. 526 :277–280. 10.1038/nature14972 26416729 Huang, X., H. Chen, and P.L. Shaffer. 2017. Crystal structures of human GlyRα3 bound to ivermectin. Structure. 25 :945–950.28479061 Jansen, M., M. Bali, and M.H. Akabas. 2008. Modular design of Cys-loop ligand-gated ion channels: functional 5-HT3 and GABA ρ1 receptors lacking the large cytoplasmic M3M4 loop. J. Gen. Physiol. 131 :137–146. 10.1085/jgp.200709896 18227272 Kelley, S.P., J.I. Dunlop, E.F. Kirkness, J.J. Lambert, and J.A. Peters. 2003. A cytoplasmic region determines single-channel conductance in 5-HT3 receptors. Nature. 424 :321–324. 10.1038/nature01788 12867984 Kouvatsos, N., P. Giastas, D. Chroni-Tzartou, C. Poulopoulou, and S.J. Tzartos. 2016. Crystal structure of a human neuronal nAChR extracellular domain in pentameric assembly: Ligand-bound α2 homopentamer. Proc. Natl. Acad. Sci. USA. 113 :9635–9640. 10.1073/pnas.1602619113 27493220 Kozuska, J.L., I.M. Paulsen, W.J. Belfield, I.L. Martin, D.J. Cole, A. Holt, and S.M. Dunn. 2014. Impact of intracellular domain flexibility upon properties of activated human 5-HT3 receptors. Br. J. Pharmacol. 171 :1617–1628. 10.1111/bph.12536 24283776 Kracun, S., P.C. Harkness, A.J. Gibb, and N.S. Millar. 2008. Influence of the M3-M4 intracellular domain upon nicotinic acetylcholine receptor assembly, targeting and function. Br. J. Pharmacol. 153 :1474–1484. 10.1038/sj.bjp.0707676 18204482 Krzywkowski, K., P.A. Davies, P.L. Feinberg-Zadek, H. Bräuner-Osborne, and A.A. Jensen. 2008. High-frequency HTR3B variant associated with major depression dramatically augments the signaling of the human 5-HT3AB receptor. Proc. Natl. Acad. Sci. USA. 105 :722–727. 10.1073/pnas.0708454105 18184810 Lansdell, S.J., V.J. Gee, P.C. Harkness, A.I. Doward, E.R. Baker, A.J. Gibb, and N.S. Millar. 2005. RIC-3 enhances functional expression of multiple nicotinic acetylcholine receptor subtypes in mammalian cells. Mol. Pharmacol. 68 :1431–1438. 10.1124/mol.105.017459 16120769 Laverty, D., R. Desai, T. Uchański, S. Masiulis, W.J. Stec, T. Malinauskas, J. Zivanov, E. Pardon, J. Steyaert, K.W. Miller, et al. 2019. Cryo-EM structure of the human α1β3γ2 GABAA receptor in a lipid bilayer. Nature. 565 :516–520. 10.1038/s41586-018-0833-4 30602789 Lipscombe, R.J., M. Sumiya, J.A. Summerfield, and M.W. Turner. 1995. Distinct physicochemical characteristics of human mannose binding protein expressed by individuals of differing genotype. Immunology. 85 :660–667.7558163 Liu, Z., G. Ramanoudjame, D. Liu, R.O. Fox, V. Jayaraman, M. Kurnikova, and M. Cascio. 2008. Overexpression and functional characterization of the extracellular domain of the human α1 glycine receptor. Biochemistry. 47 :9803–9810. 10.1021/bi800659x 18710260 Livesey, M.R., M.A. Cooper, T.Z. Deeb, J.E. Carland, J. Kozuska, T.G. Hales, J.J. Lambert, and J.A. Peters. 2008. Structural determinants of Ca2+ permeability and conduction in the human 5-hydroxytryptamine type 3A receptor. J. Biol. Chem. 283 :19301–19313. 10.1074/jbc.M802406200 18474595 Lummis, S.C. 2012. 5-HT(3) receptors. J. Biol. Chem. 287 :40239–40245. 10.1074/jbc.R112.406496 23038271 Masiulis, S., R. Desai, T. Uchański, I. Serna Martin, D. Laverty, D. Karia, T. Malinauskas, J. Zivanov, E. Pardon, A. Kotecha, . 2019. GABAA receptor signalling mechanisms revealed by structural pharmacology. Nature. 565 :454–459. 10.1038/s41586-018-0832-5 30602790 McKinnon, N.K., D.C. Reeves, and M.H. Akabas. 2011. 5-HT3 receptor ion size selectivity is a property of the transmembrane channel, not the cytoplasmic vestibule portals. J. Gen. Physiol. 138 :453–466. 10.1085/jgp.201110686 21948949 McKinnon, N.K., M. Bali, and M.H. Akabas. 2012. Length and amino acid sequence of peptides substituted for the 5-HT3A receptor M3M4 loop may affect channel expression and desensitization. PLoS One. 7 :e35563. 10.1371/journal.pone.0035563 22539982 Miller, P.S., and A.R. Aricescu. 2014. Crystal structure of a human GABAA receptor. Nature. 512 :270–275. 10.1038/nature13293 24909990 Miller, P.S., and T.G. Smart. 2010. Binding, activation and modulation of Cys-loop receptors. Trends Pharmacol. Sci. 31 :161–174. 10.1016/j.tips.2009.12.005 20096941 Mnatsakanyan, N., S.N. Nishtala, A. Pandhare, M.C. Fiori, R. Goyal, J.E. Pauwels, A.F. Navetta, A. Ahrorov, and M. Jansen. 2015. Functional chimeras of GLIC obtained by adding the intracellular domain of anion- and cation-conducting Cys-loop receptors. Biochemistry. 54 :2670–2682. 10.1021/acs.biochem.5b00203 25861708 Moon, A.F., G.A. Mueller, X. Zhong, and L.C. Pedersen. 2010. A synergistic approach to protein crystallization: combination of a fixed-arm carrier with surface entropy reduction. Protein Sci. 19 :901–913.20196072 Moraga-Cid, G., G.E. Yevenes, G. Schmalzing, R.W. Peoples, and L.G. Aguayo. 2011. A single phenylalanine residue in the main intracellular loop of α1 γ-aminobutyric acid type A and glycine receptors influences their sensitivity to propofol. Anesthesiology. 115 :464–473. 10.1097/ALN.0b013e31822550f7 21673564 Morales-Perez, C.L., C.M. Noviello, and R.E. Hibbs. 2016. X-ray structure of the human α4β2 nicotinic receptor. Nature. 538 :411–415. 10.1038/nature19785 27698419 Moroni, M., I. Biro, M. Giugliano, R. Vijayan, P.C. Biggin, M. Beato, and L.G. Sivilotti. 2011. Chloride ions in the pore of glycine and GABA channels shape the time course and voltage dependence of agonist currents. J. Neurosci. 31 :14095–14106. 10.1523/JNEUROSCI.1985-11.2011 21976494 Mowrey, D.D., T. Cui, Y. Jia, D. Ma, A.M. Makhov, P. Zhang, P. Tang, and Y. Xu. 2013. Open-channel structures of the human glycine receptor α1 full-length transmembrane domain. Structure. 21 :1897–1904. 10.1016/j.str.2013.07.014 23994010 Nishtala, S.N., N. Mnatsakanyan, A. Pandhare, C. Leung, and M. Jansen. 2016. Direct interaction of the resistance to inhibitors of cholinesterase type 3 protein with the serotonin receptor type 3A intracellular domain. J. Neurochem. 137 :528–538. 10.1111/jnc.13578 26875553 Nury, H., N. Bocquet, C. Le Poupon, B. Raynal, A. Haouz, P.J. Corringer, and M. Delarue. 2010. Crystal structure of the extracellular domain of a bacterial ligand-gated ion channel. J. Mol. Biol. 395 :1114–1127. 10.1016/j.jmb.2009.11.024 19917292 Nury, H., C. Van Renterghem, Y. Weng, A. Tran, M. Baaden, V. Dufresne, J.P. Changeux, J.M. Sonner, M. Delarue, and P.J. Corringer. 2011. X-ray structures of general anaesthetics bound to a pentameric ligand-gated ion channel. Nature. 469 :428–431. 10.1038/nature09647 21248852 Pacheco, S., I. Gómez, J. Sánchez, B.I. García-Gómez, D.M. Czajkowsky, J. Zhang, M. Soberón, and A. Bravo. 2018. Helix α-3 inter-molecular salt bridges and conformational changes are essential for toxicity of Bacillus thuringiensis 3D-Cry toxin family. Sci. Rep. 8 :10331. 10.1038/s41598-018-28753-8 29985464 Pandhare, A., P.N. Grozdanov, and M. Jansen. 2016. Pentameric quaternary structure of the intracellular domain of serotonin type 3A receptors. Sci. Rep. 6 :23921. 10.1038/srep23921 27045630 Pandhare, A., A.S. Pappu, H. Wilms, M.P. Blanton, and M. Jansen. 2017. The antidepressant bupropion is a negative allosteric modulator of serotonin type 3A receptors. Neuropharmacology. 113 (Pt A ):89–99. 10.1016/j.neuropharm.2016.09.021 27671323 Pandhare, A., A.G. Stuebler, E. Pirayesh, and M. Jansen. 2019. A modified clear-native polyacrylamide gel electrophoresis technique to investigate the oligomeric state of MBP-5-HT3A-intracellular domain chimeras. Protein Expr. Purif. 153 :45–52. 10.1016/j.pep.2018.08.010 30130580 Papke, D., and C. Grosman. 2014. The role of intracellular linkers in gating and desensitization of human pentameric ligand-gated ion channels. J. Neurosci. 34 :7238–7252. 10.1523/JNEUROSCI.5105-13.2014 24849357 Phulera, S., H. Zhu, J. Yu, D.P. Claxton, N. Yoder, C. Yoshioka, and E. Gouaux. 2018. Cryo-EM structure of the benzodiazepine-sensitive α1β1γ2S tri-heteromeric GABAA receptor in complex with GABA. eLife. 7 :e39383. 10.7554/eLife.39383 30044221 Polovinkin, L., G. Hassaine, J. Perot, E. Neumann, A.A. Jensen, S.N. Lefebvre, P.J. Corringer, J. Neyton, C. Chipot, F. Dehez, . 2018. Conformational transitions of the serotonin 5-HT3 receptor. Nature. 563 :275–279. 10.1038/s41586-018-0672-3 30401839 Rodgers, K.K., Z. Bu, K.G. Fleming, D.G. Schatz, D.M. Engelman, and J.E. Coleman. 1996. A zinc-binding domain involved in the dimerization of RAG1. J. Mol. Biol. 260 :70–84. 10.1006/jmbi.1996.0382 8676393 Skinner, J.J., S. Wang, J. Lee, C. Ong, R. Sommese, S. Sivaramakrishnan, W. Koelmel, M. Hirschbeck, H. Schindelin, C. Kisker, . 2017. Conserved salt-bridge competition triggered by phosphorylation regulates the protein interactome. Proc. Natl. Acad. Sci. USA. 114 :13453–13458. 10.1073/pnas.1711543114 29208709 Spurny, R., B. Billen, R.J. Howard, M. Brams, S. Debaveye, K.L. Price, D.A. Weston, S.V. Strelkov, J. Tytgat, S. Bertrand, . 2013. Multisite binding of a general anesthetic to the prokaryotic pentameric Erwinia chrysanthemi ligand-gated ion channel (ELIC). J. Biol. Chem. 288 :8355–8364. 10.1074/jbc.M112.424507 23364792 Thompson, A.J., H.A. Lester, and S.C. Lummis. 2010. The structural basis of function in Cys-loop receptors. Q. Rev. Biophys. 43 :449–499. 10.1017/S0033583510000168 20849671 Unterer, B., C.M. Becker, and C. Villmann. 2012. The importance of TM3-4 loop subdomains for functional reconstitution of glycine receptors by independent domains. J. Biol. Chem. 287 :39205–39215. 10.1074/jbc.M112.376053 22995908 Unwin, N. 2005. Refined structure of the nicotinic acetylcholine receptor at 4 Å resolution. J. Mol. Biol. 346 :967–989. 10.1016/j.jmb.2004.12.031 15701510 Verrall, S., and Z.W. Hall. 1992. The N-terminal domains of acetylcholine receptor subunits contain recognition signals for the initial steps of receptor assembly. Cell. 68 :23–31. 10.1016/0092-8674(92)90203-O 1370654 Wells, G.B., R. Anand, F. Wang, and J. Lindstrom. 1998. Water-soluble nicotinic acetylcholine receptor formed by α7 subunit extracellular domains. J. Biol. Chem. 273 :964–973. 10.1074/jbc.273.2.964 9422757 Wen, J., T. Arakawa, and J.S. Philo. 1996. Size-exclusion chromatography with on-line light-scattering, absorbance, and refractive index detectors for studying proteins and their interactions. Anal. Biochem. 240 :155–166. 10.1006/abio.1996.0345 8811899 Yang, J. 1990. Ion permeation through 5-hydroxytryptamine-gated channels in neuroblastoma N18 cells. J. Gen. Physiol. 96 :1177–1198. 10.1085/jgp.96.6.1177 2286832 Zhu, S., C.M. Noviello, J. Teng, R.M. Walsh Jr., J.J. Kim, and R.E. Hibbs. 2018. Structure of a human synaptic GABAA receptor. Nature. 559 :67–72. 10.1038/s41586-018-0255-3 29950725

PMC006xxxxxx/PMC6829593.txt

==== Front J Exp Med J Exp Med jem jem The Journal of Experimental Medicine 0022-1007 1540-9538 Rockefeller University Press 31601677 20190980 10.1084/jem.20190980 Research Articles Article 311 320 306 Microglia drive APOE-dependent neurodegeneration in a tauopathy mouse model APOE-dependent neurodegeneration is driven by microglia Shi Yang 1 https://orcid.org/0000-0001-7912-2016 Manis Melissa 1 https://orcid.org/0000-0003-0769-604X Long Justin 1 Wang Kairuo 1 https://orcid.org/0000-0001-8402-2634 Sullivan Patrick M. 2 https://orcid.org/0000-0002-5937-0212 Remolina Serrano Javier 1 Hoyle Rosa 1 https://orcid.org/0000-0002-3400-0856 Holtzman David M. 1 1 Department of Neurology, Hope Center for Neurological Disorders, Charles F. and Joanne Knight Alzheimer's Disease Research Center, Washington University School of Medicine, St. Louis, MO 2 Department of Medicine, Duke University School of Medicine, Durham, NC Correspondence to David M. Holtzman: holtzman@neuro.wustl.edu 04 11 2019 10 10 2019 216 11 25462561 02 6 2019 21 7 2019 08 8 2019 © 2019 Shi et al. 2019 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). Shi et al. find that microglia, instead of tau-induced direct neurotoxicity, are the driving force of neurodegeneration in a tauopathy mouse model. Microglia are also required for tau pathogenesis. In addition, apoE strongly regulates neurodegeneration in the setting of tauopathy predominantly by modulating microglial function. Chronic activation of brain innate immunity is a prominent feature of Alzheimer’s disease (AD) and primary tauopathies. However, to what degree innate immunity contributes to neurodegeneration as compared with pathological protein-induced neurotoxicity, and the requirement of a particular glial cell type in neurodegeneration, are still unclear. Here we demonstrate that microglia-mediated damage, rather than pathological tau-induced direct neurotoxicity, is the leading force driving neurodegeneration in a tauopathy mouse model. Importantly, the progression of ptau pathology is also driven by microglia. In addition, we found that APOE, the strongest genetic risk factor for AD, regulates neurodegeneration predominantly by modulating microglial activation, although a minor role of apoE in regulating ptau and insoluble tau formation independent of its immunomodulatory function was also identified. Our results suggest that therapeutic strategies targeting microglia may represent an effective approach to prevent disease progression in the setting of tauopathy. National Institutes of Health https://doi.org/10.13039/100000002 NS090934 AG047644 JPB Foundation https://doi.org/10.13039/100007457 Cure Alzheimer’s Fund https://doi.org/10.13039/100007625 Washington University School of Medicine https://doi.org/10.13039/100011912 Children’s Discovery Institute https://doi.org/10.13039/100009340 CDI-CORE-2015-505 Foundation for Barnes-Jewish Hospital https://doi.org/10.13039/100007338 3770 ==== Body pmcIntroduction Alzheimer’s disease (AD) is characterized by two types of pathologies: 1) misfolded, aggregated proteins (in particular, intraneuronal tau pathology) that are capable of inducing direct neurotoxicity; and 2) chronic activation of the brain’s innate immune system that may induce neurodegeneration via various mechanisms, such as neuronal phagoptosis by microglia, complement-mediated neuronal damage, cytokine or reactive oxygen species–induced cell toxicity, etc. (Shi and Holtzman, 2018). However, to what degree innate immunity contributes to neurodegeneration as compared with tau-induced cell-autonomous neurotoxicity is unknown. APOE, the strongest genetic risk factor for late-onset AD, encodes apolipoprotein E (apoE), which by itself is a potent immune modulator (Vitek et al., 2009; Zhu et al., 2012; Gale et al., 2014; Krasemann et al., 2017; Shi et al., 2017), implying a critical role of brain innate immunity in AD pathogenesis. We recently reported that apoE strongly regulates neurodegeneration in the setting of a tauopathy (Shi et al., 2017). The presence of apoE, particularly apoE4, significantly exacerbated neurodegeneration in P301S mice, whereas genetic ablation of murine Apoe was strongly neuroprotective. These effects are in part due to apoE impacting glial function, in addition to its potential direct regulation of tau pathogenesis (Shi et al., 2017). However, to what extent apoE regulates neurodegeneration via its immunomodulatory function, and which glial cell types are involved in this process, are still elusive. While both microglia and astrocytes are immune competent, accumulating genetic and functional evidence suggests that microglia, the brain-resident professional immune cells, may play a major role in regulating innate immunity-induced neurodegeneration (Leyns et al., 2017; Pimenova et al., 2018). A recent study shows that apoE cell-autonomously regulates microglial activation (Krasemann et al., 2017). Apoe deletion from microglia prevented microglia from acquiring a neurodegenerative phenotype that was required for neuronal cell death in a facial nerve axotomy model. This highlights a role of microglia in neurodegeneration and suggests that apoE functioning through microglia constitutes an essential mechanism regulating neurodegeneration, at least in an acute axotomy model. To understand the role of microglia in neurodegeneration in neurodegenerative diseases and how it is linked to apoE function, we depleted microglia in P301S tau transgenic mice that are homozygous for human APOE4 (TE4) or with no expression of Apoe (TEKO) from 6 to 9.5 mo of age, a critical time window when neurodegeneration occurs in this mouse model. We identified a critical role of microglia in driving both neurodegeneration and tau pathogenesis, as well as in mediating apoE’s effect on neurodegeneration. Results Chow formula and mouse sex affect microglial depletion efficiency by PLX3397 To deplete microglia, we used PLX3397, a selective CSF1R/c-kit/FLT3 inhibitor that has been shown to readily cross the blood brain barrier and eliminate microglia via oral delivery in mouse chow (Elmore et al., 2014). We initially formulated PLX3397 in a grain-based chow (Purina 5053) that has been used for our previous research. However, we found a strong chow- and sex-dependent effect on the drug efficiency in depleting microglia. PLX3397 supplemented in grain-based Purina 5053 showed a significantly lower efficiency in depleting microglia compared with PLX3397 supplemented in a purified ingredient chow (AIN-76A) that has been used in previous PLX3397 research (Fig. S1, A and B). Additionally, male mice showed a higher level of microglial reduction than females when treated with the same chow (AIN-76A; Fig. S1, A and B). These effects appeared to be caused by different levels of PLX3397 in the plasma, as plasma PLX3397 levels strongly correlated with the degree of microglial reduction in the brain (Fig. S1, C and D). We therefore selected AIN-76A chow and used only male mice for the study. To guarantee that we achieved complete microglial ablation, we formulated the chow with a higher concentration of PLX3397 (400 mg/kg) as compared with the previously reported concentration (290 mg/kg). The higher drug concentration resulted in highly efficient microglial depletion, with 7-d treatment eliminating ∼90% of microglia and 21-d treatment eliminating virtually all microglia (Fig. S2). Microglial ablation during a critical time window of neurodegeneration development completely blocks neurodegeneration We treated 6-mo-old (186 d) TE4 and TEKO mice as well as their aged matched non-tau transgenic littermates with control or PLX3397-supplemented AIN-76A chow for 3 mo (99 d), and collected mouse brains at 9.5 mo of age (285 d). Consistent with our previous findings, TE4 mice treated with control chow showed severe brain atrophy predominantly in the hippocampus, piriform/entorhinal cortex, and amygdala, accompanied by significant dilatation of the lateral ventricle (Fig. 1). In contrast, TEKO mice were largely protected from tissue loss and showed preserved brain volume (Fig. 1). Strikingly, when microglia were depleted in TE4 mice for merely 3 mo, specifically during the stage when neurodegeneration begins and rapidly deteriorates, neurodegeneration was virtually fully blocked, and the brain volume of these mice remained the same as that of non-tau transgenic mice (Fig. 1). This strongly suggests that microglia are the driving force of neurodegeneration and disease progression in this tauopathy mouse model. Figure 1. Microglial depletion fully rescues neurodegeneration in TE4 mice and eliminates apoE’s effect on neurodegeneration. (A) Representative images of mouse brain sections stained with Sudan black for 9.5-mo-old TE4, TEKO, E4, and EKO mice treated with control or PLX3397-supplemented chow. Scale bar = 1 mm for all images. (B) Quantification of brain volumes in the hippocampus and the piriform/entorhinal cortex (Ent/Piri Ctx) of the mice (TE4-Ctl: n = 21; TE4-PLX: n = 17; TEKO-Ctl: n = 19; TEKO-PLX: n = 18; E4-Ctl: n = 12; E4-PLX: n = 6; EKO-Ctl: n = 13; EKO-PLX: n = 6). Data are expressed as means ± SEM. One-way ANOVA with Tukey’s post hoc test (two-sided) was used for statistical analyses. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. The Sudan black staining and brain volume quantification was performed once. ApoE regulates neurodegeneration in P301S mice predominantly by regulating microglia-mediated innate immune response As we had described previously (Shi et al., 2017), TEKO mice consistently showed a marked reduction of brain atrophy compared with TE4 mice (Fig. 1). While there was a slight trend, PLX3397-treated TEKO mice did not show a significantly increased brain volume compared with control TEKO mice (Fig. 1). The similar protective effects of microglial depletion and Apoe deletion on neurodegeneration suggest that apoE and microglia may function in the same axis in regulating neurodegeneration. Intriguingly, by depleting microglia, the effect of apoE on neurodegeneration was abolished, with PLX3397-treated TE4 and TEKO mice showing equal brain volumes (Fig. 1). This suggests that the effect of apoE on neurodegeneration is predominantly mediated by its immunomodulatory property in regulating microglial functions. However, this does not exclude the possibility that apoE directly regulates tau pathogenesis. ApoE deficiency leads to a higher resistance of microglia to PLX3397-induced cell death and keeps the surviving microglia in a homeostatic state To confirm microglial depletion in PLX3397-treated mice, we stained mouse brain sections with ionized calcium binding adaptor molecule 1 (Iba1) to assess the total microglia population. We found that PLX3397 treatment resulted in ∼100% microglial depletion in nearly all TE4 and E4 mice throughout the brain (Fig. 2, A and B), with only a few TE4 mice retaining a small pool of microglia in the piriform/entorhinal cortex and even sparser microglia in the hippocampus (Fig. 2 D). These Iba1+ microglia almost invariably co-localized with a microglial activation marker, cluster of differentiation 68 (CD68; Fig. 2 E), indicating an activated status. A previous study showed that Csf1r expression is down-regulated in activated microglia in neurodegenerative conditions (Krasemann et al., 2017), suggesting that activated microglia are less dependent on the CSF1R signaling pathway, and therefore may be less susceptible to apoptosis induced by blockage of CSF1R signaling. Interestingly, in addition to sparse Iba1+CD68+ microglia in the gray matter, there were also significant CD68+Iba1− signals in the white matter, particularly in the fimbria (Fig. 2 E), suggesting two different types of activated microglia in the hippocampus. The retention of CD68+ microglia in these drug-treated TE4 mice correlated with a higher level of neurodegeneration in respective brain regions (Fig. 2 F), suggesting a linkage between activated microglia and neurodegeneration. Figure 2. ApoE deficiency increases homeostatic microglial survival in PLX3397 treatment. (A) Representative images of Iba1 staining for 9.5-mo-old TE4, TEKO, E4, and EKO mice treated with control (Ctl) or PLX3397-supplemented (PLX) chow. Scale bar = 1 mm for all images. (B) Quantification of Iba1-covered areas in these brain sections (TE4-Ctl: n = 21; TE4-PLX: n = 17; TEKO-Ctl: n = 19; TEKO-PLX: n = 18; E4-Ctl: n = 12; E4-PLX: n = 6; EKO-Ctl: n = 13; EKO-PLX: n = 6). (C) Quantification of Iba1-covered area in PLX3397-treated mice alone. (D) A representation image of a TE4 mouse that showed incomplete microglial depletion in the brain. Scale bar = 1 mm. (E) Co-localization of Iba1 and CD68 signals in the surviving microglia in the hippocampus and entorhinal/piriform cortex (Ent/Piri Ctx) of a TE4 mouse. Scale bar = 500 μm. (F) The presence of surviving CD68+ microglia in some PLX3397-treated TE4 mice was associated with higher levels of neurodegeneration in the hippocampus and entorhinal/piriform cortex. (G) The surviving microglia in PLX3397-treated TEKO mice were predominantly CD68− in the hippocampus, but showed a higher degree of co-localization with CD68 in the entorhinal/piriform cortex. Hippocampus and piriform/entorhinal cortex share the same scale with the respective regions in E. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. The fluorescent stainings were performed once. Interestingly, Apoe ablation, particularly in the presence of the tau transgene, increased microglial resistance to PLX3397-induced cell death, resulting in a higher retention of Iba1+ microglia not only in the hippocampus and piriform/entorhinal cortex, but also occasionally in other brain regions such as thalamus (Fig. 2, A and C). However, unlike surviving microglia in TE4 mice, the remaining microglia in TEKO mice were largely negative for CD68 (Fig. 2 G), indicating maintenance of a homeostatic status. The enhanced survival of nonactivated microglia in Apoe-deficient mice suggests that apoE regulates CSF1R signaling in microglia and likely affects downstream functions such as cell survival, proliferation, and differentiation. Apoe ablation may reprogram microglia to be less dependent on CSF1R signaling or raise the threshold for microglial apoptosis by up-regulating CSF1R expression. Therefore, microglial preservation in TE4 and TEKO mice upon PLX3397 treatment likely followed two distinct mechanisms. The retention of nonactivated microglia in TEKO mice is likely due to an intrinsically higher resistance to or a higher tolerance of PLX3397-induced apoptosis by apoE-deficient microglia, whereas the CD68+ microglia retained in TE4 mice likely came from enhanced survival of microglia that had been activated before PLX3397 treatment, owing to the accelerated disease progression in TE4 mice. The dissociation between Iba1 and CD68 signals in PLX3397-treated TEKO mice was most prominent in the hippocampus, where >95% of remaining Iba1+ microglia were CD68 negative, whereas in the piriform/entorhinal cortex, a higher and more variable degree of co-localization was observed (Fig. 2 G). The emerging CD68+ microglia in PLX3397-treated TEKO mice was most likely due to chronic activation of the initially surviving nonactivated microglia during disease progression at a later stage. The variation of the CD68+ microglia levels between the hippocampus and piriform/entorhinal cortex indicates relatively independent disease progression in these two brain regions, with the piriform/entorhinal cortex appearing to be affected earlier than the hippocampus. The activation status of microglia, but not the proliferation of microglia, is associated with neurodegeneration Interestingly, we observed no difference in the percent area covered by Iba1 immunoreactivity between TE4 and TEKO mice treated with the control chow (Fig. 2, A and B), despite a significant difference of brain volume between these mice. In contrast, there was a significant reduction of CD68 immunoreactivity in control chow-treated TEKO mice compared with TE4 mice (Fig. 3, A and B). Furthermore, the amount of CD68 immunoreactivity highly correlated with the degree of brain atrophy (Fig. 3 C). This supports the idea that the activation status of microglia, but not their absolute number, determines microglial neurodegenerative functions. On the other hand, we observed no difference in the level of glial fibrillary acidic protein (GFAP)+ astrocytes in TE4 and TEKO mice treated with either control or PLX3397-supplemented chow (Fig. S3), suggesting that the number of GFAP+ astrocytes is not directly linked with neurodegeneration. Figure 3. Activated microglia are associated with neurodegeneration. (A) Representative images of CD68 staining in the hippocampus and entorhinal/piriform cortex (Ent/Piri Ctx) of 9.5-mo-old TE4 and TEKO mice treated with control (Ctl) or PLX3397-supplemented (PLX) chow. Scale bar = 500 μm. Images of the same brain region share the same scale. (B) Quantification of CD68-covered area in all groups of mice (TE4-Ctl: n = 21; TE4-PLX: n = 17; TEKO-Ctl: n = 19; TEKO-PLX: n = 18; E4-Ctl: n = 12; E4-PLX: n = 6; EKO-Ctl: n = 13; EKO-PLX: n = 6). (C) Correlation between CD68 signal and the brain volume in the hippocampus and entorhinal/piriform cortex in control TE4 and TEKO mice. Pearson correlation analysis (two-sided). Hippocampus: R2 = 0.6872, P < 0.0001; entorhinal/piriform cortex: R2 = 0.5611, P < 0.0001. Data are expressed as means ± SEM. One-way ANOVA with Tukey’s post hoc test. **, P < 0.01; ****, P < 0.0001. The CD68 staining was performed once. Microglia are the driving force of tau pathogenesis and mediate apoE’s effect on tau pathology Microglial activation has been shown to promote tau phosphorylation by activating tau kinases via IL-1 signaling (Li et al., 2003; Kitazawa et al., 2005). To understand how tau pathology develops with little to no impact of microglia and how apoE regulates tau pathogenesis independent of its effect on microglia, we analyzed tau pathology in mouse brains. We first stained brain sections with the AT8 antibody that recognizes the Ser202 and Thr205 epitopes of phosphorylated tau (ptau). We previously identified four types of ptau staining patterns (type1–type4) in P301S tau transgenic mice that are associated with progressively more advanced pathological tau/neurodegeneration stages (Shi et al., 2017). Consistent with our previous findings, TEKO mice treated with control chow were highly enriched with the type1 pattern, which represents the earliest pathological tau stage, whereas TE4 mice treated with control chow showed predominantly type3 and type4 patterns, which are associated with advanced tau stages and neurodegeneration (Fig. 4, A and B). Remarkably, microglial depletion in TE4 mice from 6 mo of age, a time point shortly after tau pathology onset, completely blocked the progression of pathological tau stages, resulting in almost exclusively a type1 pattern in these mice (Fig. 4 B). This result strongly indicates that microglia are the primary driving force of ptau histopathology. Activated microglia either directly induce tau hyperphosphorylation by activating tau kinases through IL-1 or other signaling pathways, or trigger neurodegeneration that leads to secondary tau hyperphosphorylation and aggregation as a consequence of neuronal injury. Microglial depletion in TEKO mice did not further change the ptau staining pattern as compared with TEKO control mice, as both groups showed similar enrichment of type1 tau (Fig. 4 B). Quantification of AT8 immunoreactivity in the hippocampus revealed a significant ptau reduction in PLX3397-treated TE4 mice compared with control TE4 mice, with no difference noted between PLX3397-treated and control TEKO mice (Fig. 4 C). However, we noticed that type1 tau exhibited two sub-categories: an early-type1 stage characterized by solely mossy fiber staining and a late-type1 stage featuring emerging neuronal cell body and neuritic staining in addition to mossy fiber staining (Fig. 4 D). We found that control TEKO mice demonstrated a significant shift from the early-type1 tau toward the late subtype relative to PLX3397-treated TEKO mice (Fig. 4 E), indicating a mildly accelerated rate of disease progression. Notably, in the absence of microglia, TE4 and TEKO mice showed comparable degrees of ptau pathology with similar ptau patterns and AT8 coverage, suggesting that the effect of apoE on tau histopathology is primarily mediated by apoE modulating microglial function. Figure 4. Microglia drive tau pathogenesis and mediate apoE’s effect on tau pathogenesis. (A) Representative images of the four major ptau staining patterns in 9.5-mo-old TE4 and TEKO mice. Scale bar = 500 μm for all images. (B) Distribution of the four ptau staining patterns in 9.5-mo-old TE4 and TEKO mice treated with control (Ctl) or PLX3397-supplemented (PLX) chow (TE4-Ctl: n = 21; TE4-PLX: n = 17; TEKO-Ctl: n = 19; TEKO-PLX: n = 18). Fisher’s exact test, two-sided (all groups: P = 2.2 × 104; TE4-Ctl vs. TE4-PLX: P = 4.1 × 104; TE4-Ctl vs. TEKO-Ctl: P = 4.1 × 104). (C) Quantification of ptau (AT8)-covered area in the hippocampus of these mice. (D) Representative images of early-type1 and late-type1 ptau staining patterns within the type1 category (same image used for type1 and type1-early in A and D, respectively). Scale bar = 500 μm for both images. (E) Distribution of early-type1 and late-type1 ptau patterns in TE4 and TEKO mice harboring the type1 ptau staining pattern (TE4-Ctl: n = 6; TE4-PLX: n = 15; TEKO-Ctl: n = 11; TEKO-PLX: n = 11). Fisher’s exact test, two-sided (all groups: P = 0.0097; TEKO-Ctl vs. TEKO-PLX: P = 0.0089). Data are expressed as means ± SEM. One-way ANOVA with Tukey’s post hoc test (two-sided). **, P < 0.01; ***, P < 0.001. The AT8 staining was performed once. We next performed biochemical analysis for these samples. Mouse posterior cortical tissue was extracted sequentially in reassembly buffer (RAB; salt buffer), radioimmunoprecipitation assay buffer (RIPA; detergent buffer), and 70% formic acid (FA), which contain soluble, less soluble, and highly insoluble proteins, respectively. Ptau and human tau levels in each fraction were analyzed via quantitative ELISA. We found that PLX3397 treatment significantly reduced ptau levels in both TE4 and TEKO mice (Fig. 5 A). Surprisingly, while comparable human tau levels between groups were observed in most fractions, there was a significant increase of insoluble tau level in PLX3397-treated TE4 mice relative to control TE4 mice, despite preserved brain volume in PLX3397-treated TE4 mice (Fig. 5 B). The elevation of insoluble tau was likely a secondary effect of PLX3397 treatment and appeared to be associated with Apoe status, as no difference was observed between control and PLX3397-treated TEKO mice. The concurrence of increased insoluble tau and attenuated brain atrophy suggests that formation of insoluble tau does not directly cause neurodegeneration manifested as brain volume loss. Figure 5. PLX3397 treatment in TE4 mice results in a sharp elevation of soluble apoE level that correlates with higher ptau and insoluble human tau levels. (A and B) ELISA results showing concentrations of ptau and human tau in RAB, RIPA, and FA fractions, respectively, for 9.5-mo-old TE4 and TEKO mice treated with control (Ctl) or PLX3397-supplemented (PLX) chow (TE4-Ctl: n = 21; TE4-PLX: n = 17; TEKO-Ctl: n = 19; TEKO-PLX: n = 18). (C) Human apoE level in the RAB fraction for 9.5-mo-old TE4 and E4 mice treated with control or PLX-supplemented chow (TE4-Ctl: n = 21; TE4-PLX: n = 17; E4-Ctl: n = 12; E4-PLX: n = 6). (D) Western blot probing for human apoE in the RAB and RIPA fractions for 9.5-mo-old TE4 and E4 mice treated with control or PLX3397-supplemented chow (n = 3 per group). (E) Correlation of RAB-soluble apoE level with soluble and insoluble human tau levels, respectively. Pearson correlation analysis (two-sided). RAB apoE vs. FA tau: R2 = 0.6240, P = 0.0002; RAB apoE vs. RAB tau: R2 = 0.1581, P = 0.1. (F) Correlation of RAB-soluble apoE level with ptau levels in the RAB, RIPA, and FA fractions, respectively, in PLX3397-treated TE4 mice. Pearson correlation analysis (two-sided). RAB apoE vs. RAB ptau: R2 = 0.8039, P < 0.0001; RAB apoE vs. RIPA ptau: R2 = 0.7312, P < 0.0001; RAB apoE vs. FA ptau: R2 = 0.5369, P = 0.0008. Data are expressed as means ± SEM. One-way ANOVA with Tukey’s post hoc test. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. All ELISAs were repeated twice. PLX3397 treatment in TE4 mice induces a marked elevation of soluble apoE present in astrocytes and neurons that positively correlates with ptau and insoluble human tau levels Based on the protective effect of apoE deficiency on neurodegeneration, we reasoned that PLX3397 treatment may reduce apoE levels in TE4 mice. However, we unexpectedly observed a four to fivefold increase of soluble apoE levels in PLX3397- treatedTE4 and E4 mice relative to their control chow-treated counterparts, with TE4-PLX mice showing an even higher level of apoE compared to E4-PLX mice (Fig. 5 C). Because the apoE ELISA assay we used is incompatible with RIPA and FA fractions, we were only able to measure apoE levels in the RAB fraction. To exclude the possibility that we were detecting only a specific pool of apoE by ELISA using the monoclonal apoE antibodies and to assess apoE levels in other fractions, we performed immunoblotting for both the RAB and RIPA fractions using a polyclonal apoE antibody. We confirmed a marked elevation of apoE levels predominantly in the RAB fraction in PLX3397-treated TE4 and E4 mice as compared with control mice (Fig. 5 D), indicating a primarily soluble state of elevated apoE. To understand where the elevated apoE came from, we co-stained apoE with various cell type markers in brain sections. Compared with control TE4 and E4 mice, there was a strong increase in the intensity of apoE signal that largely co-localized with GFAP+ astrocytes in both PLX3397-treated TE4 and E4 mice (Fig. 6, A–F). Diffuse apoE staining was also enhanced (Fig. 6, A–D), and no apoE signal was observed in oligodendrocytes (Fig. 6 G). Intriguingly, in PLX3397-treated TE4 mice, apoE staining was also present in cells of neuronal morphology, the signal of which was particularly strong in the dentate gyrus (Fig. 6, A and H–J), CA3 of the hippocampus, and the piriform/entorhinal cortex (Fig. S4). This type of signal was barely observed in PLX3397-treated E4 mice except weakly in the piriform/entorhinal cortex (Fig. S4, A and B), and was not at all present in control TE4 or E4 mice. This apoE staining pattern was verified by five different monoclonal and polyclonal apoE antibodies (data not shown) and by no apoE staining in PLX3397-treated or control TEKO/EKO mice (Fig. 6 E and Fig. S4, A and B). The apoE+ neuronal shaped cells did not co-localize with neuronal nuclei (NeuN), a marker for mature neurons (Fig. 6 I and Fig. S4 D), but partially co-localized with ptau-positive neurons (Fig. 6 J), indicating that they very likely represent a subset of neurons. They may be neuronal precursor cells or immature neurons that have been reported to express CSF1R (Nandi et al., 2012), or they may be CSF1R-expressing mature neurons (Luo et al., 2013) that have switched from being NeuN+ to NeuN− via transcriptional reprogramming in response to intracellular apoE buildup (either due to de novo apoE synthesis or apoE uptake and accumulation). The number of apoE+ neurons in PLX3397-treated TE4 mice varied across samples and appeared to correlate with the amount of ptau signal in individual samples, suggesting that tau pathology likely increases the sensitivity of these neurons to CSF1R signaling and promotes neuronal apoE buildup upon CSF1R blockage. Overall, while elevated apoE in PLX3397-treated E4 and TE4 mice primarily came from astrocytes, the additional elevation of apoE in TE4-PLX mice over E4-PLX mice appeared to be due to an increased apoE level in subsets of neurons. Figure 6. Elevated apoE in PLX3397-treated TE4 mice is present in astrocytes and a subset of NeuN− neurons. (A–E) Representative images of apoE (green, first column), GFAP (red, second column), and merged apoE/GFAP (third column) staining in the hippocampus of 9.5-mo-old TE4-PLX, TE4-Ctl, E4-PLX, E4-Ctl, and TEKO-PLX mice, respectively. Scale bar = 50 μm for all images in the three columns. Far left column: zoom-in of the selected area in respective apoE staining images. Scale bar = 10 μm. (F) Co-localization of apoE with GFAP+ astrocytes in TE4-PLX mouse hippocampus. Scale bar = 10 μm. (G) Non-co-localization between apoE and CNPase+ oligodendrocytes in TE4-PLX mouse hippocampus. Scale bar = 10 μm. (H) Non-co-localization between apoE and NeuN in TE4-PLX mouse dentate gyrus. Scale bar = 50 μm. (I) Zoom-in of the selected area in H. Scale bar = 10 μm. (J) Partial co-localization of apoE with AT8+ neurons in TE4-PLX mouse dentate gyrus. Arrowheads point to some co-localized neurons. Scale bar = 10 μm. For F–J, the third column is a merged image of the first two columns. ApoE staining was repeated seven times using different antibodies. Interestingly, we found that the level of RAB-soluble apoE correlated with a lower level of soluble tau and a higher level of insoluble tau in PLX3397-treated TE4 mice (Fig. 5 E). In addition, there was a strong positive correlation between apoE level and ptau levels in all three fractions (Fig. 5 F). These correlations remained significant even after excluding the samples that showed incomplete microglial depletion (data not shown), indicating that apoE derived from astrocytes and likely neurons may directly regulate tau pathogenesis by promoting ptau generation and insoluble tau formation independent of the impact of microglia. However, this direct effect of apoE on tau pathogenesis by itself was insufficient to impact neurodegeneration. The neuroprotective effect of PLX3397 is unlikely due to its direct impact on neurons Previous studies report CSF1R expression in small subsets of neurons under normal conditions, and the expression goes up with neuronal injury (Wang et al., 1999; Luo et al., 2013). Therefore, it is possible that PLX3397 protects against neurodegeneration via a direct effect on neurons. However, neuronal CSF1R signaling has been shown to promote neuronal survival (Wang et al., 1999; Nandi et al., 2012; Luo et al., 2013; Chitu et al., 2016). Therefore, assuming PLX3397 indeed affects neuronal function, it would be likely to have a detrimental rather than protective effect on neurons. To understand whether and how PLX3397 might directly impact neurons, we cultured hippocampal neurons derived from apoE4 knockin (KI) mice in cytarabine (AraC)-supplemented medium and infected the neurons with adeno-associated virus (AAV)2/8-Syn-P301S Tau. We treated the neurons with either DMSO or 8.5 μM PLX3397 (comparable to PLX3397 concentration in drug-treated mouse brains) at plating and maintained the culture for 8 d. No astrocytes or microglia were detected by the end of the experiment. At a lower plating density (40,000 cells/well), PLX3397 did not affect neuronal cell number, but mildly inhibited neurite outgrowth (Fig. 7, A and B). At a higher density (100,000 cells/well), the inhibition of neurite outgrowth by PLX3397 became highly prominent, and a mild reduction of neuron number was observed (Fig. 7, C and D). These results were consistent with previous studies showing that CSF1R signaling promotes neuronal survival and neuronal differentiation (Wang et al., 1999; Nandi et al., 2012; Luo et al., 2013; Chitu et al., 2016), suggesting that the neuroprotective effect of PLX3397 in TE4 mice was likely not due to a direct effect of PLX3397 on neurons. Figure 7. PLX3397 does not have a direct protective effect on neurons. (A and B) Primary hippocampal neurons derived from E17 apoE4 KI mouse embryos were infected with AAV2/8-Syn-P301S tau, and cultured at a density of 40,000 or 100,000 cells/well, respectively, in 24-well plates for 8 d in AraC-supplemented medium. The neurons were treated with either DMSO or 8.5 μM PLX3397 at plating. Representative images of neurons stained with MAP2. Scale bar = 50 μm for all images. (C and D) Quantification of neuron cell numbers in each imaging field. For 40,000 cells/well: n = 8 replicates per group; for 10,000 cells/well: n = 3 replicates per group. Data are expressed as means ± SEM. Unpaired two-tailed Student’s t test. *, P < 0.05. This experiment was performed once. Effects of PLX3397 treatment on blood cell phenotyping To understand whether PLX3397 affects blood cell populations other than microglia in the brain, we performed complete blood count (CBC) for mouse whole blood as well as flow cytometry for white blood cells in all groups of mice. CBC results showed that PLX3397 treatment significantly reduced the number of RBCs (RBC and hematocrit %) and hemoglobin levels (hemoglobin and mean corpuscular hemoglobin concentration), while increasing the size of RBCs (mean corpuscular volume) in all groups (Fig. 8 A). The number of platelets was also reduced (Fig. 8 B). Interestingly, PLX3397 treatment reduced the total number of white blood cells in E4 and TE4 mice, but not in EKO and TEKO mice (Fig. 8 C), suggesting an apoE-dependent effect on white blood cells under PLX3397 treatment. Detailed analysis by flow cytometry (gating strategy summarized in Fig. S5) revealed that PLX3397 treatment reduced or showed a trend of reducing the frequency of dendritic cells (DCs), natural killer (NK) cells, and Ly6C− monocytes in all groups of mice (Fig. 8, D–F). PLX3397 preferentially targeted Ly6C− monocytes over Ly6C+ monocytes, including Ly6Chi and Ly6Clo monocytes (Fig. 8 F). We found that in control chow-treated mice, Ly6C+ monocytes expressed a significantly higher level of CSF1R (CD115) compared with Ly6C− monocytes (Fig. 8 G), suggesting a higher tolerance to CSF1R inhibition-induced cell death by Ly6C+ monocytes. Interestingly, PLX3397 treatment led to a notable reduction of CSF1R expression in Ly6C+ monocytes and a milder CSF1R reduction in Ly6C− monocytes, with all surviving monocyte populations expressing a similarly low level of CSF1R (Fig. 8 G). Therefore, while the initial CSF1R expression level may be a key factor determining the susceptibility of monocytes to PLX3397-induced cell death, blockage of CSF1R signaling down-regulated CSF1R expression in the surviving monocytes and converted them to a phenotype that was less dependent on CSF1R signaling. No difference in neutrophil levels was observed between groups, whereas eosinophils showed a trend of increased frequency upon PLX3397 treatment (data not shown). Interestingly, EKO and TEKO mice showed an overall increased frequency of multiple myeloid cell types compared with E4 and TE4 mice, suggesting that apoE deficiency is associated with a higher proliferative capacity of certain myeloid cells. On the other hand, there appeared to be an overall trend of increased T cell frequency (Fig. 8 H) upon PLX3397 treatment. This effect was most pronounced in TEKO mice and was observed in all the T cell subpopulations, including regulatory T cells (T reg cells; Fig. 8 I), CD4+ T cells, CD8+ T cells, memory T cells, and NK T cells (other data not shown). In addition, there was a trend of increased B cell frequency in control TE4 mice over other groups (Fig. 8 J). The significance of these blood cell changes and their potential influence on brain pathogenesis and neurodegeneration are still unclear, and need to be further addressed in future studies. Figure 8. PLX3397 treatment and apoE status regulate blood cell phenotyping. (A–C) Complete blood cell count results for the level of RBCs, hemoglobin, platelets, and white blood cells in 9.5-mo-old mice of all genotypes treated with control (Ctl) or PLX3397-supplemented (PLX) chow (TE4-Ctl: n = 14; TE4-PLX: n = 7; TEKO-Ctl: n = 11; TEKO-PLX: n = 10; E4-Ctl: n = 13; E4-PLX: n = 8; EKO-Ctl: n = 10; EKO-PLX: n = 7). (D–F) Flow cytometry quantification of the frequency of DCs, NK cells, Ly6Chi, Ly6Clo, and Ly6C− monocytes, respectively (TE4-Ctl: n = 7; TE4-PLX: n = 7; TEKO-Ctl: n = 7; TEKO-PLX: n = 5; E4-Ctl: n = 11; E4-PLX: n = 8; EKO-Ctl: n = 5; EKO-PLX: n = 5). (G) Median fluorescence intensity (MFI) of CSF1R (CD115) in Ly6Chi, Ly6Clo, and Ly6C− monocytes in all groups of mice (two-way ANOVA with Sidak’s post hoc test). (H–J) Population frequencies for T cells, T reg cells, and B cells, respectively. Data are expressed as means ± SEM. One-way ANOVA with Tukey’s post hoc test. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. HCT, hematocrit; HGB, hemoglobin; MCHC, mean corpuscular hemoglobin concentration; MCV, mean corpuscular volume . The CBC and flow cytometry experiments for blood cell phenotyping were performed >10 times. Discussion AD and many other neurodegenerative diseases are known as proteopathies characterized by accumulation of misfolded and aggregated proteins. For decades, research and drug development in the AD field have largely focused on the two pathological protein hallmarks of AD, amyloid-β and tau pathology. Given a high correlation between tau pathology and AD clinical readouts (Arriagada et al., 1992; Josephs et al., 2008), pathological tau-induced cell-autonomous neurotoxicity was believed to play a key role in driving neurodegeneration. It was not until recently that the role of brain innate immunity in AD and tauopathy pathogenesis became increasingly recognized. While it is widely acknowledged that chronic innate immune activation in the brain can cause neuronal loss, the interaction between innate immunity and tau pathology and their respective contributions to neurodegeneration are not well understood. This is largely due to an intertwined relationship between tau pathogenesis and innate immunity, making it difficult to separate the two effects. With complete microglial depletion, we were able to assess the impact of microglia on tau pathogenesis, as well as the effect of tau pathology on neurodegeneration, independent of the influence of microglia-mediated inflammation on neurodegeneration. Here, we demonstrated that microglia served as the driving force of neurodegeneration in a mouse model of tauopathy. Depletion of microglia in TE4 mice upon neurodegeneration onset fully blocked development of brain atrophy. Importantly, in the absence of microglia, progression of ptau pathology was halted, indicating that ptau pathogenesis is also predominantly driven by microglia-mediated inflammation. It is unclear whether the rescue of neurodegeneration in the absence of microglia resulted from a direct attenuation of inflammation-induced neuronal loss or was an indirect result of blocked ptau pathology due to lack of tau-induced neurotoxicity. While we cannot exclude the role of tau pathology in inducing neurodegeneration, our results suggest that ptau pathology progresses very slowly in the absence of microglia. Whether tau pathology could progress to an advanced stage within a normal lifespan independent of microglia to affect neurodegeneration is uncertain. In addition, PLX3397-treated TE4 mice showed a preserved brain volume despite harboring an increased level of insoluble tau compared with control TE4 mice, suggesting that certain forms of pathological tau do not directly cause neuronal death. This is consistent with previous reports that tau tangles are off the pathway of acute neuronal death (de Calignon et al., 2009, 2010; Kuchibhotla et al., 2014). Therefore, microglial activation, rather than pathological tau-induced neurotoxicity, is likely the key factor dictating disease progression in this model. We did not observe a difference of GFAP immunoreactivity between PLX3397-treated and control groups, suggesting that GFAP+ astrocytes do not directly regulate neurodegeneration. However, it is possible that astrocytes crosstalk with microglia to influence neurodegeneration. Regardless, the fact that microglial depletion is sufficient to block tau pathology and neurodegeneration suggests that microglia play a key role in mediating neurodegeneration. Interestingly, a previous study showed that a mild microglia reduction (∼30%) by PLX3397 treatment did not change tau burden or cortical atrophy in a tauopathy mouse model (Bennett et al., 2018). We found that the retention of even a small population of activated microglia was associated with more neurodegeneration compared with complete microglial eradication, suggesting a potent role of activated microglia in regulating neurodegeneration and the requirement of depletion of the vast majority of microglia in order to reveal their effects on neurodegeneration. We surprisingly found a mouse chow- and sex-dependent effect on the efficiency of microglial depletion, revealing the complexity of immune regulation that can be easily neglected. While the effects appeared to be due to differences in plasma PLX3397 levels, whether plasma PLX3397 concentration is affected by food digestion/absorption, drug catabolism regulated by blood cell composition that is potentially affected by chow type or sex, or other factors is unknown. We found that the chow formulas indeed could affect blood cell phenotypes. For instance, Purina 5053 resulted in an increase of the neutrophil population and a concomitant down-regulation of the lymphocyte population compared with AIN-76A (data not shown). These variables should be taken into consideration in future studies. We identified a critical difference between microglial proliferation and microglial activation, the concepts of which have not always been clear in previous studies. Control TEKO mice showed no difference and even a trend of higher level of Iba1+ microglia compared with control TE4 mice, despite a marked rescue of brain volume loss in TEKO mice. In contrast, the level of CD68+-activated microglia was notably lower in control TEKO mice and highly correlated with the level of neurodegeneration, suggesting that the activation status of microglia, rather than the absolute number of microglia, is functionally associated with neurodegeneration. Microglial proliferation represents one aspect of microglial biology upon stimulation, but is not necessarily linked with transcriptional programming that converts microglia to an activated neurodegenerative phenotype. ApoE deficiency does not block the proliferating capacity of microglia. In fact, it appears to increase microglial numbers under certain stimulating conditions, such as mild neurodegeneration as was observed in control TEKO mice, or LPS stimulation as was reported in previous literature (Zhu et al., 2012). We also observed a larger population of certain blood myeloid cells in EKO and TEKO mice relative to TE4 and E4 mice. In addition, TEKO and EKO mice presented a higher resistance to PLX3397-induced microglial cell death relative to TE4 and E4 mice, with preservation of predominantly nonactivated microglia. This evidence suggests that apoE deficiency promotes the proliferation and survival of certain myeloid cell lineages but may interfere with their functional maturation. In the brain, apoE deficiency prevents microglia from acquiring an activated status featuring a neurodegenerative phenotype that promotes neural injury (Krasemann et al., 2017). How apoE controls the survival, proliferation, and activation of microglia is unclear. One possibility is through regulating CSF1R signaling. However, detailed studies are needed in the future to address this question. We previously uncovered a significant role of apoE in regulating neurodegeneration in the setting of tauopathy. We proposed two major mechanisms through which apoE modulates neurodegeneration: 1) apoE regulates tau pathogenesis (i.e., progression of ptau stages); and 2) apoE regulates glia-mediated innate immunity. However, the respective contribution of the two mechanisms to apoE’s effect on neurodegeneration was unclear, given that the effect of apoE on tau pathogenesis was obscured by effects of apoE on innate immunity. Our current experimental scheme allows simultaneous assessment of the two aspects. We found that the effect of apoE on neurodegeneration was eliminated in the absence of microglia, suggesting a central role of microglia in mediating apoE’s effect on neurodegeneration. Importantly, effects of apoE on ptau stage progression and ptau immunoreactivity were also abolished upon microglial depletion, suggesting that the effect of apoE on tau pathogenesis was primarily driven by the effect of apoE on microglial function, and there was no direct effect of apoE on tau histopathology in the absence of microglia. This supports the notion that the effect of apoE on neurodegeneration in the tauopathy model predominantly results from apoE regulating microglial function. It is possible that the lack of apoE’s effect on tau pathology in PLX3397-treated TE4 and TEKO mice was complicated by the significantly higher level of retained microglia in TEKO mice relative to TE4 mice. These excessive microglia in TEKO mice, although mostly being CD68− that were not associated with neuronal loss, may still be able to affect tau pathogenesis, for instance, by producing factors to induce tau phosphorylation. To assess the effect of apoE on tau pathogenesis completely independent of the impact of microglia, we analyzed the level of apoE, tau, and ptau only in PLX3397-treated TE4 mice that showed complete elimination of microglia. Interestingly, we found a significant positive correlation of salt-soluble brain apoE levels with both ptau and insoluble tau levels in these mice, suggesting a potential direct role of apoE present in astrocytes and neurons in promoting ptau and insoluble tau formation. However, this effect of apoE on pathological tau generation did not contribute to brain volume loss and therefore did not significantly account for apoE’s effect on neurodegeneration. Our previous finding that apoE deficiency strongly protects against neurodegeneration suggests that lowering apoE levels may be beneficial. Here, we surprisingly found a marked elevation of apoE in both astrocytes and neurons in TE4 mice upon PLX3397 treatment. The elevated apoE level in PLX3397-treated TE4 mice may be related to the higher insoluble tau level in these mice, given the correlation between apoE and insoluble tau levels. However, in contrast to control TE4 mice that showed severe neurodegeneration, PLX3397-treated TE4 mice showed full rescue of brain volume despite a higher apoE level. This result illustrates a critical concept that the absolute amount of apoE is not the only key to neurodegeneration. The effect of apoE on neurodegeneration is dependent on the presence of microglia. How does apoE function through microglia to induce neurodegeneration? On the one hand, microglia-derived apoE cell-autonomously regulates the acquisition of the neurodegenerative phenotype by microglia that leads to neural injury. It is possible that the source and location of apoE, in addition to the quantity of apoE, matter for neurodegeneration. ApoE present in astrocytes and neurons may regulate ptau and insoluble tau formation that is not associated with acute neuronal loss; whereas microglia-derived apoE controls neurodegenerative microglial activation and functions, and therefore may be essential for neurodegeneration. On the other hand, apoE may serve as an opsonin coating the surface of injured or dead neurons, and recruit activated microglia via an apoE–TREM2 interaction to promote neuronal phagoptosis by microglia (Shi and Holtzman, 2018), therefore exacerbating neurodegeneration. Neuroinflammation in AD is primarily driven by microglia and astrocytes. However, it’s likely that blood-circulating immune cells crosstalk with central nervous system glia cells, for instance, via cytokine exchange, to influence disease pathogenesis. Emerging evidence that gut microbiota regulate microglial maturation and function (Erny et al., 2015; Thion et al., 2018) supports this possibility. Previous literature reported inconsistent results regarding the effect of PLX3397 on blood monocytes (Chitu et al., 2012; Sluijter et al., 2014), and detailed studies of PLX3397’s impact on other blood cell types are rare. Here, we performed blood cell phenotyping for a full panel of major blood cell types to assess the effect of PLX3397 treatment, apoE genotype, and disease status on blood cell composition. We found that at a dose of 400 mg/kg chow, PLX3397 treatment had a significant impact on the myeloid-megakaryocyte-erythroid cell lineages, leading to a reduction of RBCs, platelets, DCs, and Ly6C− monocytes. NK cells showed a similar reduction despite being a lymphoid lineage. The enlarged cell body of erythrocytes indicates a slowing of DNA synthesis that prevents normal cell division (Hesdorffer and Longo, 2015). This suggests that PLX3397 may prohibit the proliferation of committed precursors of the affected cells. Granulocytes showed either an opposite pattern or no change, suggesting a relatively independent or compensatory regulation mechanism. Corresponding to an overall reduction of myeloid cells, PLX3397 increased the frequency of T cells, particularly in TEKO mice. The effect of apoE on blood cell phenotyping was mainly reflected in increased myeloid cell populations in apoE-deficient mice, which was in line with a higher cell proliferation capacity associated with apoE deficiency. Interestingly, B cell frequency was uniquely elevated in control TE4 mice over other groups, suggesting a potential link between B cell function and neurodegeneration. However, the meaning of these blood cell changes and whether they are functionally connected to microglia in the brain requires further investigation. In summary, we demonstrated that microglial activation, rather than tau-induced direct neurotoxicity, is the driving force of neurodegeneration in the setting of tauopathy. Critically, the progression of ptau pathology by itself is driven by microglia. In addition, microglia are the key mediator of apoE’s effect on neurodegeneration. Therefore, therapeutic strategies targeting microglia may be a highly effective approach to prevent disease progression. A deeper understanding of the neurodegenerative microglial signature and how to specifically engineer and control microglial functions may be a key to the design and development of microglia-targeted therapies. Materials and methods Animals P301S tau transgenic mice (#008169; The Jackson Laboratory) expressing human P301S 1N4R tau driven by the PrP promoter were backcrossed and maintained on a C57BL/6 background. Human APOE4 KI mice (C57BL/6) were provided by P.M. Sullivan, and Apoe KO mice (C57BL/6) were purchased from The Jackson Laboratory (#002052). P301S mice were crossed to APOE4 KI or Apoe KO mice to generate TE4 and TEKO mice, respectively. All animal procedures and experiments were performed under guidelines approved by the animal studies committee at Washington University School of Medicine. PLX3397 formulation in mouse chow. PLX3397 was provided by Plexxikon Inc. For the initial chow characterization study, PLX3397 was formulated in either the Purina 5053 (Lab Supply) chow or the AIN-76A (Research Diets) chow at a concentration of 300 mg/kg chow. For the formal study, PLX3397 was formulated in the AIN-76A (Research Diets) chow at a concentration of 400 mg/kg chow. PLX3397-supplemented chow characterization Mice were fed with either AIN-76A or Purina 5053 chow supplemented with 300 mg/kg PLX3397 for 7 d. The amount of chow consumption for each mouse was measured every 24 h over the 7-d period, and was normalized to mouse weight to obtain the amount of chow consumption per gram of mouse weight each day. This value was further converted to the amount of drug consumption per day per gram of mouse weight by multiplying the drug concentration and the drug incorporation rate in diet. Plasma drug concentrations and drug incorporation rate in diet were measured by Integrated Analytical Solutions, Inc., via liquid chromatography tandem-mass spectrometry. Brain section staining For AT8 histological staining, frozen brain sections were washed three times in Tris-buffered saline (TBS) buffer followed by incubation in 0.3% hydrogen peroxide in TBS for 10 min. Sections were then washed and blocked with 3% milk in 0.25% TBS-X (Triton X-100) for 0.5 h, followed by incubation at 4°C overnight with biotinylated AT8 antibody (MN1020B; Thermo Fisher Scientific). The next day, the slices were developed using the VECTASTAIN Elite ABC HRP Kit (Vector Laboratories) following the manufacturer’s instructions. Stained sections were imaged by the NanoZoomer digital pathology system, and pathology was quantified using Image J. For immunofluorescence staining, sections were washed in TBS three times and blocked in blocking buffer (3% BSA + 3% goat serum in 0.3% TBS-X) for 0.5 h at room temperature (RT), followed by overnight incubation at 4°C with primary antibodies in stain buffer (1% BSA + 1% goat serum in 0.3% TBS-X). ApoE and 2′,3′-cyclic nucleotide 3′-phosphodiesterase (CNPase) staining required antigen retrieval in sodium citrate buffer (10 mM sodium citrate, 0.05% Tween 20, pH 6.0) at 95°C for 30 min before blocking. The next day, the sections were washed in TBS and incubated with fluorescence-labeled secondary antibodies (Molecular Probes) for 1 h at RT. The slices were then washed and mounted in Prolong Gold antifade mounting media (P36931; Molecular Probes). Images for Iba1, CD68, and GFAP staining in Fig. 2, Fig. 3, and Fig. S3 were taken using the Cytation 5 imager at 4× magnification with image stitching and quantified using Image J. Images for apoE co-localization experiments were taken with a Nikon A1Rsi Confocal Microscope at 40× magnification. Images for Iba1 staining in the chow characterization experiments were taken by an epi-fluorescence microscope at 4× or 10× magnification and quantified using MetaMorph. Primary antibodies were as follows: CD68 (MCA1957, 1:500; AbD SeroTec), GFAP (MAB3402, 1:2,000; EMD Millipore), Iba1 (019–19741, 1:5,000; Wako), ApoE (13366S, 1:500; Cell Signaling), CNPase (836404, 1:500; BioLegend), and NeuN (MAB377, 1:100; EMD Millipore). Volumetric analysis Brain volume was quantified as described (Shi et al., 2017). Briefly, the left hemisphere of brain was sectioned coronally at 50 μm, and every sixth coronal brain section (300 μm between sections) starting rostrally at bregma +2.1 mm to the dorsal end of the hippocampus at bregma −3.9 mm was mounted for each mouse. All mounted sections were stained with 0.1% Sudan black in 70% ethanol at RT for 20 min and then washed in 70% ethanol. The sections were finally washed in Milli-Q water and coverslipped with Fluoromount. The stained slices were imaged with the NanoZoomer, and areas of interest were traced and measured in each slice using the NanoZoomer Digital Pathology viewer. The volume was calculated using the following formula: volume = (sum of area) × 0.3 mm. For hippocampus and posterior lateral ventricle, quantification started from bregma −1.1 and ended at bregma −3.9. For piriform/entorhinal cortex, quantification started at bregma −2.3 and ended at bregma −3.9. Brain extraction Mouse posterior cortical tissue was processed in RAB, RIPA, and 70% FA buffer sequentially. The tissue was first weighed and homogenized using an electric pestle at 10 μl buffer/1 mg tissue in RAB buffer (100 mM 2-(N-morpholino) ethanesulfonic acid, 1 mM EGTA, 0.5 mM MgSO4, 750 mM NaCl, 20 mM NaF, 1 mM Na3VO4, pH 7.0, supplemented by protease inhibitors [Complete; Roche] and phosphatase inhibitors [PhosSTOP; Roche]). After centrifugation at 50,000 ×g for 20 min, the supernatant was taken as the RAB-soluble fraction, and the pellet was dissolved in RIPA buffer (150 mM NaCl, 50 mM Tris, 0.5% deoxycholic acid, 1% Triton X-100, 0.1% SDS, 5 mM EDTA, 20 mM NaF, 1 mM Na3VO4, pH 8.0, supplemented by Complete and PhosSTOP) at 10 μl buffer/1 mg tissue by sonication at a 30% amplitude for 1 min. After centrifugation at 50,000 ×g for 20 min, the supernatant was taken as the RIPA-soluble fraction. The pellet was sonicated in 70% FA at 10 μl buffer/1 mg tissue using the same sonicating parameter, and centrifuged at 50,000 ×g for 20 min. The supernatant was taken as the FA-soluble fraction. All fractions were stored in −80°C until analyzed. ELISA The ELISA plates were coated with 20-μg/ml coating antibody overnight at 4°C. The next day, the plates were washed in PBS for five times, and mouse brain lysates were diluted in sample buffer (human tau and ptau sample buffer: 0.25% BSA in PBS with 300 μM Tris, pH 7.4; human apoE sample buffer: 0.5% BSA in 0.025% PBS-Tween 20 [PBST]; both buffers were supplemented with protease inhibitors) and loaded into the plates in duplicates. The plates were incubated at 4°C overnight. On the third day, the plates were washed in PBS, followed by addition of the detection antibody and incubation at 37°C for 90 min. The plates were then washed with PBS and incubated with streptavidin-poly-horseradish peroxidase-40 for 90 min at RT. Plates were then washed and developed with Super Slow ELISA 3,3′,5,5′-Tetramethylbenzidine (Sigma) and read at 650 nm. For FA fractions, samples were neutralized with 3 M Tris buffer (pH 8.0) before loading into the plate. Tau5 (gift from L. Binder, Northwestern University, Chicago, IL), HJ14.5 (in house anti-ptau antibody) and HJ15.6 (in house anti-human apoE antibody) were used as coating antibodies, and biotinylated HT7 (MN1000B; Thermo Fisher Scientific), biotinylated AT8 (MN1020B; Thermo Fisher Scientific), and biotinylated HJ15.4 (in house anti-human apoE antibody) were used as detection antibodies for human tau, ptau, and human apoE ELISAs, respectively. Western blot Samples from the RAB and RIPA fractions of mouse brain lysates were run for Western blot using 4–12% NuPAGE (Invitrogen) gels in 3-(N-morpholino) propanesulfonic acid buffer and transferred to nitrocellulose membranes. The membranes were blocked with 5% milk in 0.125% PBS-X for 30 min at RT, followed by incubation of primary antibodies (human apoE [ab24139; Abcam] and α-tubulin [T5168; Sigma] overnight at 4°C. The membranes were then washed with 0.0125% PBS-X and incubated with HRP-conjugated secondary antibodies (Santa Cruz) for 1.5 h at RT. Membranes were then washed and developed using the ChemiDoc MP Imaging System (Bio-Rad). Primary neuron culture and PLX3397 treatment Primary hippocampal neurons were obtained from E17 apoE4 KI mouse fetuses. Hippocampi were dissected in calcium- and magnesium-free HBSS with careful stripping of meninges. Tissue was digested in HBSS containing 0.25% trypsin (#15090-046; GIBCO) and 0.3 mg/ml DNase (#DN-25; Sigma) at 37°C for 10 min, and was dissociated in HBSS containing 0.4 mg/ml DNase using a fire-polished Pasteur glass pipette and filtered through a 70-μm nylon mesh. Filtered material was pelleted at 1000 ×g for 5 min, washed with neurobasal medium (Neurobasal + 1× B27 + 1× penicillin/streptomycin + 1× L-glutamine) once, and infected with AAV2/8-synapsin-P301S tau virus for 3 h on ice. Cells were then pelleted, washed, and plated at a density of 40,000 cells/well or 100,000 cells/well in 500 μl neurobasal medium onto 24-well tissue culture plates that had been coated with 10 μg/ml poly-L-lysine (#P2636; Sigma). At plating, 1 μM AraC was added and replenished at day in vitro 5 to prevent glia growth. The neurons were treated with 8.5 μM PLX3397 (dissolved in DMSO) or DMSO at plating and maintained for 8 d. 200 μl fresh medium containing 8.5 μM PLX3397 or DMSO was supplemented at day in vitro 5. Immunocytochemistry At DIV8, neurons were fixed in Dulbecco’s PBS containing 4% paraformaldehyde and 4% sucrose at RT for 10 min and permeabilized with 0.3% PBST for 10 min. After blocking in 0.1% PBS-X containing 3% BSA and 3% goat serum for 30 min at RT, cells were incubated with primary antibodies in stain buffer (1% BSA + 1% goat serum in 0.1% PBS-X) overnight at 4°C. The next day, cells were washed three times with 0.1% PBST and incubated in secondary antibodies for 1 h at RT. Cells were then washed three times in 0.1% PBST and mounted in Prolong Gold antifade mounting media (P36931; Molecular Probes). Images were taken using the Cytation 5 imager at 4× magnification using a scheme covering ∼70% of the total area of a well with image stitching for quantification. Representative images were taken at 10× magnification. Image J was used for quantification of neuron number. Primary antibodies were as follows: MAP2 (AB5543, 1:4,000; EMD Millipore), GFAP (MAB3402, 1:2,000; EMD Millipore), ALDH1L1 (ab190298, 1:400; Abcam), and Iba1 (019–19741, 1:5,000; Wako). Blood cell phenotyping Mouse blood was collected via cardiac puncture from anaesthetized mice into heparin-rinsed tubes. 60 μl of the whole blood of each animal was immediately sent to Washington University Division of Comparative Medicine Research Animal Diagnostic Laboratory for CBC in a double-blinded manner. The rest of the blood was first subjected to RBC lysis using the Pharm Lyse buffer (555899; BD Biosciences) following the manufacturer’s instructions. The remaining white blood cells were washed with PBS and stained with a fixable live/dead cell staining dye (L23102; Thermo Fisher Scientific) on ice for 30 min. The cells were then spun down and washed twice with stain buffer (554656; BD Biosciences), followed by incubation in stain buffer supplemented with mouse Fcγ receptor block reagent (553141; BD Biosciences) on ice for 15 min. A master mix of all the fluorescently tagged antibodies for cell surface markers was created by diluting the antibodies in the FcγR block–containing stain buffer supplemented with the brilliant stain buffer (563794; BD Biosciences). The master mix was then equally distributed to each individual blood sample. The cells were stained on ice for 30 min. After incubation, the cells were washed twice with stain buffer and fixed in the cytofix buffer (554655; BD Biosciences) on ice for 10 min. After fixation, the cells were spun down, resuspended in stain buffer, and run through flow cytometry using a BD LSRFortessa X-20 cell analyzer. The AbC Anti-Rat/Hamster Bead Kit (A10389; Thermo Fisher Scientific), AbC Anti-Mouse Bead Kit (A10344; Thermo Fisher Scientific), and ArC Amine Reactive Compensation Bead Kit (A10346; Thermo Fisher Scientific) were used for setting flow cytometry compensation. Each channel was assigned a fluorescence minus one control to set the threshold for the positive population. Around 90% of samples were collected of 50,000 cells in the CD45+ gate, and the remaining 10% of samples were collected of 100,000 cells in the CD45+ gate. Data analysis and figure generation were performed using FlowJo v10. Antibodies used for flow cytometry are as follows: BUV395 rat anti-mouse CD45 (clone 30-F11, #564279; BD Biosciences), APC-Cy7 rat anti-CD11b (clone M1/70, #557657; BD Biosciences), BV605 hamster anti-mouse CD11c (clone N418, #744179; BD Biosciences), BV650 rat anti-mouse I-A/I-E (clone M5/114.15.2, #563415; BD Biosciences), BV786 rat anti-mouse Siglec-F (clone E50-2440, #740956; BD Biosciences), Alexa Fluor 700 rat anti-mouse Ly-6G (clone 1A8, #561236; BD Biosciences), PerCP-Cy5.5 rat anti-mouse Ly-6C (clone AL-21, #560525; BD Biosciences), BV711 rat anti-mouse CD19 (clone 1D3, #563157; BD Biosciences), Alexa Fluor 647 rat anti-mouse CD25 (clone 7D4, #563598; BD Biosciences), BUV737 rat anti-mouse CD3 Molecular Complex (clone 17A2, #564380; BD Biosciences), PE-Cy7 rat anti-mouse CD4 (clone RM4-5, #552775; BD Biosciences), FITC rat anti-mouse CD8a (clone 53-6.7, #553030; BD Biosciences), PE mouse anti-mouse NK-1.1 (clone PK136, #557391; BD Biosciences), Super Bright 436 rat anti-mouse CD115 (c-fms, clone AFS98; eBioscience, #62-1152-82; Invitrogen), and PE-Cy5 rat anti-mouse CD44 (clone IM7, #553135; BD Biosciences). Statistical analysis Unless explicitly stated, all data were shown as means ± SEM. Differences between groups were evaluated by one-way ANOVA tests with post hoc Tukey’s multiple comparisons tests. GraphPad Prism version 8 for Windows (GraphPad Software) was used for these analyses and creation of the plots. For statistical analysis of the ptau staining pattern distribution, two-sided Fisher’s exact tests were performed using the “fisher.test” function in R version 3.5.2 (The R Foundation for Statistical Computing). Online supplemental material Fig. S1 shows that microglial depletion efficiency is affected by both sex and chow formula and correlates with plasma drug concentration. Fig. S2 shows high efficiency of microglial depletion by PLX3397 supplemented in the AIN-76A chow at a 400-mg/kg dose. Fig. S3 shows no difference of GFAP+ astrocytes in PLX3397-treated or control TE4 and TEKO mice, or in their non-tau transgenic counterparts. Fig. S4 shows elevated apoE in astrocytes and NeuN− neurons in the piriform/entorhinal cortex of PLX3397-treated TE4 and E4 mice compared with control TE4 and E4 mice. Fig. S5 shows the gating strategy for isolating different blood cell types in flow cytometry. Supplementary Material Supplemental Materials (PDF) Acknowledgments We thank Plexxikon Inc. for providing PLX3397 for the study. We thank Andrey Rymar for all the PLX3397-related assistance in the study. We thank Washington University Division of Comparative Medicine Research Animal Diagnostic Laboratory for performing the CBC for the samples. This study was supported by the National Institutes of Health (NS090934 and AG047644 [to D.M. Holtzman]), the JPB Foundation (to D.M. Holtzman), and the Cure Alzheimer’s Fund (to D.M. Holtzman). Experiments were performed in part through the use of Washington University Center for Cellular Imaging supported by Washington University School of Medicine, the Children’s Discovery Institute of Washington University, and St. Louis Children’s Hospital (CDI-CORE-2015-505) and the Foundation for Barnes-Jewish Hospital (3770). D.M. Holtzman is listed as inventor on a patent licensed by Washington University to C2N Diagnostics on the therapeutic use of anti-tau antibodies. D.M. Holtzman co-founded and is on the scientific advisory board of C2N Diagnostics, LLC. C2N Diagnostics, LLC has licensed certain anti-tau antibodies to AbbVie for therapeutic development. D.M. Holtzman is on the scientific advisory board of Denali and consults for Genentech and Idorsia. The remaining authors declare no competing financial interests. Author contributions: Y. Shi conceived the study and designed the experiments. Y. Shi performed the experiments with the help of M. Manis, J. Long, K. Wang, J. Remolina Serrano, and R. Hoyle. P.M. Sullivan provided the human APOE4 KI mice. Y. Shi analyzed the data. Y. Shi and D.M. Holtzman wrote the paper. All authors reviewed and approved the manuscript. ==== Refs Arriagada, P.V., J.H. Growdon, E.T. Hedley-Whyte, and B.T. Hyman. 1992. Neurofibrillary tangles but not senile plaques parallel duration and severity of Alzheimer’s disease. Neurology. 42 :631–639. 10.1212/WNL.42.3.631 1549228 Bennett, R.E., A. Bryant, M. Hu, A.B. Robbins, S.C. Hopp, and B.T. Hyman. 2018. Partial reduction of microglia does not affect tau pathology in aged mice. J. Neuroinflammation. 15 :311. 10.1186/s12974-018-1348-5 30413160 Chitu, V., V. Nacu, J.F. Charles, W.M. Henne, H.T. McMahon, S. Nandi, H. Ketchum, R. Harris, M.C. Nakamura, and E.R. Stanley. 2012. PSTPIP2 deficiency in mice causes osteopenia and increased differentiation of multipotent myeloid precursors into osteoclasts. Blood. 120 :3126–3135. 10.1182/blood-2012-04-425595 22923495 Chitu, V., Ş. Gokhan, S. Nandi, M.F. Mehler, and E.R. Stanley. 2016. Emerging roles for CSF-1 receptor and its ligands in the nervous system. Trends Neurosci. 39 :378–393. 10.1016/j.tins.2016.03.005 27083478 de Calignon, A., T.L. Spires-Jones, R. Pitstick, G.A. Carlson, and B.T. Hyman. 2009. Tangle-bearing neurons survive despite disruption of membrane integrity in a mouse model of tauopathy. J. Neuropathol. Exp. Neurol. 68 :757–761. 10.1097/NEN.0b013e3181a9fc66 19535996 de Calignon, A., L.M. Fox, R. Pitstick, G.A. Carlson, B.J. Bacskai, T.L. Spires-Jones, and B.T. Hyman. 2010. Caspase activation precedes and leads to tangles. Nature. 464 :1201–1204. 10.1038/nature08890 20357768 Elmore, M.R., A.R. Najafi, M.A. Koike, N.N. Dagher, E.E. Spangenberg, R.A. Rice, M. Kitazawa, B. Matusow, H. Nguyen, B.L. West, and K.N. Green. 2014. Colony-stimulating factor 1 receptor signaling is necessary for microglia viability, unmasking a microglia progenitor cell in the adult brain. Neuron. 82 :380–397. 10.1016/j.neuron.2014.02.040 24742461 Erny, D., A.L. Hrabě de Angelis, D. Jaitin, P. Wieghofer, O. Staszewski, E. David, H. Keren-Shaul, T. Mahlakoiv, K. Jakobshagen, T. Buch, . 2015. Host microbiota constantly control maturation and function of microglia in the CNS. Nat. Neurosci. 18 :965–977. 10.1038/nn.4030 26030851 Gale, S.C., L. Gao, C. Mikacenic, S.M. Coyle, N. Rafaels, T. Murray Dudenkov, J.H. Madenspacher, D.W. Draper, W. Ge, J.J. Aloor, . 2014. APOε4 is associated with enhanced in vivo innate immune responses in human subjects. J. Allergy Clin. Immunol. 134 :127–134. 10.1016/j.jaci.2014.01.032 24655576 Hesdorffer, C.S., and D.L. Longo. 2015. Drug-induced megaloblastic anemia. N. Engl. J. Med. 373 :1649–1658. 10.1056/NEJMra1508861 26488695 Josephs, K.A., J.L. Whitwell, Z. Ahmed, M.M. Shiung, S.D. Weigand, D.S. Knopman, B.F. Boeve, J.E. Parisi, R.C. Petersen, D.W. Dickson, and C.R. Jack Jr. 2008. β-amyloid burden is not associated with rates of brain atrophy. Ann. Neurol. 63 :204–212. 10.1002/ana.21223 17894374 Kitazawa, M., S. Oddo, T.R. Yamasaki, K.N. Green, and F.M. LaFerla. 2005. Lipopolysaccharide-induced inflammation exacerbates tau pathology by a cyclin-dependent kinase 5-mediated pathway in a transgenic model of Alzheimer’s disease. J. Neurosci. 25 :8843–8853. 10.1523/JNEUROSCI.2868-05.2005 16192374 Krasemann, S., C. Madore, R. Cialic, C. Baufeld, N. Calcagno, R. El Fatimy, L. Beckers, E. O’Loughlin, Y. Xu, Z. Fanek, . 2017. The TREM2-APOE pathway drives the transcriptional phenotype of dysfunctional microglia in neurodegenerative diseases. Immunity. 47 :566–581.e9. 10.1016/j.immuni.2017.08.008 28930663 Kuchibhotla, K.V., S. Wegmann, K.J. Kopeikina, J. Hawkes, N. Rudinskiy, M.L. Andermann, T.L. Spires-Jones, B.J. Bacskai, and B.T. Hyman. 2014. Neurofibrillary tangle-bearing neurons are functionally integrated in cortical circuits in vivo. Proc. Natl. Acad. Sci. USA. 111 :510–514. 10.1073/pnas.1318807111 24368848 Leyns, C.E.G., J.D. Ulrich, M.B. Finn, F.R. Stewart, L.J. Koscal, J. Remolina Serrano, G.O. Robinson, E. Anderson, M. Colonna, and D.M. Holtzman. 2017. TREM2 deficiency attenuates neuroinflammation and protects against neurodegeneration in a mouse model of tauopathy. Proc. Natl. Acad. Sci. USA. 114 :11524–11529. 10.1073/pnas.1710311114 29073081 Li, Y., L. Liu, S.W. Barger, and W.S. Griffin. 2003. Interleukin-1 mediates pathological effects of microglia on tau phosphorylation and on synaptophysin synthesis in cortical neurons through a p38-MAPK pathway. J. Neurosci. 23 :1605–1611. 10.1523/JNEUROSCI.23-05-01605.2003 12629164 Luo, J., F. Elwood, M. Britschgi, S. Villeda, H. Zhang, Z. Ding, L. Zhu, H. Alabsi, R. Getachew, R. Narasimhan, . 2013. Colony-stimulating factor 1 receptor (CSF1R) signaling in injured neurons facilitates protection and survival. J. Exp. Med. 210 :157–172. 10.1084/jem.20120412 23296467 Nandi, S., S. Gokhan, X.M. Dai, S. Wei, G. Enikolopov, H. Lin, M.F. Mehler, and E.R. Stanley. 2012. The CSF-1 receptor ligands IL-34 and CSF-1 exhibit distinct developmental brain expression patterns and regulate neural progenitor cell maintenance and maturation. Dev. Biol. 367 :100–113. 10.1016/j.ydbio.2012.03.026 22542597 Pimenova, A.A., T. Raj, and A.M. Goate. 2018. Untangling genetic risk for Alzheimer’s disease. Biol. Psychiatry. 83 :300–310. 10.1016/j.biopsych.2017.05.014 28666525 Shi, Y., and D.M. Holtzman. 2018. Interplay between innate immunity and Alzheimer disease: APOE and TREM2 in the spotlight. Nat. Rev. Immunol. 18 :759–772. 10.1038/s41577-018-0051-1 30140051 Shi, Y., K. Yamada, S.A. Liddelow, S.T. Smith, L. Zhao, W. Luo, R.M. Tsai, S. Spina, L.T. Grinberg, J.C. Rojas, . Alzheimer’s Disease Neuroimaging Initiative. 2017. ApoE4 markedly exacerbates tau-mediated neurodegeneration in a mouse model of tauopathy. Nature. 549 :523–527. 10.1038/nature24016 28959956 Sluijter, M., T.C. van der Sluis, P.A. van der Velden, M. Versluis, B.L. West, S.H. van der Burg, and T. van Hall. 2014. Inhibition of CSF-1R supports T-cell mediated melanoma therapy. PLoS One. 9 :e104230. 10.1371/journal.pone.0104230 25110953 Thion, M.S., D. Low, A. Silvin, J. Chen, P. Grisel, J. Schulte-Schrepping, R. Blecher, T. Ulas, P. Squarzoni, G. Hoeffel, . 2018. Microbiome influences prenatal and adult microglia in a sex-specific manner. Cell. 172 :500–516.e16. 10.1016/j.cell.2017.11.042 29275859 Vitek, M.P., C.M. Brown, and C.A. Colton. 2009. APOE genotype-specific differences in the innate immune response. Neurobiol. Aging. 30 :1350–1360. 10.1016/j.neurobiolaging.2007.11.014 18155324 Wang, Y., O. Berezovska, and S. Fedoroff. 1999. Expression of colony stimulating factor-1 receptor (CSF-1R) by CNS neurons in mice. J. Neurosci. Res. 57 :616–632. 10.1002/(SICI)1097-4547(19990901)57:5<616::AID-JNR4>3.0.CO;2-E 10462686 Zhu, Y., E. Nwabuisi-Heath, S.B. Dumanis, L.M. Tai, C. Yu, G.W. Rebeck, and M.J. LaDu. 2012. APOE genotype alters glial activation and loss of synaptic markers in mice. Glia. 60 :559–569. 10.1002/glia.22289 22228589

PMC006xxxxxx/PMC6829661.txt

==== Front J Cell Biol J Cell Biol jcb jcb The Journal of Cell Biology 0021-9525 1540-8140 Rockefeller University Press 31575603 201905097 10.1083/jcb.201905097 Research Articles Article 34 19 ARFRP1 functions upstream of ARL1 and ARL5 to coordinate recruitment of distinct tethering factors to the trans-Golgi network ARFRP1 coordinates tethering at the TGN Ishida Morié https://orcid.org/0000-0002-5673-6370 Bonifacino Juan S. Cell Biology and Neurobiology Branch, Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, MD Correspondence to Juan S. Bonifacino: juan.bonifacino@nih.gov 04 11 2019 01 10 2019 218 11 36813696 13 5 2019 09 8 2019 27 8 2019 This is a work of the U.S. Government and is not subject to copyright protection in the United States. Foreign copyrights may apply. 2019 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). Ishida and Bonifacino show that the small GTPase ARFRP1 functions upstream of two other small GTPases, ARL1 and ARL5, to coordinate the recruitment of two distinct classes of tethering factors, golgins and GARP, respectively, to the trans-Golgi network. SNARE-mediated fusion of endosome-derived transport carriers with the trans-Golgi network (TGN) depends on the concerted action of two types of tethering factors: long coiled-coil tethers of the golgin family, and the heterotetrameric complex GARP. Whereas the golgins mediate long-distance capture of the carriers, GARP promotes assembly of the SNAREs. It remains to be determined, however, how the functions of these tethering factors are coordinated. Herein we report that the ARF-like (ARL) GTPase ARFRP1 functions upstream of two other ARL GTPases, ARL1 and ARL5, which in turn recruit golgins and GARP, respectively, to the TGN. We also show that this mechanism is essential for the delivery of retrograde cargos to the TGN. Our findings thus demonstrate that ARFRP1 is a master regulator of retrograde-carrier tethering to the TGN. The coordinated recruitment of distinct tethering factors by a bifurcated GTPase cascade may be paradigmatic of other vesicular fusion events within the cell. Eunice Kennedy Shriver National Institute of Child Health and Human Development https://doi.org/10.13039/100009633 ZIA HD001607 Japan Society for the Promotion of Science https://doi.org/10.13039/501100001691 ==== Body pmcIntroduction The TGN is a tubular-reticular organelle associated with the distal face of the Golgi stack that serves as a hub for multiple forward and retrograde pathways in the endomembrane system of eukaryotic cells (Griffiths and Simons, 1986; Guo et al., 2014). At the TGN, biosynthetic cargos coming from the Golgi stack are sorted to various post-Golgi compartments such as endosomes, lysosomes, secretory granules, different domains of the plasma membrane, and the extracellular space. The TGN also receives cargos by retrograde transport from endolysosomal compartments (Lu and Hong, 2014; Hierro et al., 2015). The delivery of retrograde cargos into the TGN occurs by SNARE-dependent fusion with tubular/vesicular carriers derived from endosomes. The SNAREs involved in this process include the Q-SNAREs syntaxin 6, syntaxin 16, and VTI1A, and an R-SNARE, VAMP4 or VAMP3, which assemble into a heterotetrameric trans-SNARE complex to enable merger of the carrier and TGN membranes (Mallard et al., 2002). In vivo, the specificity and efficiency of SNARE-dependent fusion events depend on two distinct classes of tethering factors known as homodimeric long coiled-coil proteins and multisubunit tethering complexes (MTCs; Yu and Hughson, 2010). At the mammalian TGN, four homodimeric long coiled-coil proteins named Golgin-245 (Yoshino et al., 2005), Golgin-97 (Lu et al., 2004), GCC185 (Reddy et al., 2006; Derby et al., 2007), and GCC88 (Lieu et al., 2007; belonging to a family termed “golgins,” reviewed by Cheung and Pfeffer, 2016; Fig. 1 A) and at least one MTC, the heterotetrameric Golgi-associated retrograde protein (GARP) complex (Siniossoglou and Pelham, 2002; Conibear et al., 2003; Liewen et al., 2005; Quenneville et al., 2006; Pérez-Victoria et al., 2008; Bonifacino and Hierro, 2011; Fig. 1 B), promote the delivery of retrograde cargos to the TGN. Each class of tethering factors plays a distinct role in the fusion event: while the golgins mediate long-distance capture of the carriers (Wong and Munro, 2014; Cheung et al., 2015), GARP coordinates the assembly of the trans-SNARE complex (Siniossoglou and Pelham, 2002; Conibear et al., 2003; Quenneville et al., 2006; Pérez-Victoria and Bonifacino, 2009). The physiological importance of mammalian GARP is underscored by the embryonic lethality of mice with null mutations in GARP subunit genes (Bennett and Dunn, 1958; Schmitt-John et al., 2005; Sugimoto et al., 2012; Karlsson et al., 2013), the motor neuron degeneration caused by a hypomorphic mutation in the VPS54 subunit of GARP in the wobbler mouse (Schmitt-John et al., 2005; Pérez-Victoria et al., 2010a), and the neurodevelopmental abnormalities in human patients with hypomorphic mutations in GARP subunit genes (Feinstein et al., 2014; Hady-Cohen et al., 2018; Gershlick et al., 2019; Uwineza et al., 2019). Figure 1. Characteristics of TGN tethering factors. (A) Schematic representation of golgin coiled-coil tethers associated with the TGN (Golgin-245, Golgin-97, GCC88, and GCC185) and the late Golgi apparatus (TMF1). The location of a GRIP domain near the C terminus (black segment), a RAB6-binding region (RBD; light gray segment), and the total number of amino acids in each human protein are indicated. For additional details, see Cheung and Pfeffer (2016). (B) Schematic representation of GARP and EARP MTCs. GARP is composed of VPS51, VPS52, VPS53, and VPS54, whereas EARP is composed of VPS50, VPS51, VPS52, and VPS53. Amino acid numbers in the human proteins are indicated. For additional details, see Schindler et al. (2015). In addition to SNAREs and tethering factors, a third type of proteins involved in the fusion of endosome-derived carriers with the TGN consists of small GTPases of the ARF-like (ARL; Donaldson and Jackson, 2011; Sztul et al., 2019) and RAB families (Pfeffer, 2017). ARL1 is required for the recruitment of Golgin-245 and Golgin-97 to the TGN through an interaction with an ∼45-residue GRIP domain at or near the C terminus of these golgins (Lu and Hong, 2003; Panic et al., 2003a; Fig. 1 A). ARL1 appears less important for the recruitment of GCC185 and GCC88 to the TGN in mammals, despite the fact that these proteins also have a GRIP domain at their C termini (Derby et al., 2004; Burguete et al., 2008; Houghton et al., 2009; Torres et al., 2014). In the case of GCC185, an interaction of RAB6A with a region upstream of the GRIP domain (Fig. 1 A) promotes binding to ARL1 and thus contributes to association of GCC185 with the TGN (Burguete et al., 2008). In contrast to the golgins, it is unclear how GARP is recruited to the TGN. In the yeast Saccharomyces cerevisiae, GARP was shown to interact with the GTP-bound forms of the Arl1p and Ypt6p orthologues of mammalian ARL1 and RAB6 (the latter of which occurs as RAB6A and RAB6B paralogs), respectively (Siniossoglou and Pelham, 2001; Panic et al., 2003b). However, deletion of yeast ARL1 had no effect on the association of GARP with the late Golgi (equivalent of the mammalian TGN; Panic et al., 2003b), and deletion of yeast YPT6 resulted in a more dispersed but still particulate distribution of GARP (Siniossoglou and Pelham, 2001). In higher eukaryotes, including Drosophila melanogaster and human cells, recruitment of GARP to the TGN was shown to depend on another member of the ARL family, ARL5 (Rosa-Ferreira et al., 2015). In humans, there are two ARL5 paralogs, ARL5A and ARL5B, both of which localize to the TGN (Houghton et al., 2012). siRNA-mediated knockdown (KD) of ARL5B, but not ARL5A, reduced retrograde transport from endosomes to the TGN (Houghton et al., 2012) as well as association of GARP with the TGN (Rosa-Ferreira et al., 2015). The specific requirement of ARL5B is probably due to its greater abundance relative to ARL5A in the HeLa cells that were used in those experiments (Rosa-Ferreira et al., 2015). Nevertheless, the dissociation of GARP from the TGN upon KD of ARL5B, alone or in combination with ARL5A, was incomplete, with 30–45% of the KD cells still retaining GARP staining at the TGN (Rosa-Ferreira et al., 2015). The uncertainty about the factors that regulate the recruitment of mammalian GARP to the TGN prompted us to test the involvement of several Golgi-localized ARL and RAB GTPases by knocking out their genes in human HeLa cells and examining the localization of GARP by immunofluorescence microscopy. The use of knock-out (KO) cells avoided the incomplete depletion and off-target effects that are typical of siRNA-mediated KD. Using this approach, we found that double KO of ARL5A and ARL5B (hereafter referred to as ARL5 KO), but not ARL1 or double KO of RAB6A and RAB6B (referred to as RAB6 KO), abrogated the association of GARP with the TGN. Surprisingly, we observed that KO of another ARL-family member, ARFRP1, also abolished the recruitment of GARP to the TGN. Further experiments showed that ARFRP1 functions upstream of both ARL1 and ARL5 to promote the recruitment of three TGN golgins (Golgin-245, Golgin-97, and GCC88) and GARP, respectively, to the TGN. Our studies thus revealed that ARFRP1 functions as a master regulator of retrograde-carrier tethering by enabling the coordinated recruitment of two types of tethering factors to the TGN. Results ARL5 and ARFRP1 are required for GARP localization to the TGN This project was initiated to identify small GTPases that regulate the recruitment of GARP to the TGN, by taking advantage of the ability to KO individual genes in HeLa cells using CRISPR/Cas9 technology. The mammalian GARP complex comprises four subunits named VPS51 (originally known as ANG2), VPS52, VPS53, and VPS54 (Liewen et al., 2005; Pérez-Victoria et al., 2008, 2010b; Fig. 1 B). A related complex named “endosome-associated recycling protein” (EARP) shares the VPS51, VPS52, and VPS53 subunits with GARP, but has an alternative subunit named VPS50 (also known as syndetin or VPS54L) in place of VPS54 (Gillingham et al., 2014; Schindler et al., 2015; Fig. 1 B). Unlike GARP, EARP is largely associated with an endosomal compartment and was not studied here (Gillingham et al., 2014; Schindler et al., 2015; Gershlick et al., 2019). Our attempts to detect endogenous GARP by immunofluorescence microscopy and immunoblotting using several commercially available antibodies to the GARP-specific VPS54 subunit were unsuccessful, probably due to the low abundance of this protein in HeLa cells (∼700 molecules per cell; Kulak et al., 2014). To overcome this limitation, we transfected HeLa cells with a plasmid encoding human VPS54 tagged at its C terminus with 13 copies of the Myc epitope (VPS54-13Myc). As previously reported (Schindler et al., 2015), this construct localized to a juxtanuclear structure characteristic of the TGN (Fig. S1 A, WT). This localization was abrogated by KO of VPS51, VPS52, or VPS53, but not VPS50, indicating that it reflected the distribution of the whole GARP complex (Fig. S1, A–C). Next, we examined the effect of knocking out several TGN-localized small GTPases on the association of VPS54-13Myc with the TGN. We observed that KO of ARL1, ARL5 (i.e., both ARL5A and ARL5B paralogs), RAB6 (i.e., both RAB6A and RAB6B paralogs), or another member of the ARL family, ARFRP1 (Schürmann et al., 1995), completely abolished the expression of the corresponding proteins, as demonstrated by immunoblotting (Fig. 2 A and Fig. S2, A and B). KO of ARL1 or RAB6 had little or no effect on the localization of VPS54-13Myc to the TGN, but KO of ARL5 resulted in displacement of VPS54-13Myc to the cytosol (Fig. 2, B and C). This latter result was even more dramatic than that obtained by siRNA-mediated KD of ARL5 (Rosa-Ferreira et al., 2015), as it was observed in 100% of the cells. Surprisingly, KO of ARFRP1 also prevented association of VPS54-13Myc with the TGN (Fig. 2, B and C). From these experiments we concluded that ARL5 and ARFRP1 are both required for association of GARP with the TGN. Figure 2. Both ARL5 and ARFRP1 are required for localization of GARP to the TGN. (A) KO of TGN-localized small GTPases in HeLa cells confirmed by immunoblot analysis with antibodies to the proteins indicated on the right. In this figure and subsequent figures, ARL5 KO represents KO of both ARL5A and ARL5B, and RAB6 KO represents KO of both RAB6A and RAB6B. α-Tubulin was used as a loading control. The positions of molecular mass markers are indicated on the left. (B) Immunofluorescence microscopy of WT and KO HeLa cells transfected with a plasmid encoding VPS54-13Myc and stained for the Myc epitope (red), giantin (green), and nuclei (DAPI; blue). Scale bars: 10 μm. Insets are magnified views of the boxed areas. Inset scale bars: 5 μm. (C) Quantification of the percentage of cells exhibiting VPS54-13Myc staining at the TGN. Values are the mean ± SEM from three independent experiments. More than 100 cells per sample were counted in each experiment. The statistical significance of the differences relative to WT cells was determined using Dunnett’s test. **, P < 0.01; ***, P < 0.001. ARFRP1 functions upstream of ARL5 in GARP recruitment to the TGN To investigate the functional relationship of ARL5 and ARFRP1 in the recruitment of GARP to the TGN, we performed a series of reciprocal rescue experiments. Expression of WT ARL5B-GFP in ARL5-KO cells restored the localization of VPS54-13Myc to the TGN (Fig. 3, A and B). Likewise, expression of WT ARFRP1-GFP in ARFRP1-KO cells rescued (albeit partially) the association of VPS54-13Myc with the TGN (Fig. 3, C and D). These experiments ruled out that the effects of the KOs on VPS54-13Myc localization were due to off-target effects. Like other small GTPases, ARL5 and ARFRP1 cycle between GTP-bound, active and GDP-bound, inactive forms (Donaldson and Jackson, 2011; Sztul et al., 2019). Rescue of VPS54-13Myc association with the TGN was also observed upon expression of constitutively active (Q70L) but not inactive (T30N) ARL5B-GFP in ARL5-KO cells (Fig. 3, A and B). Similarly, TGN localization of VPS54-13Myc was rescued by expression of constitutively active (Q79L) but not inactive (T31N) ARFRP1-GFP in ARFRP1-KO cells (Fig. 3, C and D). These results indicated that association of VPS54-13Myc with the TGN depends on the GTP-bound, active forms of ARL5 and ARFRP1. Importantly, whereas expression of WT or Q79L ARFRP1-GFP in ARL5-KO cells did not rescue the association of VPS54-13Myc with the TGN (Fig. 3, A and B), expression of WT or Q70L ARL5B-GFP in ARFRP1-KO cells partially rescued VPS54-13Myc association (Fig. 3, C and D). The ability of WT ARL5B-GFP to associate with the TGN and rescue VPS54-Myc localization in ARFRP1-KO cells (Fig. 3, C and D) suggests that a substantial fraction of this protein exists in the GTP-bound, active form under the overexpression conditions of these experiments. Taken together, these results are consistent with ARFRP1 acting upstream of ARL5 to recruit GARP to the TGN. Figure 3. ARFRP1 functions upstream of ARL5 for the recruitment of GARP to the TGN. (A) GFP-tagged ARL5B (WT, Q70L, or T30N) or ARFRP1 (WT, Q79L, or T31N) were co-expressed with VPS54-13Myc in ARL5-KO cells. Cells were stained with DAPI (blue) and examined for GFP distribution by confocal microscopy. Scale bars: 10 μm. Insets are magnified views of the boxed areas. Inset scale bars: 5 μm. (B) The percentage of cells with VPS54-13Myc staining at the TGN was quantified as described in Fig. 2 C. Values are the mean ± SEM from three independent experiments. More than 100 cells per sample were counted in each experiment. ***, P < 0.001, in comparison to ARL5 KO cells only expressing VPS54-13Myc (control) using Dunnett’s test. (C) GFP-tagged ARL5B (WT, Q70L, or T30N) or ARFRP1 (WT, Q79L, or T31N) were co-expressed with VPS54-13Myc in ARFRP1-KO cells. Cells were analyzed as in A. (D) Quantification of cells having VPS54-13Myc staining at the TGN as described in Fig. 2 C. Values are the mean ± SEM from three independent experiments. More than 100 cells per sample were counted in each experiment. ***, P < 0.001, in comparison to ARFRP1 KO cells only expressing VPS54-13Myc (control) using Dunnett’s test. (E) Subcellular fractionation of WT, ARFRP1-KO, and ARFRP1-KO-rescue HeLa cells. Whole cells and cytosolic and membrane (Memb) fractions obtained as described in Materials and methods were analyzed by SDS-PAGE and immunoblotting for the proteins indicated on the right. The positions of molecular mass markers are indicated on the left. (F) Quantification of the ratio of membrane to the sum of membrane and cytosolic ARL1 and ARL5 protein from experiments such as that described in E. Ratios for ARFRP1-KO cells and ARFRP1-KO cells rescued with tag-less ARFRP1 were normalized to WT cells. Values are the mean ± SEM from three independent experiments. The statistical significance of the differences relative to WT cells was determined using Dunnett’s test. *, P < 0.05; **, P < 0.01. We next wished to address whether endogenous ARL5 (as opposed to overexpressed WT ARL5B-GFP; Fig. 3 C) required ARFRP1 for association with the TGN. However, the antibodies to ARL5 that we tested worked for immunoblotting but not for immunofluorescence microscopy. Therefore, we fractionated cells into cytosol and membranes and immunoblotted for ARL5, ARL1, and RAB6. We observed that KO of ARFRP1 decreased the association of ARL5 and ARL1, but not RAB6, with membranes (Fig. 3, E and F). Rescue of these cells by expression of ARFRP1 restored normal levels of ARL5 and ARL1 membrane association. These analyses thus demonstrated that ARFRP1 is required for association of both ARL5 and ARL1 with membranes. Impaired retrograde transport in ARL5-KO and ARFRP1-KO cells GARP is required for delivery of retrograde cargos such as the TGN resident protein TGN46 and the Shiga toxin B subunit (STxB) from endosomes to the TGN (Pérez-Victoria et al., 2008, 2010a,b; Pérez-Victoria and Bonifacino, 2009). To determine if ARL5 and ARFRP1 are similarly required, we examined the distribution of TGN46 and the internalization of Cy3-conjugated STxB (Cy3-STxB) in ARL5-KO and ARFRP1-KO cells (Fig. 4). Indeed, we observed that KO of ARL5 or ARFRP1 caused dispersal of TGN46 from the TGN similar to that caused by VPS54 KO (Fig. 4 A), indicative of a defect in retrograde transport. As expected, TGN46 dispersal was prevented by expression of the corresponding GFP-tagged GTPase or VPS54 (Fig. 4 B). Interestingly, immunoblot analysis showed an increase in TGN46 species with faster electrophoretic mobility in cells with KO of ARL5, ARFRP1, or GARP subunits, but not VPS50 (Fig. 4 C). These differences were likely due to decreased carbohydrate modifications in the KO cells, as previously observed in cells from a patient with mutations in VPS51 (Gershlick et al., 2019). KO of ARL5, ARFRP1, or VPS54 also impaired the delivery of internalized Cy3-STxB (Johannes and Goud, 1998) to the Golgi complex (Fig. 4, D and E). These experiments thus demonstrated that ARL5, ARFRP1, and GARP are similarly required for retrograde transport to the TGN, consistent with the role of these small GTPases in the regulation of GARP. Figure 4. Altered localization of TGN46 and impaired transport of STxB to the Golgi complex in ARL5-KO and ARFRP1-KO cells. (A) Immunofluorescence microscopy of WT, ARFRP1-KO, ARL5-KO, and VPS54-KO cells immunostained for endogenous TGN46 and counterstained with DAPI (blue). Scale bars: 10 μm. (B) ARFRP1-KO, ARL5-KO, or VPS54-KO cells were rescued by expression of the corresponding GFP-tagged proteins, stained for endogenous TGN46, transgenic GFP (only for VPS54-GFP), and nuclei (DAPI; blue) and imaged by confocal microscopy (GFP fluorescence was directly observed for ARL5B-GFP and ARFRP1-GFP). Scale bars: 10 μm. Arrows indicate rescued cells. (C) SDS-PAGE and immunoblot analysis of endogenous TGN46 and α-tubulin (loading control) in WT and the indicated KO cells. The positions of molecular mass markers are indicated on the left. (D) Live WT, ARL5-KO, ARFRP1-KO, or VPS54-KO cells were incubated for 15 min with Cy3-STxB and chased for 1 h in regular culture medium at 37°C, after which cells were fixed, immunostained for endogenous GM130, and imaged by confocal microscopy. Scale bars: 10 μm. Insets are magnified views of the boxed areas. Inset scale bars: 5 μm. (E) Quantification of cells having Cy3-STxB staining at the TGN as described in Fig. 2 C. Values are the mean ± SEM from three independent experiments. More than 100 cells per sample were counted in each experiment. ***, P < 0.001, in comparison to WT cells using Dunnett’s test. ARFRP1 and ARL1, but not ARL5 or RAB6, are required for recruitment of Golgin-245, Golgin-97, and GCC88 to the TGN Next, we investigated the possibility that the recruitment of GARP to the TGN is co-regulated with that of the TGN golgins. Previous studies showed that mammalian ARFRP1 and the orthologous yeast Arl3p promote recruitment of mammalian ARL1 (Shin et al., 2005; Zahn et al., 2006) and yeast Arl1p (Panic et al., 2003b; Setty et al., 2003), respectively, to the TGN. Thus, indirectly, ARFRP1/Arl3p contributes through ARL1/Arl1p to the recruitment of the GRIP domain–containing proteins Golgin-245 and Golgin-97 in mammals (Shin et al., 2005; Zahn et al., 2006) and Imh1p in yeast (Panic et al., 2003b; Setty et al., 2003). To develop a more global understanding of the small GTPase requirements for the recruitment of tethering factors to the TGN, we examined the effect of knocking out ARL1, ARL5, ARFRP1, or RAB6 on the association of Golgin-245, Golgin-97, GCC185, and GCC88 (Fig. 1 A) with the TGN. We observed that KO of ARL1 or ARFRP1 abrogated the association of Golgin-245 and GCC88, and, partially, Golgin-97, with the TGN (Fig. 5 A). In contrast, these KOs had little or no effect on the recruitment of the TGN golgin GCC185 and the Golgi-stack golgin TMF1 (used here as a control; Fig. 1 A and Fig. 5 A). Interestingly, and in line with these observations, immunoblot analysis showed that KO of ARL1 or ARFRP1 reduced the levels of Golgin-245, Golgin-97, and GCC88, but not GCC185 (Fig. 5 B), indicating that association with the TGN is required for stability of these golgins. In contrast, KO of ARL5 or RAB6 had no effect on the TGN association (Fig. 5 A) or levels of any of these golgins (Fig. 5 B). RAB6 KO abolished the association of GCC185 with the TGN and of TMF1 with the Golgi stack (Fridmann-Sirkis et al., 2004; Fig. 5 A). The association of GCC185 and TMF1 with the Golgi complex could be rescued by expression of GFP-Rab6A (mouse orthologue) in RAB6-KO cells (Fig. 5 C). These experiments therefore indicated that ARFRP1 and ARL1 regulate the recruitment of Golgin-245, Golgin-97, and GCC88, whereas RAB6 regulates the recruitment of GCC185, to the TGN. Figure 5. Small GTPases required for localization of golgins to the TGN. (A) Immunofluorescence microscopy of WT, ARL1-KO, ARFRP1-KO, ARL5-KO, and RAB6-KO cells immunostained for endogenous Golgin-245, Golgin-97, GCC88, GCC185, or TMF1 and counterstained with DAPI (blue). Scale bars: 10 μm. Insets are magnified views of the boxed areas. Inset scale bars: 5 μm. Notice that ARL1 KO or ARFRP1 KO caused complete disappearance of Golgin-245 and GCC88, and a partial decrease in the intensity of Golgin-97, at the TGN; quantification in 10 cells per sample in three independent experiment showed that Golgin-97 decrease was 75.6% ± 2.4% in ARL1-KO cells and 55.0% ± 1.9% in ARFRP1-KO cells. (B) SDS-PAGE and immunoblot analysis of endogenous golgins and α-tubulin (loading control) in WT and KO cells. The positions of molecular mass markers are indicated on the left. (C) Immunofluorescence microscopy of RAB6-KO cells transfected with a plasmid encoding GFP-tagged mouse Rab6A (green), immunostained for endogenous GCC185 and TMF1 (red), and counterstained with DAPI (blue). Cells were examined for GFP fluorescence by confocal microscopy. Scale bars: 10 μm. ARFRP1 functions upstream of ARL1 in the recruitment of GCC88 to the TGN Our finding that GCC88 requires ARFRP1 and ARL1 for recruitment to the TGN is at odds with the previous observation that GCC88 does not bind ARL1 in vivo (Derby et al., 2004). This discrepancy prompted us to analyze in further detail the functional interaction of GCC88 with ARFRP1 and ARL1. We observed that the association of GCC88 with the TGN in ARL1-KO cells could be rescued by expression of WT ARL1-GFP or active ARL1-GFP Q71L but not inactive ARL1-GFP T31N (Fig. 6, A and B). In contrast, ARFRP1-GFP failed to rescue GCC88 association with the TGN in ARL1-KO cells, irrespective of its activation state (Fig. 6, A and B). GCC88 localization to the TGN in ARFRP1-KO cells could also be restored upon expression of WT ARFRP1-GFP and active ARFRP1-GFP Q79L, but not inactive ARFRP1-GFP T31N (Fig. 6, C and D). In this case, however, expression of active ARL1-GFP Q71L, but not the other forms of ARL1-GFP, in ARFRP1-KO cells also enabled association of GCC88 with the TGN (Fig. 6, C and D). Finally, we found that, whereas the localization of endogenous ARFRP1 was not altered by ARL1 KO, endogenous ARL1 lost its TGN localization upon ARFRP1 KO (Fig. 6 E). These results were consistent with the decreased association of endogenous ARL1 with membranes in ARFRP1-KO cells analyzed by subcellular fractionation (Fig. 3, E and F). Therefore, ARFRP1 functions upstream of ARL1 to recruit GCC88 to the TGN. Figure 6. ARFRP1 functions upstream of ARL1 in the recruitment of GCC88 to the TGN. (A) ARL1-KO cells were transfected with plasmids encoding GFP-tagged ARL1 (WT, Q71L, or T31N) or GFP-tagged ARFRP1 (WT, Q79L, or T31N), immunostained for endogenous GCC88, counterstained with DAPI (blue), and imaged by confocal microscopy. (B) The percentage of cells having GCC88 staining at the TGN from experiments such as that in panel A was quantified as described in Fig. 2 C. Values are the mean ± SEM from three independent experiments. More than 100 cells per sample were counted in each experiment. ***, P < 0.001, in comparison to untransfected ARL1 KO cells using Dunnett’s test. (C) ARFRP1-KO cells were transfected and analyzed as described in A. (D) The percentage of cells having GCC88 staining at the TGN from experiments such as that in C was quantified as described in Fig. 2 C. Values are the mean ± SEM from three independent experiments. More than 100 cells per sample were counted in each experiment. ***, P < 0.001, in comparison to untransfected ARFRP1 KO cells using Dunnett’s test. (E) WT, ARL1-KO, or ARFRP1-KO cells were immunostained for endogenous ARL1 or ARFRP1, counterstained with DAPI (blue), and imaged by confocal microscopy. Scale bars: 10 μm. Insets are magnified views of the boxed areas. Inset scale bars: 5 μm. Taken together, all of the above experiments indicated that ARFRP1 is the source of two GTPase cascades leading to the ARL5-dependent recruitment of GARP and ARL1-dependent recruitment of a subset of golgins to the TGN. SYS1 regulates the ARFRP1-mediated recruitment of GARP to the TGN The targeting of yeast Arl3p (ARFRP1 orthologue) to the late Golgi relies on acetylation of its N terminus and interaction with the transmembrane protein Sys1p (Behnia et al., 2004; Setty et al., 2004). To investigate if human SYS1 plays a similar role in the recruitment of ARFRP1, ARL1, GARP, and golgins, we examined the localization of these proteins in SYS1-KO HeLa cells (Fig. 7 A). We observed that KO of SYS1 caused dissociation of endogenous ARFRP1, ARL1, and GCC88 and transgenic VPS54-13Myc from the TGN (Fig. 7, B–F). The TGN localization of VPS54-13Myc and GCC88 could be restored by expression of GFP-tagged SYS1 in SYS1-KO cells (Fig. 7, C–F). Subcellular fractionation/immunoblotting experiments showed that SYS1 KO decreased the association of ARL5 as well as ARL1 and ARFRP1 with membranes (Fig. 7, G and H). These decreases could be reversed by re-expression of SYS1 in the SYS1-KO cells (Fig. 7, G and H). Unexpectedly, we found that SYS1 KO also decreased the total amounts of ARFRP1 (Fig. 7, G and I), indicating that SYS1 stabilizes the ARFRP1 protein. In contrast to these effects, SYS1 KO did not affect the localization of the RAB6-dependent GCC185 and TMF1 golgins (Fig. S3). From these experiments, we concluded that SYS1 functions upstream of ARFRP1 to initiate the small GTPase cascades that lead to the recruitment of both golgins and GARP to the TGN. Figure 7. SYS1 is required for the association of ARFRP1, ARL1, GCC88, and GARP with the TGN. (A) Due to the lack of good antibodies to SYS1, KO of the SYS1 gene was demonstrated by Sanger sequencing (top) and genomic PCR (bottom). KO resulted in a 34- and 36-bp deletion. The 34-bp deletion is predicted to result in a frameshift causing premature termination at amino acid 28. The 36-bp deletion is not predicted to result in a frameshift, but to cause a deletion of more than half of the first transmembrane domain. (B) WT and SYS1-KO cells were immunostained for endogenous ARL1 or ARFRP1, counterstained with DAPI (blue), and imaged by confocal microscopy. (C) SYS1-KO cells were co-transfected with plasmids encoding VPS54-13Myc and either GFP alone or SYS1-GFP as indicated in the figure, immunostained for the Myc epitope, counterstained with DAPI (blue), and imaged by confocal microscopy. Insets are magnified views of the boxed areas. Inset scale bars: 5 μm. (D) SYS1-KO cells were transfected with plasmids encoding GFP or SYS1-GFP as indicated in the figure, immunostained for endogenous GCC88, counterstained with DAPI (blue), and imaged by confocal microscopy. Scale bars in B, C, and D: 10 μm. (E) Quantification of cells having VPS54-13Myc staining at the TGN as described in Fig. 2 C. Values are the mean ± SEM from three independent experiments. More than 100 cells per sample were counted in each experiment. (F) Quantification of cells having GCC88 staining at the TGN as described in Fig. 2 C. Values are the mean ± SEM from three independent experiments. More than 100 cells per sample were counted in each experiment. In E and F, the statistical significance of the differences relative to SYS1-KO cells only expressing VPS54-13Myc (control; E) or untransfected SYS1-KO cells (F) was determined using Dunnett’s test. ***, P < 0.001. (G) Subcellular fractionation of WT, SYS1-KO, and SYS1-rescue (Res) HeLa cells performed as described in Fig. 3 E. (H) The ratio of membrane to the sum of membrane and cytosolic ARL1 and ARL5 protein from the experiments described in G was quantified as described in Fig. 3 F. Values are the mean ± SEM from three independent experiments. *, P < 0.05; ***, P < 0.001. (I) Quantification of the abundance of ARFRP1 normalized to the abundance of α-tubulin from the experiments described in G. Band intensities in whole cells were measured, and values for SYS1-KO cells and SYS1-rescue cells were normalized to WT cells. Values are the mean ± SEM from three independent experiments. The statistical significance of the differences relative to WT cells was determined using Dunnett’s test. *, P < 0.05. Integrity of the TGN in KO cells Finally, to ascertain that loss of localization of GARP, golgins, TGN46, and internalized Cy3-STxB to the TGN was not due to complete disappearance of this organelle in the various KO cells used in this study, we examined the localization of the TGN-associated adaptor protein Myc-GGA2 (Dell’Angelica et al., 2000), expressed by transient transfection in WT, ARL1-KO, ARL5-KO, ARFRP1-KO, RAB6-KO, and SYS1-KO cells (Fig. S4). We observed that Myc-GGA2 localized to the TGN in all these cell lines, although in some KO cells the TGN appeared slightly dispersed (Fig. S4). These results were consistent with the TGN association of GARP in ARL1-KO cells (Fig. 2 B), the TGN golgins in ARL5-KO cells (Fig. 5 A), and GCC185 in ARFRP1-KO (Fig. 5 A) and SYS1-KO cells (Fig. S3). Therefore, the KOs made in this study impaired the association of tethering factors and the retrograde transport of specific cargos to the TGN without fundamentally altering the integrity of this organelle. Discussion The results of our study demonstrate that SYS1 and ARFRP1 regulate not only the previously reported ARL1-dependent recruitment of a subset of golgins (Panic et al., 2003b; Setty et al., 2003; Shin et al., 2005; Zahn et al., 2006) but also the ARL5-dependent recruitment of GARP to the TGN (Fig. 8). SYS1 and ARFRP1 thus coordinate the recruitment of two structurally and functionally distinct classes of tethering factors to the TGN. This coordination likely ensures that the long-distance capture of retrograde transport carriers by the golgins is synchronized with the transfer of the carriers to GARP for their subsequent SNARE-dependent fusion with the TGN. The coordinated regulation of long coiled-coil tethers and MTCs enabled by this mechanism may be paradigmatic for other membrane fusion events that take place in the endomembrane system of eukaryotic cells. Figure 8. Model for the function of ARFRP1 in the coordinated recruitment of golgins and GARP to the TGN. This model is based on results shown in this article and previous publications cited in the text. The multispanning membrane protein SYS1 recruits ARFRP1 to the TGN, possibly by acting as a GEF that converts ARFRP1-GDP to ARFRP1-GTP. This process involves the N-terminally acetylated amphipathic α-helix of ARFRP1. ARFRP1 then promotes the recruitment and/or activation of both ARL1 and ARL5 to the TGN. ARFRP1 could do so either by acting as a GEF or by recruiting specific GEFs for ARL1 or ARL5. The resulting ARL1-GTP and ARL5-GTP associate with the TGN membrane via their N-terminally myristoylated amphipathic α-helices. ARL1 in turn recruits three golgins to the TGN (Golgin-245, Golgin-97, and GCC88), which capture retrograde transport carriers containing specific cargos. ARL5, on the other hand, recruits the GARP complex. The golgins then undergo a conformational collapse (Cheung et al., 2015) that brings the carriers close to the TGN, enabling the transfer of the carriers to GARP. Finally, GARP promotes the assembly of the trans-SNARE complex that allows fusion of the carrier and TGN membranes, resulting in delivery of the specific cargos to the TGN. SYS1 and ARFRP1 thus function upstream of two other small GTPases, enabling the recruitment of distinct types of tethering factors to the TGN. A fourth golgin, GCC185, is recruited to the TGN by RAB6. It remains to be determined if and how RAB6-GCC185 coordinates with ARFRP1-ARL5-GARP to tether transport carriers to the TGN. The role of SYS1, ARFRP1, and ARL5 in the recruitment of GARP to the TGN Previous studies of small GTPases that regulate the recruitment of GARP to the TGN in different organisms had painted a complicated picture involving RAB6, ARL1, ARL5, and their orthologues and paralogs (Siniossoglou and Pelham, 2001; Panic et al., 2003b; Liewen et al., 2005; Rosa-Ferreira et al., 2015). Following on the finding that RAB4 may regulate the recruitment of the related EARP complex to endosomes (Gillingham et al., 2014; Schindler et al., 2015), we spent considerable effort performing siRNA and dominant-negative interference screens to identify RAB GTPases that mediate the recruitment of GARP to the TGN. This effort, however, failed to produce any reliable hits. We thus switched our approach to testing for the role of ARL-family GTPases using CRISPR/Cas9 KO cells. This approach succeeded in revealing that ARL5, but not ARL1 and RAB6, is required for GARP recruitment to the TGN in human cells. The effect of ARL5 KO was more complete than that of ARL5 KD used in previous studies (Rosa-Ferreira et al., 2015), providing strong confirmation that this GTPase is critical for GARP recruitment to the TGN. The fact that Drosophila ARL5 physically interacts with GARP (Rosa-Ferreira et al., 2015) makes it likely that GARP is a direct effector of ARL5. A surprising finding was that a second ARL-family GTPase, ARFRP1, was also required for GARP recruitment to the TGN. This GTPase had previously been shown to localize to the TGN and to mediate both forward and retrograde transport processes at this organelle (Shin et al., 2005; Zahn et al., 2008; Nishimoto-Morita et al., 2009; Guo et al., 2013; Ma et al., 2018). Most importantly for our study, ARFRP1 and its yeast orthologue Arl3p had been shown to mediate recruitment of ARL1/Arl1p and their effector golgins to the TGN (Panic et al., 2003b; Setty et al., 2003; Shin et al., 2005; Zahn et al., 2006). This raised the possibility that ARFRP1 could similarly function upstream of ARL5 to recruit GARP to the TGN. Indeed, we found that ARFRP1-KO decreased the association of ARL5 with membranes and that the dissociation of GARP from the TGN in ARFRP1-KO cells could be suppressed by overexpression of constitutively active but not inactive ARL5, indicating that ARFRP1-GTP is required for activation of ARL5. Since yeast Sys1p had been shown to target Arl3p to the late Golgi (Behnia et al., 2004; Setty et al., 2004), we also tested whether the orthologous human SYS1 was required for ARFRP1 recruitment to the TGN. We found that not only was this the case, but also that SYS1 was required for ARL1, ARL5, GCC88, and GARP recruitment to the TGN. Moreover, we observed that SYS1 was required for the maintenance of normal levels of ARFRP1, suggesting that formation of a complex between SYS1 and ARFRP1 stabilizes the ARFRP1 protein. The SYS1-ARFRP1 ensemble thus regulates both the ARL1-golgin and ARL5-GARP arms of vesicle tethering at the TGN (Fig. 8). Small GTPases that regulate the recruitment of golgins to the TGN The Arl3p-Arl1p arm is well known to regulate the recruitment of the only GRIP domain–containing golgin in yeast, Imh1p (Panic et al., 2003b; Setty et al., 2003). Similarly, there is agreement that ARFRP1 and ARL1 regulate the recruitment of Golgin-245 and Golgin-97 to the TGN (Lu and Hong, 2003; Derby et al., 2004; Wu et al., 2004; Shin et al., 2005; Zahn et al., 2006). In addition to confirming these findings, we observed that association of GCC88 with the TGN is dependent on both ARFRP1 and ARL1, but not ARL5 and RAB6. There is less agreement in the literature about the specific GTPase requirement for recruitment of GCC185 to the TGN. Whereas one study showed that ARL1 and RAB6A/A′ (splice variants of RAB6A) cooperate to recruit GCC185 to the TGN (Burguete et al., 2008), another study argued that the localization of GCC185 to the TGN is independent of both ARL1 and RAB6A/A′ (Houghton et al., 2009). Our analyses using KO cells now demonstrate that GCC185 recruitment to the TGN is independent of ARFRP1, ARL1, and ARL5, but dependent on RAB6. The difference with previous studies could be due to the KO of RAB6B in addition to RAB6A in our study. Our results lead us to conclude the existence of two distinct mechanisms for the recruitment of golgins to the TGN: ARFRP1-ARL1 for Golgin-245, Golgin-97, and GCC88, and RAB6 for GCC185. Although these are the predominant mechanisms, we cannot rule out that other small GTPases contribute to some extent to the recruitment or function of these TGN golgins. The existence of cooperative or alternative mechanisms for tethering to the TGN is supported by the ability to suppress defects in yeast Arl1p mutants by overexpression of Ypt6p (Wakade et al., 2017) and vice versa (Chen et al., 2019). How does ARFRP1 mediate recruitment and function of ARL5 and ARL1? A plausible explanation for the functional relationships among the ARL GTPases examined in our study is that ARFRP1 functions as a guanine nucleotide exchange factor (GEF) for exchange of GTP for GDP on both ARL5 and ARL1. A precedent for such a mechanism is the activity of ARL13B as a GEF for ARL3 in the process of ciliogenesis (Gotthardt et al., 2015; Ivanova et al., 2017). However, ARL13B is atypical in that, in addition to its G-domain, it comprises an ∼20-kD C-terminal extension that is required for its ARL3-GEF activity and that is not present in ARFRP1. An alternative explanation would be that ARFRP1 recruits GEFs for ARL5 and ARL1. This would be analogous to the role of the yeast RAB Ypt32p in recruiting the Sec4p GEF for another RAB, Sec2p (Ortiz et al., 2002), and of the mammalian RAB33B in recruiting the RIC1-RGP1 GEF for RAB6A (Pusapati et al., 2012). Yeast Mon2p was proposed to act as a GEF for Arl1p (Jochum et al., 2002) and to function together with Arl3p and Arl1p in an alternative pathway to that mediated by Ypt6p (Setty et al., 2003). However, KO of MON2 did not affect Arl1p and Imh1p recruitment to the late Golgi (Panic et al., 2003b; Setty et al., 2003). To date, no other candidate GEFs for ARL5 and ARL1 have been identified. Further studies will thus be required to determine whether ARFRP1 has intrinsic GEF activity toward ARL5 and ARL1, or whether it functions to recruit distinct GEFs for these ARLs. In addition to elucidating a mechanism that enables the coordinated regulation of two types of vesicle-tethering factor, our findings have uncovered a novel principle in which a single small GTPase is the origin of two small GTPase cascades involved in protein trafficking. As GEFs and GAPs for other small GTPases are identified, it will be of interest to determine if a similar principle applies to other vesicle-tethering events or to other trafficking processes that depend on simultaneous engagement of distinct sets of proteins by multiple small GTPases. Materials and methods Antibodies The following antibodies were used for immunoblotting and/or immunofluorescence microscopy: rabbit anti-VPS51 (HPA039650; Atlas Antibodies), rabbit anti-VPS52 made in our laboratory (Pérez-Victoria et al., 2008), rabbit anti-VPS53 (HPA024446; Atlas Antibodies), mouse anti-VPS50 (FLJ20097, monoclonal antibody M01, 2D11; Abnova), mouse anti-Myc epitope (9E10; Santa Cruz Biotechnology), sheep anti-TGN46 (AHP500G; Bio-Rad), mouse anti–β-actin (G043; Applied Biological Materials), rabbit anti-giantin (ab80864; Abcam), mouse HRP-conjugated anti-α-tubulin (DM1A; Santa Cruz Biotechnology), rabbit anti-ARL1 (16012-1-AP; Proteintech), rabbit anti-ARFRP1 (PA5-50606; Thermo Fisher Scientific), mouse anti-ARL5A (sc-514680; Santa Cruz Biotechnology), rabbit anti-RAB6A (GTX110646; GeneTex), rabbit anti-GCC88 (HPA021323; Sigma-Aldrich), rabbit anti-GCC185 (HPA035849; Sigma-Aldrich), mouse anti-Golgin-245 (611281; BD Biosciences), mouse anti-GM130 (610822; BD Biosciences), mouse anti–Golgin-97 (A-21270; Thermo Fisher Scientific), rabbit anti-TMF1 (HPA008729, Sigma-Aldrich), monoclonal HRP-conjugated anti-GFP (Miltenyi Biotec Inc.), rabbit anti-GFP (A-11122; Thermo Fisher Scientific), HRP-conjugated goat anti-rabbit and donkey anti-mouse antibodies (Jackson ImmunoResearch), HRP-conjugated donkey anti-sheep (R&D Systems), and Alexa Fluor–conjugated secondary antibodies for immunostaining (Thermo Fisher Scientific). Plasmid constructs A plasmid encoding human VPS54 with a C-terminal 13Myc tag (pCI-neo-hVPS54-13Myc) was described previously (Gershlick et al., 2019). Human VPS54 cDNA was purchased (Origene) and subcloned into the pEGFP-N1 vector (Clontech, Takara Bio Inc.). The purchased VPS54 cDNA contains a 120-bp insertion at the 3′ coding region (5′-ACA​GGG​TCT​CAT​GGT​CAC​ACA​GGC​TGG​AGG​GCA​GTG​GTG​TGA​TCA​TGG​CTC​ACT​GCA​GCC​TCA​ACC​TCC​GTG​GCT​CAA​GCA​ATC​CTC​CCA​CTG​CAG​ACT​CCC​CAA​TAA​CTG​CGA​CTA​CAG-3′) that is likely to derive from alternative splicing. As a result, 40 amino acids (QGLMVTQAGGQWCDHGSLQPQPPWLKQSSHCRLPNNCDYR) are inserted after Gly943 of the human VPS54 sequence registered in UniProt (Q9P1Q0). Sequences encoding N-terminal (amino acid residues 1–557) and C-terminal halves (amino acid residues 558–1,017) of VPS54 were amplified and subcloned into a pCI-neo-13Myc backbone vector (Schindler et al., 2015). Human ARL5B (#67400) and ARFRP1 (#67471) cDNAs were obtained from Addgene (deposited by Richard Kahn, Emory University, Atlanta, GA) and subcloned into pEGFP-N1. The ARFRP1 cDNA was also subcloned into a pQCXIP vector (Clontech). Constitutively active and inactive forms of ARL1, ARL5B, and ARFRP1 tagged with EGFP (herein referred to as GFP) were generated by site-directed mutagenesis using Q5 High-Fidelity DNA Polymerase (New England Biolabs). pGFP-C1-mouse Rab6A and Rab6B were the kind gift of Mitsunori Fukuda (Tohoku University, Sendai, Japan; Matsui et al., 2011). Human SYS1 cDNA was obtained from the ORFeome v8.1 collection (Dharmacon) and subcloned into the pGFP-N1 and pQCXIP vectors. A plasmid encoding Myc-GGA2 (pCR3.1-Myc-hGGA2) was previously described (Dell’Angelica et al., 2000). The sequence of all constructs was confirmed by Sanger sequencing. siRNAs The following FlexiPlate siRNAs (Qiagen) were used in our experiments: human ARL5A (5′-CAA​GTT​AAT​GGC​ATT​GAT​TTA/CAC​GTT​TCC​TAA​TGT​GGG​ATA-3′), human ARL5B (5′-CTC​CCG​GAT​TGG​TGT​GAG​ATA/CTG​CCT​TCT​TGT​ATC​TAG​TAA-3′), human RAB6A (5′-CAG​ATG​GGT​CAT​ATT​CTT​TAT/TAC​GGT​CTT​CTT​TGA​GGT​CAA-3′), human RAB6B (5′-CGG​CTG​TTA​CTT​AAA​CAA​CTA/ATC​CAT​GTT​CTT​AGA​GCC​TCA-3′). Control siRNA (5′-CAG​TCG​CGT​TTG​CGA​CTG​GTT-3′) was purchased from Dharmacon. CRISPR/Cas9 KO VPS50-KO, VPS51-KO, VPS52-KO, VPS53-KO, VPS54-KO, ARL1-KO, ARL5-KO, ARFRP1-KO, RAB6-KO, and SYS1-KO HeLa cells were generated using CRISPR/Cas9 (Cong et al., 2013). The targeting sequences for VPS50 (5′-TGA​ACA​AGT​ATA​TTT​TTC​TG/TGA​TAT​TGT​TAA​ATA​TGA​GC/ATA​AGC​TCT​TGT​TCA​GCT​TG-3′), VPS51 (5′-CCT​AGC​CCG​GGG​TCT​GGA​CC/TGC​GCC​TTC​CGC​CGA​CGC​TC/TCG​CAT​CAG​CGC​CAC​GCT​GC/CGT​CAC​CCT​CCG​TCC​CCC​AG-3′), VPS52 (5′-CTC​TGA​GAT​CCG​GAC​ACT​GC/TGG​GGA​GCT​TGT​TGA​TGG​TC-3′), VPS53 (5′-GGA​GGA​GGA​ACT​GGA​GTT​CG/GGT​GCA​GCT​GGC​CAT​CGA​GC-3′), VPS54 (5′-ACT​GCC​AGA​TGT​GTG​TCC​CA/GAG​GCA​CTG​GTG​AAG​AAC​TG-3′), ARL1 (5′-TTT​CCA​GTC​TGT​TTG​GAA​CT/TTG​TAA​TCT​GTA​CAA​AAT​TG-3′), ARL5A (5′-TAA​TAC​ACG​TTT​CCT​AAT​GT/CAC​TTA​CTA​TAC​TAA​CAC​AG-3′), ARL5B (5′-CAA​AGT​AAT​TAT​AGT​GGG​AC/TAA​ACA​CCC​ATA​CTT​ACA​AT-3′), ARFRP1 (5′-TCC​TGG​GCC​TGG​ACA​ATG​CT/GTA​CAA​GTA​CAT​GTT​TCA​GA-3′), RAB6A (5′-ATG​TCC​ACG​GGC​GGA​GAC​TT/GGA​GCA​AAG​CGG​TGA​GTG​CG-3′), RAB6B (5′-ATG​TCC​GCA​GGG​GGA​GAT​TT/TTG​GGC​GGC​GGG​GGT​CGA​CT-3′), and SYS1 (5′-ACC​GTG​TAT​TAC​GGC​TCG​CT/GCA​TGA​GGA​CGA​TCT​GCG​AC-3′) were cloned separately into pSpCas9 (BB)-2A-GFP plasmid (#48138; Addgene; deposited by Feng Zhang, Massachusetts Institute of Technology, Cambridge, MA). HeLa cells were transfected with two to four plasmids containing the different targeting sequences for the same gene, and GFP-positive cells were isolated by flow cytofluorometry after 24 h and single-cell cloned in 96-well plates. KO of each clone was confirmed by immunoblotting using specific antibodies. Because we were not able to obtain antibodies that detect VPS54 and SYS1, we confirmed genomic deletions in these genes by Sanger sequencing of the genomic PCR products. The following primers were used to amplify each fragment: VPS54, 5′-TCA​ATT​TTC​CCA​ATT​AAG​AGC​AA-3′ (forward), 5′-CAC​CAT​GTT​AGC​CAG​GAT​GA-3′ (reverse); SYS1, 5′-GAC​TCT​TGG​AAT​GGG​CTC​AC-3′ (forward), 5′-TTC​CCA​GGG​CTA​CAA​AGA​AA-3′ (reverse). To generate ARFRP1-rescue and SYS1-rescue cells, retrovirus particles were prepared by transfecting HEK293T cells with pQCXIP-ARFRP1 or pQCXIP-SYS1 and retrovirus packaging plasmids (Clontech). Medium was collected 48 h after transfection and centrifuged for 10 min at 1,000 ×g to remove debris. The ARFRP1-KO or SYS1-KO HeLa cells were immediately infected with the corresponding virus, and stably transduced cells were selected with 2 μg/ml puromycin. Cell culture and transfection Unless otherwise specified, most experiments were done in HeLa cells (ATCC). H4 and HEK293T cells (ATCC) were used in some experiments. All cells were cultured in DMEM supplemented with 10% FBS and MycoZap Plus-CL (Lonza). Plasmids were transfected using Lipofectamine 2000 (Thermo Fisher Scientific), and cells were analyzed 24–48 h after transfection. siRNAs were transfected with Lipofectamine RNAiMAX (Thermo Fisher Scientific), and cells were analyzed 72 h after transfection. Immunoblot analysis Cells were scraped off the dish and lysed in buffer containing 1% Triton X-100; 50 mM Tris, pH 7.4; 150 mM NaCl; and cOmplete EDTA-free Protease Inhibitor (Roche). Laemmli SDS-PAGE sample buffer (Bio-Rad) containing 2.5% 2-mercaptoethanol was added to the lysate and incubated for 5 min at 98°C. Immunoblotting was performed using SDS-PAGE separation and subsequent transfer to polyvinyl difluoride or nitrocellulose membranes. Membranes were blocked for 0.5–1 h with 3% nonfat milk (Bio-Rad) in TBS-T (TBS supplemented with 0.05% Tween 20; Sigma-Aldrich) before being incubated overnight with primary antibody diluted in TBS-T with 3% nonfat milk. Membranes were washed three times for 20 min in TBS-T before being incubated for 2–3 h in HRP-conjugated secondary antibody (1:5,000) diluted in TBS-T with 3% nonfat milk. Membranes were washed three times in TBS-T and visualized using either Clarity ECL Western Blot substrate (Bio-Rad) or a 1:1 mixture of homemade ECL solution (Solution 1: 0.25 mM luminol, 0.37 mM p-coumaric acid, and 0.1 M Tris–HCl, pH 8.5; and Solution 2: 0.0192% hydrogen peroxide and 0.1 M Tris-HCl, pH 8.5). Immunofluorescent staining Cells for immunofluorescence microscopy were plated onto fibronectin-coated cover glasses, transfected as described above, and fixed using 4% paraformaldehyde in PBS. Cells were permeabilized with 0.1% Triton-X in PBS at RT for 5 min and incubated in blocking buffer (1% BSA in PBS) for 30 min. Primary antibodies were diluted in blocking buffer and incubated on cells for 1 h at RT. Alexa Fluor secondary antibodies were diluted in blocking buffer containing DAPI, and cells were incubated for 1 h at RT. The coverslips were mounted on glass slides using ProLong Gold Antifade (Thermo Fisher Scientific), and the cells were imaged on a confocal microscope (LSM710 or LSM880; Carl Zeiss) with an oil-immersion 63×/1.40 NA Plan-Apochromat Oil DIC M27 objective lens (Carl Zeiss). Image settings (i.e., gain, laser power, and pinhole) were kept constant for images presented for comparison. Images were acquired using ZEN 2012 software (Carl Zeiss) and processed by Fiji (https://fiji.sc), including brightness adjustment, contrast adjustment, channel merging, and cropping. Uptake of STxB Uptake of Cy3-STxB (Amessou et al., 2006; kind gift of David Gershlick, University of Cambridge, Cambridge, UK) was performed as previously described (Pérez-Victoria et al., 2008). Briefly, cells were incubated for 30 min in STxB uptake medium (DMEM without FBS plus 1% BSA). Cells were subsequently incubated with 0.5 μg/ml Cy3-STxB in STxB uptake medium for 15 min and washed with PBS once before being incubated for 1 h with prewarmed complete DMEM. Cells were then fixed with 4% paraformaldehyde in PBS and immunostained for GM130 as a marker for the Golgi complex as described above. Cytosol-membrane fractionation A cell fractionation kit (#9038; Cell Signaling Technology) was used to fractionate proteins into cytosolic and membrane fractions according to the manufacturer's instructions. Briefly, cells on a culture dish were trypsinized, and 2.5 million cells were aliquoted into a 1.5-ml tube. Cells were washed with PBS once and resuspended in 0.5 ml PBS. 0.1 ml of the suspension was aliquoted into another 1.5-ml tube for analysis of whole-cell lysate. The rest of the suspension was centrifuged for 5 min at 500 ×g at 4°C, and the pellet was resuspended in 0.5 ml of cytoplasmic isolation buffer. After incubation on ice for 5 min, the suspension was centrifuged for 5 min at 500 ×g at 4°C, and the supernatant was saved as the cytosolic fraction. The pellet was resuspended in 0.5 ml of membrane isolation buffer and incubated on ice for 5 min. The suspension was centrifuged for 5 min at 8,000 ×g at 4°C. The supernatant was saved as the membrane fraction. Laemmli sample buffer (Bio-Rad) containing 2.5% 2-mercaptoethanol was added to each fraction, and all samples were heated for 5 min at 98°C. Samples were subjected to SDS-PAGE followed by immunoblotting as described above. Quantification and statistics Quantification of the percentage of cells with VPS54-13Myc (Fig. 2 C; Fig. 3, B and D; and Fig. 7 E), Cy3-StxB (Fig. 4 E), or GCC88 (Fig. 6, B and D; and Fig. 7 F) on the TGN was performed by counting >100 cells per experiment per sample in three independent experiments. Percentages were calculated with Excel (Microsoft) and, wherever indicated, statistically analyzed by Dunnett's multiple comparison test (Prism 8 for macOS; GraphPad Software). Quantification of the ratio of membrane to the sum of membrane and cytosolic ARL1 and ARL5 in subcellular fractionation experiments (Fig. 3 F and Fig. 7 H), the protein abundance of ARFRP1 (Fig. 7 I), and the mean intensity of Golgin-97 after background subtraction in immunofluorescent images (Fig. 5 A) was performed in Fiji. Percentages were calculated with Excel (Microsoft) and statistically analyzed by Dunnett’s multiple comparison test (Prism 8 for macOS). All quantitative data are expressed as the mean ± SEM, and all graphs were drawn using Prism 8 for macOS. Asterisks indicate the calculated statistical significance, as follows: NS, P > 0.05; *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001. Online supplemental material Fig. S1 shows validation of epitope-tagged VPS54 as a surrogate for GARP in immunofluorescence microscopy experiments. Fig. S2 shows specificity of antibodies to ARL5 and RAB6 paralogs. Fig. S3 shows intracellular localization of GCC185 and TMF1 in WT and SYS1-KO cells. Fig. S4 shows intracellular localization of Myc-GGA2 in WT, ARL1-KO, ARL5-KO, ARFRP1-KO, RAB6-KO, and SYS1-KO cells. Supplementary Material Supplemental Materials (PDF) Acknowledgments We thank M. Fukuda, D. Gershlick, R. Kahn, and F. Zhang for kind gifts of reagents, and members of the Bonifacino laboratory for helpful discussions. This work was funded by the Intramural Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development (ZIA HD001607). M. Ishida was the recipient of a fellowship from the Japan Society for the Promotion of Science. The authors declare no competing financial interests. Author contributions: M. Ishida and J.S. Bonifacino conceived the project. M. Ishida performed all the experiments. M. Ishida and J.S. Bonifacino analyzed and interpreted the data. M. Ishida and J. S. Bonifacino wrote the manuscript. ==== Refs Amessou, M., V. Popoff, B. Yelamos, A. Saint-Pol, and L. Johannes. 2006. Measuring retrograde transport to the trans-Golgi network. Curr. Protoc. Cell Biol. Chapter15 :15.10. Behnia, R., B. Panic, J.R. Whyte, and S. Munro. 2004. Targeting of the Arf-like GTPase Arl3p to the Golgi requires N-terminal acetylation and the membrane protein Sys1p. Nat. Cell Biol. 6 :405–413. 10.1038/ncb1120 15077113 Bennett, D., and L.C. Dunn. 1958. Effects on embryonic development of a group of genetically similar lethal alleles derived from different populations of wild house mice. J. Morphol. 103 :135–157. 10.1002/jmor.1051030106 Bonifacino, J.S., and A. Hierro. 2011. Transport according to GARP: Receiving retrograde cargo at the trans-Golgi network. Trends Cell Biol. 21 :159–167. 10.1016/j.tcb.2010.11.003 21183348 Burguete, A.S., T.D. Fenn, A.T. Brunger, and S.R. Pfeffer. 2008. Rab and Arl GTPase family members cooperate in the localization of the golgin GCC185. Cell. 132 :286–298. 10.1016/j.cell.2007.11.048 18243103 Chen, Y.T., I.H. Wang, Y.H. Wang, W.Y. Chiu, J.H. Hu, W.H. Chen, and F.S. Lee. 2019. Action of Arl1 GTPase and golgin Imh1 in Ypt6-independent retrograde transport from endosomes to the trans-Golgi network. Mol. Biol. Cell. 30 :1008–1019. 10.1091/mbc.E18-09-0579 30726160 Cheung, P.Y., and S.R. Pfeffer. 2016. Transport vesicle tethering at the trans Golgi network: Coiled coil proteins in action. Front. Cell Dev. Biol. 4 :18. 10.3389/fcell.2016.00018 27014693 Cheung, P.Y., C. Limouse, H. Mabuchi, and S.R. Pfeffer. 2015. Protein flexibility is required for vesicle tethering at the Golgi. eLife. 4 :e12790. 10.7554/eLife.12790 26653856 Cong, L., F.A. Ran, D. Cox, S. Lin, R. Barretto, N. Habib, P.D. Hsu, X. Wu, W. Jiang, L.A. Marraffini, . 2013. Multiplex genome engineering using CRISPR/Cas systems. Science. 339 :819–823. 10.1126/science.1231143 23287718 Conibear, E., J.N. Cleck, and T.H. Stevens. 2003. Vps51p mediates the association of the GARP (Vps52/53/54) complex with the late Golgi t-SNARE Tlg1p. Mol. Biol. Cell. 14 :1610–1623. 10.1091/mbc.e02-10-0654 12686613 Dell’Angelica, E.C., R. Puertollano, C. Mullins, R.C. Aguilar, J.D. Vargas, L.M. Hartnell, and J.S. Bonifacino. 2000. GGAs: A family of ADP ribosylation factor-binding proteins related to adaptors and associated with the Golgi complex. J. Cell Biol. 149 :81–94. 10.1083/jcb.149.1.81 10747089 Derby, M.C., C. van Vliet, D. Brown, M.R. Luke, L. Lu, W. Hong, J.L. Stow, and P.A. Gleeson. 2004. Mammalian GRIP domain proteins differ in their membrane binding properties and are recruited to distinct domains of the TGN. J. Cell Sci. 117 :5865–5874. 10.1242/jcs.01497 15522892 Derby, M.C., Z.Z. Lieu, D. Brown, J.L. Stow, B. Goud, and P.A. Gleeson. 2007. The trans-Golgi network golgin, GCC185, is required for endosome-to-Golgi transport and maintenance of Golgi structure. Traffic. 8 :758–773. 10.1111/j.1600-0854.2007.00563.x 17488291 Donaldson, J.G., and C.L. Jackson. 2011. ARF family G proteins and their regulators: Roles in membrane transport, development and disease. Nat. Rev. Mol. Cell Biol. 12 :362–375. 10.1038/nrm3117 21587297 Feinstein, M., H. Flusser, T. Lerman-Sagie, B. Ben-Zeev, D. Lev, O. Agamy, I. Cohen, R. Kadir, S. Sivan, E. Leshinsky-Silver, . 2014. VPS53 mutations cause progressive cerebello-cerebral atrophy type 2 (PCCA2). J. Med. Genet. 51 :303–308. 10.1136/jmedgenet-2013-101823 24577744 Fridmann-Sirkis, Y., S. Siniossoglou, and H.R. Pelham. 2004. TMF is a golgin that binds Rab6 and influences Golgi morphology. BMC Cell Biol. 5 :18. 10.1186/1471-2121-5-18 15128430 Gershlick, D.C., M. Ishida, J.R. Jones, A. Bellomo, J.S. Bonifacino, and D.B. Everman. 2019. A neurodevelopmental disorder caused by mutations in the VPS51 subunit of the GARP and EARP complexes. Hum. Mol. Genet. 28 :1548–1560. 10.1093/hmg/ddy423 30624672 Gillingham, A.K., R. Sinka, I.L. Torres, K.S. Lilley, and S. Munro. 2014. Toward a comprehensive map of the effectors of rab GTPases. Dev. Cell. 31 :358–373. 10.1016/j.devcel.2014.10.007 25453831 Gotthardt, K., M. Lokaj, C. Koerner, N. Falk, A. Gießl, and A. Wittinghofer. 2015. A G-protein activation cascade from Arl13B to Arl3 and implications for ciliary targeting of lipidated proteins. eLife. 4 :e11859. 10.7554/eLife.11859 26551564 Griffiths, G., and K. Simons. 1986. The trans Golgi network: Sorting at the exit site of the Golgi complex. Science. 234 :438–443. 10.1126/science.2945253 2945253 Guo, Y., G. Zanetti, and R. Schekman. 2013. A novel GTP-binding protein-adaptor protein complex responsible for export of Vangl2 from the trans Golgi network. eLife. 2 :e00160.23326640 Guo, Y., D.W. Sirkis, and R. Schekman. 2014. Protein sorting at the trans-Golgi network. Annu. Rev. Cell Dev. Biol. 30 :169–206. 10.1146/annurev-cellbio-100913-013012 25150009 Hady-Cohen, R., H. Ben-Pazi, V. Adir, K. Yosovich, L. Blumkin, T. Lerman-Sagie, and D. Lev. 2018. Progressive cerebello-cerebral atrophy and progressive encephalopathy with edema, hypsarrhythmia and optic atrophy may be allelic syndromes. Eur. J. Paediatr. Neurol. 22 :1133–1138. 10.1016/j.ejpn.2018.07.003 30100179 Hierro, A., D.C. Gershlick, A.L. Rojas, and J.S. Bonifacino. 2015. Formation of tubulovesicular carriers from endosomes and their fusion to the trans-Golgi network. Int. Rev. Cell Mol. Biol. 318 :159–202. 10.1016/bs.ircmb.2015.05.005 26315886 Houghton, F.J., P.L. Chew, S. Lodeho, B. Goud, and P.A. Gleeson. 2009. The localization of the Golgin GCC185 is independent of Rab6A/A′ and Arl1. Cell. 138 :787–794. 10.1016/j.cell.2009.05.048 19703403 Houghton, F.J., S.A. Bellingham, A.F. Hill, D. Bourges, D.K. Ang, T. Gemetzis, I. Gasnereau, and P.A. Gleeson. 2012. Arl5b is a Golgi-localised small G protein involved in the regulation of retrograde transport. Exp. Cell Res. 318 :464–477. 10.1016/j.yexcr.2011.12.023 22245584 Ivanova, A.A., T. Caspary, N.T. Seyfried, D.M. Duong, A.B. West, Z. Liu, and R.A. Kahn. 2017. Biochemical characterization of purified mammalian ARL13B protein indicates that it is an atypical GTPase and ARL3 guanine nucleotide exchange factor (GEF). J. Biol. Chem. 292 :11091–11108. 10.1074/jbc.M117.784025 28487361 Jochum, A., D. Jackson, H. Schwarz, R. Pipkorn, and B. Singer-Krüger. 2002. Yeast Ysl2p, homologous to Sec7 domain guanine nucleotide exchange factors, functions in endocytosis and maintenance of vacuole integrity and interacts with the Arf-Like small GTPase Arl1p. Mol. Cell. Biol. 22 :4914–4928. 10.1128/MCB.22.13.4914-4928.2002 12052896 Johannes, L., and B. Goud. 1998. Surfing on a retrograde wave: How does Shiga toxin reach the endoplasmic reticulum? Trends Cell Biol. 8 :158–162. 10.1016/S0962-8924(97)01209-9 9695830 Karlsson, P., A. Droce, J.M. Moser, S. Cuhlmann, C.O. Padilla, P. Heimann, J.W. Bartsch, A. Füchtbauer, E.M. Füchtbauer, and T. Schmitt-John. 2013. Loss of vps54 function leads to vesicle traffic impairment, protein mis-sorting and embryonic lethality. Int. J. Mol. Sci. 14 :10908–10925. 10.3390/ijms140610908 23708095 Kulak, N.A., G. Pichler, I. Paron, N. Nagaraj, and M. Mann. 2014. Minimal, encapsulated proteomic-sample processing applied to copy-number estimation in eukaryotic cells. Nat. Methods. 11 :319–324. 10.1038/nmeth.2834 24487582 Lieu, Z.Z., M.C. Derby, R.D. Teasdale, C. Hart, P. Gunn, and P.A. Gleeson. 2007. The golgin GCC88 is required for efficient retrograde transport of cargo from the early endosomes to the trans-Golgi network. Mol. Biol. Cell. 18 :4979–4991. 10.1091/mbc.e07-06-0622 17914056 Liewen, H., I. Meinhold-Heerlein, V. Oliveira, R. Schwarzenbacher, G. Luo, A. Wadle, M. Jung, M. Pfreundschuh, and F. Stenner-Liewen. 2005. Characterization of the human GARP (Golgi associated retrograde protein) complex. Exp. Cell Res. 306 :24–34. 10.1016/j.yexcr.2005.01.022 15878329 Lu, L., and W. Hong. 2003. Interaction of Arl1-GTP with GRIP domains recruits autoantigens Golgin-97 and Golgin-245/p230 onto the Golgi. Mol. Biol. Cell. 14 :3767–3781. 10.1091/mbc.e03-01-0864 12972563 Lu, L., and W. Hong. 2014. From endosomes to the trans-Golgi network. Semin. Cell Dev. Biol. 31 :30–39. 10.1016/j.semcdb.2014.04.024 24769370 Lu, L., G. Tai, and W. Hong. 2004. Autoantigen Golgin-97, an effector of Arl1 GTPase, participates in traffic from the endosome to the trans-golgi network. Mol. Biol. Cell. 15 :4426–4443. 10.1091/mbc.e03-12-0872 15269279 Ma, T., B. Li, R. Wang, P.K. Lau, Y. Huang, L. Jiang, R. Schekman, and Y. Guo. 2018. A mechanism for differential sorting of the planar cell polarity proteins Frizzled6 and Vangl2 at the trans-Golgi network. J. Biol. Chem. 293 :8410–8427. 10.1074/jbc.RA118.001906 29666182 Mallard, F., B.L. Tang, T. Galli, D. Tenza, A. Saint-Pol, X. Yue, C. Antony, W. Hong, B. Goud, and L. Johannes. 2002. Early/recycling endosomes-to-TGN transport involves two SNARE complexes and a Rab6 isoform. J. Cell Biol. 156 :653–664. 10.1083/jcb.200110081 11839770 Matsui, T., T. Itoh, and M. Fukuda. 2011. Small GTPase Rab12 regulates constitutive degradation of transferrin receptor. Traffic. 12 :1432–1443. 10.1111/j.1600-0854.2011.01240.x 21718402 Nishimoto-Morita, K., H.W. Shin, H. Mitsuhashi, M. Kitamura, Q. Zhang, L. Johannes, and K. Nakayama. 2009. Differential effects of depletion of ARL1 and ARFRP1 on membrane trafficking between the trans-Golgi network and endosomes. J. Biol. Chem. 284 :10583–10592. 10.1074/jbc.M900847200 19224922 Ortiz, D., M. Medkova, C. Walch-Solimena, and P. Novick. 2002. Ypt32 recruits the Sec4p guanine nucleotide exchange factor, Sec2p, to secretory vesicles; evidence for a Rab cascade in yeast. J. Cell Biol. 157 :1005–1015. 10.1083/jcb.200201003 12045183 Panic, B., O. Perisic, D.B. Veprintsev, R.L. Williams, and S. Munro. 2003 a. Structural basis for Arl1-dependent targeting of homodimeric GRIP domains to the Golgi apparatus. Mol. Cell. 12 :863–874. 10.1016/S1097-2765(03)00356-3 14580338 Panic, B., J.R. Whyte, and S. Munro. 2003 b. The ARF-like GTPases Arl1p and Arl3p act in a pathway that interacts with vesicle-tethering factors at the Golgi apparatus. Curr. Biol. 13 :405–410. 10.1016/S0960-9822(03)00091-5 12620189 Pérez-Victoria, F.J., and J.S. Bonifacino. 2009. Dual roles of the mammalian GARP complex in tethering and SNARE complex assembly at the trans-Golgi network. Mol. Cell. Biol. 29 :5251–5263. 10.1128/MCB.00495-09 19620288 Pérez-Victoria, F.J., G.A. Mardones, and J.S. Bonifacino. 2008. Requirement of the human GARP complex for mannose 6-phosphate-receptor-dependent sorting of cathepsin D to lysosomes. Mol. Biol. Cell. 19 :2350–2362. 10.1091/mbc.e07-11-1189 18367545 Pérez-Victoria, F.J., G. Abascal-Palacios, I. Tascón, A. Kajava, J.G. Magadán, E.P. Pioro, J.S. Bonifacino, and A. Hierro. 2010 a. Structural basis for the wobbler mouse neurodegenerative disorder caused by mutation in the Vps54 subunit of the GARP complex. Proc. Natl. Acad. Sci. USA. 107 :12860–12865. 10.1073/pnas.1004756107 20615984 Pérez-Victoria, F.J., C. Schindler, J.G. Magadán, G.A. Mardones, C. Delevoye, M. Romao, G. Raposo, and J.S. Bonifacino. 2010 b. Ang2/fat-free is a conserved subunit of the Golgi-associated retrograde protein complex. Mol. Biol. Cell. 21 :3386–3395. 10.1091/mbc.e10-05-0392 20685960 Pfeffer, S.R. 2017. Rab GTPases: Master regulators that establish the secretory and endocytic pathways. Mol. Biol. Cell. 28 :712–715. 10.1091/mbc.e16-10-0737 28292916 Pusapati, G.V., G. Luchetti, and S.R. Pfeffer. 2012. Ric1-Rgp1 complex is a guanine nucleotide exchange factor for the late Golgi Rab6A GTPase and an effector of the medial Golgi Rab33B GTPase. J. Biol. Chem. 287 :42129–42137. 10.1074/jbc.M112.414565 23091056 Quenneville, N.R., T.Y. Chao, J.M. McCaffery, and E. Conibear. 2006. Domains within the GARP subunit Vps54 confer separate functions in complex assembly and early endosome recognition. Mol. Biol. Cell. 17 :1859–1870. 10.1091/mbc.e05-11-1002 16452629 Reddy, J.V., A.S. Burguete, K. Sridevi, I.G. Ganley, R.M. Nottingham, and S.R. Pfeffer. 2006. A functional role for the GCC185 golgin in mannose 6-phosphate receptor recycling. Mol. Biol. Cell. 17 :4353–4363. 10.1091/mbc.e06-02-0153 16885419 Rosa-Ferreira, C., C. Christis, I.L. Torres, and S. Munro. 2015. The small G protein Arl5 contributes to endosome-to-Golgi traffic by aiding the recruitment of the GARP complex to the Golgi. Biol. Open. 4 :474–481. 10.1242/bio.201410975 25795912 Schindler, C., Y. Chen, J. Pu, X. Guo, and J.S. Bonifacino. 2015. EARP is a multisubunit tethering complex involved in endocytic recycling. Nat. Cell Biol. 17 :639–650. 10.1038/ncb3129 25799061 Schmitt-John, T., C. Drepper, A. Mussmann, P. Hahn, M. Kuhlmann, C. Thiel, M. Hafner, A. Lengeling, P. Heimann, J.M. Jones, . 2005. Mutation of Vps54 causes motor neuron disease and defective spermiogenesis in the wobbler mouse. Nat. Genet. 37 :1213–1215. 10.1038/ng1661 16244655 Schürmann, A., S. Massmann, and H.G. Joost. 1995. ARP is a plasma membrane-associated Ras-related GTPase with remote similarity to the family of ADP-ribosylation factors. J. Biol. Chem. 270 :30657–30663. 10.1074/jbc.270.51.30657 8530503 Setty, S.R., M.E. Shin, A. Yoshino, M.S. Marks, and C.G. Burd. 2003. Golgi recruitment of GRIP domain proteins by Arf-like GTPase 1 is regulated by Arf-like GTPase 3. Curr. Biol. 13 :401–404. 10.1016/S0960-9822(03)00089-7 12620188 Setty, S.R., T.I. Strochlic, A.H. Tong, C. Boone, and C.G. Burd. 2004. Golgi targeting of ARF-like GTPase Arl3p requires its Nα-acetylation and the integral membrane protein Sys1p. Nat. Cell Biol. 6 :414–419. 10.1038/ncb1121 15077114 Shin, H.W., H. Kobayashi, M. Kitamura, S. Waguri, T. Suganuma, Y. Uchiyama, and K. Nakayama. 2005. Roles of ARFRP1 (ADP-ribosylation factor-related protein 1) in post-Golgi membrane trafficking. J. Cell Sci. 118 :4039–4048. 10.1242/jcs.02524 16129887 Siniossoglou, S., and H.R. Pelham. 2001. An effector of Ypt6p binds the SNARE Tlg1p and mediates selective fusion of vesicles with late Golgi membranes. EMBO J. 20 :5991–5998. 10.1093/emboj/20.21.5991 11689439 Siniossoglou, S., and H.R. Pelham. 2002. Vps51p links the VFT complex to the SNARE Tlg1p. J. Biol. Chem. 277 :48318–48324. 10.1074/jbc.M209428200 12377769 Sugimoto, M., M. Kondo, M. Hirose, M. Suzuki, K. Mekada, T. Abe, H. Kiyonari, A. Ogura, N. Takagi, K. Artzt, . 2012. Molecular identification of t(w5): Vps52 promotes pluripotential cell differentiation through cell-cell interactions. Cell Reports. 2 :1363–1374. 10.1016/j.celrep.2012.10.004 23142660 Sztul, E., P.W. Chen, J.E. Casanova, J. Cherfils, J.B. Dacks, D.G. Lambright, F.S. Lee, P.A. Randazzo, L.C. Santy, A. Schürmann, . 2019. ARF GTPases and their GEFs and GAPs: Concepts and challenges. Mol. Biol. Cell. 30 :1249–1271. 10.1091/mbc.E18-12-0820 31084567 Torres, I.L., C. Rosa-Ferreira, and S. Munro. 2014. The Arf family G protein Arl1 is required for secretory granule biogenesis in Drosophila. J. Cell Sci. 127 :2151–2160. 10.1242/jcs.122028 24610947 Uwineza, A., J.H. Caberg, J. Hitayezu, S. Wenric, L. Mutesa, Y. Vial, S. Drunat, S. Passemard, A. Verloes, V. El Ghouzzi, . 2019. VPS51 biallelic variants cause microcephaly with brain malformations: A confirmatory report. Eur. J. Med. Genet. 62 :103704. 10.1016/j.ejmg.2019.103704 31207318 Wakade, R., H. Labbaoui, D. Stalder, R.A. Arkowitz, and M. Bassilana. 2017. Overexpression of YPT6 restores invasive filamentous growth and secretory vesicle clustering in a Candida albicans arl1 mutant. Small GTPases. 29 :1–7. 10.1080/21541248.2017.1378157 Wong, M., and S. Munro. 2014. The specificity of vesicle traffic to the Golgi is encoded in the golgin coiled-coil proteins. Science. 346 :1256898. 10.1126/science.1256898 25359980 Wu, M., L. Lu, W. Hong, and H. Song. 2004. Structural basis for recruitment of GRIP domain golgin-245 by small GTPase Arl1. Nat. Struct. Mol. Biol. 11 :86–94. 10.1038/nsmb714 14718928 Yoshino, A., S.R. Setty, C. Poynton, E.L. Whiteman, A. Saint-Pol, C.G. Burd, L. Johannes, E.L. Holzbaur, M. Koval, J.M. McCaffery, . 2005. tGolgin-1 (p230, golgin-245) modulates Shiga-toxin transport to the Golgi and Golgi motility towards the microtubule-organizing centre. J. Cell Sci. 118 :2279–2293. 10.1242/jcs.02358 15870108 Yu, I.M., and F.M. Hughson. 2010. Tethering factors as organizers of intracellular vesicular traffic. Annu. Rev. Cell Dev. Biol. 26 :137–156. 10.1146/annurev.cellbio.042308.113327 19575650 Zahn, C., A. Hommel, L. Lu, W. Hong, D.J. Walther, S. Florian, H.G. Joost, and A. Schürmann. 2006. Knockout of Arfrp1 leads to disruption of ARF-like1 (ARL1) targeting to the trans-Golgi in mouse embryos and HeLa cells. Mol. Membr. Biol. 23 :475–485. 10.1080/09687860600840100 17127620 Zahn, C., A. Jaschke, J. Weiske, A. Hommel, D. Hesse, R. Augustin, L. Lu, W. Hong, S. Florian, A. Scheepers, . 2008. ADP-ribosylation factor-like GTPase ARFRP1 is required for trans-Golgi to plasma membrane trafficking of E-cadherin. J. Biol. Chem. 283 :27179–27188. 10.1074/jbc.M802108200 18662990

PMC006xxxxxx/PMC6888974.txt

==== Front J Exp Med J Exp Med jem jem The Journal of Experimental Medicine 0022-1007 1540-9538 Rockefeller University Press 31537641 20190678 10.1084/jem.20190678 Research Articles Article 310 Concomitant PIK3CD and TNFRSF9 deficiencies cause chronic active Epstein-Barr virus infection of T cells Dual PIK3CD and TNFRSF9 deficiency Rodriguez Rémy 12* https://orcid.org/0000-0002-4189-996X Fournier Benjamin 12* Cordeiro Debora Jorge 12* Winter Sarah 12* https://orcid.org/0000-0003-1080-0936 Izawa Kazushi 1 https://orcid.org/0000-0003-4211-7373 Martin Emmanuel 1 Boutboul David 12 Lenoir Christelle 1 Fraitag Sylvie 3 Kracker Sven 24 https://orcid.org/0000-0001-7897-4890 Watts Tania H. 5 Picard Capucine 1267 Bruneau Julie 23 Callebaut Isabelle 8 https://orcid.org/0000-0003-4025-5983 Fischer Alain 27910 Neven Bénédicte 27 https://orcid.org/0000-0001-8238-4391 Latour Sylvain 12 1 Laboratory of Lymphocyte Activation and Susceptibility to EBV Infection, Institut National de la Santé et la Recherche Médicale, Unité Mixte de Recherche 1163, Paris, France 2 University Paris Descartes Sorbonne Paris Cité, Imagine Institute, Paris, France 3 Department of Pathology, Necker-Enfants Malades Hospital, Assistance Publique-Hôpitaux de Paris, Paris, France 4 Laboratory of Human Lymphohematopoiesis, Institut National de la Santé et la Recherche Médicale, Unité Mixte de Recherche 1163, Paris, France 5 Department of Immunology, University of Toronto, Toronto, Canada 6 Centre d’Etude des Déficits Immunitaires, Necker-Enfants Malades Hospital, Assistance Publique-Hôpitaux de Paris, Paris, France 7 Department of Pediatric Immunology, Hematology and Rheumatology, Necker-Enfants Malades Hospital, Assistance Publique-Hôpitaux de Paris, Paris, France 8 Sorbonne Université, Muséum National d’Histoire Naturelle, Centre National de la Recherche Scientifique Unité Mixte de Recherche 7590, Institut de Minéralogie, de Physique des Matériaux et de Cosmochimie, Paris, France 9 Collège de France, Paris, France 10 Institut National de la Santé et la Recherche Médicale, Unité Mixte de Recherche 1163, Paris, France Correspondence to Sylvain Latour: sylvain.latour@inserm.fr * R. Rodriguez, B. Fournier, D. Jorge Cordeiro, and S. Winter contributed equally to this paper. K. Izawa's current address is Dept. of Pediatrics, Graduate School of Medicine, Kyoto University, Japan. 02 12 2019 19 9 2019 216 12 28002818 15 4 2019 23 7 2019 29 8 2019 © 2019 Rodriguez et al. 2019 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). Identification of biallelic loss-of-function mutations in TNFRSF9 and PIK3CD in a kindred with chronic active Epstein-Barr virus infection of T cells (CAEBV) suggests that CAEBV is the consequence of factors providing growth advantage to EBV-infected T cells combined with defective cell immunity toward EBV-infected cells. Infection of T cells by Epstein-Barr virus (EBV) causes chronic active EBV infection (CAEBV) characterized by T cell lymphoproliferative disorders (T-LPD) of unclear etiology. Here, we identified two homozygous biallelic loss-of-function mutations in PIK3CD and TNFRSF9 in a patient who developed a fatal CAEBV. The mutation in TNFRSF9 gene coding CD137/4-1BB, a costimulatory molecule expressed by antigen-specific activated T cells, resulted in a complete loss of CD137 expression and impaired T cell expansion toward CD137 ligand–expressing cells. Isolated as observed in one sibling, CD137 deficiency resulted in persistent EBV-infected T cells but without clinical manifestations. The mutation in PIK3CD gene that encodes the catalytic subunit p110δ of the PI3K significantly reduced its kinase activity. Deficient T cells for PIK3CD exhibited reduced AKT signaling, while calcium flux, RAS-MAPK activation, and proliferation were increased, suggestive of an imbalance between the PLCγ1 and PI3K pathways. These skewed signals in T cells may sustain accumulation of EBV-infected T cells, a process controlled by the CD137–CD137L pathway, highlighting its critical role in immunity to EBV. Graphical Abstract Ministère de la Recherche Ligue Contre le Cancer https://doi.org/10.13039/501100004099 Agence Nationale de la Recherche https://doi.org/10.13039/501100001665 Fondation Bettencourt Schueller https://doi.org/10.13039/501100007492 Fondation pour la Recherche Médicale https://doi.org/10.13039/501100002915 FDM20170638301 Ligue Contre le Cancer https://doi.org/10.13039/501100004099 Institut National de la Santé et de la Recherche Médicale https://doi.org/10.13039/501100001677 Rare Diseases Fondation Agence Nationale de la Recherche https://doi.org/10.13039/501100001665 ANR-14-CE14-0028-01 ANR-18-CE15-0025-01 ANR-10-IAHU-01 European Research CouncilERC-2009-AdG_20090506 n°FP7-249816 ==== Body pmcIntroduction The Epstein-Barr virus (EBV) is an oncogenic virus that infects most of humans with a marked tropism for epithelial cells and B lymphocytes (Taylor et al., 2015). The primary infection is self-limited, while latent EBV-infected B cells persist lifelong. In immune-compromised hosts, primary infection and persistence of proliferative EBV-infected B cells result in severe and often fatal lymphoproliferative diseases, including hemophagocytic lymphohistiocytosis (HLH) and nonmalignant and malignant B cell proliferations (Taylor et al., 2015; Tangye et al., 2017; Latour and Winter, 2018). Several inherited forms of susceptibility to develop EBV-driven B cell lymphoproliferative disorders have been identified, including gene defects in SH2D1A, CORO1A, CTPS1, MAGT1, ITK, CD27, and CD70 (Cohen, 2015; Tangye et al., 2017; Latour and Winter, 2018). Heterozygous gain-of-function (GOF) mutations in PIK3CD have also been associated with impaired immune control of EBV-infected B cells (Coulter et al., 2017; Edwards et al., 2019). Patients with these primary immunodeficiencies are characterized by their inability to mount efficient virus-specific T cell responses to eliminate EBV-infected B cells. Studies of these primary immunodeficiencies allowed the identification of several key factors required for T cell expansion and cell cytotoxicity and highlighted the role of T–B interactions in the control of EBV-infected B cells. EBV can also occasionally infect T lymphocytes and/or natural killer (NK) cells, leading to EBV-driven T/NK cell lymphoproliferative diseases, also termed as chronic active EBV infection (CAEBV; Fujiwara et al., 2014; Park and Ko, 2014). T/NK cell lymphoproliferative diseases are rare diseases of childhood and young adults often observed in populations from Asia and Central South America. CAEBV is characterized by persistence of EBV-infected T or/and NK cells that may progress to life-threatening lymphoproliferative disorders including T and NK lymphomas frequently associated with HLH symptoms (Kimura et al., 2001, 2012). Patients typically exhibit a high EBV load that can persist over years. They also often develop characteristic cutaneous manifestations, namely hypersensitivity to mosquito bites and hydroa vacciniforme–like lymphoma (Okano et al., 2005; Quintanilla-Martinez et al., 2013). The pathogenesis of T/NK lymphoproliferative disorders is poorly understood. It is associated in some cases with somatic mutations in DDX3X (Okuno et al., 2019). However, it is strongly suspected that these patients may suffer from an immune defect that results in inefficient CD8+ T cell responses toward EBV-infected T cells (Fujiwara et al., 2014; Taylor et al., 2015). Of note, few cases of B cell type of CAEBV associated with B cell lymphoproliferations, chronic viremia, and HLH have been also reported (Kimura and Cohen, 2017). It is not clear whether these cases differ from “classical” EBV-associated B cell disorders. Herein, we report the first identification of germline mutations causing CAEBV. We identified two homozygous loss-of-function (LOF) mutations in PIK3CD and TNFRSF9 in a patient presenting lethal CAEBV. Studies of the functional consequences of these mutations indicate that occurrence of CAEBV may be viewed as a consequence of factors providing a growth advantage to EBV-infected T cells combined with defective cell immunity toward EBV-infected cells. Results Identification of a family with a susceptibility to infection of T cells by EBV We investigated the case of a male child born to consanguineous parents originating from Pakistan diagnosed with an EBV-associated T cell lymphoproliferative disease (Fig. 1 A). He had recurrent upper respiratory tract and skin infections that started at 3 mo of age necessitating recurrent hospitalizations. Those included episodes of panaritium (at the age of 6, 7, and 9.5 yr) and boils requiring antibiotic intravenous infusion. Candida albicans, Staphylococcus aureus, and HSV were found in skin specimens. At the age of 10 yr, he received a prophylactic antibiotic treatment, and since then, he had not developed infections (with the exception of EBV infection [see below]) requiring hospitalization and antibiotic intravenous infusion. At the age of 9 yr and 10 mo, he came to our attention for persistent fever, weight loss, hepatosplenomegaly, and recurrent lymphadenopathies associated with persistent high blood load of EBV (6.5 × 106 copies/ml; Fig. 1 B). Episodes resolved spontaneously without treatment in a few weeks. High EBV loads were also detected in a liver biopsy (data not shown). Immunohistochemistry of the liver biopsy showed mild portal infiltration with CD3+ CD8+ mononuclear cells (Fig. 1 D). In situ hybridization with Epstein-Barr encoding region (EBER) probe was mainly positive in CD3+ cells, while most of CD20+ B cells were negative indicative of liver infiltration by EBV-infected T cells. Hydroa vacciniforme–like lesions also contained accumulation of EBV-containing CD8+ T cells that hybridized with EBER (Fig. S1). The presence of EBV in T cells was also examined by PCR on purified CD3+ cells from blood showing a high EBV load (5.3 × 106 copies/µg DNA) in CD3+ cells similar to the level detected in whole blood at that time (5.5 × 106 copies/µg DNA), whereas purified B cells were negative. Blood EBV load remained stably high over the years (between 5.5 × 106 and 6.5 × 106 copies/ml) despite treatment with the anti-CD20 monoclonal antibody (Fig. 1 B). Between the ages of 10 and 14 yr, symptoms recurred one or twice a year. At the age of 14 yr, he developed a fatal acute episode of HLH. Figure 1. Identification of a homozygous mutation in TNFRSF9 in two siblings with chronic EBV viremia and EBV-infected T cells. (A) Family pedigree with DNA electropherograms and the TNFRSF9 c.170DelG (−) genotype of each individual. The proband is indicated by an arrow. The healthy sister with chronic EBV corresponds to the gray circle. (B and C) EBV blood loads in the patient (B) and his sister (C) with EBV copies at different time points (black circles; y, years; m, months). Arrows indicate anti-CD20 treatments. (D) Patient liver biopsy. Staining with anti-CD3 (CD3) or anti-CD20 (CD20) showing portal and sinusoidal inflammatory infiltrate composed of CD3+ T lymphocytes and CD20+ B cells. Double staining with anti-CD3 antibody and EBER probe (CD3+EBER) detects EBV in the nucleus of most of CD3+ T cells. Magnification, ×200. Scale bar, 60 µm. (E) Schematic representation of TNFRSF9 intron–exon organization (coding regions in dark gray) and protein structure (extracellular [EC], transmembrane [TM], and intracytoplasmic [IC] domains) above. The mutation is indicated in red. Immunological investigations showed no major abnormalities. The patient had normal T cell counts with fluctuating CD8+ lymphopenia associated with decreased naive CD8+ T cells and an increased proportion of memory CD8+ T cells (Table 1). Prior to anti-CD20 therapy, B cell counts were normal, but there was a nearly complete lack of circulating memory and switched B cells. Ig levels were normal to high for IgG and low to normal for IgM and IgA. Low IgM levels persisted after recovery of normal B cell numbers following anti-CD20 treatment. NK, invariant NKT (iNKT), and mucosal-associated invariant T cell counts were reduced. T cell proliferation in response to CD3, PHA, and antigen stimulations was elevated. Based on the severity of the clinical phenotype, a gene defect leading to T and B cell immunodeficiency was suspected to be the cause of his condition. Table 1. Immunological features of PBMCs from the patient and his sister Patient Sister Age-matched normal values 10/12/14 yr 6/10 yr Leukocytes (cells mm−3) (4,400–15,500) 6,800/ND/4,900 6,600/6,000 Neutrophils (cells mm−3) (1,800–8,000) 1,400/ND/2,200 3,800/ND Monocytes (cells mm−3) (200–1,000) 400/ND/200 400/ND Lymphocytes (cells mm−3) (1,900–3,700) 1,900/2,000/2,200 2,100/1,543 T cells CD3+ (cells mm−3) (1,200–2,600) 1,501/1,440/1,474 1,596/1,096 CD4+ (cells mm−3) (650–1,500) 1,102/1,140/1,056 987/663 CD8+ (cells mm−3) (370–1,100) 342/260/396 462/370 CD4/CD8 ratio (0.9–2.6) 3.2/4.4/2.7 2.1/2.9 TCRγ/δ (%) (0.2–14) 4/ND/ND 1.4/ND CD31+CD45RA+/CD4+ (recent naive thymic emigrant; %) (43–55) 23/35/34 50 CD45RO+/CD4+ (memory; %) (13/30) 69/53/52 32/40 CCR7+CD45RA+/CD8+ (naive; %) (52/68) 23/29/25 62/58 CCR7+CD45RA−/CD8+ (central memory; %) (2–4) 6/5/4 4/4.5 CCR7−CD27−CD45RA−/CD8+ (effector memory; %) (11/20) 67/48/67 17/21 CCR7+CD27−CD45RA+/CD8+ (exhausted effector memory; %) (1–18) 6/18/8 0.7 CD127lowCD25+/CD4+ (regulatory; %) (4–20) 5.2/ND/ND 6.6/ND Va7.2+CD161+/CD3+ (MAIT; %) (1–8) 0.2/0.7/0.1 2.9/ND Va24+Vb11+CD161+/CD3+ (iNKT; %) (>0.02) 0.06/0.09/0.05 0.03/ND T cell proliferation (cpm 10−3) PHA (6.25 mg ml−1) (>50) 150/ND/ND 75/ND OKT3 (10 ng ml−1) (>30) 65.5/ND/ND ND Candida (>10) 23/ND/ND ND Tetanus toxoid (>10) 12/ND/ND ND NK cells CD16+CD56+ (cells mm−3) (100–480) 57/40/2 147/123 CD16+CD56+ (%) (4–17) 3/2/1 7/8 B cells CD19+ (cells mm−3) (270–860) 342/520/704 357/332 CD19+ (%) (13–27) 18/26/32 17/23 CD21+CD27+/CD19+ (memory; %) (>10) 2/1/1 3/1 IgD−IgM−/CD19+ CD21+CD27+ (switched; %) (21–49) ND/0.3/0.2 ND Ig levels (g liter−1; 16 yr old) IgG (6.6–12.8) 13.9/14.21/8.92 ND/12.5 IgM (0.5–2.1) 0.99/0.74/0.28 ND/0.27 IgA (0.7–3.4) 0.69/0.56/0.45 ND/0.81 Different immunological parameters of the patient were tested from blood and PBMCs: numbers of blood cell populations, different T cell subsets, B cell subsets, and NK cells from PBMCs (tested by flow cytometry), T cell proliferation in response to different stimuli (evaluated by incorporation of [3H]thymidine), and serum (Ig subclasses). Values in bold correspond to values below the age-matched normal values. MAIT, mucosal-associated invariant T cell. The other healthy siblings were also evaluated repeatedly for the presence of EBV during the study. All had EBV-positive serology with IgG anti-VCA and anti-EBNA. No detectable EBV load was found in the three brothers. In contrast, the asymptomatic 8-yr-old sister exhibited a persistent and elevated blood EBV load (>106 copies/ml) once she encountered EBV at the age of 6 yr (Fig. 1 C). This high EBV load was only slightly reduced by treatment with anti-CD20, indicating that persistent EBV-infected cells were not B lymphocytes. This was confirmed by PCR from purified CD3+ cells, which detected a high content of EBV (5.9 × 106 copies/µg DNA), as well as by EBER staining of peripheral blood mononuclear cells (PBMCs; not shown). Even though she maintained a persistent high viremia, she remained asymptomatic since. She had normal immunological parameters, except that the proportion of memory B cells that was low before the anti-CD20 treatment remained low 2 yr after treatment (Table 1). These results suggest that the sister is affected by the same immunodeficiency, with the possibility of an up-to-now incomplete clinical penetrance. Identification of a homozygous LOF mutation in TNFRSF9 in both siblings with EBV-infected T cells To identify a gene defect responsible for the disease in the patient and his sister, whole-exome sequencing (WES) was performed in all family members. Because of the consanguinity, an autosomal recessive inheritance pattern of the gene defect was applied to filter WES data with a focus on homozygous genetic variations. One significant homozygous variation in TNFRSF9 was identified in exon 4 (NM_001561.5), corresponding to a frameshift point deletion that results in a premature stop codon (chromosome 1, g.7998819delC, c.170delG, p.G57fsX91; Fig. 1 E). It predicted to remove a large part of the extracellular, transmembrane, and intracytoplasmic domains of the protein. Sanger sequencing confirmed the WES results; only the patient and his sister were homozygous carriers of the mutation, while the parents were heterozygous (Fig. 1 A). Other siblings had heterozygous or WT genotypes. This mutation was not present in public WES and whole-genome sequencing databases and our own databases at IMAGINE Institute. TNFRSF9 encodes CD137 (also known as 4-1BB), which belongs to the TNF receptor (TNFR) family (Wortzman et al., 2013). CD137 is a costimulatory molecule that is only expressed by activated T cells. When engaged by its ligand (CD137L), CD137 enhances antigen-specific T cell responses, including proliferation, IL-2 secretion, survival, and cytolytic activity (DeBenedette et al., 1997, 1999; Shuford et al., 1997; Wen et al., 2002; Bukczynski et al., 2004). As reported, CD137 expression was only detectable on control T cells that were activated with anti-CD3 antibody, anti-CD3/CD28–coated beads, PHA, or PMA + ionomycin, but not on resting T cells (Fig. 2, A–C). In contrast, T cells of the patient and his sister failed to express CD137 in these conditions. The expression of the full transcript of CD137 was also examined in PHA-activated T cells by semiquantitative RT-PCR and found to be markedly decreased in cells of the patient and his sister, while cells were activated similarly as control cells based on CD25 expression up-regulation (Fig. 2 C). This suggests that the CD137 transcript is subject to nonsense mutation–mediated RNA decay. Furthermore, no alternative spliced CD137 transcripts were detected in cells from the patient and his sister. These findings indicate that the premature stop codon p.G57fsX91 led to the absence of CD137 expression. Figure 2. The G57fsX91 mutation prevents CD137 expression in activated T cells. (A) FACS histograms of CD137 of T cell blasts from the patient (Pat.), his sister, and a control (Ctrl.) unstimulated (No stim.) or stimulated with immobilized anti-CD3 antibody or PMA+ionomycin (PMA+Iono) for 7 d. Isotype control in gray. (B) FACS histograms of CD137 and CD27 of PBMCs unstimulated (gray) or stimulated with anti-CD3/CD28–coated beads (red) for 3 d. (C) Expression of CD137 full-length transcript (768 pb) and GAPDH (258 pb) by qRT-PCR in 72 h PHA-stimulated blasts. GAPDH as normalization controls for cDNA samples diluted as indicated, with the DNA ladder (kilobases) on the right. Bottom: FACS histograms of CD137 (isotype control in gray) and CD25 (non- and PHA-stimulated cells in red and blue, respectively). (D) FACS histograms of CD137L expression on P815 expressing CD137L (P815-CD137L) or not (P815-empty). Isotype control is shown in blue. (E) FACS dot-plots of proliferation assays of PBMCs labeled with the CellTrace Violet dye and cocultured or not (/) for 5 d with irradiated P815-CD137L or P815-empty cells preincubated with anti-CD3. Staining with anti-CD25 antibody as activation marker and gating on CD4+ or CD8+ T cells. Proliferating and nonproliferating T cells in the upper left and right lower gates, respectively. (F) FACS histograms of CD137 in proliferating (blue) or nonproliferating (red) CD4+ or CD8+ T cells gated from E. Numbers in A and E correspond the percentage of cells in gates. (D) One representative experiment of three. (E and F) experiments done in duplicate with the same results. One replicate is shown. The costimulatory function of CD137 was next examined in T cells of the asymptomatic mutated sister only, since no material was available to test her affected brother. T cell proliferation was assessed following coculture with mouse mastocytoma P815 cells stably expressing or not CD137L (P815-CD137L) in the presence of soluble anti-CD3 antibody at a low concentration (Wen et al., 2002; Fig. 2 D). Coculture with P815-CD137L cells, but not with CD137L-negative P815 (P815-empty), triggered proliferation and up-regulation of the activation marker CD25 of both control CD4+ and CD8+ T cells from a healthy donor (Fig. 2 E, left panels). Importantly, dividing control CD25+ T cells expressed CD137, whereas nonproliferating cells were negative for CD137 (Fig. 2 F, upper panels). In contrast, T cells of the sister failed to proliferate when cocultured with CD137L-expressing P815 (Fig. 2 E, right panels), while CD137 was not detected (Fig. 2 F, lower panels). However, CD3/CD28-dependent proliferation of T cells of the sister (as well as the patient) was preserved and thus unaffected by CD137 deficiency (Fig. 6 E and Fig. 7, A and B). To formally prove that the premature stop codon p.G57fsX91 in CD137 was responsible for the defect of CD137L-triggered proliferation, CD137 expression was restored in T cells of the sister by transfecting PBMCs with a CD137-containing lentiviral vector or an empty vector (Figs. 3 and S2). CD137 expression was achieved in >40% of CD8+ T cells of the sister (Fig. 3 A). Restored CD137 expression by both CD8+ and CD4+ T cells was associated with their ability to proliferate and to express CD25 when cocultured with P815-CD137L, but not with P815-empty (Figs. 3 B and C and S2 and data not shown). Overexpression of CD137 by control CD8+ and CD4+ T cells only resulted in a slight increase of proliferating CD137+ T cells. These results demonstrate that the mutation in TNFRSF9 identified in the patient and his sister behaves as an amorphic mutation leading to impaired T cell expansion toward cells expressing CD137L. Thus, the loss of costimulatory function of CD137 in activated T cells of the patient and his sister could impair clearance of EBV-infected cells resulting from defective expansion of EBV-specific T cells. Figure 3. Correction of CD137 expression in CD137-deficient T cells restores their capacity to proliferate in response to CD137L-expressing cells. (A) FACS histograms of CD137 expression of T cells from the sister infected or not (NI) with a lentiviral vector containing or not (pLVX empty) a cDNA for CD137 (pLVX-CD137). (B and C) FACS dot-plots of proliferation assays from T cells shown in A (same as cells shown in Fig. 2 E). Cells were stained with an anti-CD25 antibody (B) as an activation marker or with an anti-CD137 antibody (C) and CellTrace Violet dilution analyzed by flow cytometry after gating on CD8+ T cells. Numbers correspond the percentage of proliferating cells in gates. Experiments in A–C were done in duplicate with the same results. One replicate is shown. Identification of a homozygous missense LOF mutation in PIK3CD in the patient, but not in his sister Because the patient’s sister has not developed any clinical manifestation despite the persistence of EBV viremia and circulating EBV-infected T cells, we postulated that the patient could be a carrier of an additional genetic factor that might account for the overt clinical phenotype. First, we looked for an allelic reversion event of the TNFRSF9 mutation in the patient’s sister that could have explained why she has no clinical signs. We failed to detect any reversion event in PBMCs and PHA-activated T cells from the sister (as well as from the patient) using Sanger and next-generation sequencing (Fig. S3). Therefore, WES results were reanalyzed considering the sister as not affected. One significant missense biallelic mutation in PIK3CD was identified in the patient at position 9783218 (rs573872848) on chromosome 1, (NM_005026.3 c.2462G>A) in exon 20 that led to a R821H substitution (Fig. 4 A). In public WES and whole-genome sequencing databases and own databases, the mutation was only found as heterozygous with a frequency of 3.8 × 10−5. The mutation was further verified by Sanger sequencing confirming the autosomal recessive inheritance in the kindred (Fig. 4 B). Both parents were heterozygous carriers of the mutation as well as the sister and one brother, while the other siblings were not mutated. The patient was thus the only homozygous carrier of the mutation. PIK3CD encodes p110δ, the catalytic subunit of the phosphoinositide 3-kinase (PI3K). PIK3CD deficiency was recently reported to cause immunodeficiency in four patients, but the functional consequences of the PIK3CD defect were not studied (Sogkas et al., 2018; Cohen et al., 2019; Swan et al., 2019). p110δ is highly expressed in lymphocytes and is the main active isoform of the PI3K catalytic subunit in human T cells (Okkenhaug et al., 2007; Okkenhaug and Fruman, 2010). p110δ associates with the regulatory p85α subunit which is also shared with other isoforms of PI3K catalytic subunit, including the p110γ and p110α (Engelman et al., 2006). Figure 4. Identification in the patient of a homozygous mutation in PIK3CD affecting the catalytic site. (A) Schematic representation of PIK3CD coding the p110δ. Intron–exon organization (coding regions in dark gray) and corresponding protein structure. ABD, adaptor-binding domain; RBD, Ras-binding domain. The mutation is indicated in red, and the previously GOF mutation E1021K in gray. (B) Family pedigree with DNA electropherograms and the PIK3CD c.2462 G>A genotype of each individual (also see Fig. 1 A). (C) Ribbon representation of the p110δ–p85α complex (PDB: 5T8F), highlighting possible bonds between the p110δ kinase domain and p85α, in addition to contacts made by C2 and ABD. (D) Comparison of the experimental three-dimensional structures of PIK3CD (PI3KD; PDB: 5T8F; Castanedo et al., 2017) and PIK3CA (PI3KA; PDB: 2RD0; Huang et al., 2007) in complex with p85α. The catalytic residues K779 (PI3KD) and K802 (PI3KA) are shown and specific contacts with p85α (in purple) are highlighted. (E) Alignment of isoforms of p110 within the kinase domain. The observed secondary structures of p110δ/PIK3CD (PDB: 5T8F) and p110α/PIK3CA (PDB: 2RD0) are reported above and below the alignment, respectively (UniProtKB accession numbers PIK3CD_HUMAN: O00329, PK3CB_HUMAN: P42338, PK3CG_HUMAN: P48736, and PK3CA_HUMAN: P42336). Reduced kinase activity of the mutant PIK3CD/p110δ R821H The R821H mutation is located in a highly conserved region of the p110δ catalytic domain. To characterize the impact of this mutation, analysis of the three-dimensional structures of p110δ–p85α complexes was undertaken. p110δ associates with its p85α subunit through interactions outside the kinase domain and involves the adaptor-binding domain (ABD) and the C2 domains (Huang et al., 2007; Vadas et al., 2011; Burke, 2018; Fig. 4 C). The R821 is located in the β6-β7 loop of the kinase domain in the vicinity of the active site that includes the gatekeeper I825 and catalytic lysine K779 (Fig. 4 D, upper panel). R821 and D820 are the only amino acids in the kinase domain able to establish bonds, after side-chain rotamer change, with p85α Q475 (H-bond) and E468 (salt-bridges). Thus, these interactions provide a specific contact point between the p110δ kinase domain and p85α, in addition to the well-recognized contacts made between p85α and the ABD and the C2 domain, which are shared by all isoforms. The former specific contact is lost in the p110δ R821H mutant protein. Such an interaction does not exist between p110α and p85α (Huang et al., 2007), where the equivalent amino acid to R821 in p110δ is a cysteine (C844), which makes an endogenous H-bond with the G802 residue, thereby contributing to the structure of the catalytic site (Fig. 4 D, lower panel; and Fig. 4 E). Similarly, the R821 likely contributes to the structure of the catalytic site through a salt bridge formed with the E468 of p85α. When this specific conformation is disrupted, kinase activity is likely impaired. We therefore tested the respective consequences of the R821H substitution on both p110δ enzymatic activity and binding to p85α (Fig. 5). First, whole-cell extracts lysates from control and patient cells showed comparable levels of p110δ and p85α, indicating that the mutation had no impact on p110δ expression (Fig. 5 A). Similarly, immunoprecipitations (IPs) of p85α or p110δ recovered from both donor and patient cell extracts showed comparable amounts of p110δ and p85α, showing that the R821H mutation in p110δ did not disrupt association with p85α. Consequences of the R821H mutation on p110δ kinase activity was next examined in HEK293T cells that transiently expressed WT p110δ, the R821H mutant, or an overactive E1021K p110δ mutant and were cotransfected or not with p85α. Total cell extracts showed comparable expression of both WT and mutated forms of p110δ (Fig. 5 B). In IP experiments, recovery of WT and mutated p110δ was similar and comparable amounts of p85α were found to be associated with p110δ. Immunoprecipitates were then tested for PI3K activity. No production of PIP3 was detected in cells expressing p110δ alone in the absence of p85α, highlighting the need of the p85α subunit for actual kinase activity, as previously reported (Yu et al., 1998). Cotransfection of p85α and WT p110δ led to a significant production of PIP3 that was significantly diminished when p85α was coexpressed with the R821H p110δ (Fig. 5 C). As expected, cotransfection of p85α with the overactive E1021K p110δ further increased PIP3 production. Similar diminution in PIP3 production was observed from T cell blast protein extracts of the patient when compared with protein extracts of control cells (Fig. 5, D and E). These data demonstrate that the p110δ R821H mutation behaves as an LOF hypomorphic mutation resulting in a profound defect in PI3K activity. Figure 5. The R821H mutation in PIK3CD impairs p110δ kinase activity, but not binding to p85α. (A and B) Analysis of the p85α–p110δ association. (A) T cell lysates from the patient (Pat.) or a control (Ctrl.) after IP with control IgG (Rabbit IgG) or with antibody to p85α or p110δ followed by WB with anti-p85α or anti-p110δ. Total lysates (input) are shown on the left; molecular weight markers are shown on the right. (B) Lysates of HEK293T cells ectopically expressing or not p85α and/or WT, R821H, or E1021K p110δ, followed or not (input controls) by IP with anti-p110δ antibody (IP: p110δ) and WB with anti-p110δ, -p85α, and -KU-70 antibody as loading control. HEK293T expressed small amount of p85α that is not detected in the input but detectable in the IP of the p110δ, p110δ R821H, and p110δ E1021K (without ectopic p85α expression). (C) In vitro PI3K assays after IP, as in B, followed by quantification of produced PIP3 (left). Statistical analyses are shown on the right. (D) Same as A, except that total lysates are shown (input) and WB with anti–β-actin was used as a loading control. (E) In vitro PI3K assay of IPs obtained as in D followed by quantification of produced PIP3. Data are representative of two independent experiments (A, B, and C, left, and D) or are mean and SD of two independent experiments with three (C, right) or two (E) kinase assay replicates per group. ns, not significant; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001 (one-way ANOVA with post hoc Bonferroni t test). Reduced PI3K-dependent AKT/S6K signaling in patient T cells Activation of the AKT and p70-S6K kinase pathway is dependent on PI3K activity in T cells (Okkenhaug et al., 2007). Increased phosphorylation of AKT and p70-S6K in T cells is considered a hallmark of patients carrying PIK3CD GOF mutations associated with activated PI3K delta syndrome type 1 (Angulo et al., 2013; Lucas et al., 2014). Phosphorylation of p70-S6K and AKT on serine 473 and threonine 308 (not shown), respectively, were found to be significantly reduced in T cell blasts of the patient in response to TCR/CD3 stimulation in comparison to control cells (Fig. 6, A–C). However, tyrosine phosphorylation of ZAP-70 (Fig. 6 A) and whole tyrosine phosphorylation signals (data not shown) that are independent of PI3K activity were equivalent between patient and control cells. Residual AKT phosphorylation in patient cells is likely explained by the hypomorphic nature of the mutation and/or by the activity of other PI3K catalytic subunit isoforms. Of note, the p110δ-specific pharmacological inhibitor IC87114 reduced TCR/CD3–induced AKT phosphorylation in control cells to a level comparable to that seen in cells of the patient (Fig. 6 C). These results show that decreased kinase activity of the mutant R821H p110δ leads to reduced AKT and p70-S6K activation in patient T cells. Figure 6. The R821H mutation in PIK3CD is associated with diminished activation of AKT and increased T cell proliferation and calcium flux. (A) Immunoblots of phosphorylated AKT (P-AKT), p70 S6K (P-p70 S6K), and ZAP70 (P-ZAP70) in T cell blasts from a control (Ctrl.) and the patient (Pat.) stimulated with anti-CD3 antibody for 0, 1, 2, 4, and 8 min. Total AKT and ACTIN were used as loading controls. Molecular weight markers are shown on the right. (B) Densitometry quantifications of immunoblots of P-AKT of T cells stimulated as in A for 0, 5, and 10 min. (C) FACS histograms of intracellular P-AKT in activated T cells not stimulated (no stim.) or stimulated with anti-CD3 in the presence or not of the p110δ pharmacological inhibitor IC87114 (IC). (D and E) FACS dot-plots and corresponding histograms of proliferation assays (after gating on CD3+ cells) assessed with the CFSE dye. T cell blasts (D) or PBMCs (E) stimulated or not with immobilized 1 µg ml−1 anti-CD3 (D, left), different doses of anti-CD3 (D, right) or anti-CD3/CD28 coated beads (E). Staining with anti-CD25 antibody as an activation marker. Numbers ahead each peak correspond to the number of divisions. Calculated proliferation indexes from FACS data are shown on the right. (F) Intracellular Ca2+ levels by real time flow cytometry in T cell blasts or PBMCs of Ctrl. (gray) or Pat. (black line) after stimulation with anti-CD3 and a cross-linker. (G) Same as A, with immunoblots of phosphorylated ERK (P-ERK). (H) Normalized densitometry quantifications of P-ERK immunoblots. Stimulation of patient cells at 5 min was used as 100%. *, P < 0.1; ***, P < 0.001; ****, P < 0.0001 with unpaired (B) or paired (D, E, and H) Student’s t test from data of five (B), two (0.1 and 10 µg ml−1; D), or four (0 and 1 µg ml−1; D) or four (E and H) independent experiments with mean and SD. In D, at 10 µg ml−1, P = 0.052 as indicated. Data are representative of three independent experiments for A, C, and F. A.U., arbitrary units. Increased proliferation of patient T cells associated with an elevated Ca2+ influx PI3K-dependent signaling is known to regulate many lymphocyte functions, including cell survival and proliferation. Hence, we assessed T cell proliferation using a vital dye tracer. It was striking to observe that patient T cells directly tested from PBMCs or T cell blasts in culture exhibited a significantly increased proliferation in comparison to control cells in response to TCR/CD3 stimulation (Fig. 6, D and E). However, T cell blasts from the patient exhibited normal degranulation and activation-induced cell death upon anti-CD3 stimulation as well as normal apoptosis upon anti-FAS stimulation, indicating that not all T cell functions were increased (Fig. S4, A–C). We also examined intracellular Ca2+ mobilization and the MAPK/ERK pathway in response to TCR activation, which are known to be downstream and dependent of phosphatidylinositol-4,5-bisphosphate (PIP2) metabolism in T cells. Ca2+ flux and phospho-ERK to anti-CD3 stimulation were found to be increased in patient T cells when compared with control cells (Fig. 6, F–H). Consistent with increased Ca2+ flux, phospho-PLCγ1 was found to be enhanced in T cells of the patient upon anti-CD3 stimulation (Fig. S4 D). T cells of the sister, who is heterozygous for the PIK3CD mutation, were also evaluated, showing an intermediate decrease of TCR-dependent AKT phosphorylation and Ca2+ flux upon CD3 stimulation when compared with the patient (Fig. 7 D; and Fig. S4, E and F), while T cell proliferation upon stimulation with CD3 alone or CD3/CD28–coated beads was not significantly different from that of controls cells (Figs. 7 A and S4 G). Taken together, these results indicate that the LOF of R821H mutation in PIK3CD independently of the mutation in TNFRSF9 is associated with increased TCR-dependent proliferation and calcium mobilization in T cells. Figure 7. Increased proliferation of activated T cells from the patient is not observed in his sister. (A and B) FACS histograms from plots of proliferation assays of T cell blasts (A) or PBMCs (B) from a control (Ctrl.), the patient (Pat.), or his sister assessed with the CellTrace Violet dye. Numbers ahead each peak correspond to the number of divisions. Cells were (A) stimulated or not (0) with immobilized 1 or 10 µg ml−1 anti-CD3 antibody or anti-CD3/CD28–coated beads or (B) stimulated by coculture or not (no stim.) with LCLs expressing (LCL-CD70+) or not CD70 (LCL-CD70−) after preincubation with anti-CD3 antibody. (C) FACS dot-plots of CD27 and CD3 of PBMCs. Numbers represent the percentage of CD27+ CD3+ cells in the gates. (D) Intracellular Ca2+ levels analyzed by real-time flow cytometry in T cell blasts from the control (black line), the patient (red line), or his sister (blue line) after stimulation with anti-CD3 and a cross-linker. Data are from a single experiment (in which the cells of the patient and his sister were compared) but representative of several independent experiments in which the patient and his sister were tested separately and compared with controls (see Fig. 6, D and E; and Fig. S4 G). We previously reported that CD27, another costimulatory molecule of the TNFR superfamily expressed by T cells, is required for expansion of EBV-specific T cells toward EBV-infected B lymphoblastoid cell lines (LCLs) expressing CD70, the ligand of CD27 (Izawa et al., 2017). CD27-dependent T cell proliferation toward LCL-expressing CD70 (LCL-CD70+) or CD70-negative LCL (LCL-CD70−) was analyzed with T cells from the patient and his sister (Fig. 7 B). LCL-CD70+, but not LCL-CD70−, triggered proliferation of T cells from the sister that was normal or slightly increased compared with control cells. Proliferation of T cells from the patient was increased in the same conditions. CD27 was found normally expressed on T cells from the sister and the patient (Fig. 7 C). The increased CD27-dependent proliferation in the patient might suggest a compensatory mechanism for the absence of the CD137 pathway and/or related to the increased proliferation rate of activated T cells associated with the R821H mutation in PIK3CD. Imbalanced phospholipase C gamma 1 (PLCγ1)–dependent signals in PIK3CD-deficient T cells Because of the lack of available material from the patient and in order to further demonstrate the effect of the R821H p110δ mutation on T cell responses, two Jurkat T cell lines deficient for p110δ (PIK3CD−/−) were generated by the CRISPR-Cas9 technology using RNA guides targeting respectively exon 4 or exon 5 of PIK3CD (referred as CRISPR-Ex4 and CRISPR-Ex5). They were then reconstituted with the R821H mutant, the GOF E1021K mutant, or WT p110δ/PIK3CD (Figs. 8 and S5). Absence of p110δ expression in CRISPR-Ex4 and CRISPR-Ex5 cell lines was confirmed by Western blot (WB) analysis (Fig. 8 A). CRISPR-Ex4 and CRISPR-Ex5 cell lines transduced with WT, R821H p110δ, or E1021K p110δ coding lentiviruses showed comparable levels of p110δ (Figs. 8 B and S5). Constitutive AKT phosphorylation was examined and found to be increased in cells reconstituted with the activating E1021K mutant and reduced in cells expressing the inactive R821H mutant or in cells transduced with the empty vector (Figs. 8 C and S5). Of note, in Jurkat cells, AKT phosphorylation/activation and proliferation are constitutive and uncoupled to the TCR activation, in contrast to Ca2+ mobilization, which still requires TCR-CD3 stimulation (Shan et al., 2000; Astoul et al., 2001; Abraham and Weiss, 2004). When compared with E1021K p110δ–expressing cells, proliferation of p110δ-deficient cells and R821H p110δ–expressing cells was significantly increased (Fig. 8, E and D; and Fig. S5). Anti-CD3-induced calcium flux was also found to be increased (Figs. 8 F and S5). The abnormal T lymphocyte phenotypes of the patient were thus recapitulated in these Jurkat cell lines. Taken together, these results demonstrate that impaired PI3K activity leads to an increased calcium flux and cell proliferation, while increased PI3K activity results in the opposite phenotypes. Figure 8. Imbalanced PLCγ1–dependent signals in PIK3CD-deficient Jurkat T cells. (A and B) Immunoblots of p110δ expression (A) of PIK3CD-deficient (PIK3CD−/−) Jurkat cell lines obtained from CRISPR-Cas9 targeting of exon 4 (Ex. 4) or exon 5 (Ex. 5) of PIK3CD or nontargeted (WT) Jurkat cell lines (B) of WT or exon 4–targeted PIK3CD-deficient Jurkat cells (CRISPR-Ex4-PIK3CD−/−) reconstituted with an empty vector or a vector coding for WT (p110δ WT), R821H, or E1021K p110δ. (C) FACS histograms of intracellular phosphorylated AKT at serine 473 (Ser473 P-AKT) in the different exon 4 PIK3CD−/− Jurkat cell lines shown in B. Isotype controls are in gray. (D) FACS histograms from plots of proliferation assays of the different CRISPR–exon 4–PIK3CD−/− Jurkat cell lines shown in B assessed with the CellTrace Violet dye. Numbers ahead each peak correspond to the number of divisions. The red arrow indicates the peak of the fifth division in each histogram. (E) Indexes of proliferation calculated from D. (F) Intracellular Ca2+ levels analyzed by real-time flow cytometry in the different CRISPR–exon 4–PIK3CD−/− Jurkat cell lines shown in B stimulated with anti-CD3 antibody. (G) Immunoblots of phosphorylated PLCγ1 (P- PLCγ1) and ERK1/2 (P-ERK1/2) in WT Jurkat (WT) or CRISP–exon 4–PIK3CD−/− Jurkat (PIK3CD−/−) stimulated with anti-CD3 for 0, 1, 2, 5, 10, and 20 min. Total amounts of PLCγ1, PIK3CD (p110δ), KU80, and ERK1/2 are shown. (H) Same as G in nonstimulated PIK3CD−/− Jurkat (exon 4) reconstituted cells shown in B. (G and H) One representative experiment of two is shown. Similar experiments done with cells PIK3CD−/− cells obtained by targeting exon 5 showing the same results (see Fig. S4). Molecular weight markers are shown on the right (A and B) and left (G and H). Data are representative of three (A and B) or two (C, D, and F) independent experiments with three (D) or one (C and F) biological replicate per group in each. Two-way ANOVA test in E with mean and SD of two independent experiments with three and two replicates. ns, not significant; ***, P < 0.001, n = 5. Calcium entry and ERK1/2 activation are initiated by PLCγ1, which shares with p110δ the same PIP2 substrate (Rhee, 2001; Engelman et al., 2006; Okkenhaug et al., 2007). TCR-dependent activation of PLCγ1 results in the cleavage of PIP2 into inositol-1,4,5-triphosphate (IP3) and diacylglycerol. IP3 and diacylglycerol are second messengers that activate Ca2+ mobilization and the MAPK pathway, respectively. Thus, our findings strongly suggest the existence of a balance between PLCγ1–dependent and PI3K-dependent signaling pathways that could compete for PIP2 availability. To test this hypothesis, we examined PLCγ1 and MAPK activation in response to TCR-CD3 stimulation in WT Jurkat cells compared with PIK3CD-deficient Jurkat cell lines (Figs. 8 G and S5). Phosphorylation of PLC-γ1 and ERK1/2 kinases, which are known to reflect PLCγ1 and MAPK activation, was tested in response to anti-CD3 stimulation. Strikingly, a constitutive phosphorylation of PLCγ1 and ERK1/2 was strongly detectable in PIK3CD-deficient cell lines, but not in WT cells. In PIK3CD-deficient cells, TCR-CD3 activation led to a further increased ERK1/2 phosphorylation that was more intense, fast, and blunted more rapidly compared with WT cells. Activation did not further increase PLCγ1 phosphorylation, which was already very high in unstimulated cells. In p110δ-deficient cells reconstituted with R821H, a marked constitutive phosphorylation of PLCγ1 and ERK1/2 was detectable, while in cells reconstituted with the GOF E1021K mutant, constitutive PLCγ1 and ERK1/2 phosphorylation was low (Figs. 8 H and S5). Therefore, these data show that in the absence of p110δ or impaired p110δ activity, PLCγ1 is more active, leading to increased calcium flux and activation of the downstream MAPK pathway. This may likely result in increased proliferation. Discussion We have identified herein an immunodeficient patient carrier of two homozygous LOF mutations in PIK3CD and TNFRSF9 with recurrent infections and an uncontrolled EBV-induced T cell proliferative disorder with features of CAEBV. His asymptomatic sister, who poorly controlled EBV replication, shared the same homozygous TNFRSF9 LOF mutation. These differential genotypes contributed to dissect the potential roles of each gene alteration. Recently, four patients with severe infections and inflammatory colitis associated with hypogammaglobulinemia and loss of B cells have been reported to be carriers of homozygous frameshift mutations in PIK3CD resulting in premature stop codons leading to absence of protein or removing the catalytic domain (Sogkas et al., 2018; Cohen et al., 2019; Swan et al., 2019). These patients had normal T cell subsets and counts, and T cell proliferation in response to various stimuli was reported to be normal or enhanced (like in our patient). No EBV infection was reported in these patients, but it was not known whether these patients had encountered EBV. However, the clinical phenotype of our patient partially differs from those patients. There are several explanations for this difference. First, the R821H mutation of our patient is hypomorphic and associated with a normal expression and residual PIK3CD enzymatic activity, in contrast to the mutations in PIK3CD-deficient patients, which are either null or lead to a truncated protein. Thus, this may explain that in our patient, the B cell phenotype is attenuated. Second, our patient did not develop colitis. CD137 deficiency may impair T cell responses avoiding the occurrence of colitis in our patient. Indeed, the inflammatory colitis in one patient was examined and found to be associated with a substantial increase in activated effector CD8+ T compartments in the lamina propria, suggesting an abnormal expansion of effector CD8+ T cells (Swan et al., 2019). Of note, CD137 costimulation is well characterized in humans to induce CD8 T cell expansion with effector functions (see below), thus supporting that in our patient, such an expansion of CD8+ T cells in the lamina propria would be impaired. Interestingly, two patients with dual PIK3CD and small kinetochore-associated protein deficiency have been reported (Sharfe et al., 2018). Notably, these patients did not develop inflammatory colitis and exhibited strong T cell defects, including proliferation, in contrast to the other patients with PIK3CD deficiency. These T cell defects are likely caused by the small kinetochore-associated protein deficiency that impaired kinetochore/spindle microtubule dynamics during mitosis and might also explain why these two patients did not develop inflammatory colitis. Heterozygous activating GOF mutations in PIK3CD have also been reported. They are responsible for activated PI3K delta syndrome type 1 syndrome, a complex immunodeficiency associated with variable clinical phenotypes (Angulo et al., 2013; Lucas et al., 2014; Coulter et al., 2017). To our knowledge, EBV-associated T/NK-cell lymphoproliferative disorders have not been reported in patients with GOF (and LOF) mutations in PIK3CD. Functional features associated with PIK3CD GOF mutations are increased activation-induced T cell death, low T cell proliferation, and increased phosphorylation of AKT and p70-S6K kinase (Angulo et al., 2013; Lucas et al., 2014; Carpier and Lucas, 2018). The T cell abnormalities detected herein in the patient are the opposite, consisting of increased T cell proliferation, decreased phosphorylation of AKT and p70-S6K kinase, normal senescent CD57+ T cells, and preserved activation-induced cell death. Therefore, these functional abnormalities (associated with the PIK3CDR821H mutation) correspond to dysfunctions that would have been expected to be caused by PIK3CDLOF mutations (in opposition to those associated with PIK3CD GOF mutations). The observed association of the PI3KCDR821H LOF mutation with increased proliferation, ERK1/2 activation, and calcium flux comes as a relative surprise given the known role of the PI3K–AKT pathway in cell survival. However, an identical phenotype was reproduced in p110δ-deficient Jurkat cell lines obtained by CRISPR-Cas9, demonstrating that these abnormalities are not the consequence of unrelated genetic variations in the patient and/or indirect consequence of EBV infection of T cells. These results suggest that there is a physiological balance between PLCγ1 and PI3K-dependent pathways in T cells. Indeed, PLCγ1 and PI3Kδ (p110δ) activities share the same substrate, PIP2, and hence are competing to access to it (Okkenhaug et al., 2007). We propose that the skewed signaling toward PLC-γ1 resulting from PIK3CD deficiency may provide a permissive cellular background in T cells to EBV infection or/and for the persistence and expansion of infected T cells because of the related proliferation advantage. The mechanisms involved in the entry and maintenance of EBV in T cells are not known (Fujiwara et al., 2014). However, a recent study identified somatic cancer driver mutations in EBV-positive infected T cells of CAEBV patients (Okuno et al., 2019). These mutations accumulated when the disease progressed and were associated with a shorter overall survival. The homozygous germline PIK3CDR821H found in the patient might be viewed as a driver mutation allowing EBV-infected T cells to persist and acquire additional somatic mutations, leading to uncontrolled proliferation. Thus, a growth selective advantage could be a very significant contributing factor to the onset of CAEBV originating from germline (the present report) or somatic mutations. The persistent EBV replication and the presence of circulating EBV-infected T cells in the healthy sister led us to the discovery of a homozygous deleterious mutation in TNFRSF9 shared by the patient and his sister. It is unlikely that the CD137 deficiency plays a role in the functional abnormalities detected in T cells of the patient, as these functional phenotypes were reproduced in p110δ-deficient Jurkat cells. Rather, it is likely that CD137 deficiency accounts for the impaired immune control of EBV-infected T cells. CD137 is a highly potent costimulatory molecule of T cell responses during viral infections (Wortzman et al., 2013). It has been shown that CD137 ligation sustains antigen-specific CD8+ T cell responses (Shuford et al., 1997; DeBenedette et al., 1999; Takahashi et al., 1999; Lee et al., 2002; Maus et al., 2002). Furthermore, targeting CD137 with agonistic antibodies also demonstrated potent effects on antitumoral and viral responses (Chester et al., 2018). Chimeric antigen receptor (CAR) T cell therapy using CARs containing the intracytoplasmic domain of CD137 was also shown to be the most efficient to sustained CAR T cell responses (Li et al., 2018). CD27 is a well-known costimulatory molecule belonging to the TNFR super family, like CD137 (Watts, 2005). In contrast to CD137, CD27 is highly expressed on both resting and activated T cells. Importantly, primary immunodeficiencies caused by mutations in CD27 or CD70, the ligand of CD27, have been reported in patients with a high susceptibility to develop EBV-driven B cell lymphoproliferative disorders (van Montfrans et al., 2012; Salzer et al., 2013; Alkhairy et al., 2015; Abolhassani et al., 2017; Izawa et al., 2017). We showed that CD70 expression on EBV-infected B cells is required for expansion of EBV-specific T cells via cosignals deliver by CD27 (Izawa et al., 2017). Similarly, CD137-CD137 ligand (CD137L) interactions with antigen presenting cells possibly including EBV-infected T cells or B cells, may play an important similar nonredundant role in the expansion of EBV-specific T cells notably required to target EBV-infected T cells. Supporting this assumption, CD137L is expressed by T cells and was shown to induce T cell trans-costimulation resulting in potent tumor rejection (Stephan et al., 2007). Furthermore, EBV-unrelated cutaneous T cell lymphomas have been shown to express significant amount of CD137L (Kamijo et al., 2018). The recent description of EBV-driven B cell lymphomas in two patients with CD137 deficiency is strongly in favor of a key role of CD137 in immunity to EBV (Alosaimi et al., 2019), although the clinical asymptomatic feature of the sister (in our study) suggests incomplete penetrance. In this study, expansion and function of antigen-specific CD137-deficient CD8+ T lymphocytes in coculture with B-LCLs were impaired. With our observations, this supports that CD137 delivers to EBV-specific T cells cosignals required for their expansion. In conclusion, we report the first genetically well-documented case of EBV-driven T/NK cell lymphoproliferative disorder associated with homozygous mutations in PIK3CD and TNFRSF9, highlighting the potential role of these pathways in the immunity to EBV when EBV infects T cells. Additional genetic and/or environmental factors may also participate to this susceptibility. These results provide the first evidence of the multifactorial genetic inheritance underlying EBV-associated T/NK cell lymphoproliferative disorders, that can be now considered as the result in part of a primary immune defect. This observation also constitutes one of the first examples of a recessive immunodeficiency resulting from two gene defects with mechanisms likely to independently contribute (and even synergize) to cause the disease. Materials and methods Ethics Written informed consent was obtained from all humans in this study in accordance with the Helsinki declaration and local legislation and ethical guidelines from the Comité de Protection des Personnes de l’Ile de France II, Hôpital Necker-Enfants Malade, Paris. Blood from healthy donors was obtained at Etablissement français du sang under approved protocols (convention 15/EFS/012). WES Genomic DNA was extracted from whole blood according to standard methods, and WES and data analysis were performed as previously described (Martin et al., 2014; Izawa et al., 2017; Winter et al., 2018). Genomic DNA regions flanking PIK3CD and TNFRSF9 mutations were amplified using the forward primer 5′-CTT​GAC​CAT​GCC​ATT​TGC-3′ and reverse primer 5′-CTT​GGA​CTT​CAG​CCA​GTT​G-3′ (for PIK3CD) or the forward primer 5′-CGT​GTA​CCA​CTA​TTG​TCT​GCC-3′ and the reverse primer 5′-GAA​CTC​ATA​CCT​TTA​CAC​TGC​C-3′ (for TNFRSF9) with Platinum Taq DNA Polymerase (Invitrogen) according to manufacturer’s recommendations, gel purified with the High Pure PCR Product Purification Kit (Roche), sequenced with the BigDye terminator v3.1 Cycle Sequencing Kit (Applied Biosystems), and analyzed on a 3500xL Genetic Analyzer (Applied Biosystems). All collected sequences were analyzed using DNADynamo (BlueTractorSoftware). To search for reversion events, amplicons of TNFRSF9 obtained by RT-PCR were also sequenced by WES methods to analyze single reads. Cell cultures PBMCs were isolated by Ficoll-Paque (Lymphoprep; Proteogenix) density-gradient centrifugation and washed and resuspended at a density of 106 cells/ml in complete Panserin 401 medium (Pan Biotech) containing 5% human male AB serum (BioWest), 100 U/ml penicillin, and 100 µg/ml streptomycin (Gibco). T cell blasts were expanded by incubating PBMCs for 72 h with 2.5 µg ml−1 PHA (Sigma-Aldrich); dead cells were then removed by Ficoll-Paque density-gradient centrifugation (Lymphoprep), and T cell blasts were cultured in complete Panserin 401 medium culture supplemented with 100 IU/ml recombinant human IL-2 (R&D Systems). Murine mastocytoma P815 cell lines, EBV-infected B-LCLs, and Jurkat T cell lines were cultured in complete RPMI 1640 GlutaMAX medium (Invitrogen) containing 10% heat-inactivated fetal calf serum (Gibco), 100 U/ml penicillin, and 100 µg ml−1 streptomycin (Gibco). HEK293T cells were cultured in complete DMEM, high glucose, GlutaMAX supplement, pyruvate medium (Gibco) containing 10% heat-inactivated fetal calf serum (Gibco), 100 U/ml penicillin, and 100 µg ml−1 streptomycin (Gibco). Constructs and RT-PCR for TNFRSF9 WT PIK3CD and TNFRSF9 coding sequences were obtained by RT-PCR from control T cell blasts with the forward primer 5′-CAC​CAT​GCC​CCC​TGG​GGT​GGA​CTG-3′ and the reverse primer 5′-GGG​AGG​AGC​CAC​TAC​TGC​CTG​TTG-3′ for PIK3CD and with the forward primer 5′-CAC​CAT​GGG​AAA​CAG​CTG​TTA​CAA-3′ and the reverse primer 5′-TCA​CAG​TTC​ACA​TCC​TCC​TTC​TTC-3′ for TNFRSF9. The PIK3CD and TNFRSF9 cDNAs were inserted into the plasmid pcDNA3.1 directional TOPO (Invitrogen) according to manufacturer’s instructions. The c.2462G>A (p.R821H) and c.3061G>A (p.E1021K) mutants were obtained with a Q5 site-directed mutagenesis kit (Promega) with the following primers: 2462GA-F, 5′-ACC​GGG​GAC​CAC​ACA​GGC​CTC-3′; 2462GA-R, 5′-GGG​GAG​GCA​GCC​ATA​GGG-3′; 3061GA-F, 5′-GAA​GTT​TAA​CAA​AGC​CCT​CCG​TG-3′; and 3061GA-R, 5′-ACT​CGG​AAG​TGC​TTC​AGT​G-3′. Coding sequences were confirmed by Sanger sequencing and subcloned into the bicistronic lentiviral plasmid (pLenti7.3/V5 TOPO; Invitrogen) or modified by replacing puromycin coding sequences by those of GFP as the reporter gene (pLenti7.3-GFP). p85α coding expression plasmid (p3XFLAG-CMV-p85α) was previously reported (Deau et al., 2014). CD137/TNFRSF9 transcript expression was analyzed by semiquantitativePCR with the forward primer 5′-ATG​GGA​AAC​AGC​TGT​TAC​AA-3′ and the reverse primer 5′-TCA​CAG​TTC​ACA​TCC​TCC​TTC​TTC-3′ for TNFRSF9 and with the forward primer 5′-ATG​CCA​TCA​CTG​CCA​CCC​AG-3′ and the reverse primer 5′-CCT​GCT​TCA​CCA​CCT​TCT​TG-3′ for GAPDH as a control for amplification and normalization. CRISPR-Cas9 genome editing The pSpCas9(BB)-2A-GFP (pX458) plasmid was a gift from Feng Zhang (Massachusetts Institute of Technology, Boston, MA; plasmid 48138; Addgene). All sgRNAs were designed using the Massachusetts Institute of Technology CRISPR Design Tool (http://crispr.mit.edu). 24-bp oligonucleotides containing the target sequences in exon 4 and exon 5 of PIK3CD were synthesized (Eurogentec) with a 4-bp overhang (later shown in lowercase) to enable cloning into the BbsI site. sgRNA sequences are as follows: sgEx4F, 5′-cac​cgA​GGT​GAA​CAC​ATA​GGC​CTC​G-3′; sgEx4R, 5′aaa​cCG​AGG​CCT​ATG​TGT​TCA​CCT​c-3′; sgEx5F, 5′-cac​cgG​TAC​CCT​GCG​GCT​CCC​GAA​C-3′; sgEx5R, 5′-aaa​cGT​TCG​GGA​GCC​GCA​GGG​TAC​c-3′. Pairs of synthetized oligos were annealed, phosphorylated, ligated to linearized vector, and transformed into Stbl3 bacteria (Life Technologies). sgRNA insertion was confirmed by Sanger sequencing using sequencing primer (5′-TTT​CTT​GGG​TAG​TTT​GCA​GTT​TT-3′). The plasmids were serially transfected three times into Jurkat T cells with Nepa21 electroporator (Nepagene). Cells were subcloned by limit dilution and cultured for 3 wk. DNA was extracted, and regions flanking the targeted site were amplified and sequenced using the following primers: Ex4F, 5′-ACT​GTC​CCT​CCA​GCT​GCT​GT-3′; Ex4R, 5′-GAG​CTA​CCT​TTG​CCG​ATG​AGG​AGG; Ex5F, 5′-CCG​CAG​GCC​TCC​ACG​AGT​TT-3′; Ex5R, 5′-CTC​ACC​TCG​CTG​CCC​TCA​AAC​T. Two clones presenting biallelic frameshift mutations in exon 4 (g.1:9715598delG, c.199delG, p.A67fsX42) or exon 5 (g.1:9716036_046, c.558_568del, p.P186PfsX) were selected. Lentiviral gene transfer PIK3CD-deficient Jurkat cell lines with empty pLenti-7.3-GFP or pLenti-7.3-GFP encoding WT, p.R821H, and p.E1021K p110δ were performed by lentivirus-mediated gene transfer. HEK293T cells were cotransfected with the lentiviral plasmid together with the plasmids pVSVG and pGag-pol at a molar ratio of 1:1:1 with Lipofectamine 2000 (Invitrogen) according to supplier’s instructions. Supernatants containing virus particles were harvested 36 h after transfection and freshly used for infection of Jurkat cells. All Jurkat cell lines expressed comparable levels of CD3 and CD28. Transductions of CD137-deficient T cells or control T cells with the pLenti-7.3-CD137 as previously described (Martin et al., 2014). Immunoblotting Stimulation of cells, protein extraction, and immunoblotting protocols have been described previously (Hauck et al., 2012; Martin et al., 2014; Winter et al., 2018). Briefly, cells (5 × 106 cells per ml) were stimulated by anti-CD3 antibody (1 µg/ml, clone OKT3) cross-linked with a rabbit anti-mouse IgG (2 µg/ml) for the different time periods. Cells were washed in PBS, and proteins were extracted with cell lysis buffer (1% NP-40 [NP-40 alternative; Calbiochem], 50 mM Tris, pH 8, 150 mM NaCl, 20 mM EDTA, 1 mM Na3VO4, 1 mM NaF, complete protease inhibitor cocktail [Roche], and phosphatase inhibitor cocktails 2 and 3 [Sigma]). Proteins were denatured by boiling 10 min with sample buffer (125 mM Tris, pH 6.8, 3% SDS, 10% glycerol, 5% 2β-mercaptoethanol, and 0.01% bromophenol blue), separated by SDS-PAGE, and transferred on polyvinylidene fluoride membrane (Millipore). Membranes were blocked with milk or BSA-based buffer before incubation with antibodies. The following antibodies were used for immunoblotting: anti-AKT (11E7, #4685S), anti-AKT phosphorylated on T308 (244F9, #4056S), anti-AKT phosphorylated on S473 (193H12, #4058S), anti-Erk1/2 (p44/42 MAPK, 137F5, #4695S), anti-phosphorylated Erk1/2 (phospho-p44/42 MAPK, 20G11, #4376S), anti-Ku70 (D35, #4103S), anti-p110α (C73F8, #4249S), anti-p85α(#4292), anti-p70 S6K (#9202S), anti-phosphorylated p70 S6K (108D2, #9234S), anti-PLC-γ1 (#2822S), anti-phosphorylated PLCγ1 (D6M9S, #140008S), anti-PTEN (138G6, #9559S), anti-phosphorylated tyrosine (P-Tyr-100, #9411S), anti-ZAP-70 (99F2, #2705S), and anti-phosphorylated ZAP-70 (#2704S; all from Cell Signaling Technology), anti-PI3K p110δ (H-219, sc-7176; Santa Cruz), and anti–β-actin (A2066, Sigma-Aldrich). Membranes were then washed and incubated with secondary anti-mouse (GE-Healthcare) or anti-rabbit (Cell Signaling Technology) HRP-linked antibodies. Pierce ECL WB substrate and Bio-Rad Clarity ECL substrate were used for revelation. For reprobing, membranes were incubated with Restore WB stripping buffer (Thermo Scientific) before blocking. Densitometry analyses were performed using ImageJ software and normalized to loading WB controls. Immunochemistry Staining and in situ hybridization were performed on an automated stainer (Bond Max; Leica Biosystems). The presence of EBV was demonstrated by in situ hybridization for the small RNA–encoding regions 1 and 2 (EBER). Antibodies used and dilution were anti-CD3 (polyclonal anti-CD3, 1/200; DAKO), anti-CD20 (polyclonal anti-CD3, 1/200; DAKO), and anti-CD8 (clone C8/144B; 1/200; DAKO). PI3 kinase assay and IPs HEK293T cells were transfected with the indicated expression plasmids using Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer’s instructions. T cell blasts or transfected (HEK293T cells 48 h after transfection) were lysed as described above. 500 mg protein was incubated with 2 µg polyclonal rabbit anti-anti-p110δ from Santa-Cruz (clone H-219) or normal rabbit IgG from Cell Signaling Technology (#2729) for 1 h, and immune complexes were recovered with Protein A Sepharose beads (GE Healthcare) at 4°C. The beads were then washed three times with 20 mM Tris-HCl, pH 7.4, 137 mM NaCl, 1 mM CaCl2, 1 mM MgCl2, and 1 mM sodium orthovanadate; three times with 0.1 M Tris-HCl, pH 7.4, 5 mM LiCl, and 1 mM sodium orthovanadate; and two times with 10 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 5 mM EDTA. For immunoblot, proteins were eluted from beads by boiling with SDS sample buffer for 10 min and were processed as described above. Kinase assay was performed with PI3K Activity ELISA:pico kit (Echelon) according to supplier recommendations. Briefly, immunoprecipitates were incubated with PI(4–5)P2 substrate for 6 h, and enzyme reaction and PIP3 standards were then incubated with a PIP3-binding protein. The mixture was transferred on a PIP3-coated microplate for competitive binding. Afterwards, a peroxidase-linked detector was added, allowing colorimetric detection of the amount of PIP3 produced by comparison with the standard curve. Flow cytometry Cell membrane staining and the flow cytometry–based phenotypic analyses were performed according to standard flow cytometry methods. The following validated antibodies were used: anti-CD3 (UCHT1), anti-CD4 (OKT4), anti-CD8 (RPA-T8), anti-CD14 (M5E2), anti-CD19 (HIB19), anti-CD25 (BC96), anti-CD27 (O323), anti-CD28 (CD28.2), anti-CD56 (HCD56), anti-CD57 (NK.1), anti-CD107a/b (H4A3/HAB4), anti-CD137 (4B4-1), anti-CD161 (HP-3G10), anti-TCR Vα7.2 (3C10), and anti-IgM (MHM88; all purchased from BioLegend); anti-CD16 (3G8), anti-CD45RA (HI100), anti-CD45RO (UCHL1), anti-CD197/CCR7 (3D12), anti-Perforin (dG9), anti-TCRαβ (OT31), and anti-IgD (IA6-2; BD Biosciences); anti-CD21 (BL13), anti-TCRγδ (IMMY510), anti-TCR Vα24 (C15), and anti-TCR Vβ11 (X21; Beckman Coulter). These antibodies were conjugated to FITC, PE, phycoerythrin-cyanin5 (PE-Cy5), PE-Cy5.5, PE-Cy7, peridinin-chlorophyll, peridinin-chlorophyll-Cy5.5, allophycocyanin, allophycocyanin-Cy7, allophycocyanin-Vio7, Brilliant Violet 421 (BV421), BV510, BV605, BV650, BV711, or BV785. For phosflow staining, 106 cells were incubated in 1 ml PBS with 1 µg/ml anti-CD3, pelleted, resuspended in 100 µl hot PBS containing 2 µg/ml rabbit anti-mouse cross-linker, incubated at 37°C for 5 min, fixed with 1 ml preheated phosflow fix buffer 1 (BD Biosciences), and then permeabilized with phosflow perm buffer 3 (BD Biosciences) according to manufacturer’s instructions. Cells were then incubated with Alexa Fluor 647–conjugated anti-AKT phosphorylated on S473 antibody (D9E; Cell Signaling) or Alexa Fluor 647–conjugated anti-AKT phosphorylated on T308 antibody (C31E5E; Cell Signaling). All data were collected on an LSRFortessa X-20 cytometer (BD Biosciences). Proliferation assays T cell blasts were washed and cultured without IL-2 for 72 h to synchronize the cells. T cell blasts or PBMCs were labeled with CellTrace Violet dye (Invitrogen) or CFSE (Invitrogen) according to the manufacturer’s instructions. Cells were then cultured for 4 to 7 d in complete Panserin 401 medium alone or in the presence of 0.1, 1, or 10 µg/ml immobilized anti-CD3 antibody (clone OKT3; eBioscience), dynabeads Human T-Activator CD3/CD28 (Invitrogen), 100 IU/ml IL-2, 2.5 µg ml−1 PHA, or 10−5 M ionomycin (Sigma-Aldrich) plus 10−7 M PMA (Sigma-Aldrich). Cells were surface stained for CD3, CD4, CD8, and CD25 detection and analyzed by flow cytometry (LSRFortessa X-20; BD Biosciences). Proliferation assays with coculture of T cells with the P815 cells or EBV B cells have been previously described elsewhere (Wen et al., 2002; Izawa et al., 2017). Briefly, irradiated P815 expressing or not CD137L or irradiated EBV B cell lines (LCLs) expressing CD70 were preincubated with soluble 0.25 µg ml−1 anti-CD3 antibody, washed, and cocultured with PBMCs or T cell blasts labeled with CellTrace Violet dye. The proliferation index (corresponding to the total number of divisions divided by the number of cells that went into division) was calculated using FlowJo software (TreeStar). Calcium flux analysis Ca2+ influx was assessed by real-time flow cytometry, as previously described (Martin et al., 2014). Briefly, cells were loaded with 5 µM Indo-1 AM (Molecular Probes) in presence of 2.5 mM of probenecid (PowerLoad; Molecular Probes), washed, and surface stained for CD4 and CD8 detection. Cells were analyzed in real time with a FACS ARIA II flow cytometer (BD Biosciences). During acquisition, 1 µg ml−1 anti-CD3 antibody was added to the cells, followed by 10 µg ml−1 of F(ab′)2 rabbit–anti-mouse IgG cross-linker (Jackson ImmunoResearch) and finally incubated with a calcium ionophore (1 mM ionomycin; Sigma). Changes in the intracellular calcium concentration are quantified by a shift in the indo-1 emission peak from 485 nm (indo-blue) for unbound dye to 405 nm (indo-violet) when the indo-1 molecule is bound to calcium. Data were analyzed using kinetic tool of FlowJo software (TreeStar). Intracellular Ca2+ levels correspond to the normalized ratio of 405 nm/485 nm indo-1 emission peaks. Apoptosis assay T cell blasts were left unstimulated or stimulated for 12 h with 0.01, 0.1, 1, and 10 µg ml−1 immobilized anti-CD3 (clone OKT3) or cross-linked anti-FAS antibody (clone Apo1.3) as previously described (Rigaud et al., 2006). Cells were then washed and stained for viability (viaprobe; BD Biosciences); surface expression of CD3, CD4, and CD8; and surface localization of phosphatidylserines using PE-conjugated Annexin-V (BD Biosciences) and analyzed by flow cytometry. Apoptotic cells corresponding to Annexin V+/viaprobe− cells. Degranulation assay T cell blasts were stimulated for 4 h with 0.3, 3, and 30 µg ml−1 of immobilized anti-CD3 in the presence of PE-conjugated anti-LAMP-1/2 (H4A3, H4B4; BD Biosciences) as previously described (Martin et al., 2014). Cells were then washed and stained for surface expression of CD3 and CD8 and analyzed by flow cytometry. Statistical analyses Data were analyzed using paired and unpaired Student’s t test or one-way ANOVA with a post hoc Bonferroni t test using PRISM software (GraphPad). Online supplemental material Fig. S1 shows the infiltrate of EBV-infected CD3+CD8+ T cells in hydroa vacciniforme–like lesions from a skin biopsy of the patient. Fig. S2 shows that correction of CD137 expression in CD4+ CD137-deficient T cells from the sister restores their capacity to proliferate in response to CD137L-expressing cells. Fig. S3 shows that sequencing of the mutation by Sanger or NGS does not reveal any reversion genetic event in lymphocytes from the patient and his sister. Fig. S4 shows additional functional analyses of T cells from the patient and his sister, including apoptosis, degranulation, phospho-PLC-γ1, phospho-AKT, calcium flux, and proliferation. Fig. S5 shows signaling analyses of PIK3CD/p110δ-deficient Jurkat T cells obtained by targeting PIK3CD exon 5 by CRISPR-Cas9 and reconstituted or not with WT, R821H, or E1021K p110δ. Supplementary Material Supplemental Materials (PDF) Acknowledgments We acknowledge the patient, his family, and the healthy donors for cooperation and blood gifts. S. Latour is a senior scientist at the Centre National de la Recherche Scientifique (France). R. Rodriguez was supported by the Ministère de la Recherche (France) and Ligue Contre le Cancer (France), D. Boutboul by the Agence Nationale de la Recherche (France), S. Winter and D. Jorge Cordeiro by the Imagine Institut PhD program funded by the Fondation Bettencourt Schueller, and B. Fournier by the Fondation pour la Recherche Médicale (FDM20170638301). This work was supported by grants from the Ligue Contre le Cancer Equipe Labelisée (France; S. Latour), Institut National de la Santé et de la Recherche Médicale (France), Rare Diseases Fondation (France; S. Latour), the Agence Nationale de la Recherche (ANR-14-CE14-0028-01 and ANR-18-CE15-0025-01 [S. Latour] and ANR-10-IAHU-01 [Imagine Institut]), and the European Research Council (ERC-2009-AdG_20090506 n°FP7-249816; A. Fischer). The authors declare no competing financial interests. Author contributions: R. Rodriguez, D. Jorge Cordeiro, S. Winter, S. Latour, and B. Fournier designed the research and performed experiments and analyzed the data. C. Picard, D. Boutboul, K. Izawa, C. Lenoir, E. Martin, J. Bruneau, S. Fraitag, and I. Callebaut performed experiments and analyzed the data. B. Neven identified the patients and provided clinical data. B. Neven, S. Latour, and A. Fischer analyzed clinical data. S. Kracker and T.H. Watts provide key reagents. A. Fischer and R. Rodriguez participated in the writing. S. Latour wrote the manuscript and supervised the research. ==== Refs Abolhassani, H., E.S. Edwards, A. Ikinciogullari, H. Jing, S. Borte, M. Buggert, L. Du, M. Matsuda-Lennikov, R. Romano, R. Caridha, . 2017. Combined immunodeficiency and Epstein-Barr virus-induced B cell malignancy in humans with inherited CD70 deficiency. J. Exp. Med. 214 :91–106. 10.1084/jem.20160849 28011864 Abraham, R.T., and A. Weiss. 2004. Jurkat T cells and development of the T-cell receptor signalling paradigm. Nat. Rev. Immunol. 4 :301–308. 10.1038/nri1330 15057788 Alkhairy, O.K., R. Perez-Becker, G.J. Driessen, H. Abolhassani, J. van Montfrans, S. Borte, S. Choo, N. Wang, K. Tesselaar, M. Fang, . 2015. Novel mutations in TNFRSF7/CD27: Clinical, immunologic, and genetic characterization of human CD27 deficiency. J. Allergy Clin. Immunol. 136 :703–712.e10. 10.1016/j.jaci.2015.02.022 25843314 Alosaimi, M.F., M. Hoenig, F. Jaber, C.D. Platt, J. Jones, J. Wallace, K.M. Debatin, A. Schulz, E. Jacobsen, P. Möller, . 2019. Immunodeficiency and EBV-induced lymphoproliferation caused by 4-1BB deficiency. J. Allergy Clin. Immunol. 144 :574–583.e5. 10.1016/j.jaci.2019.03.002 30872117 Angulo, I., O. Vadas, F. Garçon, E. Banham-Hall, V. Plagnol, T.R. Leahy, H. Baxendale, T. Coulter, J. Curtis, C. Wu, . 2013. Phosphoinositide 3-kinase δ gene mutation predisposes to respiratory infection and airway damage. Science. 342 :866–871. 10.1126/science.1243292 24136356 Astoul, E., C. Edmunds, D.A. Cantrell, and S.G. Ward. 2001. PI 3-K and T-cell activation: limitations of T-leukemic cell lines as signaling models. Trends Immunol. 22 :490–496. 10.1016/S1471-4906(01)01973-1 11525939 Bukczynski, J., T. Wen, K. Ellefsen, J. Gauldie, and T.H. Watts. 2004. Costimulatory ligand 4-1BBL (CD137L) as an efficient adjuvant for human antiviral cytotoxic T cell responses. Proc. Natl. Acad. Sci. USA. 101 :1291–1296. 10.1073/pnas.0306567101 14745033 Burke, J.E. 2018. Structural Basis for Regulation of Phosphoinositide Kinases and Their Involvement in Human Disease. Mol. Cell. 71 :653–673. 10.1016/j.molcel.2018.08.005 30193094 Carpier, J.M., and C.L. Lucas. 2018. Epstein-Barr Virus Susceptibility in Activated PI3Kδ Syndrome (APDS) Immunodeficiency. Front. Immunol. 8 :2005. 10.3389/fimmu.2017.02005 29387064 Castanedo, G.M., N. Blaquiere, M. Beresini, B. Bravo, H. Brightbill, J. Chen, H.F. Cui, C. Eigenbrot, C. Everett, J. Feng, . 2017. Structure-Based Design of Tricyclic NF-κB Inducing Kinase (NIK) Inhibitors That Have High Selectivity over Phosphoinositide-3-kinase (PI3K). J. Med. Chem. 60 :627–640. 10.1021/acs.jmedchem.6b01363 28005357 Chester, C., M.F. Sanmamed, J. Wang, and I. Melero. 2018. Immunotherapy targeting 4-1BB: mechanistic rationale, clinical results, and future strategies. Blood. 131 :49–57.29118009 Cohen, J.I. 2015. Primary Immunodeficiencies Associated with EBV Disease. Curr. Top. Microbiol. Immunol. 390 :241–265.26424649 Cohen, S.B.W., W. Bainter, J.L. Johnson, T.Y. Lin, J.C.Y. Wong, J.G. Wallace, J. Jones, S. Qureshi, F. Mir, F. Qamar, . 2019. Human primary immunodeficiency caused by expression of a kinase-dead p110δ mutant. J. Allergy Clin. Immunol. 143 :797–799.e2. 10.1016/j.jaci.2018.10.005 30336224 Coulter, T.I., A. Chandra, C.M. Bacon, J. Babar, J. Curtis, N. Screaton, J.R. Goodlad, G. Farmer, C.L. Steele, T.R. Leahy, . 2017. Clinical spectrum and features of activated phosphoinositide 3-kinase δ syndrome: A large patient cohort study. J. Allergy Clin. Immunol. 139 :597–606.e4. 10.1016/j.jaci.2016.06.021 27555459 Deau, M.C., L. Heurtier, P. Frange, F. Suarez, C. Bole-Feysot, P. Nitschke, M. Cavazzana, C. Picard, A. Durandy, A. Fischer, and S. Kracker. 2014. A human immunodeficiency caused by mutations in the PIK3R1 gene. J. Clin. Invest. 124 :3923–3928. 10.1172/JCI75746 25133428 DeBenedette, M.A., A. Shahinian, T.W. Mak, and T.H. Watts. 1997. Costimulation of CD28- T lymphocytes by 4-1BB ligand. J. Immunol. 158 :551–559.8992967 DeBenedette, M.A., T. Wen, M.F. Bachmann, P.S. Ohashi, B.H. Barber, K.L. Stocking, J.J. Peschon, and T.H. Watts. 1999. Analysis of 4-1BB ligand (4-1BBL)-deficient mice and of mice lacking both 4-1BBL and CD28 reveals a role for 4-1BBL in skin allograft rejection and in the cytotoxic T cell response to influenza virus. J. Immunol. 163 :4833–4841.10528184 Edwards, E.S.J., J. Bier, T.S. Cole, M. Wong, P. Hsu, L.J. Berglund, K. Boztug, A. Lau, E. Gostick, D.A. Price, . 2019. Activating PIK3CD mutations impair human cytotoxic lymphocyte differentiation and function and EBV immunity. J. Allergy Clin. Immunol. 143 :276–291.e6. 10.1016/j.jaci.2018.04.030 29800648 Engelman, J.A., J. Luo, and L.C. Cantley. 2006. The evolution of phosphatidylinositol 3-kinases as regulators of growth and metabolism. Nat. Rev. Genet. 7 :606–619. 10.1038/nrg1879 16847462 Fujiwara, S., H. Kimura, K. Imadome, A. Arai, E. Kodama, T. Morio, N. Shimizu, and H. Wakiguchi. 2014. Current research on chronic active Epstein-Barr virus infection in Japan. Pediatr. Int. 56 :159–166. 10.1111/ped.12314 24528553 Hauck, F., C. Randriamampita, E. Martin, S. Gerart, N. Lambert, A. Lim, J. Soulier, Z. Maciorowski, F. Touzot, D. Moshous, . 2012. Primary T-cell immunodeficiency with immunodysregulation caused by autosomal recessive LCK deficiency. J. Allergy Clin. Immunol. 130 :1144–1152.e11. 10.1016/j.jaci.2012.07.029 22985903 Huang, C.H., D. Mandelker, O. Schmidt-Kittler, Y. Samuels, V.E. Velculescu, K.W. Kinzler, B. Vogelstein, S.B. Gabelli, and L.M. Amzel. 2007. The structure of a human p110alpha/p85alpha complex elucidates the effects of oncogenic PI3Kalpha mutations. Science. 318 :1744–1748. 10.1126/science.1150799 18079394 Izawa, K., E. Martin, C. Soudais, J. Bruneau, D. Boutboul, R. Rodriguez, C. Lenoir, A.D. Hislop, C. Besson, F. Touzot, . 2017. Inherited CD70 deficiency in humans reveals a critical role for the CD70-CD27 pathway in immunity to Epstein-Barr virus infection. J. Exp. Med. 214 :73–89. 10.1084/jem.20160784 28011863 Kamijo, H., T. Miyagaki, N. Shishido-Takahashi, R. Nakajima, T. Oka, H. Suga, M. Sugaya, and S. Sato. 2018. Aberrant CD137 ligand expression induced by GATA6 overexpression promotes tumor progression in cutaneous T-cell lymphoma. Blood. 132 :1922–1935. 10.1182/blood-2018-04-845834 30194255 Kimura, H., and J.I. Cohen. 2017. Chronic Active Epstein-Barr Virus Disease. Front. Immunol. 8 :1867. 10.3389/fimmu.2017.01867 29375552 Kimura, H., Y. Hoshino, H. Kanegane, I. Tsuge, T. Okamura, K. Kawa, and T. Morishima. 2001. Clinical and virologic characteristics of chronic active Epstein-Barr virus infection. Blood. 98 :280–286. 10.1182/blood.V98.2.280 11435294 Kimura, H., Y. Ito, S. Kawabe, K. Gotoh, Y. Takahashi, S. Kojima, T. Naoe, S. Esaki, A. Kikuta, A. Sawada, . 2012. EBV-associated T/NK-cell lymphoproliferative diseases in nonimmunocompromised hosts: prospective analysis of 108 cases. Blood. 119 :673–686. 10.1182/blood-2011-10-381921 22096243 Latour, S., and S. Winter. 2018. Inherited Immunodeficiencies With High Predisposition to Epstein-Barr Virus-Driven Lymphoproliferative Diseases. Front. Immunol. 9 :1103. 10.3389/fimmu.2018.01103 29942301 Lee, H.W., S.J. Park, B.K. Choi, H.H. Kim, K.O. Nam, and B.S. Kwon. 2002. 4-1BB promotes the survival of CD8+ T lymphocytes by increasing expression of Bcl-xL and Bfl-1. J. Immunol. 169 :4882–4888. 10.4049/jimmunol.169.9.4882 12391199 Li, G., J.C. Boucher, H. Kotani, K. Park, Y. Zhang, B. Shrestha, X. Wang, L. Guan, N. Beatty, D. Abate-Daga, and M.L. Davila. 2018. 4-1BB enhancement of CAR T function requires NF-κB and TRAFs. JCI Insight. 3 :e121322. 10.1172/jci.insight.121322 30232281 Lucas, C.L., H.S. Kuehn, F. Zhao, J.E. Niemela, E.K. Deenick, U. Palendira, D.T. Avery, L. Moens, J.L. Cannons, M. Biancalana, . 2014. Dominant-activating germline mutations in the gene encoding the PI(3)K catalytic subunit p110δ result in T cell senescence and human immunodeficiency. Nat. Immunol. 15 :88–97. 10.1038/ni.2771 24165795 Martin, E., N. Palmic, S. Sanquer, C. Lenoir, F. Hauck, C. Mongellaz, S. Fabrega, P. Nitschké, M.D. Esposti, J. Schwartzentruber, . 2014. CTP synthase 1 deficiency in humans reveals its central role in lymphocyte proliferation. Nature. 510 :288–292. 10.1038/nature13386 24870241 Maus, M.V., A.K. Thomas, D.G. Leonard, D. Allman, K. Addya, K. Schlienger, J.L. Riley, and C.H. June. 2002. Ex vivo expansion of polyclonal and antigen-specific cytotoxic T lymphocytes by artificial APCs expressing ligands for the T-cell receptor, CD28 and 4-1BB. Nat. Biotechnol. 20 :143–148. 10.1038/nbt0202-143 11821859 Okano, M., K. Kawa, H. Kimura, A. Yachie, H. Wakiguchi, A. Maeda, S. Imai, S. Ohga, H. Kanegane, S. Tsuchiya, . 2005. Proposed guidelines for diagnosing chronic active Epstein-Barr virus infection. Am. J. Hematol. 80 :64–69. 10.1002/ajh.20398 16138335 Okkenhaug, K., and D.A. Fruman. 2010. PI3Ks in lymphocyte signaling and development. Curr. Top. Microbiol. Immunol. 346 :57–85.20563708 Okkenhaug, K., K. Ali, and B. Vanhaesebroeck. 2007. Antigen receptor signalling: a distinctive role for the p110delta isoform of PI3K. Trends Immunol. 28 :80–87. 10.1016/j.it.2006.12.007 17208518 Okuno, Y., T. Murata, Y. Sato, H. Muramatsu, Y. Ito, T. Watanabe, T. Okuno, N. Murakami, K. Yoshida, A. Sawada, . 2019. Defective Epstein-Barr virus in chronic active infection and haematological malignancy. Nat. Microbiol. 4 :404–413. 10.1038/s41564-018-0334-0 30664667 Park, S., and Y.H. Ko. 2014. Epstein-Barr virus-associated T/natural killer-cell lymphoproliferative disorders. J. Dermatol. 41 :29–39. 10.1111/1346-8138.12322 24438142 Quintanilla-Martinez, L., C. Ridaura, F. Nagl, M. Sáez-de-Ocariz, C. Durán-McKinster, R. Ruiz-Maldonado, G. Alderete, P. Grube, C. Lome-Maldonado, I. Bonzheim, and F. Fend. 2013. Hydroa vacciniforme-like lymphoma: a chronic EBV+ lymphoproliferative disorder with risk to develop a systemic lymphoma. Blood. 122 :3101–3110. 10.1182/blood-2013-05-502203 23982171 Rhee, S.G. 2001. Regulation of phosphoinositide-specific phospholipase C. Annu. Rev. Biochem. 70 :281–312. 10.1146/annurev.biochem.70.1.281 11395409 Rigaud, S., M.C. Fondanèche, N. Lambert, B. Pasquier, V. Mateo, P. Soulas, L. Galicier, F. Le Deist, F. Rieux-Laucat, P. Revy, . 2006. XIAP deficiency in humans causes an X-linked lymphoproliferative syndrome. Nature. 444 :110–114. 10.1038/nature05257 17080092 Salzer, E., S. Daschkey, S. Choo, M. Gombert, E. Santos-Valente, S. Ginzel, M. Schwendinger, O.A. Haas, G. Fritsch, W.F. Pickl, . 2013. Combined immunodeficiency with life-threatening EBV-associated lymphoproliferative disorder in patients lacking functional CD27. Haematologica. 98 :473–478. 10.3324/haematol.2012.068791 22801960 Shan, X., M.J. Czar, S.C. Bunnell, P. Liu, Y. Liu, P.L. Schwartzberg, and R.L. Wange. 2000. Deficiency of PTEN in Jurkat T cells causes constitutive localization of Itk to the plasma membrane and hyperresponsiveness to CD3 stimulation. Mol. Cell. Biol. 20 :6945–6957. 10.1128/MCB.20.18.6945-6957.2000 10958690 Sharfe, N., A. Karanxha, H. Dadi, D. Merico, D. Chitayat, J.A. Herbrick, S. Freeman, S. Grinstein, and C.M. Roifman. 2018. Dual loss of p110δ PI3-kinase and SKAP (KNSTRN) expression leads to combined immunodeficiency and multisystem syndromic features. J. Allergy Clin. Immunol. 142 :618–629. 10.1016/j.jaci.2017.10.033 29180244 Shuford, W.W., K. Klussman, D.D. Tritchler, D.T. Loo, J. Chalupny, A.W. Siadak, T.J. Brown, J. Emswiler, H. Raecho, C.P. Larsen, . 1997. 4-1BB costimulatory signals preferentially induce CD8+ T cell proliferation and lead to the amplification in vivo of cytotoxic T cell responses. J. Exp. Med. 186 :47–55. 10.1084/jem.186.1.47 9206996 Sogkas, G., M. Fedchenko, A. Dhingra, A. Jablonka, R.E. Schmidt, and F. Atschekzei. 2018. Primary immunodeficiency disorder caused by phosphoinositide 3-kinase δ deficiency. J. Allergy Clin. Immunol. 142 :1650–1653.e2. 10.1016/j.jaci.2018.06.039 30040974 Stephan, M.T., V. Ponomarev, R.J. Brentjens, A.H. Chang, K.V. Dobrenkov, G. Heller, and M. Sadelain. 2007. T cell-encoded CD80 and 4-1BBL induce auto- and transcostimulation, resulting in potent tumor rejection. Nat. Med. 13 :1440–1449. 10.1038/nm1676 18026115 Swan, D.J.D., D. Aschenbrenner, C.A. Lamb, K. Chakraborty, J. Clark, S. Pandey, K.R. Engelhardt, R. Chen, A. Cavounidis, Y. Ding, . 2019. Immunodeficiency, autoimmune thrombocytopenia and enterocolitis caused by autosomal recessive deficiency of PIK3CD-encoded phosphoinositide 3-kinase δ. Haematologica.:haematol.2018.208397. 10.3324/haematol.2018.208397 Takahashi, C., R.S. Mittler, and A.T. Vella. 1999. Cutting edge: 4-1BB is a bona fide CD8 T cell survival signal. J. Immunol. 162 :5037–5040.10227968 Tangye, S.G., U. Palendira, and E.S. Edwards. 2017. Human immunity against EBV-lessons from the clinic. J. Exp. Med. 214 :269–283. 10.1084/jem.20161846 28108590 Taylor, G.S., H.M. Long, J.M. Brooks, A.B. Rickinson, and A.D. Hislop. 2015. The immunology of Epstein-Barr virus-induced disease. Annu. Rev. Immunol. 33 :787–821. 10.1146/annurev-immunol-032414-112326 25706097 Vadas, O., J.E. Burke, X. Zhang, A. Berndt, and R.L. Williams. 2011. Structural basis for activation and inhibition of class I phosphoinositide 3-kinases. Sci. Signal. 4 :re2. 10.1126/scisignal.2002165 22009150 van Montfrans, J.M., A.I. Hoepelman, S. Otto, M. van Gijn, L. van de Corput, R.A. de Weger, L. Monaco-Shawver, P.P. Banerjee, E.A. Sanders, C.M. Jol-van der Zijde, . 2012. CD27 deficiency is associated with combined immunodeficiency and persistent symptomatic EBV viremia. J. Allergy Clin. Immunol. 129 :787–793.e6. 10.1016/j.jaci.2011.11.013 22197273 Watts, T.H. 2005. TNF/TNFR family members in costimulation of T cell responses. Annu. Rev. Immunol. 23 :23–68. 10.1146/annurev.immunol.23.021704.115839 15771565 Wen, T., J. Bukczynski, and T.H. Watts. 2002. 4-1BB ligand-mediated costimulation of human T cells induces CD4 and CD8 T cell expansion, cytokine production, and the development of cytolytic effector function. J. Immunol. 168 :4897–4906. 10.4049/jimmunol.168.10.4897 11994439 Winter, S., E. Martin, D. Boutboul, C. Lenoir, S. Boudjemaa, A. Petit, C. Picard, A. Fischer, G. Leverger, and S. Latour. 2018. Loss of RASGRP1 in humans impairs T-cell expansion leading to Epstein-Barr virus susceptibility. EMBO Mol. Med. 10 :188–199. 10.15252/emmm.201708292 29282224 Wortzman, M.E., D.L. Clouthier, A.J. McPherson, G.H. Lin, and T.H. Watts. 2013. The contextual role of TNFR family members in CD8(+) T-cell control of viral infections. Immunol. Rev. 255 :125–148. 10.1111/imr.12086 23947352 Yu, J., C. Wjasow, and J.M. Backer. 1998. Regulation of the p85/p110alpha phosphatidylinositol 3′-kinase. Distinct roles for the n-terminal and c-terminal SH2 domains. J. Biol. Chem. 273 :30199–30203. 10.1074/jbc.273.46.30199 9804776

PMC006xxxxxx/PMC6970427.txt

==== Front Learn Mem Learn Mem learnmem Learning & Memory 1072-0502 1549-5485 Cold Spring Harbor Laboratory Press 31949037 10.1101/lm.050344.119 LM050344Fuj Research Depotentiation depends on IP3 receptor activation sustained by synaptic inputs after LTP induction Depotentiation depends on synaptic activity Depotentiation depends on synaptic activity Fujii Satoshi 12 Yamazaki Yoshihiko 1 Goto Jun-ichi 12 Fujiwara Hiroki 1 Mikoshiba Katsuhiko 2 1 Department of Physiology, Yamagata University School of Medicine, Yamagata 990-9585, Japan 2 Laboratory for Developmental Neurobiology, Riken Brain Science Institute, Wako, Saitama 351-0198, Japan Corresponding author: sfujii@med.id.yamagata-u.ac.jp 2 2020 27 2 5266 26 7 2019 29 10 2020 © 2020 Fujii et al.; Published by Cold Spring Harbor Laboratory Press 2020 https://creativecommons.org/licenses/by-nc/4.0/ This article is distributed exclusively by Cold Spring Harbor Laboratory Press for the first 12 months after the full-issue publication date (see http://learnmem.cshlp.org/site/misc/terms.xhtml). After 12 months, it is available under a Creative Commons License (Attribution-NonCommercial 4.0 International), as described at http://creativecommons.org/licenses/by-nc/4.0/. In CA1 neurons of guinea pig hippocampal slices, long-term potentiation (LTP) was induced in field excitatory postsynaptic potentials (EPSPs) or population spikes (PSs) by the delivery of high-frequency stimulation (HFS, 100 pulses at 100 Hz) to CA1 synapses, and was reversed by the delivery of a train of low-frequency stimulation (LFS, 1000 pulses at 2 Hz) at 30 min after HFS (depotentiation), and this effect was inhibited when test synaptic stimulation was halted for a 19-min period after HFS or for a 20-min period after LFS or applied over the same time period in the presence of an antagonist of N-methyl-D-aspartate receptors (NMDARs), group I metabotropic glutamate receptors (mGluRs), or inositol 1, 4, 5-trisphosphate receptors (IP3Rs). Depotentiation was also blocked by the application of a Ca2+/calmodulin-dependent protein kinase II (CaMKII) inhibitor or a calcineurin inhibitor applied in the presence of test synaptic input for a 10-min period after HFS or for a 20-min period after LFS. These results suggest that, in postsynaptic neurons, the coactivation of NMDARs and group I mGluRs due to sustained synaptic activity following LTP induction results in the activation of IP3Rs and CaMKII, which leads to the activation of calcineurin after LFS and depotentiation of CA1 synaptic responses. JSPS KAKENHI24500434 17K01971 ==== Body pmcThe stimulation of group I metabotropic glutamate receptors (mGluRs) on hippocampal neurons activates phospholipase C, which hydrolyzes the inositol lipid precursor in the postsynaptic plasma membrane to form inositol 1, 4, 5-trisphosphate (IP3) and diacylglycerol; the former opens IP3 receptor (IP3R) channels and the latter activates protein kinase C (Ben-Ari et al. 1992; Nakanishi 1992). IP3Rs act as IP3-gated Ca2+-release channels on the endoplasmic reticulum (ER) of a variety of cells (Miyazaki et al. 1992; Berridge 1993; Mikoshiba 1993). Type 1 IP3Rs are the major members of the IP3R family in the central nervous system and are predominantly enriched in hippocampal neurons (Furuichi et al. 1989; Nakanishi et al. 1991). Prior synaptic activity can influence the subsequent induction of synaptic plasticity in the hippocampus. A type of synaptic plasticity, named depotentiation, in which low-frequency afferent stimulation at 1–2 Hz reverses preestablished long-term potentiation (LTP), has been observed at mossy fiber-CA3 pyramidal neuron synapses (Chen et al. 2001; Yamazaki et al. 2011) and at Schaffer collateral/commissural pathway CA1 synapses (Fujii et al. 1991). Using type 1 IP3R-deficient mice (Matsumoto et al. 1996), we found that the LTP or depotentiation induced at CA1 synapses was increased or attenuated, respectively, in which the mean magnitude of the responses after the delivery of either a short period of high-frequency stimulation (HFS, 10 pulses at 100 Hz) or low-frequency stimulation (LFS, 1000 pulses at 1 Hz) was significantly greater than that observed in wild-type mice (Fujii et al. 2000). Previously (Sugita et al. 2016), we found that bath application of 2-aminoethoxydiphenyl borate (2-APB), an antagonist of IP3Rs and/or store-operated calcium channels (Maruyama et al. 1997; Iwasaki et al. 2001; Bootman et al. 2002; Peppiatt et al. 2003), during priming HFS significantly decreased the magnitude of depotentiation in hippocampal CA1 neurons, and suggested that IP3Rs remain activated after HFS and that the modulation of IP3R activity induced during and/or after the subsequent LFS dephosphorylates postsynaptic proteins, leading to a decrease in LTP amplitude. We also found that the induction of depotentiation at CA1 synapses was inhibited by the application of FK506, a calcineurin inhibitor (Liu et al. 1991), to CA1 neurons for 20 min from the end of LFS, and suggested that the dephosphorylation of postsynaptic proteins as a result of calcineurin activation, which occurs during LFS, is maintained by test synaptic stimulation after LFS (Sugita et al. 2016). Studies on the hippocampal CA1 region have shown that N-methyl-D-aspartate receptor (NMDAR) activation occurs in postsynaptic cells when test synaptic stimulation is given at a frequency as low as 0.05 Hz (Yang 2000; Yamazaki et al. 2012), and the coactivation of NMDARs and mGluRs by test synaptic stimulation of postsynaptic cells determines whether LTP or long-term depression (LTD) is induced at CA1 synapses (Fujii et al. 2003, 2004). In our previous studies (Fujii et al. 1991, 1996, 2000), we reported a type of synaptic plasticity in hippocampal CA1 neurons that we referred to as “LTP suppression,” in which a train of LFS given prior to the delivery of HFS suppresses LTP induction. We recently demonstrated that, in this LFS-induced LTP suppression, stopping test synaptic stimulation after priming LFS negates the effects of LFS on subsequent LTP induction and can be replicated by perfusion with an antagonist for NMDARs, group I mGluRs, or IP3Rs or with a calcineurin inhibitor (Fujii et al. 2016). Previously (Yamazaki et al. 2012), we reported a type of depotentiation in CA1 neurons in which LTP induced in field excitatory postsynaptic potentials (EPSPs) by the delivery of LFS (80 pulses at 1 Hz) was reversed by the reapplication of the same LFS at 20 min later, and suggested that the induction of depotentiation depends on the coactivation of NMDARs and IP3Rs in postsynaptic neurons. However, when preestablished LTP induced by HFS (100 Hz at 100 pulses) is depotentiated in CA1 neurons by the subsequent delivery of LFS (1000 pulses at 2 Hz), it is not known whether the delivery of test synaptic stimulation after priming HFS determines if depotentiation is induced in CA1 neurons. In the present study, we used a pharmacological approach to study the effects of test synaptic stimulation on the depotentiation induced at CA1 synapses in hippocampal slices from mature guinea pigs. Results LTP induction does not require test synaptic stimulation after HFS LTP of the synaptic responses was induced in hippocampal CA1 neurons by the delivery of HFS (a tetanus of 100 pulses at 100 Hz) in standard perfusate when test synaptic stimulation at 0.05 Hz was continued throughout the experiment or withheld for a 19-min period immediately after HFS. Figure 1A and B show an example time course and summarized time course, respectively, for the change in the slope of the field EPSP (S-EPSP) or amplitude of the population spike (A-PS) in response to HFS when test synaptic stimulation at 0.05 Hz was either continued throughout the experiment (control, n = 7) or was stopped for the 19-min period from 1 to 20 min after HFS (stimulation off, n = 6). Figure 1. Effect of stopping test synaptic stimulation immediately after HFS on LTP induction. (A) Sample waveforms and a typical time course of LTP in the S-EPSP (left panel) or A-PS (right panel) induced by HFS (100 pulses at 100 Hz) when test synaptic stimulation at 0.05 Hz was either continued throughout the experiment (control) or was stopped for the 19-min period of 1 to 20 min after HFS (stimulation off). The sample waveforms were taken at the times indicated as a and b in the time course figure. (B) Summarized time course for the change in the S-EPSP (left panel) or A-PS (right panel) induced by HFS (100 pulses at 100 Hz) when test synaptic stimulation was either continued throughout the experiment (control, n = 7) or was stopped for the 19-min period of 1 to 20 min after HFS (stimulation off, n = 6). The horizontal bar marks the stimulation-off period and the arrow represents the application of HFS. In these, and all subsequent time course figures, the ordinate shows the S-EPSP or A-PS expressed as a percentage of the averaged value measured during the 10-min period before HFS. The symbols and bars represent the mean ± S.E.M. When test synaptic stimulation at 0.05 Hz was continued throughout the experiment, mean LTP in the S-EPSP measured at 25–30 or 95–100 min after HFS was 168.3 ± 5.6% or 156.4 ± 6.0%, respectively, of the pre-HFS levels (Fig. 1B, left panel), while mean LTP in the A-PS (Fig. 1B, right panel) measured at 25–30 or 95–100 min after HFS was 166.1 ± 3.7% or 168.8 ± 4.7%, respectively, of the pre-HFS levels. When test synaptic stimulation at 0.05 Hz was stopped for the 19-min period from 1 to 20 min after HFS, but was applied before and after this period (stimulation off), mean LTP in the S-EPSP measured at 25–30 or 95–100 min after HFS was 165.9 ± 9.0% or 164.9 ± 7.1%, respectively, of the pre-HFS levels (Fig. 1B, left panel), neither result being significantly different from the corresponding value for control LTP. Mean LTP in the A-PS measured at 25–30 or 95–100 min after HFS was 158.1 ± 13.3% or 174.3 ± 16.9%, respectively, of the pre-HFS levels (Fig. 1B, right panel), again not significantly different from the corresponding value for control LTP. Thus, stable LTP was induced in both the S-EPSP and A-PS when test synaptic stimulation at 0.05 Hz was stopped for the 19-min period from 1 to 20 min after HFS, showing that the induction of LTP triggered by HFS does not require test synaptic stimulation immediately after HFS. LTP induction does not depend on CaMKII activated by test synaptic stimulation after HFS In hippocampal CA1 neurons, HFS (100 pulses at 100 Hz) activates Ca2+/calmodulin-dependent protein kinase II (CaMKII), which phosphorylates either CaMKII itself or the GluA1 subunit of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs) to induce LTP (Bliss and Collingridge 1993; Griffith 2004). When test synaptic stimulation at 0.05 Hz was continued throughout the experiment and the slices were perfused with 10 µM KN-62, a specific inhibitor of CaMKII, for the 10-min period from 9 min before HFS to 1 min after HFS (n = 6), as shown in Figure 2, LTP induction was inhibited in the S-EPSP (left panel (1)) and A-PS (right panel (1)). Mean LTP in the S-EPSP measured at 25–30 or 95–100 min after HFS was 112.0 ± 6.6% or 112.0 ± 7.9%, respectively, of the pre-HFS levels (empty circles in left panel). Mean LTP in the A-PS measured at 25–30 or 95–100 min after HFS was 118.7 ± 8.0% or 112.5 ± 6.5%, respectively, of the pre-HFS levels (empty circles in right panel). These results indicate that LTP induction in hippocampal CA1 neurons requires the activity of CaMKII phosphorylated by HFS. Figure 2. Effect of a CaMKII inhibitor applied for a 10-min period on LTP induction. Summarized time course for the change in the S-EPSP (left panel) or A-PS (right panel) induced by HFS (100 pulses at 100 Hz) when test synaptic stimulation was continued throughout the experiment and the slices were perfused with 10 µM KN-62 for the 10-min period either from 9 min before HFS to 1 min after HFS (n = 6, empty bar and circles (1)) or from 1 to 11 min after HFS (n = 6, filled bar and circles (2)). The horizontal bar marks the period of KN-62 application and the arrow represents the delivery of HFS. The symbols and bars represent the mean ± S.E.M. We studied the effects of CaMKII activated by test synaptic stimulation after HFS on LTP induction in CA1 neurons by applying HFS (100 pulses at 100 Hz) to the slices in standard solution, then perfusing the slices with 10 µM KN-62 for the 10-min period from 1 to 11 min after HFS in the presence of test electrical stimulation at 0.05 Hz. When the slices were perfused with 10 µM KN-62 during this period (n = 6), as shown in Figure 2, LTP was induced in the S-EPSP (left panel (2)) and A-PS (right panel (2)). Mean LTP in the S-EPSP measured at 25–30 or 95–100 min after HFS was 150.3 ± 8.1% or 163.5 ± 14.6%, respectively, of the pre-HFS levels (filled circles in left panel). Mean LTP in the A-PS measured at 25–30 or 95–100 min after HFS was 158.3 ± 13.3% or 159.6 ± 7.4%, respectively, of the pre-HFS levels (filled circles in right panel). Since KN-62 does not affect the activity of autophosphorylated CaMKII (Tokumitsu et al. 1990) and since LTP induction in hippocampal CA1 neurons requires the activity of CaMKII autophosphorylated during and/or after HFS (Bliss and Collingridge 1993), this result suggests that LTP induction in hippocampal CA1 neurons does not depend on the subsequent activation of CaMKII by test synaptic stimulation during the 10-min period immediately after HFS. Effects of test synaptic stimulation delivered immediately after 2-Hz LFS on synaptic responses We examined the effects of test synaptic stimulation delivered immediately after LFS (1000 pulses at 2 Hz) on the synaptic responses of CA1 neurons when LFS was delivered to naïve Schaffer collateral/commissural pathway-CA1 neuron synapses. Figure 3A shows sample wave forms (top traces) and a typical example of the time course of the changes in the S-EPSP (left bottom panel) and A-PS (right one) in response to LFS when test synaptic stimulation was continued (filled circles) or stopped (empty circles) for the 20-min period from 0 to 20 min after the end of LFS (stimulation off). When test synaptic stimulation was continued throughout the experiment (filled circles), we observed that the responses were depressed; these slowly recovered toward pre-LFS control levels, reaching a plateau within 20 min. However, when test synaptic stimulation was stopped for 20 min after LFS (empty circles), its resumption induced responses at the pre-LFS control levels. Figure 3. Effect of stopping test synaptic stimulation immediately after LFS on synaptic responses. (A) Sample waveforms and a typical time course for the synaptic responses in the S-EPSP (left panel) or A-PS (right panel) before and after LFS (1000 pulses at 2 Hz) when test synaptic stimulation was either delivered at 0.05 Hz throughout the experiment (filled circles) or stopped for the 20-min period from 0 to 20 min after the end of LFS (unfilled circles). The sample waveforms were taken at the times indicated as a and b in the time course figure. (B) Summarized time course for the change in the S-EPSP (left panel) or A-PS (right panel) induced by LFS (1000 pulses at 2 Hz) when test synaptic stimulation was either delivered throughout the experiment (filled circles, n = 6) or stopped for the 20-min period from 0 to 20 min after the end of LFS (unfilled circles, n = 6). The unfilled horizontal bar marks the stimulation-off period and the filled horizontal bar represents the delivery of LFS. In these time course figures, the ordinate shows the S-EPSP or A-PS expressed as a percentage of the averaged value measured during the 10-min period before the delivery of LFS. The symbols and bars represent the mean ± S.E.M. The filled circles in Figure 3B show the summarized results for six experiments in which test synaptic stimulation at 0.05 Hz was continued throughout the experiment. In these experiments, the mean value of the S-EPSP (left panel) or A-PS (right panel) measured at 55–60 min after the end of LFS was 100.4 ± 4.5% or 100.1 ± 6.0% of the pre-LFS levels, respectively, showing that LFS delivery induced no significant change in field EPSPs or PSs for up to 60 min. The empty circles in Figure 3B show the summarized results for six experiments in which test synaptic stimulation was stopped for 20 min after LFS (stimulation off). In these experiments, the mean value of the S-EPSP or A-PS (right panel) measured at 55–60 min after LFS was 104.7 ± 3.1% or 106.2 ± 4.2% of the pre-LFS levels, respectively, neither result being significantly different from the corresponding control value. Thus, halting test synaptic stimulation for 20 min immediately after LFS did not affect the level of responses measured at 55–60 min after LFS (1000 pulses at 2 Hz) in naïve CA1 synaptic pathways. Effects of NMDARs, mGluRs, or IP3Rs activated immediately after 2-Hz LFS on synaptic responses When LFS (1000 pulses at 2 Hz) was delivered to naïve CA1 synaptic pathways, it was possible that test synaptic stimulation delivered immediately after LFS could have activated NMDARs, mGluRs, or IP3Rs at CA1 synapses, while they did not affect the responses measured at 55–60 min after LFS (Fig. 3). Therefore, in the following experiments shown in Figure 4A, we measured the LFS-induced changes in the S-EPSP (left panel) or A-PS (right panel) for up to 60 min after LFS when test synaptic stimulation at 0.05 Hz was continued throughout the experiment, but an antagonist for each of these receptors was added to the perfusate for the 20-min period from 0 to 20 min after the end of LFS. Figure 4. Effects of NMDARs, mGluRs, group I mGluRs, IP3Rs, CaMKII, or calcineurin activated immediately after LFS on synaptic responses. (A) Summarized results for the time course of changes in the S-EPSP (left panel) or A-PS (right panel) when LFS (1000 pulses at 2 Hz) was delivered to naïve CA1 synaptic inputs and 50 µM AP5 (n = 7, unfilled circles), 100 µM S-4CPG (n = 6, filled circles), or 10 µM 2-APB (n = 6, filled triangles) was applied to the perfusate (gray bar) in the presence of test synaptic inputs during the 20-min period from 0 to 20 min after the end of LFS (black bar). (B) Summarized results for the time course of change in the S-EPSP (left panel) or A-PS (right panel) when LFS (1000 pulses at 2 Hz) was delivered to naïve CA1 synaptic inputs and 10 µM KN-62 (n = 6) or 1 µM FK506 (n = 6) was applied (hatched bar) to the perfusate in the presence of test synaptic input during the 20-min period from 0 to 20 min after the end of LFS (gray bar). In these time course figures, the ordinate shows the S-EPSP or A-PS expressed as a percentage of the averaged value measured during the 10-min period before the delivery of LFS. The symbols and bars represent the mean ± S.E.M. When test synaptic stimulation was delivered at 0.05 Hz in the presence of 50 µM AP5, an NMDAR inhibitor, for 20 min after LFS (empty circles in Fig. 4A, n = 7), test synaptic stimulation delivered just after LFS depressed the responses; these slowly recovered toward pre-LFS control levels, reaching a plateau at 40–50 min after the end of LFS. The mean magnitude of the S-EPSP or A-PS of these responses measured at 55–60 min after LFS was 104.7 ± 4.7% or 100.1 ± 9.8%, respectively, of the pre-LFS levels, neither result being significantly different from the corresponding control value (filled circles in Fig. 3B). When test synaptic stimulation was delivered in the presence of 100 µM S-4CPG, a specific group I mGluR antagonist, for 20 min after LFS (filled circles in Fig. 4A, n = 6), test synaptic stimulation delivered just after LFS depressed the responses; these responses recovered toward pre-LFS control levels, reaching a plateau within 40 min after the end of LFS. The mean magnitude of the S-EPSP or A-PS of these responses measured at 55–60 min after LFS was 102.7 ± 5.9% or 102.2 ± 5.9%, respectively, of the pre-LFS levels, neither result being significantly different from the corresponding control value (filled circles in Fig. 3B). When test synaptic stimulation was delivered in the presence of 10 µM 2-APB, an IP3R antagonist, for 20 min after LFS (filled triangles in Fig. 4A, n = 6), the responses that were depressed just after LFS recovered toward pre-LFS control levels, reaching a plateau within 40 min after the end of LFS. The mean magnitude of the S-EPSP or A-PS of the responses measured at 55–60 min after LFS was 100.0 ± 7.0% or 105.1 ± 8.4%, respectively, of the pre-LFS levels, neither result being significantly different from the corresponding control value (filled circles in Fig. 3B). From all of the results shown in Figure 4A, we conclude that the activity of NMDARs, group I mGluRs, and/or IP3Rs in CA1 neurons for the 20-min period after LFS is not included in the responses measured at 55–60 min after LFS (1000 pulses at 2 Hz) of naïve CA1 synaptic input pathways. Effects of CaMKII or calcineurin activated immediately after 2-Hz LFS on synaptic responses We studied whether an inhibitor of CaMKII or calcineurin applied over the same period immediately after LFS (1000 pulses at 2 Hz) affected the synaptic responses when LFS was delivered to naïve CA1 synaptic pathways. In the following experiments, shown in Figure 4B, we observed the LFS-induced changes in the S-EPSP (left panel) or A-PS (right panel) when test synaptic stimulation at 0.05 Hz was continued throughout the experiment, but an inhibitor of CaMKII or calcineurin was applied to the perfusate for 20 min from 0 to 20 min after the end of LFS. When test synaptic stimulation was delivered in the presence of 10 µM KN-62 for 20 min after LFS, the responses that were depressed just after LFS recovered toward pre-LFS control levels, reaching a plateau at 40–50 min after the end of LFS (filled squares in Fig. 4B, n = 6). The mean magnitude of the S-EPSP or A-PS of the responses measured from 55 to 60 min after LFS was 105.0 ± 5.7% or 105.7 ± 9.4%, respectively, of the pre-LFS levels, neither result being significantly different from the corresponding control value (filled circles in Fig. 3B). In addition, when test synaptic stimulation was delivered in the presence of 1 µM FK506, a specific inhibitor of calcineurin, for 20 min after LFS (empty squares in Fig. 4B, n = 6), the responses that were depressed just after LFS recovered toward pre-LFS control levels, reaching a plateau within 40 min after the end of LFS. The mean magnitude of the S-EPSP or A-PS of the responses measured from 55 to 60 min after HFS was 97.0 ± 7.0% or 98.0 ± 9.3%, respectively, of the pre-LFS levels, neither result being significantly different from the corresponding control value (filled circles in Fig. 3B). Therefore, we conclude that the activity of CaMKII or calcineurin in CA1 neurons for 20 min immediately after LFS is not included in the level of responses measured at 55–60 min after LFS (1000 pulses at 2 Hz) of naïve CA1 synaptic input pathways. The activation of CaMKII or calcineurin occurs downstream from NMDAR activation and Ca2+/calmodulin complex formation in the signaling cascade of LTP or LTD in hippocampal CA1 neurons (Bliss and Collingridge 1993; Bear and Abraham 1996). Therefore, from the results shown in Figure 4A and B, we suggest that activation of the signaling pathway, NMDARs activation—Ca2+/calmodulin complex formation—CaMKII activation and/or—calcineurin activation does not occur in CA1 neurons during the 20-min period immediately after LFS, and does not affect the level to which the responses of CA1 neurons recover to at 55–60 min after the end of LFS (1000 pulses at 2 Hz) of naïve CA1 synaptic pathways. Depotentiation depends on test synaptic stimulation after HFS or LFS We studied the induction of depotentiation in hippocampal CA1 neurons by LFS (1000 pulses at 2 Hz) applied at 30 min after the delivery of HFS (100 pulses at 100 Hz). In subsequent experiments (Figs. 5–7), we measured the S-EPSP and A-PS during the 5-min period from 55 to 60 min after the end of LFS or the reduction in LTP in the S-EPSP and A-PS (Tables 1–4), and compared these values with the corresponding values for the depotentiation induced in a standard solution by LFS given at 30 min after HFS (first row in Table 1). Figure 5. Effect of stopping test synaptic stimulation after HFS or LFS on depotentiation. Typical example (A) and summarized time course (B) for the change in the S-EPSP (left panel) or A-PS (right panel) in response to a single train of LFS (1000 pulses at 2 Hz) (horizontal black bar) applied at 30 min after HFS (100 pulses at 100 Hz) (arrow). The sample traces above the main panel were taken at the times indicated as a, b, and c in A. In the control, test synaptic stimulation was applied throughout the experiment (filled circles), while the horizontal gray or white bar represents, respectively, stopping test synaptic stimulation at 0.05 Hz for the 19-min period from 1 to 20 min after HFS (gray squares, stimulation off-1) or for the 20-min period from 0 to 20 min after the end of LFS (unfilled triangles, stimulation off-2). In (B), n = 7 for the control and n = 6 for the other two traces. (C) Summarized time course for the change in the S-EPSP (left panel) or A-PS (right panel) in response to a single train of LFS (1000 pulses at 2 Hz) (horizontal black bar) applied at 30 min after HFS (100 pulses at 100 Hz) (arrow). The horizontal white bar represents stopping test synaptic stimulation for the 20-min period at 30–50 min after HFS (stimulation off-3). n = 6. X, Y, or Z in A represents the averaged value for the 10-min period immediately prior to HFS or LFS, the averaged value at 5–0 min immediately before LFS, and the stable level at 55–60 min after the end of LFS of the responses, respectively. Figure 6. Involvement of NMDARs, mGluRs, group I mGluRs, IP3Rs, CaMKII, or calcineurin activated after HFS in the induction of depotentiation. (A) Summarized results for the time course of depotentiation in the S-EPSP (left panel) or A-PS (right panel) when 50 µM AP5 (n = 5, unfilled circles), 100 µM S-4CPG (n = 5, filled circles), or 10 µM 2-APB (n = 5, filled triangles) was applied to the perfusate (hatched bar) in the presence of test synaptic stimulation during the 19-min period from 1 to 20 min after HFS (arrow) before the subsequent LFS (black bar). (B) Summarized results for the time course of depotentiation in the S-EPSP (left panel) or A-PS (right panel) when 10 µM KN-62 (n = 6, filled circles) or 1 µM FK506 (n = 6, unfilled circles) was applied to the perfusate (hatched bar) in the presence of test synaptic stimulation during the 10-min period from 1 to 11 min after the end of LFS (black bar). Figure 7. Effects of NMDARs, mGluRs, group I mGluRs, IP3Rs, CaMKII, or calcineurin activated after LFS on the induction of depotentiation. (A) Summarized results for the time course of depotentiation in the S-EPSP (left panel) or A-PS (right panel) when 50 µM AP5 (n = 5, unfilled circles), 100 µM S-4CPG (n = 5, filled circles), or 2-APB (n = 7, filled triangles) was applied to the perfusate (gray bar) in the presence of test synaptic stimulation during the 20-min period from 0 to 20 min after the end of LFS (black bar). (B) Summarized results for the time course of depotentiation in the S-EPSP (left panel) or A-PS (right panel) when 10 µM KN-62 (n = 6, filled circles) or 1 µM FK506 (n = 6, unfilled circles) was applied to the perfusate (hatched bar) in the presence of test synaptic stimulation during the 20-min period from 0 to 20 min after the end of LFS (black bar). Table 1. Effects of stopping the test synaptic stimulation at different times after HFS on depotentiation Table 2. Effects of the application of an NMDAR, mGluR, or IP3R antagonists for the 19-min period from 1 to 20 min after HFS on depotentiation Table 3. Effects of the application of a protein kinase inhibitor or calcineurin inhibitor for the 10-min period from 1 to 11 min after HFS on depotentiation Table 4. Effect of the application of an NMDAR, mGluR, or IP3R antagonist, protein kinase inhibitor, or calcineurin inhibitor for the 20-min period from 0 to 20 min after LFS on depotentiation Figure 5A shows sample wave forms (top traces) and a typical example of the time course of depotentiation in the S-EPSP (left panel) or A-PS (right panel), while Figure 5B shows the summarized results (n = 7). When test synaptic stimulation at 0.05 Hz was continued throughout the experiment (control, filled circles), test synaptic stimulation delivered just after LFS depressed the responses; these recovered toward a level below pre-LFS levels, reaching a plateau within 20 min. Table 1 shows the summarized results for seven experiments for the mean percentage reduction in LTP of the S-EPSP or A-PS (first row), which represents the magnitude of depotentiation in the responses. In hippocampal CA1 neurons, LFS (1000 pulses at 2 Hz) given at 30 min after the delivery of HFS (100 pulses at 100 Hz) caused a reduction in the LTP induced in the S-EPSP and A-PS (control in Figure 5A,B), while LFS itself did not induce a significant change in responses (control in Fig. 3). Therefore, these results indicate that the delivery of 100-Hz HFS has a preconditioning effect on LFS-induced depotentiation in hippocampal CA1 neurons. The preconditioning effect of HFS on the depotentiation of hippocampal CA1 neurons involves processes that are disrupted by stopping the delivery of test synaptic stimulation at 0.05 Hz after HFS. As shown in a typical example (gray squares in Fig. 5A) or the summarized results for six experiments (gray squares in Fig. 5B), when test synaptic stimulation at 0.05 Hz was stopped for the 19-min period of 1 to 20 min after priming HFS (stimulation off-1), LTP was induced in the S-EPSP or A-PS, but depotentiation was not induced in CA1 neurons. In these cells, the mean magnitude of the S-EPSP or A-PS measured at 25–30 min after priming HFS was not different from the corresponding value for control depotentiation (second row in Table 1), showing that the LTP induced in these cells was not affected by stopping test synaptic stimulation for 19 min immediately after HFS. However, Table 1 shows that the mean magnitude of the S-EPSP or A-PS measured at 55–60 min after LFS and the mean percentage reduction in LTP in the S-EPSP or A-PS were significantly different from the corresponding values for control depotentiation (second row), indicating no induction, or significant attenuation, of depotentiation in the S-EPSP or A-PS. Since stopping test synaptic stimulation at 0.05 Hz immediately after HFS did not affect LTP induction in hippocampal CA1 neurons (Fig. 1), but inhibited depotentiation at CA1 synapses (Fig. 5A), we suggest that the delivery of 100-Hz HFS has a preconditioning effect on LFS-induced depotentiation at CA1 synapses and that this effect depends on test synaptic stimulation after HFS to CA1 synapses. The induction of depotentiation in hippocampal CA1 neurons also depends on test synaptic stimulation immediately after LFS. As shown in a typical example (empty triangles in Fig. 5A) or the summarized results for six experiments (empty triangles in Fig. 5B), when test synaptic stimulation at 0.05 Hz was stopped for the 20-min period from 0 to 20 min after the end of LFS (stimulation off-2), LFS failed to induce depotentiation in the S-EPSP or A-PS. As shown in the third row of Table 1, the mean S-EPSP or A-PS measured at 25–30 min after HFS was not significantly different from the corresponding value for control depotentiation, whereas the mean S-EPSP or A-PS and the percentage reduction in LTP of the S-EPSP or A-PS measured at 55–60 min after LFS were significantly different from the corresponding values for control depotentiation. These results show that a preconditioning effect of HFS on depotentiation in hippocampal CA1 neurons involves processes that are disrupted by stopping test synaptic stimulation immediately after LFS. Since stopping test synaptic stimulation immediately after LFS did not affect synaptic transmission at 55–60 min after LFS at CA1 synapses (Fig. 3), but inhibited the depotentiation of hippocampal CA1 neurons (Fig. 5A), we conclude that the preconditioning effect of HFS on the LFS-induced depotentiation at CA1 synapses also depends on test synaptic stimulation for 20 min immediately after LFS. Until when does preconditioning HFS affect the induction of depotentiation? We studied the length of time in which preconditioning HFS continued to affect the induction of depotentiation in CA1 neurons. As shown by the summarized results for six experiments in Figure 5C, when test synaptic stimulation was stopped for 20 min from 30 to 50 min after the end of LFS (from 58 to 78 min after HFS) (stimulation off-3), depotentiation in the S-EPSP or A-PS was successfully induced at 55–60 min after the end of LFS in CA1 neurons. As shown in the bottom row of Table 1, the mean S-EPSP or A-PS measured at 25–30 min after HFS, the mean S-EPSP or A-PS measured at 55–60 min after LFS, and the percentage reduction in LTP of the S-EPSP or A-PS measured at 55–60 min after LFS were not significantly different from the corresponding values for control depotentiation. Thus, the depotentiation induced by LFS (1000 pulses at 2 Hz) was sensitive to the interval between priming HFS and the start of the halt of test synaptic stimulation, and was greater when this interval was <68 min. From the results shown in Figure 5, we conclude that the preconditioning effect of HFS on the LFS-induced depotentiation at CA1 synapses was maintained until 0–20 min after LFS (38–58 min after HFS), but ended at the 30- to 50-min period after the end of LFS (at 68–88 min after HFS). Effect of NMDARs, mGluRs, or IP3Rs activated immediately after priming HFS on depotentiation In order to determine whether test synaptic stimulation delivered after priming HFS activated NMDARs, group I mGluRs, and/or IP3Rs, which are involved in the induction of depotentiation in CA1 neurons, we examined whether the effect of stopping test synaptic stimulation for the 19-min period from 1 to 20 min after HFS could be replicated by perfusion with antagonists for these receptors. As shown in Figure 6A, when the slices were perfused with 50 µM AP5 (n = 5, unfilled circles) or 100 µM S-4CPG (n = 5, filled squares) for the same 19-min period after HFS in the presence of test synaptic stimulation of Schaffer collaterals, LTP was induced in the S-EPSP (left panel) and A-PS (right panel), but depotentiation was inhibited. As shown in Table 2, the mean values for the S-EPSP and A-PS at 25–30 min after HFS in the presence of these inhibitors were not significantly different from the corresponding values measured at 25–30 min after HFS for control LTP (filled circles in Fig. 1B) or control depotentiation (filled circles in Fig. 5B). However, in both cases, LFS at 30 min after priming HFS failed to induce depotentiation in the S-EPSP or A-PS; as shown in Table 2, the percentage reduction in LTP of the S-EPSP or A-PS measured at 55–60 min after LFS was significantly lower than the corresponding control value (Control in Table 1). These results show that test synaptic stimulation after priming HFS induces the coactivation of NMDARs and group I mGluRs in postsynaptic CA1 neurons and that this is required for the induction of depotentiation at CA1 synapses. Figure 6A and Table 2 show that similar results were obtained when 10 µM 2-APB (n = 5, filled triangles) was used, indicating that test synaptic stimulation after priming HFS activates IP3Rs and that this is also required for the induction of depotentiation in hippocampal CA1 neurons. Thus, for the induction of depotentiation in CA1 neurons, the effect of stopping test synaptic stimulation for the 19-min period of 1–20 min after priming HFS was replicated by perfusion with antagonists to NMDARs, group I mGluRs, or IP3Rs. Since IP3Rs act downstream from group I mGluRs in the signaling cascade in hippocampal neurons, we conclude that the mechanism of depotentiation at CA1 synapses involves the coactivation of NMDARs and IP3Rs in postsynaptic neurons caused by test synaptic stimulation delivered after priming HFS. It is possible that an increase of the postsynaptic intracellular Ca2+ concentration ([Ca2+]i) during this period, as a result of Ca2+ influx through NMDARs and Ca2+ efflux through IP3Rs into the cytosol of postsynaptic cells, is involved in the mechanism of LFS-induced depotentiation at CA1 synapses. Activation of CaMKII and calcineurin after HFS is necessary for the induction of depotentiation We hypothesized that the formation of Ca2+/calmodulin complexes in hippocampal CA1 neurons due to a postsynaptic increase in [Ca2+]i during the period immediately after priming HFS could activate either CaMKII or calcineurin in postsynaptic cells and induce depotentiation, while LTP induction does not depend on CaMKII activation during the 10-min period after HFS (100 pulses at 100 Hz) (filled bars (2) and circles in Fig. 2). We first studied the effects of CaMKII activation after priming HFS on the induction of depotentiation in CA1 neurons by applying HFS (100 pulses at 100 Hz) in a standard solution, then perfusing the slices with 10 µM KN-62 for the 10-min period from 1 to 11 min after priming HFS in the presence of test electrical stimulation at 0.05 Hz. As shown by the summarized results for six experiments in Figure 6B (filled squares), the delivery of LFS at 30 min after preconditioning HFS failed to induce depotentiation in the S-EPSP or A-PS in CA1 neurons. As shown in the upper row of Table 3, the mean value for the S-EPSP or A-PS at 25–30 min after HFS in slices perfused with a CaMKII inhibitor was not significantly different from the corresponding value measured at 25–30 min for control LTP (filled circles in Fig. 1B) or control depotentiation (filled circles in Fig. 5B and Control in Table 1). However, in these slices, the delivery of LFS at 30 min after priming HFS failed to induce depotentiation in the S-EPSP or A-PS; as shown in Table 3, the percentage reduction in the LTP of the S-EPSP or A-PS measured at 55–60 min after LFS was significantly lower than the corresponding control value (Control in Table 1). We examined whether the activation of calcineurin immediately after priming HFS was involved in the mechanism of depotentiation in CA1 neurons. As shown in Figure 6B, when the slices (n = 6) were perfused with 1 µM FK506 for the 10-min period of 1 to 11 min after priming HFS in the presence of test synaptic inputs (unfilled circles), the delivery of LFS at 30 min after preconditioning HFS failed to induce depotentiation in the S-EPSP (left panel) and A-PS (right panel). As shown in the lower row of Table 3, the mean value for the S-EPSP or A-PS at 25–30 min after HFS in the presence of 1 µM FK506 was not significantly different from the corresponding value measured at 25–30 min for control LTP (filled circles in Fig. 1B) or control depotentiation (filled circles in Fig. 5B and Control in Table 1). However, in these cases, the percentage reduction in the LTP of the S-EPSP or A-PS measured at 55–60 min after LFS was significantly lower than the corresponding control value (Control in Table 1). On the basis of these results, we conclude that CaMKII and calcineurin activated immediately after priming HFS are required for the induction of depotentiation at CA1 synapses. Effects of NMDARs, mGluRs, or IP3Rs activated immediately after LFS on depotentiation Since the delivery of test synaptic stimulation immediately after LFS was necessary for the induction of depotentiation in hippocampal CA1 neurons (stimulation off-2 in Fig. 5), we thought it possible that the coactivation of NMDARs and group I mGluRs and/or IP3Rs due to test synaptic stimulation also occurred during the period immediately after LFS during the depotentiation of hippocampal CA1 neurons. Therefore, we examined whether the effect of stopping test synaptic stimulation for the 20-min period from 0 to 20 min after LFS could be replicated by perfusion with antagonists of these receptors. As shown in Figure 7A, when LFS was delivered in the standard perfusate, but 50 µM AP5 (n = 5, unfilled circles), 100 µM S-4CPG (n = 5, filled circles), or 10 µM 2-APB (n = 7, filled triangles) was applied for the 20-min period from 0 to 20 min after the end of LFS, the LFS-induced depotentiation in the S-EPSP (left panel) and A-PS (right panel) was attenuated. In these slices (first to third rows of Table 4), each percentage reduction of LTP in the S-EPSP or A-PS measured at 55–60 min after LFS was significantly lower than the corresponding value for control depotentiation (Table 1). These results suggest that, for the induction of depotentiation in CA1 neurons, the effect of stopping test synaptic stimulation for the 20-min period immediately after LFS could be replicated by perfusion with each antagonist of NMDARs, group I mGluRs, or IP3Rs. Application of the antagonists for NMDARs, group I mGluRs, or IP3Rs to CA1 neurons for the 20-min period immediately after LFS had no effect on the level of responses measured at 55–60 min after LFS of naïve CA1 synaptic input pathways and test synaptic stimulation at 0.05 Hz delivered throughout the experiment (Fig. 4A). Therefore, we suggest that the preconditioning effect of HFS on the induction of depotentiation in hippocampal CA1 neurons involves the coactivation of NMDARs and IP3Rs, the latter of which act downstream from group I mGluRs in the signaling cascade, in postsynaptic neurons during the 20-min period immediately after LFS. Activation of CaMKII and calcineurin after LFS is necessary for depotentiation In LFS-induced depotentiation at CA1 synapses, we thought it possible that the coactivation of NMDARs and IP3Rs, which occurred immediately after LFS in postsynaptic CA1 neurons, could increase postsynaptic [Ca2+]i, leading to the formation of Ca2+/calmodulin complexes and the activation of CaMKII and/or calcineurin in postsynaptic CA1 neurons. Thus, we studied the effects of CaMKII or calcineurin activated immediately after LFS on the induction of depotentiation in CA1 neurons by applying LFS in a standard solution, then perfusing the slices with 10 µM KN-62 (n = 6) or 1 µM FK506 (n = 6) for the 20-min period from 0 to 20 min after the end of LFS in the presence of test synaptic stimulation. As shown in Figure 7B, LFS-induced depotentiation in the S-EPSP (left panel) or A-PS (right panel) was attenuated by applying 10 µM KN-62 (filled squares) during the 20-min period immediately after LFS. In these slices (fourth row in Table 4), the mean magnitude of the S-EPSP and A-PS and the percentage reduction in the LTP of the S-EPSP or A-PS measured at 55–60 min after LFS were all significantly different from the corresponding control values. In addition, as shown in Figure 7B, LFS-induced depotentiation in the S-EPSP (left panel) and A-PS (right panel) was attenuated by applying 1 µM FK506 (unfilled squares) during the 20-min period immediately after LFS. In these slices (bottom row in Table 4), the mean magnitude of the S-EPSP and A-PS and percentage reduction in the LTP of the S-EPSP or A-PS measured at 55–60 min after LFS were also significantly different from the corresponding control values (Control in Table 1). Since the application of CaMKII or calcineurin inhibitors to CA1 neurons for the 20-min period immediately after LFS did not affect the level of responses measured at 55–60 min after LFS of naïve CA1 synaptic input pathways and test electrical stimulation at 0.05 Hz delivered throughout the experiment (Fig. 4B), the results shown in Figure 7B indicate that the conditioning effect provided by priming HFS for the induction of depotentiation in CA1 neurons involves both CaMKII and calcineurin activated during the 20-min period immediately after LFS. The results shown so far indicate that that priming HFS triggers postsynaptic cellular events necessary for the depotentiation of CA1 neurons, including the activation of group I mGluRs and/or IP3Rs, which is maintained by test synaptic inputs at 0.05 Hz, at least until 68 min after priming HFS. Although the role of CaMKII is still unclear in the mechanism of LFS-induced depotentiation in hippocampal CA1 neurons, we suggest that the activation of IP3Rs by preconditioning HFS results in the activation of calcineurin after LFS in postsynaptic neurons, leading to the dephosphorylation of postsynaptic proteins and a decrease in the amplitude of LTP induced by priming HFS in hippocampal CA1 neurons. Discussion In the depotentiation of hippocampal CA1 neurons, LFS (1000 pulses at 2 Hz) given at 30 min after the delivery of HFS (100 pulses at 100 Hz) caused a reduction in LTP (control in Fig. 5), while 2-Hz LFS itself induced no significant change in responses (control in Fig. 3). In these neurons, LTP induction did not rely on the formation of depotentiation, but instead depended on the activity of postsynaptic neurons after priming HFS, since stopping test synaptic stimulation at 0.05 Hz for 20 min after priming HFS did not affect LTP induction (Fig. 1), but did disrupt the formation of depotentiation (Fig. 5A). In addition, stopping test synaptic stimulation at 0.05 Hz for 20 min immediately after LFS did not affect the synaptic responses in naïve synaptic pathways (Fig. 3), but did disrupt the formation of depotentiation (Fig. 5A). These results indicate that 100-Hz HFS has a conditioning effect for the induction of depotentiation, and this effect is maintained by test synaptic stimulation at 0.05 Hz for a 20-min period after LFS. Since the effect of stopping test synaptic stimulation after HFS or LFS could be replicated by perfusion with AP5, S-4CPG, or 2-APB for the induction of depotentiation (Figs. 6A, 7A), we suggest that the preconditioning effect of HFS on the induction of depotentiation at CA1 synapses involves NMDARs, group I mGluRs, or IP3Rs in postsynaptic neurons activated by test synaptic stimulation after priming HFS. Since the application of CaMKII or calcineurin inhibitors to CA1 neurons for the 10-min period immediately after HFS or 20-min period immediately after LFS inhibited the induction of depotentiation in the presence of test synaptic stimulation (Figs. 6B, 7B), we suggest that the conditioning effect provided by priming HFS for the induction of depotentiation at CA1 synapses involves CaMKII and calcineurin in postsynaptic cells, which are activated immediately after priming HFS or subsequent LFS. Our previous study using hippocampal CA1 neurons from IP3R1-deficient mice demonstrated that HFS (100 pulses at 100 Hz) induces normal LTP, but a train of LFS (1000 pulses at 1 Hz), delivered at 60 min after HFS, fails to induce depotentiation (Fujii et al. 2000). In CA1 neurons of mature guinea pigs, HFS (100 pulses at 100 Hz) given in the presence of 2-APB induces similar LTP to that seen in the absence of 2-APB and is maintained for at least 60 min (Taufiq et al. 2005; Fujii et al. 2016), while a train of LFS (1000 pulses at 2 Hz) delivered at 30 min after HFS in the presence of 2-APB induces significantly lower depotentiation than that seen in the absence of 2-APB (Sugita et al. 2016). These observations indicate that the conditioning effect provided by priming HFS for the induction of depotentiation at CA1 synapses involves the activation of IP3Rs in postsynaptic neurons. LTP induction at CA1 synapses requires sufficient depolarization of the postsynaptic membrane to activate NMDARs (Collingridge et al. 1988a, 1988b; Alford et al. 1993) and/or voltage-gated calcium channels (Ito et al. 1995) to increase postsynaptic [Ca2+]i for the activation of CaMKII (Bliss and Collingridge 1993). In this study, we have shown the effect of CaMKII activated and autophosphorylated during or after HFS on LTP induction and the effect of CaMKII activated after HFS on the depotentiation of CA1 neurons by applying KN-62 to the perfusate during or after HFS (Figs. 2, 6B). While KN-62 and Ca2+/calmodulin complexes bind competitively to the regulatory domain of CaMKII (Tokumitsu et al. 1990), KN-62 does not affect LTP induction that depends on the activity of CaMKII autophosphorylated during and/or after HFS (Bliss and Collingridge 1993). Therefore, we suggest that the preconditioning effect of HFS on the induction of depotentiation consists of IP3Rs activated by HFS and CaMKII activated after HFS due to the binding of Ca2+/calmodulin complexes to CaMKII after HFS, but does not include CaMKII autophosphorylated during and/or after HFS at CA1 synapses. The effect of stopping test synaptic stimulation during the 20-min period after priming HFS or after the subsequent LFS on the induction of depotentiation (Fig. 5) could be replicated by perfusion with each antagonist of NMDARs, group I mGluRs, and IP3Rs for the same period (Figs. 6A, 7A). This suggests that test synaptic stimulation sustains the preconditioning effect of HFS for the induction of depotentiation, causing the coactivation of NMDARs and group I mGluRs and/or IP3Rs in postsynaptic CA1 neurons for a 20-min period after the subsequent LFS. Given that the increase of postsynaptic [Ca2+]i is caused by the activation of NMDARs and/or mGluRs in CA1 neurons (Pin and Duvoisin 1995; Otani and Connor 1998; Skeberdis et al. 2001) and that IP3Rs act downstream from group I mGluRs in the signaling cascade (Mikoshiba 1993), it is possible that the increase of postsynaptic [Ca2+]i due to Ca2+ influx through NMDARs/channels and Ca2+ efflux through IP3Rs caused by test synaptic stimulation after HFS is involved in the signaling mechanism of the preconditioning effect of HFS on LFS-induced depotentiation in CA1 neurons. The activation of either CaMKII or calcineurin occurs downstream from the increase of postsynaptic [Ca2+]i and Ca2+/calmodulin complex formation in the signaling cascade of hippocampal synaptic plasticity in CA1 neurons (Bliss and Collingridge 1993; Bear and Abraham 1996; Griffith 2004). In this study, we demonstrated that CaMKII or calcineurin, either activated by test synaptic stimulation during a 10-min period after priming HFS or during a 20-min period after subsequent LFS, was involved in the mechanism of LFS-induced depotentiation at CA1 synapses (Figs. 6B, 7B). Thus, we suggest that the preconditioning effect of priming HFS on the induction of depotentiation involves CaMKII and calcineurin, with the activity of both enzymes being sustained by test synaptic stimulation, but also being modified by LFS and contributing to the dephosphorylation of postsynaptic proteins after LFS to induce depotentiation in hippocampal CA1 neurons. Since IP3Rs activated by test synaptic stimulation after HFS did not affect LTP amplitude (Fig. 6A), while those activated after LFS caused a decrease in LTP amplitude (Fig. 7A), it is possible that the delivery of LFS modulates the activity of IP3Rs to switch these two enzymes from the phosphorylation to dephosphorylation of AMPARs on postsynaptic neurons, decreases LTP amplitude after LFS, and induces depotentiation at CA1 synapses. The delivery of a 2-Hz LFS train to CA1 synapses, which by itself does not induce NMDAR-dependent LTD in hippocampal CA1 neurons (Fujii et al. 2010), induces depotentiation after the prior activation of IP3Rs at CA1 synapses (Sugita et al. 2016). Previous studies have demonstrated that LFS-induced depotentiation in CA1 neurons is blocked by the application, during LFS, of either the NMDAR antagonist AP5 (Fujii et al. 1991; Huang et al. 2001, Sugita et al. 2016) or IP3R inhibitor 2-APB (Yamazaki et al. 2012; Sugita et al. 2016). This implies that the increase of postsynaptic [Ca2+]i during a 2-Hz LFS train is a result of Ca2+ influx through NMDARs and release through IP3Rs is involved in LFS-induced depotentiation in CA1 neurons. In hippocampal CA1 neurons, the homosynaptic LTD induced by 1-Hz LFS requires NMDAR activation and a moderate increase in postsynaptic [Ca2+]i, the latter triggering the activation of calcineurin (Linden 1994; Bear and Abraham 1996). In the mechanism of LTD in CA1 neurons, postsynaptic calcineurin dephosphorylates and inactivates inhibitor-1, allowing protein phosphatase 1 to act on Thr286 in the catalytic domain of CaMKII or Ser831 in the AMPAR GluA1 subunit (Lisman 1994; Mulkey et al. 1994; Barria et al. 1997; Lee et al. 2000). Group I mGluRs are positively coupled to phospholipase C and their activation is required for the induction of LTD in hippocampal CA1 neurons (Palmer et al. 1997; Reyes-Harde and Stanton 1998). Taufiq et al. (2005) showed that IP3R activation, which occurs downstream from group I mGluRs and phospholipase C in the signaling cascade, plays an important role in facilitating NMDAR-dependent LTD induced by 1-Hz LFS in hippocampal CA1 neurons. Since IP3R activation is increased by a decrease in the cytoplasmic levels of free Ca2+/calmodulin complexes (Michikawa et al. 1999), the activation of CaMKII due to the binding of free Ca2+/calmodulin complexes to the regulatory domain of its α-subunit (CaMKIIα) (Griffith 2004) and the activation of calcineurin due to the binding of free Ca2+/calmodulin complexes to the catalytic domain of calcineurin (Ye et al. 2008; Baumgärtel and Mansuy 2015) may decrease the levels of free Ca2+/calmodulin complexes and increase IP3R activation in the dendritic spine of postsynaptic CA1 neurons. The results shown in Figures 6 and 7 indicate that the preconditioning effect of HFS on the induction of depotentiation consists of IP3Rs, CaMKII, and calcineurin activated by test synaptic stimulation after HFS in postsynaptic CA1 neurons. Therefore, we think it possible that the activation of CaMKII and calcineurin due to the coactivation of NMDARs and IP3Rs caused by test synaptic stimulation after priming HFS (Fig. 6A,B) plays a role in modulating IP3R activation during and/or after 2-Hz LFS and contributes to the formation of NMDAR-dependent LTD after the delivery of 2-Hz LFS to CA1 synapses (Fig. 7A,B). IP3R-binding protein released with IP3 (IRBIT) (Ando et al. 2003) binds to IP3Rs and inhibits their activity by blocking the access of IP3 to a common binding site (Ando et al. 2003, 2006). The phosphorylation of IRBIT is essential for its binding to IP3Rs to prevent their activation when the concentration of IP3 is low (Ando et al. 2006). Kawaai et al. (2015) recently demonstrated that IRBIT binds to the regulatory domain of CaMKIIα and inhibits its activity in the central nervous system. They also demonstrated that an excess of Ca2+/calmodulin complexes resulted in the dissociation of IRBIT from the regulatory domain of CaMKIIα because IRBIT and Ca2+/calmodulin complexes bind competitively to this domain (Kawaai et al. 2015). Therefore, we think it possible, in LFS-induced depotentiation in hippocampal CA1 neurons, that IRBIT phosphorylated by priming HFS remains phosphorylated after HFS or LFS and limits IP3R activation and CaMKIIα activity by blocking the access of IP3 to a common binding site of IP3Rs and the binding of the Ca2+/calmodulin complexes to CaMKIIα. In this study, we have shown the effects of CaMKII activated after HFS on LTP induction and the formation of depotentiation in CA1 neurons by applying HFS (100 pulses at 100 Hz) in a standard solution, then perfused the slices with a CaMKII inhibitor during the 10-min period immediately after HFS in the presence of test synaptic stimulation at 0.05 Hz (Figs. 2, 6B). Thus, as illustrated in Figure 8A and B, we suggest the following signaling mechanism for depotentiation during and after HFS in postsynaptic CA1 neurons: the postsynaptic increase in [Ca2+]i and free Ca2+/calmodulin complex levels due to the coactivation of NMDARs/channels and group I mGluRs during the period of test synaptic stimulation after priming HFS stimulates CaMKII activation while IRBIT phosphorylated and released during and after HFS inhibits further activation of IP3R or CaMKII after priming HFS (Kawaai et al. 2015). Therefore, HFS induces a state that maintains LTP, provided there is no subsequent LFS (Figs. 1, 5). Figure 8. Schematic depiction of the mechanism of LFS-induced depotentiation in hippocampal CA1 neurons. (A) Synaptic inputs after HFS or LFS activate NMDARs and mGluRs on postsynaptic CA1 pyramidal neurons. The activation of IP3Rs on the membrane of the ER occurs downstream from group I mGluRs and phospholipase C in the signaling cascade. The increase in [Ca2+]i due to the coactivation of NMDARs and IP3Rs induces an increase in the free cytoplasmic levels of Ca2+/calmodulin complexes and IRBIT in the postsynaptic dendritic spines of CA1 neurons. (B) CaMKII autophosphorylated during and after priming HFS may continue to phosphorylate AMPA-type glutamate receptors on the dendritic spines of CA1 neurons to potentiate synaptic responses (to induce LTP), while IRBIT that is released from IP3Rs into the cytoplasm and is phosphorylated may block the binding of Ca2+/calmodulin complexes to the regulatory domain of CaMKII and inhibit further CaMKII activation in postsynaptic CA1 neurons. CaMKII is active when Ca2+/calmodulin complexes are bound to its regulatory domain. (C) During synaptic transmission after LFS, the levels of free Ca2+/calmodulin complexes, which are not sequestrated by CaMKII in the presence of IRBIT, are increased to a point at which calcineurin is activated in the dendritic spines to dephosphorylate CaMKII and AMPARs at CA1 synapses, leading to a reduction in synaptic potentiation (depotentiation) in CA1 neurons. When LFS of 1000 pulses at 2 Hz was delivered to naïve CA1 synaptic pathways and test synaptic stimulation was delivered throughout the experiment, the activation of IP3Rs during and after the application of a 2-Hz LFS train to CA1 synapses did not decrease responses (Fig. 4A; Fujii et al. 2010). However, since the activation of IP3Rs during and after the LFS actually decreased them in depotentiation in hippocampal CA1 neurons (Fig. 7A; Sugita et al. 2016), it is possible that IRBIT phosphorylated by priming HFS remains phosphorylated during and after the 2-Hz LFS train and limits the activation of IP3Rs at CA1 synapses. Stopping the test electrical stimulation of CA1 neurons during the 20-min period from 0 to 20 min after the end of the 2-Hz LFS train significantly attenuated the induction of depotentiation in CA1 neurons (Fig. 5), and this effect could be replicated by perfusion with 50 µM AP5, 100 µM S-4CPG, 10 µM 2-APB, or 2 µM FK506 (Fig. 7). In our previous study of LFS-induced depotentiation at CA1 synapses (Sugita et al. 2016), we suggested that the activation of calcineurin that occurs during LFS is maintained after LFS by test synaptic inputs applied to CA1 synapses. Therefore, we suggest the following signaling mechanism for depotentiation during and after LFS in postsynaptic CA1 neurons (Fig. 8C): the postsynaptic increase of [Ca2+]i due to the coactivation of NMDARs and group I mGluRs, caused by the delivery of LFS and/or test synaptic stimulation after LFS, might stimulate calcineurin activation in postsynaptic CA1 neurons. Calcineurin has a high affinity for Ca2+/calmodulin complexes (Meyer et al. 1992; Ye et al. 2008). Since IRBIT binds to the regulatory domain of CaMKIIα and inhibits CaMKII kinase activity and since IRBIT and Ca2+/calmodulin complexes bind competitively to the regulatory domain of CaMKIIα (Kawaai et al. 2015), after LFS, the levels of free Ca2+/calmodulin, which is not sequestrated by CaMKII in the presence of IRBIT, may be increased to a point at which calcineurin is activated in the dendritic spines, resulting in the dephosphorylation of AMPARs in postsynaptic cells and the reduction of LTP at CA1 synapses. Recently, Park et al. (2019) studied depotentiation induced by the LFS train (2 Hz for 10 min) in adult rat hippocampal CA1 neurons and found a sensitive form of LTP to the effect of LFS and the effective timing of LFS after LTP induction. In their study, LTP was induced by a compressed theta burst stimulation (TBS) and a spaced TBS induction protocol where the only difference was in the interepisode interval of three bursts (10 sec vs.10 min). They showed a pronounced effect of LFS on the LTP induced by the compressed TBS when the timing between the induction of LTP and the delivery of LFS was from between 5 to 60 min and suggested that a type of transcriptionally independent form of LTP induced by the compressed TBS is sensitive to depotentiation. Since the depotentiation of CA1 neurons is significantly attenuated when the interval between HFS and LFS is 60 min or longer (Fujii et al. 1991; Sugita et al. 2016), the depotentiation of CA1 neurons is considered to be sensitive to the transcriptionally independent form of LTP. Angiotensin II (ANG II) stimulates the renal tubular reabsorption of NaCl by activating Na+/H+ exchanger type 3 (NHE3). In cultured opossum kidney proximal tubule cells, the activation of NHE3 by ANG II is mediated by the binding of IRBIT to NHE3, a process stimulated by CaMKII (He et al. 2010). The addition of ANGII to this cell line increases the binding of IRBIT to NHE3 after 5 min, but the bound IRBIT is released after 45 min, and at least 15 min of ANG II treatment is required to increase NHE3 activity and surface expression (He et al. 2010). Since the depotentiation of CA1 neurons is significantly attenuated when the interval between HFS and LFS is 60 min or longer (Fujii et al. 1991; Sugita et al. 2016) or when the interval between priming HFS and the start of the halt of test synaptic stimulation is 68 min or longer (Fig. 5C), we think it possible that IRBIT that is phosphorylated and released from IP3Rs during and/or after HFS inhibits the further activation of CaMKII during a period of <68 min after HFS in postsynaptic CA1 neurons. Concluding remarks From the results of the present study, we suggest that the postsynaptic cellular events involved in the induction of LFS-induced depotentiation are as follows: (1) depotentiation triggered by priming HFS involves the coactivation of NMDARs and group I mGluRs, which is sustained by test synaptic stimulation after HFS (Fig. 8A); (2) coactivation of NMDARs and mGluRs caused by test synaptic stimulation induces Ca2+ influx through NMDARs/channels, Ca2+ efflux due to Ca2+-induced Ca2+ release from intracellular stores, and Ca2+ efflux through IP3Rs in the dendrites of postsynaptic CA1 neurons (Fig. 8A); (3) an increase of postsynaptic [Ca2+]i due to Ca2+ influx through NMDARs/channels and Ca2+ efflux through IP3Rs results in the formation of free Ca2+/calmodulin complexes that activate CaMKII, while IRBIT released from IP3Rs due to IP3R activation inhibits CaMKII activation further in the dendritic cytoplasm of postsynaptic CA1 neurons (Fig. 8B); (4) in the presence of IRBIT, which binds to the regulatory domain of CaMKIIα and inhibits its kinase activity, free Ca2+/calmodulin complexes are not sequestrated by CaMKII; (5) the moderate increase in the levels of [Ca2+]i caused by the coactivation of NMDARs and IP3Rs during and/or after the 2-Hz LFS results in an increase in the levels of free Ca2+/calmodulin complexes in postsynaptic CA1 neurons; and (6) the levels of free Ca2+/calmodulin complexes, which are not sequestrated by CaMKII in the presence of IRBIT, are increased to a point at which calcineurin is activated in the dendritic spines, resulting in the dephosphorylation of AMPARs in postsynaptic CA1 neurons and the reduction of LTP at CA1 synapses (Fig. 8C). Therefore, we conclude that sustained synaptic activity after LTP induction, which continuously stimulates group I mGluRs and/or IP3Rs in postsynaptic CA1 neurons, is essential for LFS-induced depotentiation in CA1 neurons. In LFS-induced LTP suppression at CA1 synapses, the coactivation of NMDARs and group I mGluRs caused by sustained synaptic activity after priming LFS results in the activation of IP3Rs, which leads to the failure of LTP induction (Fujii et al. 2016). Thus, in hippocampal CA1 neurons, sustained synaptic activity after priming HFS or LFS, which continuously activates postsynaptic IP3Rs, determines the direction of LTP expression after the subsequent application of LFS or HFS. Materials and Methods Ethics approval The animals used were maintained and handled according to the guidelines of the Animal Care and Use Committee of Yamagata University School of Medicine. Slice preparation Male Hartley guinea pigs (3–6 wk old; Funabashi Farm Co.) were decapitated and the hippocampi were removed rapidly and cut into 500 µm-thick transverse slices. The slices were preincubated for a minimum of 1 h at 30°C in a 95% O2/5% CO2 atmosphere in a standard solution ([mM] NaCl, 124; KCl, 5.0; NaH2PO4, 1.25; MgSO4, 2.0; CaCl2, 2.5; NaHCO3, 22.0; and d-glucose, 10.0, pH ∼7.4 at 30°C) before being placed in a 1-mL recording chamber and completely submerged in standard solution perfused continuously at a rate of 2–3 mL/min; the temperature in the recording chamber was maintained at 30°C–32°C. Electrophysiology A bipolar stimulating electrode was placed in the stratum radiatum to stimulate the input pathways to the CA1 neurons. One recording electrode was positioned in the stratum radiatum and another in the pyramidal cell body layer of the CA1 region to record field EPSPs and population spikes (PSs), respectively, and a test electrical stimulus with a pulse duration of 0.1 msec was applied every 20 sec (test synaptic stimulation). The slope of field EPSPs (S-EPSP) and the amplitude of PSs (A-PS) were measured and plotted automatically. At the beginning of each experiment, the strength of the stimulus pulse was adjusted to elicit PSs with an amplitude of 40%–60% of maximal and was fixed at this level. After checking the stability of the S-EPSP and A-PS for more than 15 min, a conditioning stimulus of tetanus or LFS was delivered to induce synaptic plasticity at CA1 neurons. To induce LTP, HFS consisting of 100 pulses at 100 Hz (tetanus) was used. To induce depotentiation, a train of LFS consisting of 1000 pulses at 2 Hz was applied at 30 min after HFS delivery. The mean value of the S-EPSP or A-PS during the 10-min period immediately before HFS delivery was defined as the 100% level, while the other responses were expressed as a mean percentage ± standard error of the mean (S.E.M.) of this control level. To evaluate the control effects of LFS on the synaptic transmission of naïve synaptic pathways, the mean value of the S-EPSP or A-PS during the 10-min period immediately before LFS delivery was defined as the 100% level, while the other responses were expressed as a mean percentage ± S.E.M. of this control level. Changes in responses after HFS or LFS were calculated as follows: (i) the percentage change in the responses after HFS was calculated as (Y/X) × 100; (ii) the percentage change in the responses after LFS was calculated as (Z/X) × 100; and (iii) the percentage reduction in LTP after LFS (depotentiation) was calculated as (Y − Z)/(Y − X) × 100, where X is the averaged value for the 10-min period immediately prior to HFS or LFS, Y is the averaged value at 5–0 min immediately before LFS, and Z is the stable level at 55–60 min after the end of LFS (left panel in Fig. 5A). Using the equation given in (iii), 100 or 0% indicate a complete reduction to the pre-HFS control level or no induction of depotentiation, respectively. When the delivery of test synaptic stimulation at 0.05 Hz was halted for the 19-min period from 1 to 20 min after HFS delivery, for the 20-min period from 0 to 20 min after the end of LFS, or the 20-min period from 30 to 50 min after the end of LFS, the mean magnitude of the S-EPSP or A-PS was not measured during these periods and was omitted from the time course figures (Figs. 1, 3, 5). S-4-Carboxyphenylglycine (S-4CPG) was purchased from Tocris Cookson Ltd., while 2-APB, D, L-2-amino-5-phosphonovalerate (AP5), 1-[N, O-bis(5-isoquinolinesulfonyl)-N-methyl-L-tryosyl]-4-phenylpiperazine (KN-62), and FK506 (tacrolimus) were purchased from Sigma. Each of these test reagents was added to the perfusate when test synaptic stimulation at 0.05 Hz was continued throughout the experiment. AP5, 2-APB, or S-4CPG was applied for the 19-min period from 1 to 20 min after HFS or the 20-min period from 0 to 20 min after the end of LFS. KN-62 was applied for the 10-min period from 9 min before to 1 min after HFS, the 10-min period from 1 to 11 min after HFS, or the 20-min period after LFS. FK506 was applied for the10-min period from 1 to 11 min after HFS or the 20-min period after LFS. Statistical analysis The results were analyzed for statistical significance using the two-tailed Student's t-test, taking a P-value <0.05 as significant. Acknowledgments We thank K. Kaneko for technical assistance with the experiments. This work was supported by JSPS KAKENHI 24500434 and 17K01971. Article is online at http://www.learnmem.org/cgi/doi/10.1101/lm.050344.119. ==== Refs References AlfordS, FrenguelliBG, SchofieldJG, CollingridgeGL. 1993. Characterization of Ca2+ signals induced in hippocampal CA1 neurons by the synaptic activation of NMDA receptors. J Physiol 469 : 693–716. 10.1113/jphysiol.1993.sp019838 8271224 AndoH, MizutaniA, Matsu-uraT, MikoshibaK. 2003. IRBIT, a novel inositol 1,4,5-trisphosphate (IP3) Receptor-binding protein, is released from the IP3 receptor upon IP3 binding to the receptor. J Biol Chem 278 : 10602–10612. 10.1074/jbc.M210119200 12525476 AndoH, MizutaniA, KieferH, TsuzurugiD, MichikawaT, MikoshibaK. 2006. IRBIT suppresses IP3 receptor activity by competing with IP3 for the common binding site on the IP3 receptor. Mol Cell 22 : 795–806. 10.1016/j.molcel.2006.05.017 16793548 BarriaA, MullerD, DerkachV, GriffithLC, SoderlingTR. 1997. Regulatory phosphorylation of AMPA-type glutamate receptors by CaMKII during long-term potentiation. Science 276 : 2042–2045. 10.1126/science.276.5321.2042 9197267 BaumgärtelK, MansuyIM. 2015. Neural functions of calcineurin in synaptic plasticity and memory. Learn Mem 19 : 375–384. 10.1101/lm.027201.112 BearMF, AbrahamWC. 1996. Long-term synaptic depression. Annu Rev Neurosci 19 : 437–462. 10.1146/annurev.ne.19.030196.002253 8833450 Ben-AriY, AnikstzenL, BregestovskiP. 1992. Protein kinase C modulation of NMDA currents: an important link for LTP induction. Trends Neurosci 5 : 333–339. 10.1016/0166-2236(92)90049-E BerridgeMJ. 1993. Inositol trisphosphate and calcium signalling. Nature 361 : 315–325. 10.1038/361315a0 8381210 BlissTVP, CollingridgeGL. 1993. A synaptic model of memory: long-term potentiation in the hippocampus. Nature 361 : 31–39. 10.1038/361031a0 8421494 BootmanMD, CollinsTJ, MackenzieL, RoderickHL, BerridgeMJ, PeppiattCM. 2002. 2-aminoethoxydiphenyl borate (2-APB) is a reliable blocker of store-operated Ca2+ entry but an inconsistent inhibitor of InsP3-induced Ca2+ release. FASEB J 16 : 1145–1150. 10.1096/fj.02-0037rev 12153982 ChenY-L, HuangC-C, HsuK-S. 2001. Time-dependent reversal of long-term potentiation by low-frequency stimulation at the hippocampal mossy fiber-CA3 synapses. J Neurosci 21 : 3705–3714. 10.1523/JNEUROSCI.21-11-03705.2001 11356857 CollingridgeGL, HerronCE, LesterRA. 1988a. Synaptic activation of N-methyl-D-aspartate responses in the Schaffer collateral-commissural pathway of rat hippocampus. J Physiol 399 : 283–300. 10.1113/jphysiol.1988.sp017080 2900332 CollingridgeGL, HerronCE, LesterRA. 1988b. Frequency-dependent N-methyl-D-aspartate receptor mediated synaptic transmission in rat hippocampus. J Physiol 399 : 301–312. 10.1113/jphysiol.1988.sp017081 2900333 FujiiS, SaitoK, ItoK-I, MiyakawaH, KatoH. 1991. Reversal of long-term potentiation (depotentiation) induced by tetanus stimulation of the input to CA1 neurons of guinea pig hippocampal slices. Brain Res 555 : 112–122. 10.1016/0006-8993(91)90867-U 1681992 FujiiS, KurodaY, MiuraM, FuruseH, SasakiH, KanekoK, ItoK-I, KatoH. 1996. The long-term suppressive effect of prior activation of synaptic inputs by low-frequency stimulation on induction of long-term potentiation in CA1 neurons of guinea-pig hippocampal slices. Exp Brain Res 111 : 305–312.8911925 FujiiS, MatsumotoM, IgarahsiK, KatoH, MikoshibaK. 2000. Synaptic plasticity in hippocampal CA1 neurons of mice lacking type1 inositol-1,4,5-trisphoshate receptors. Learn Mem 7 : 312–320. 10.1101/lm.34100 11040263 FujiiS, MikoshibaK, KurodaY, AhmedTM, KatoH. 2003. Cooperativity between activation of metabotropic glutamate receptors and NMDA receptors in the induction of LTP in hippocampal CA1 neurons. Neurosci Res 46 : 509–521. 10.1016/S0168-0102(03)00162-7 12871773 FujiiS, SasakiH, MikoshibaK, YamazakiY, KurodaY, AhmedTM, KatoH. 2004. A chemical LTP induced by co-activation of metabotropic and N-methyl-D-aspartate glutamate receptors in hippocampal CA1 neurons. Brain Res 999 : 20–28. 10.1016/j.brainres.2003.11.058 14746918 FujiiS, YamazakiY, KurodaY, MikoshibaK. 2010. Involvement of inositol-1,4,5-trisphosphate receptors in the bidirectional synaptic plasticity induced in hippocampal CA1 neurons by 1–10 Hz low-frequency stimulation. Neuroscience 168 : 346–358. 10.1016/j.neuroscience.2010.03.033 20347013 FujiiS, YamazakiY, GotoJ, FujiwaraH, MikoshibaK. 2016. Prior activation of inositol 1,4,5-trisphosphate receptors suppresses the subsequent induction of long-term potentiation in hippocampal CA1 neurons. Learn Mem 23 : 208–220. 10.1101/lm.041053.115 27084928 FuruichiT, YoshikawaS, MiyawakiA, WadaK, MaedaN, MikoshibaK. 1989. Primary structure and functional expression of the inositol 1,4,5-trisphosphate-binding protein P400. Nature 342 : 32–38. 10.1038/342032a0 2554142 GriffithLC. 2004. Regulation of calcium/calmodulin-dependent protein kinase II activation by intramolecular and intermolecular interactions. J Neurosci 24 : 8394–8398. 10.1523/JNEUROSCI.3604-04.2004 15456810 HeP, KleinJ, YunCC. 2010. Activation of Na+/H+ exchanger NHE3 by angiotensin II is mediated by inositol 1,4,5-triphosphate (IP3) receptor-binding protein released with IP3 (IRBIT) and Ca2+/calmodulin-dependent protein kinase II. J Biol Chem 285 : 27869–27878. 10.1074/jbc.M110.133066 20584908 HuangCC, LiangYC, HsuKS. 2001. Characterization of the mechanism underlying the reversal of long term potentiation by low frequency stimulation at hippocampal CA1 synapses. J Biol Chem 276 : 48108–48117. 10.1074/jbc.M106388200 11679581 ItoK-I, MiuraM, FuruseH, ChenZ, KatoH, YasutomiD, InoueT, MikoshibaK, KimuraT, SakakibaraS, 1995. Voltage-gated Ca2+ channel blockers, (-Aga IV and Ni2+, suppress the induction of (-burst induced long-term potentiation in guinea pig hippocampal CA1 neurons. Neurosci Lett 183 : 112–115. 10.1016/0304-3940(94)11127-5 7746467 IwasakiH, MoriY, HaraY, UchidaK, ZhouH, MikoshibaK. 2001. 2-Aminoethoxydiphenyl borate (2-APB) inhibits capacitative calcium entry independently of the function of inositol 1, 4, 5-trisphosphate receptors. Receptors Channels 7 : 429–439.11918346 KawaaiK, MizutaniA, ShojiH, OgawaN, EbisuE, KurodaY, WakanaS, MiyakawaT, HisatsumeC, MikoshibaK. 2015. IRBIT regulates CaMKIIα activity and contributes catecholamine homeostasis through tyrosine hydroxylase phosphorylation. Proc Natl Acad Sci 112 : 5515–5520. 10.1073/pnas.1503310112 25922519 LeeH-K, BarbarosieM, KameyamaK, BearMF, HuganirR. 2000. Regulation of distinct AMPA receptor phosphorylation sites during bidirectional synaptic plasticity. Nature 405 : 955–959. 10.1038/35016089 10879537 LindenDJ. 1994. Long-term synaptic depression in the mammalian brain. Neuron 12 : 452–457. 10.1016/0896-6273(94)90205-4 LismanJ. 1994. The CaMkinase II hypothesis for the storage of synaptic memory. Trends Neurosci 17 : 406–412. 10.1016/0166-2236(94)90014-0 7530878 LiuJ, FarmerJDJr, LaneWS, FriedmanJ, WeissmanI, SchreiberSL. 1991. Calcineurin is a common target of cyclophilin-cyclosporin A and FKBP-FK506 complexes. Cell 66 : 807–815. 10.1016/0092-8674(91)90124-H 1715244 MaruyamaT, KanajiT, NakadeS, KannoT, MikoshibaK. 1997. 2APB, 2-aminoethoxydiphenyl borate, a membrane penetrable modulator of Ins(1,4,5)P3-induced Ca2+ release. J Biochem 122 : 498–505. 10.1093/oxfordjournals.jbchem.a021780 9348075 MatsumotoM, NakagawaT, InoueT, NagataE, TanakaK, TakanoH, MinowaO, KunoJ, SakakibaraS, YamadaM, 1996. Ataxia and epileptic seizures in mice lacking type 1 inositol 1,4,5-trisphosphate receptor. Nature 379 : 168–171. 10.1038/379168a0 8538767 MeyerT, HansonPI, StryerL, SchulmanH. 1992. Calmodulin trapping by calcium-calmodulin-dependent protein kinase. Science 256 : 1199–1202. 10.1126/science.256.5060.1199 1317063 MichikawaT, HirotaJ, KawanoS, HiraokaM, YamadaM, FuruichiT, MikoshibaK. 1999. Calmodulin mediates calcium-dependent inactivation of the cerebellar type 1 inositol 1,4,5-trisphosphate receptor. Neuron 23 : 799–808. 10.1016/S0896-6273(01)80037-4 10482245 MikoshibaK. 1993. Inositol 1,4,5-trisphosphate receptor. Trends Pharmacol Sci 14 : 86–89. 10.1016/0165-6147(93)90069-V 8387706 MiyazakiS, YuzakiM, NakadaK, ShirakawaH, NakanishiS, NakadeS, MikoshibaK. 1992. Block of Ca2+ wave and Ca2+ oscillation by antibody to the inositol 1,4,5-trisphosphate receptor in fertilized hamster eggs. Science 257 : 251–255. 10.1126/science.1321497 1321497 MulkeyRM, EndoS, ShenolikarsS, MalenkaRC. 1994. Involvement of a calcineurin/inhibitor-1 phosphatase cascade in hippocampal long-term depression. Nature 369 : 486–488. 10.1038/369486a0 7515479 NakanishiS. 1992. Molecular diversity of glutamate receptors and implication for brain function. Science 258 : 597–603. 10.1126/science.1329206 1329206 NakanishiS, MaedaN, MikoshibaK. 1991. Immunohistochemical localization of an inositol 1,4,5-trisphosphate receptor, P400, in neural tissue: studies in developing and adult mouse brain. J Neurosci 11 : 2075–2086. 10.1523/JNEUROSCI.11-07-02075.1991 1648604 OtaniS, ConnorJA. 1998. Requirement of rapid Ca2+ entry and synaptic activation of metabotropic glutamate receptors for the induction of long-term depression in adult rat hippocampus. J Physiol 511 : 761–770. 10.1111/j.1469-7793.1998.761bg.x 9714858 PalmerMJ, IrvingAJ, SeabrookGR, JaneDE, CollingridgeGL. 1997. The group I mGlu receptor agonist DHPG induces a novel form of LTD in the CA1 region of the hippocampus. Neuropharmacology 36 : 1517–1532. 10.1016/S0028-3908(97)00181-0 9517422 ParkP, SandersonTM, BortolottoZA, GeorgiouJ, ZhuoM, KaangK, CollingridgeGL. 2019. Differential sensitivity of three forms of hippocampal synaptic potentiation to depotentiation. Mol Brain 12 : 30. 10.1186/s13041-019-0451-6 30943994 PeppiattCM, CollinsTJ, MackenzieL, ConwaySJ, HolmesAB, BootmanMD, BerridgeMJ, SeoJT, RoderickHL. 2003. 2-Aminoethoxydiphenyl borate (2-APB) antagonises inositol 1,4,5-trisphosphate-induced calcium release, inhibits calcium pumps and has a use-dependent and slowly reversible action on store-operated calcium entry channels. Cell Calcium 34 : 97–108. 10.1016/S0143-4160(03)00026-5 12767897 PinJ-P, DuvoisinR. 1995. The metabotropic glutamate receptors: structures and functions. Neuropharmacology 34 : 1–26. 10.1016/0028-3908(94)00129-G 7623957 Reyes-HardeM, StantonPK. 1998. Postsynaptic phospholipase C activity is required for the induction of homosynaptic long-term depression in rat hippocampus. Neurosci Lett 252 : 155–158. 10.1016/S0304-3940(98)00496-0 9739984 SkeberdisJL, LanJ, OptizT, ZhengX, BennettMV, ZukinS. 2001. mGluR1-mediated potentiation of NMDA receptors involves a rise in intracellular calcium and activation of protein kinase C. Neuropharmacology 40 : 856–865. 10.1016/S0028-3908(01)00005-3 11378156 SugitaM, YamazakiY, GotoJI, FujiwaraH, AiharaT, MikoshibaK, FujiiS. 2016. Role of postsynaptic inositol 1,4,5-trisphosphate receptors in depotentiation in guinea pig hippocampal CA1 neurons. Brain Res 1642 : 154–162. 10.1016/j.brainres.2016.03.033 27018292 TaufiqAM, FujiiS, YamazakiY, SasakiH, KanekoK, LiJ, KatoH, MikoshibaK. 2005. Involvement of IP3 receptors in LTP and LTD induction in guinea pig hippocampal CA1 neurons. Learn Mem 12 : 594–600. 10.1101/lm.17405 16287718 TokumitsuH, ChijiwaT, HagiwaraM, MizutaniA, TerasawaM, HidakaH. 1990. KN-62, 1-[N, O-bis(5-isoquinolinesulfonyl)-N-methyl-L-tryosyl]-4-phenylpiperazine, a specific inhibitor of Ca2+/Ca2+-dependent protein kinase II. J Biol Chem 265 : 4315–4320.2155222 YamazakiY, SugiharaT, GotoJ-I, ChidaK, FujiwaraH, KanekoK, FujiiS, MikoshibaK. 2011. Role of inositol 1,4,5-trisphosphate receptors in the postsynaptic expression of guinea pig hippocampal mossy fiber depotentiation. Brain Res 1387 : 19–28. 10.1016/j.brainres.2011.02.088 21382354 YamazakiY, FujiiS, AiharaT, MikoshibaK. 2012. Activation of inositol 1,4,5-trisphosphate receptors during preconditioning low-frequency stimulation leads to reversal of LTP in hippocampal CA1 neurons. Neuroscience 207 : 1–11. 10.1016/j.neuroscience.2012.01.045 22330836 YangSN. 2000. Ceramide-induced sustained depression of synaptic currents mediated by ionotropic glutamate receptors in the hippocampus: an essential role of postsynaptic protein phosphatases. Neuroscience 96 : 253–258. 10.1016/S0306-4522(99)00582-5 10683565 YeQ, WangH, ZhengJ, WeiQ, JiaZ. 2008. The complex structure of calmodulin bound to a calcineurin peptide. Proteins 73 : 19–27. 10.1002/prot.22032 18384083

PMC007xxxxxx/PMC7055006.txt

==== Front J Cell Biol J Cell Biol jcb The Journal of Cell Biology 0021-9525 1540-8140 Rockefeller University Press 31968357 jcb.201904090 10.1083/jcb.201904090 Tools Biochemistry Systems and Computational Biology Organelles Trafficking A reference library for assigning protein subcellular localizations by image-based machine learning Image-based assignment of subcellular localization Schormann Wiebke 1* Hariharan Santosh 1* https://orcid.org/0000-0002-9266-7157 Andrews David W. 123 1 Biological Sciences, Sunnybrook Research Institute, Toronto, Canada 2 Department of Biochemistry, University of Toronto, Toronto, Canada 3 Department of Medical Biophysics, University of Toronto, Toronto, Canada Correspondence to David. W. Andrews: david.andrews@sri.utoronto.ca * W. Schormann and S. Hariharan contributed equally to this paper. 02 3 2020 22 1 2020 219 3 e20190409016 4 2019 30 9 2019 15 12 2019 © 2020 Schormann et al. 2020 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). Schormann et al. provide a reference library of confocal micrographs of key organelles in live epithelial cells as landmarks and a derived feature set that can be used to assign protein localization throughout the secretory pathway and to key organelles via a quantitative unbiased image-based classifier. Confocal micrographs of EGFP fusion proteins localized at key cell organelles in murine and human cells were acquired for use as subcellular localization landmarks. For each of the respective 789,011 and 523,319 optically validated cell images, morphology and statistical features were measured. Machine learning algorithms using these features permit automated assignment of the localization of other proteins and dyes in both cell types with very high accuracy. Automated assignment of subcellular localizations for model tail-anchored (TA) proteins with randomly mutated C-terminal targeting sequences allowed the discovery of motifs responsible for targeting to mitochondria, endoplasmic reticulum, and the late secretory pathway. Analysis of directed mutants enabled refinement of these motifs and characterization of protein distributions in within cellular subcompartments. Graphical Abstract Canadian Institutes of Health Research http://dx.doi.org/10.13039/501100000024 FDN 143312 Canada Research Chair http://dx.doi.org/10.13039/501100001804 ==== Body pmcIntroduction Subcellular localization of proteins is a key feature of eukaryotes. Understanding subcellular localization has long been a goal for cell biologists interested in basic mechanisms of protein sorting and for understanding the generation of organelles with distinct compositions and morphologies. Moreover, there is much to be learned about disease processes, mechanisms of signal transduction, and cellular metabolism that is directly linked to subcellular localization. Traditionally, subcellular localization of a protein of interest has been assigned by visual comparison with one or more proteins of known localization (antibody based or fluorescence protein tagged) or with organelle-specific dyes (e.g., Mitotracker) in fluorescence microscope images by an experimentalist. However, human visual inspection is prone to both drift and bias. Therefore, machine learning tools have been developed to automate the analysis of subcellular localization. Early classifiers built to distinguish subcellular structures in fluorescence micrographs in HeLa cells based on features tailored specifically for subcellular location studies functioned well with small datasets (Boland and Murphy, 2001). In addition to the newly designed region of interest (ROI) features, the well-known textural features by Haralick et al. (1973) and Zernike moments features (Zernike, 1934) were used. A new set of statistical features called threshold adjacency statistics (TAS) is faster to calculate than other commonly used statistical features (Hamilton et al., 2007), and also shows good performance (Nanni and Lumini, 2008). As a result, several supervised classification strategies to distinguish subcellular structures of the main subcellular locations (e.g., cytoplasm, nucleus, Golgi apparatus, mitochondria, and ER) have been published (Hamilton et al., 2007; Conrad et al., 2004; Li et al., 2012). A major limitation to using automated methods of image analysis for determining subcellular localization is the paucity of high-quality images with explicit annotations. This is particularly problematic for proteins that transit between organelles or those at steady-state that are located at more than one subcellular location. For such proteins, accurate annotation based on biological experiments or images of subcellular distributions can be very difficult. One approach to dealing with this problem has been to use semi-supervised methods to assign subcellular localization from lower-quality data together with multi-label classification (Xu et al., 2016). A similar approach was used to detect mis-localization of proteins in cancer cells using images from the human protein atlas (Xu et al., 2015, 2019). However, in these cases, automated analysis was limited to detection of relatively coarse localization changes such as between the cytoplasm and nucleus or mitochondria. More recently, deep learning approaches alone and together with crowd sourcing have been used to tackle the problem of classifying subcellular localization of proteins in yeast (Chong et al., 2015; Pärnamaa and Parts, 2017) and in human cell lines (Sullivan et al., 2018). A major advance in these studies was the use of hundreds of thousands of images to overcome differences inherent in the data deposited in repositories such as the human protein atlas as well as the cell to cell variations inherent in normal biology. Using multiple markers in the same cell, it was possible to automatically classify a number of subcellular structures, particularly subnuclear spot types (Sullivan et al., 2018); however, automated identification of the compartments within the secretory pathway has not been achieved. Improving automated assignment of localization requires a large dataset of high-quality images and an alternative approach to the problem of proteins having multiple subcellular localizations. Our approach was to generate a reference library of 789,011 and 523,319 optically validated landmark-based localization images in murine and human epithelia, respectively. Numerical analyses identified 160 features most useful for assignment of localization while minimizing the effects of expression levels. Rather than forcing assignment to predefined organelles, we classify images by similarity to landmarks that may themselves localize to multiple destinations. We then used this library of images to examine localization of model tail-anchor proteins. We show that automated analysis outperformed even highly trained human observers and enabled the identification multiple morphologically distinct distributions of TA proteins. Our results demonstrate the utility of the reference library of images, derived features, and machine learning to provide unbiased assignment of subcellular localizations with high accuracy in individual living cells. Results Distinct subcellular phenotypes identified from multidimensional descriptors of cell images of landmark proteins Landmark proteins were generated by fusing EGFP to proteins selected from previous reports establishing localization at specific subcellular organelles (Table 1). A reference library used for automated assignment of subcellular localization in individual cells was generated by calculating multidimensional descriptors of micrographs of cells expressing these proteins. The reference library contains 789,011 and 523,319 quality-validated cell images of the EGFP-landmark fusion proteins expressed in normal murine mammary gland (NMuMG) cells and in MCF10A cells, respectively. For organelles with recognized subdomains (mitochondria, ER, and Golgi apparatus) multiple landmark proteins were used (Fig. 1 A and Table 1). Figure 1. Accurate assignment of subcellular localization from images of NMuMG and MCF10A cells expressing EGFP-tagged landmark proteins by RF classification. (A) Representative microscope images of cells expressing the EGFP-tagged landmark proteins (Table 1). Scale bars, 25 µm. (B) Confusion matrices after RF classification of murine (NMuMG) and human (MCF10A) cells from landmark data not used in training the classifier, averaged over five independent classifications. White and red numbers show the percentage of cells assigned the highest and second highest classification landmark, respectively. SD < 1% (see Materials and methods). In addition to forward transit, proteins recycle between different organelles or within compartments of the same organelle (e.g., Golgi), which increases the heterogeneity of distribution. To capture the varying phenotypes caused by mobility of TA proteins, thousands of images were acquired for each landmark (Table 1). Prior to feature extraction, all images undergo automated quality control to remove out-of-focus cells and segmentation artifacts (Fig. S1 and Materials and methods). Figure S1. Image processing pipeline. (I) Image segmentation of cell images (scale bar, 25 µm) includes identifying the nucleus, cytoplasm, ROI, and spots. Acapella software by Perkin Elmer was used for all steps. (II and III) Prior to feature extraction, all images undergo a quality control process to (II) remove out-of-focus images by using a specific classifier (see Materials and methods) and (III) remove incorrectly segmented cell images by computing two parameters (see equations for R1 and R2 in Materials and methods). Representative images of proper cell segmentation (nuclei and cytoplasm) and poor cell segmentation (scale bars, 15 µm). (IV) Selection of optimal cell images improves classification sensitivity. (V) Final feature calculation included 160 morphology and texture features. For random forests classification, only features calculated from the EGFP channel were used. (VI) Sample training images and images of cells classified to the training landmark MAO (outer mitochondrial membrane). Scale bars, 15 µm. To visualize in two dimensions the separation of these landmark proteins in 160-dimensional feature space, we used t-distributed stochastic neighbor embedding (t-SNE) projection (Van der Maaten and Hinton, 2008) to compress the data. As displaying over 500,000 multidimensional points in two dimensions is not meaningful, a subset comprising the 200 nearest-neighbor data points to the centroid of each landmark was plotted. Landmarks in the dataset are clearly separated from each other, with only a few points distributed to other areas within the t-SNE landscape (Fig. 2 A). As another way to represent the distribution of the data and demonstrate the separation of landmarks, data points for individual landmarks were over-clustered using Phenograph (Levine et al., 2015; Fig. 2 B). As a third measure of the relationships between the multidimensional classes, the Euclidean distances between landmark centroids were tabulated and plotted as a heatmap (Fig. S2). Figure 2. Images of cell landmarks are well separated in 160-dimensional feature space. (A) Two-dimensional visualization of 160-dimensional image data for both murine and human cells from individual landmarks using the t-SNE algorithm. Each landmark is represented by the 200 cell images (dots) closest to the centroid. (B) Two-dimensional t-SNE projection of landmark clusters belonging to the secretory pathway (mouse image library). To represent the 160-dimensional space encompassed by the landmarks with a number of data points practical for dimension reduction by t-SNE, individual landmarks were overclustered using Phenograph by setting the number of neighbors to five (see Materials and methods). Figure S2. Euclidean distances between centroids of the landmarks in NMuMG cells. Top panel: Heat map of distances colored as indicated to the right. Bottom panel: Numerical data. Distances are unitless because they represent Z-scored multidimensional data but can be thought of as the number of SDs between landmarks. An ensemble of Random Forest (RF) classifiers generated from a subset (less than half) of the landmark images was used to assign localization of images not used for training to generate confusion matrices (Fig. 1 B). The high accuracy of the resulting classification of individual cells (78%) demonstrated that even though many of the images for different landmarks appear visually similar, a classifier based on the image features table accurately separates and identifies fluorescent proteins targeted to different subcellular organelles for individual cells without a colocalization marker (Fig. 1). The confusion matrix indicates that even for the two most similar and visually indistinguishable landmarks, monoamine oxidase A (MAO) and cytochrome c oxidase (CCO; outer and inner mitochondrial membranes, respectively), <15% of cells were misclassified to another single landmark (Fig. 1 B). Control experiments in which we deliberately impair the classifier by training on images of the highest or lowest intensity and then classify images of different intensity levels demonstrated that the expression level (intensity) of the landmarks neither contributed to nor confounded classification (Fig. 3). Indeed, the classification accuracy drops only in the extreme cases such as when a classifier trained using images of only the lowest intensity (highest shot noise) was used to classify images from the highest-intensity quartile (Fig. 3) and the errors introduced were confined to the most closely related patterns. This effect is negligible in our other experiments as all intensity levels were used in training. Furthermore, the SD in the assignments across 20 classifiers was <1% of cells and is therefore not specified in the figures (see Materials and methods). Taken together, these results suggest that an assignment of >20% of cells to a specific landmark location is a conservative definition for significance. Figure 3. Classification results for intensity quartile analysis. Murine image data for landmarks, indicated to the left and by the color as in Fig. 2 A, were divided into quartiles based on the average pixel intensity within the cell region. RF classification was performed using randomly selected images from the quartile indicated above the panel for training and the rest of the data for testing. The thickness of the vertical bars indicates the fraction of the cells assigned to each landmark per quartile. Each vertical line corresponds to one quartile (quartile 1 to quartile 4, left to right). The dots serve as a location guide that is color-coded to facilitate identification of the row and columns. At steady-state, many of the landmarks have some cells reproducibly assigned to other localizations. This type of “mis-localization” to other organelles actually reflects normal organelle and protein dynamics. Thus, assignment of a protein to a particular subcellular location is never realistically 100%. To our surprise, different landmarks ostensibly targeted to the same organelle were efficiently discriminated by the RF classifier. For example, the images of EGFP fused to the TA sequences of the ER localized TA proteins Bcl-2 interacting killer (Bik; Germain et al., 2002) and cytochrome b5 (Cytb5) are visually similar to each other and to images of EGFP fused to Calr-KDEL, another ER landmark (Fig. 1 A). However, the cell images were well-separated from each other in multidimensional space in both cell lines (Fig. 1 B and Fig. 2 A). As a result, in NMuMG cells, only 8% of Cytb5-expressing cells were classified as Bik (Fig. 1 B); 5% of Bik-expressing cells were classified as Cytb5, while 9% were classified as the nearest Euclidean neighbor, phosphatidylserine synthase 1 (PTDSS1; mitochondria-associated ER membrane [MAM] and ER; Table 1); and 9% were classified as the ER–Golgi intermediate compartment (ERGIC) protein 53. Classification of Bik localization to these landmarks in a small fraction of cells is highly plausible as both compartments are related to the ER. Importantly, only 2% of cells expressing Cytb5 were classified as VAMP5, the nearest Euclidean neighbor. Thus, images that are not assigned to their own landmark are assigned according to features representing their biology, and not simply by centroid Euclidean distances. Controls (Fig. 3 and described below) demonstrated that intensity differences do not account for the different classifications. It is possible that classification differences for landmarks ostensibly located at the same organelle represent different steady-state distributions within the organelle, and different extents to which the landmarks localize to different organelles, as seen previously (Xu et al., 2016). As discussed below, different steady-state distributions within an organelle may also represent functional or physical subdomains. Because the different landmarks for the same organelle are well separated, we refer to the various subcellular locations by the landmark protein rather than the organelle name. In this way, the variability of localization for each landmark is preserved in the classification. When organelle names are used, it is to designate multiple landmarks in aggregate. The localization of novel proteins and dyes can be identified across species using reference libraries To validate our multiparametric definition of the subcellular landscape, we used a new set of marker proteins previously reported to target to specific organelles and classified them using our existing set of NMuMG landmarks (Fig. 4 A). The ER-specific dye (BODIPY-thapsigargin) was predominantly classified with the ER marker Calr-KDEL (47%), but also with ER markers Cytb5 (15%) and Bik (20%) for a total ER classification of 82%. Similarly, the cis-Golgi–localized TA protein Golgi SNAP receptor complex member 2 (Membrin) was classified as mostly Golgin84 (62%) with a component assigned to the Golgi protein GalT (31%). Moreover, the mitochondrial stain Mitotracker and the mitochondrial matrix protein ornithine transcarbamylase (OTC) were both classified with the mitochondrial landmarks MAO (55% and 27%, respectively) and CCO (34% and 62%, respectively). In addition, the classifier successfully assigned outer membrane mitochondrial localization to the outer mitochondrial membrane protein Harakiri fused to the red fluorescence protein mLumin (66% outer mitochondrial MAO and 25% inner membrane CCO). The recycling endosome marker (Rab11) was classified with Rab5 (42%). Figure 4. Validation of the NMuMG RF classifier using images of cells expressing novel landmarks not used for training and images of MCF10A cells expressing selected landmarks. (A) EGFP or mLumin fusions to targeting sequences from proteins with well-characterized localizations (Table 1) or organelle-specific dyes were used as queries for classification. (B) MCF10A cells expressing landmarks were used as queries using the NMuMG classifier. (C) Images of NMuMG and MCF10A cells expressing EGFP-tagged Golgin84. Numbers indicate the percentage of cells assigned to the most prevalent assigned landmark. Scale bars, 25 µm. In addition to analyzing a new set of marker proteins, we also tested the performance of the NMuMG cell classifier by using human MCF10A cells expressing five of the landmarks (Rab5, ERGIC53, LAMP-1, Bik, and Golgin84). These proteins were stably expressed in MCF10A cells and imaged, and the resulting micrographs were analyzed using the classifier derived from images of NMuMG cells expressing all the landmarks. The classifier correctly assigned the localization of the landmark proteins expressed in MCF10A cells (Fig. 4 B). However, the ER protein Bik was mostly classified as the other ER-localized TA protein, Cytb5. Moreover, only 41% of Golgin84 in MCF10A was assigned to the NMuMG Golgin84 compartment, suggesting that there is a difference in the morphology of these compartments between the two cell lines. This hypothesis was verified by visual comparison of the corresponding images in the two cell lines (Fig. 4 C). This result demonstrates that subtle changes in morphology can be detected by the classifier, suggesting utility for a wide variety of genetic and chemical perturbation studies. However, for automated classification in more divergent cell types, it will be necessary to build a new library of landmark images. We generated a second landmark library in the human cell line MCF10A (Fig. 1) that includes additional landmarks (Table 1) and resulted in accurate assignment (77%) of subcellular localizations for images of individual cells not used in training (Fig. 1 B). Comparison of the five query proteins expressed in MCF10A revealed that as expected, the accuracy was somewhat (on average <20%) higher when assigned using the MCF10A instead of the NMuMG library. As expected, the improvement was dramatic for Golgin84 (41% to 84%; Fig. 4, B and C; and Fig. 1 B, respectively). These results demonstrate that our analysis captures the subcellular landscape with sufficient resolution to accurately assign the localization of proteins and organelle specific dyes not used for training, regardless of whether they expressed RFP or GFP fusions or in human or murine epithelial cells, attesting to the robustness of the approach. Furthermore, the analysis is sensitive enough to separate landmarks ostensibly targeted to the same organelle (e.g., Cytb5, Bik, and Calr-KDEL) and thereby identify different protein distributions within organelles (Cytb5 and Bik) and/or discriminate resident versus recycling proteins (Cytb5 vs. Calr-KDEL). Image-based analysis of the determinants of subcellular localization for TA proteins As an example of one of the uses for the localization libraries, we systematically examined the targeting behavior of TA proteins. In multiple cell types, VAMP1 targets to the ER and then transits through the secretory pathway to the cell surface (Raptis et al., 2005; Chen and Scheller, 2001). When EGFP-TA was expressed in NMuMG cells, the images of 62,000 of 104,000 cells (∼60%) were classified as VAMP5 localization. Although VAMP5 has been reported to be located primarily at the plasma membrane (Hong, 2005) due to the dynamic trafficking of proteins, images of the VAMP5 landmark in our library show protein localization throughout the secretory pathway. Thus, our results indicate that EGFP-TA has a distribution compatible with the biology of the protein the TA sequence was derived from. We used EGFP-TA to investigate the features of the TA sequence that determine the specificity of localization within the secretory pathway and to other intracellular membranes. To analyze one region of the sequence systematically, we performed random mutagenesis on the codons for the last five amino acids of the TA sequence (IYFFT), which constitute the C-terminal sequence (CTS). Sequencing revealed 995 unique sequences, each of which was individually expressed in NMuMG cells. As expected, when classified using the reference library, many of the mutants were assigned VAMP5 localization. However, a few were localized primarily at the ER (n = 8) or mitochondria (n = 13) and were therefore selected for further analysis. Our initial focus was the mutants localized at mitochondria because the requirements for TA targeting to this organelle have been analyzed extensively, yet a clearly defined consensus has not emerged. Targeting to the mitochondria is determined by positions of amino acid properties in the CTS Mitochondria are a well-known targeting destination for TA proteins that play a crucial role in apoptosis (Cory and Adams, 2002), protein import (Horie et al., 2002), organelle and vesicle fission and fusion (Scott and Youle, 2010), and other functions. While several approaches have been used to decipher the sequence requirements for localization, it is still unclear what comprises a TA mitochondrial targeting sequence. 12 of the 13 CTS sequences that resulted in mitochondrial localization of EGFP-TA were five amino acids long. The one sequence with a four–amino acid CTS due to random incorporation of a termination codon was assigned to the outer mitochondrial membrane (MAO), but not examined further. Of the five amino acid CTSs, eight were assigned only to mitochondria, all primarily to the outer mitochondrial membrane localization MAO landmark as expected for proteins anchored to mitochondria by a TA sequence. For three of these mutants, a smaller fraction was also assigned as the inner membrane landmark CCO. Of the four sequences not primarily assigned to MAO, none was primarily assigned to CCO. Even though 12 sequences is a small number, it was possible to visualize amino acid enrichment using a pLogo representation (O’Shea et al., 2013). The most overrepresented amino acid in sequences that localized EGFP-TA to mitochondria was arginine (R) at CTS position 2 (Fig. 5 A). Moreover, lysine (K) at CTS position 1 was also significantly overrepresented. There was a high occurrence of arginine at positions 3 and 5, but individually this did not reach statistical significance (Fig. 5 A). By carrying out an enrichment analysis for the number rather than position of positive charges, we observed that mitochondrial localization of mutants with at least three positively charged residues (K or R) was statistically significant (three positive charged residues, P = 5.6 e−5; four positive charged residues, P = 4.9 e−9, Fig. S3 A). This result is consistent with reports that positively charged residues are frequently observed in the CTS of mitochondrial TA proteins (Rapaport, 2003). However, previous publications (Marty et al., 2014; Rapaport, 2003) suggested that a net charge of the CTS region of +2 or greater is sufficient to target TA sequences to mitochondria. In contrast, for the TA protein Fis1, it has been reported that at least four positive charges are required for targeting to mitochondria (Rao et al., 2016). With our large datasets of random mutants, we can evaluate these rules systematically by examining sequences that target EGFP-TA to mitochondria and how frequently sequences that match the consensus do not target to mitochondria. When we examined the frequency of EGFP-TA mutants with five-residue CTS sequences with KR or RR as the first two amino acids, or net +2 or greater charges in the CTS targeting to mitochondria, the specificity of these simple rules was 0.86 or higher. However, the sensitivities and positive predictive values (PPV) are low (table in Fig. 5 A). Only 12 out of 133 +2 or greater charged CTS sequences (9.0%) targeted EGFP-TA to the mitochondria. Previous studies have not reported the frequency of sequences that adhere to the predicted rules but do not target as predicted, presumably due to the difficulty of manually assigning localization for large numbers of mutants, a task that is easily achievable with our automated reference library. Figure 5. Targeting to the mitochondria is determined by both the position and number of positive charges in the CTS. (A) Representation using pLogo of over- and underrepresented amino acids at each position of the CTS derived from all the sequences assigned to one or both of the classes of mitochondrial markers (MAO and CCO). The red horizontal bars correspond to P = 0.05. (B) Log-likelihood calculated using a PWM for all the sequences with at least two net positive charges. Bar height indicates log-likelihood, while the bar color indicates the organelle assignment by image classification. Classification threshold for the PWM (dotted line) based on the obvious breakpoint in the sequences (RHRAN). Due to the small number of EGFP-TA mutants that localized at mitochondria, mutants (underlined) were designed to test the performance of the PWM and simple rules with sequences not used for training. The table compares the performance of the different simple rules. X is any amino acid. + represents positively charged amino acids in the CTS. Figure S3. Amino acid enrichment within CTS. (A) Statistical enrichment for the number of positive charges in the CTS of EGFP-TA mutants assigned to different localizations. A hypergeometric distribution was used to calculate statistical enrichment. Asterisks highlight statistical significance at α 0.05 after Bonferroni correction. The columns indicate individual landmarks or organelles defined by merging landmarks as ΔTMD-VAMP1 (whole cell), ER (Cytb5, Calr-KDEL, Bik), ERGIC (ERGIC53), endosome (Rab5, Rab7), Golgi (Golgin84, GalT), mitochondria (MAO, CCO), MAM (PTDSS1), VAMP5, and VAMP2. (B) Statistical enrichment in number of negative charges within the CTS for EGFP-TA mutants assigned as localized at the indicated organelles. (C) Statistical enrichment for hydrophobic length added by amino acids in the CTS for EGFP-TA mutants assigned as localized at the indicated organelles. Hydrophobic length is the number of hydrophobic amino acids in the CTS before the first hydrophilic amino acid. Numbers indicate the negative log of the P value for significant enrichments (length 0, mitochondria; length 5, Golgi). Significance cut-off for negative log of P value after Bonferroni correction is 6.98. To further test the importance of positive charges, we generated mutants with a high proportion of positive amino acids (e.g., RRRNR, QRRNR, TRRNR, and SRRNR), all of which contain at least three positive charges and the over-represented R at position 2 (Fig. 5 B). The micrographs of EGFP-TA with CTS sequences RRRNR and QRRNR were assigned 89% and 80% mitochondrial localization, primarily to outer mitochondrial membrane localization (MAO, 65% and 62%), respectively. However, image-based classification assigned the mutants SRRNR and TRRNR to the ER landmark Calr-KDEL and Golgi, respectively, clearly indicating that three positive charges with an R at position 2 is not sufficient to target EGFP-TA to mitochondria. To generate a rule for predicting localization of EGFP-TA proteins at mitochondria, we used the sequence data to generate a position weight matrix (PWM) based on amino acid properties. The PWM enabled calculating a likelihood value for localizing at mitochondria for all of the mutant sequences with at least two positive charges (Fig. 5 B). Visual inspection of this data suggests that a threshold of 5.68 (equal to that of the sequence RHRAN) optimally identifies CTSs that target EGFP-TA to mitochondria. With that threshold, the PWM correctly assigns the mutants RRQVH and RHRAN that are missed by the other rules. Moreover, it correctly assigns the designed mutants RRRNR and QRRNR to mitochondria and rejects the SRRNR and TRRNR sequences. The PPV, sensitivity, and specificity demonstrate that the PWM predictive rule clearly outperformed any of the other rules (Fig. 5, table). Thus, the locations of the biochemical properties of the amino acids contribute to the targeting signal within the CTS that specifies mitochondrial outer membrane localization. To evaluate a CTS sequence predicted by the PWM to target to mitochondria in the context of a different protein, we examined the TA sequence of Bcl-2 fused to EGFP. Wild-type EGFP-TA-Bcl-2 targets to the ER, MAM, and mitochondria, as cell images were classified 45% Calr-KDEL, 20% PTDSS1, and 17% MAO (Fig. 6), consistent with previous reports demonstrating Bcl-2 localization at both ER and mitochondria (Zhu et al., 1996). Localization at multiple organelles also demonstrates that the TA sequence of Bcl-2 is permissive for insertion into multiple membranes. Figure 6. CTS motifs that target the TA-Bcl-2 sequence to different landmark localizations. Heat map: RF classifier assignments of localization according to the scale below. The most frequent assignment for each protein is indicated as a percentage within the heat map. Lower panels: Sample micrographs of the indicated EGFP-fusion proteins expressed in NMuMG cells. Scale bar, 25 µm. After removing the SHK sequence that constitutes the CTS of Bcl-2 (Henderson et al., 2007), the protein Bcl-2-ΔSHK was localized almost exclusively to the ERGIC compartment, confirming that the CTS can determine localization for this protein. Replacing the SHK sequence with the CTS sequence KRRNR generated the protein EGFP-TA-Bcl-2-ΔSHK-KRRNR, which, as predicted from the PWM, was classified exclusively (96%) as MAO (outer mitochondrial membrane). Furthermore, when the CTS of EGFP-TA-Bcl-2 was replaced with the sequences FPCVN or WTNFK that localized EGFP-TA to the ER, the resulting proteins were primarily classified as targeting to the ER (~60% of the cell images). Similar to the EGFP-TA versions, most individual images were classified as Bik, but some were assigned to the other ER subdomains defined by Cytb5 and Calr-KDEL (Fig. 6). Taken together, these results indicate that for permissive TA sequences, targeting to the mitochondrial outer membrane can be achieved by CTS sequences defined by the PWM. The PWM-defined motif is the first demonstration that the position of amino acid properties within the CTS rather than sequence identity determines mitochondrial localization. The ensemble of RF classifier correctly assigned cell images of several TA proteins that were not used in training to the outer mitochondrial rather than inner mitochondrial membrane. These included Harakiri, EGFP-TA-Bcl-2-ΔSHK-KRRNR, EGFP-TA-RRRNR, and EGFP-TA-QRRNR. This is remarkable considering a human observer would be unable to accurately determine outer from inner mitochondrial localization from the images (Fig. S4). Consistent with the classifier efficiently distinguishing localization at the outer and inner mitochondrial membrane, the confusion matrix (Fig. 1 B) indicates <15% overlap between targeting assignments for MAO and CCO. An alternative explanation is that an intensity difference between the inner and outer mitochondrial membrane protein landmarks (CCO and MAO) was sufficient to result in classification of queries as MAO based on intensity despite our selecting features less influenced by intensity. To test this hypothesis, individual cells were selected for training that expressed CCO or MAO with a similar defined intensity range (Fig. 7). This intensity filter was also applied to cell images of OTC and some of the EGFP-TA mutants that targeted to the mitochondria for use as a query set. The classification result (Fig. 7) of this image data resulted in all of the tail-anchored proteins correctly assigned as localized in the outer membrane, while OTC was correctly assigned localization to the inner membrane. Thus, the classifier accurately distinguished inner and outer mitochondrial membranes for six query proteins not used in training, independent of intensity (Fig. 7, heatmap). Figure S4. Images of different mitochondrial localized proteins correctly assigned by classification are not reliably distinguishable visually. Representative cell images (NMuMG) expressing MAO and CCO (top section, scale bars, 25 µm). Selected single-cell image of each mitochondrial landmark is magnified (middle section, scale bars, 10 µm; inset scale bars, 1 µm). Representative images of mutants classified as MAO (bottom lower section, scale bar, 25 µm). Figure 7. Image classification as outer or inner mitochondrial membrane does not depend on image intensity. Intensity graphs for images randomly selected from the full intensity range or the intensity range of 300–500 arbitrary intensity units plotted as density to facilitate comparison as continuous curves. Data are for the mitochondrial landmarks (MAO, CCO) and for the query proteins including EGFP-TA mutants and images of OTC. Heat map shows classification result for images with full vs. 300–500 intensity range (see Materials and methods). ER localization and secretory pathway motifs can be identified from a sparse dataset The 995 mutants constitute a very sparse dataset representing <0.03% of the ∼3.36 million possible five–amino acid CTS sequences. The more specific the requirements are for a targeting sequence for any particular subcellular location, the fewer examples there will be in a sparse dataset. As a consequence in our dataset, even a single sequence might represent a class of related sequences that localize similarly (a motif). To test this hypothesis, we examined further the eight EGFP-TA mutants that were assigned only ER localization. These sequences already suggest that there are sequence dependencies for the localizations defined by the landmarks Bik, Cytb5, and Calr-KDEL. Of the eight sequences, three (FPCVN, WTNFK, and DPTDS) localized EGFP-TA primarily (45–59%) to the localization defined by Bik, while the other five sequences (DEPGH, PEHVS, PKWVT, PSNHQ, and RVRPG) were assigned (43–55%) to the localization identified by Cytb5. Consistent with those representing distinct distributions within or subcompartments of the ER, none of these proteins were assigned significant localization to any other locations, including the other ER landmarks (Fig. S5 B). Figure S5. Assignment of the localizations of “CVN” mutants to different distributions within the ER. (A) Random forests classification results for “CVN” mutants to localizations defined by the landmarks Cytb5, Calr-KDEL, and Bik. Localization of at least 20% of the cell images is highlighted by numbers and gray shades. Minus (−) indicates the mutant was not localized in the post-ER secretory pathway, and plus (+) designates the mutant was also assigned localization within the post-ER secretory pathway (localization, see Fig. 8 for details). (B) Random forests classification results for EGFP-TA mutants that are restricted to either only Cytb5 or Bik localization. (C) EGFP-TA mutants classified as ER-localized independent of intensity. Random forests classification results for selected “CVN” mutants using full intensity range compared with limited range (300–500). Localization is indicated by gray shades, and numbers are provided for localizations of at least 20% of the cell images. To assess the importance of individual amino acids in determining ER localization, we chose one mutant, FPCVN, for further study; because of the 995 sequences, it is the only one ending in CVN. Images of this mutant were assigned 61% to the ER landmarks (46% Bik localization) with <20% assigned to any other location (Fig. 8 and Fig. S5 B). Figure 8. The first two CTS amino acids determine localization of EGFP-TA at ER or enable transit to the plasma membrane for a CTS ending with CVN. RF classification of localization for EGFP-TA mutants with CTS sequences derived from the ER localized mutant FPCVN arranged by the average hydrophobicity (Kyte-Doolittle) of positions 1 and 2. Assignment of the cell images (≥20%) for any one mutant to a single localization is indicated numerically and by gray shade. Localization of <20% of the cells is indicated by gray shade only. Organelles are defined as follows: ER (Cytb5, Calr-KDEL, Bik), ERGIC (ERGIC53), Golgi (Golgin84, GalT), and endosome (Rab5, Rab7). The number of cells analyzed for each mutant is indicated to the right. To determine if the CVN sequence is part of a motif, the first two residues of the FPCVN sequence were mutated to create a new set of CTS sequences on EGFP-TA that differ by one or two amino acids. We did not design “CVN” mutants with negative charges because mutants without negatively charged amino acids are more likely to escape the ER/post-ER compartments and localize at the plasma membrane (Fig. S3 B). Previous work on membrane proteins suggests that the length of the transmembrane domain (TMD) is a key feature that determines localization to the Golgi versus plasma membrane (Sharpe et al., 2010). To test this notion further, the changes were selected to alter the hydrophobicity at positions 1 and 2, thereby potentially extending the length of the hydrophobic region of the TA sequence by up to four residues as the C and V of the CVN sequence are also hydrophobic. Consistent with our sparse data-motif hypothesis, 14 of the 22 “X-X-C-V-N” sequences resulted in mutants that were classified as localizing primarily to the ER (Cytb5, Bik, or Calr-KDEL). This is a remarkable result considering that only 8 of 995 random mutants were localized similarly. To demonstrate that assignment to ER localization was not driven by image intensity, we repeated the analysis with selected images for training and testing with similar distributions of image intensities as shown above (Fig. 7). Restricting the distribution of intensities made no significant difference to the image classification results (Fig. S5 C). Plotting the average hydrophobicity (Kyte and Doolittle, 1982) of the two positions (Fig. 8) demonstrated that sequences with hydrophobicity <0.6 (i.e., hydrophilic) together with CVN constitute a new motif for targeting EGFP-TA to the ER. All of the mutants with this motif were assigned significant localization at the ER with variable localization also at ERGIC and endosomes. Of the seven sequences with intermediate hydrophobicity (>0.6 and <1.7), four (FPCVN, SLCVN, GLCVN, and SVCVN) were retained in the ER. The exceptions (LRCVN, LWCVN, and LGCVN) that were assigned significant VAMP5 localization demonstrate that hydrophobicity is not required to progress through the secretory pathway. Of the seven mutants with hydrophobicity (1.7–1.8), LGCVN, WICVN, and IWCVN were all >48% assigned VAMP5, while GLCVN and SVCVN were primarily assigned to the ER with only partial localization at endosomes and no significant localization with VAMP5. Furthermore, the sequences LWCVN, WICVN, IWCVN, and CICVN that exhibited the strongest preference, >50% assigned localization to the VAMP5 compartment, ranged in CTS sequence hydrophobicity from modest (1.45) to high (3.8). Finally, some mutants suggest that rather subtle sequence features are involved in retention or export from the ER. For example, RWCVN was assigned to ER, yet switching the order of the first two amino acids to generate the CTS sequence WRCVN resulted in assigned localization throughout the secretory pathway with equivalent numbers of cell images assigned to ER (29%) and endosomes (30%). Thus, while our results confirm a weak correlation between hydrophilicity and retention at the ER, it is clear that hydrophobicity is not the only driving force for sorting to or exclusion from the plasma membrane or endosomes. To our surprise, some of the mutants with ER-assigned localizations were classified as similar to Bik, while others were assigned to Cytb5, suggesting significant differences in their distribution in the ER. Consistent with this interpretation, several other mutants were assigned primarily to one or the other of these landmarks (Fig. S5, A and C). Discussion Our development and characterization of reference datasets, each containing more than 500,000 optically validated individual cell images, enabled the identification of 160 features useful for assigning subcellular localization with an ensemble of RF classifier. These reference sets of optically validated images, computed features, classification results, and cell clones (available from Addgene) are tools that can be used to interrogate subcellular localization, for analysis of diversity in organelle morphologies, and as a reference standard for algorithm development. For example, our automated analysis of the images highlighted both morphological similarities (mitochondria) and differences (Golgi) between premalignant murine and human mammary epithelia cells grown in monolayers (Fig. 4, B and C). Furthermore, by assigning localizations to landmarks rather than specifying a specific organelle, our approach accounts for proteins having multiple localizations. For example, VAMP5 is generally referred to as a plasma membrane protein, but here the VAMP5 localization encompasses much of the secretory pathway because at steady-state, much of the protein is in transit. Because the cell lines and the derived clones have relatively stable genomes, they are expected to constitute a platform for image-based analyses of genetic manipulations, cell signaling events, and quantitative measurement of cellular responses to environmental and extracellular matrix changes as well as responses to drugs and other perturbations. Here we have used the images and feature sets as tools to examine the sequence requirements for motifs that regulate TA protein subcellular localization. Motifs for localization at mitochondria Our dataset included only 12 mutants with a five–amino acid CTS that targeted EGFP-TA to mitochondria. However, comparison with the almost 1,000 sequences that did not result in localization at mitochondria was sufficient to identify the shared characteristics of proteins assigned mitochondrial localization as a PWM for the amino acids within the CTS. Thus, in contrast to previous results, our data directly demonstrate that the presence of positively charged amino acids is not sufficient for targeting mitochondria. Indeed, 133 EGFP-TA mutants with positively charged CTS sequences (+2 or greater) did not localize to mitochondria (Fig. 5). CTS sequences that targeted to mitochondria did not include those with a longer hydrophobic core sequence, a result consistent with other studies using smaller datasets (Costello et al., 2017). However, a shorter hydrophobic core was also not sufficient to target a TA protein to mitochondria (Fig. S3 C). Indeed, even when the ≥2 positive charges and minimal hydrophobic core were combined, there were 121 proteins targeted to nonmitochondrial locations compared with 12 assigned to the mitochondria. Automated assignment of localization will facilitate future analyses of more sequences to determine if there are multiple independent motifs for mitochondrial localization of TA proteins. Classification based on the computed feature data clearly enables automated assignment of outer mitochondrial membrane localization even though such assignment is not possible based on visual inspection of the images (Fig. 7 and Fig. S4). Our approach will also enable generation of a more robust PWM for such motifs to fully capture the rules governing localization of TA proteins at mitochondria. ER-localized TA proteins Our observation that different ER-localized TA proteins have distinct distributions suggests that the ER may be composed of multiple morphologically distinct subcompartments. A related explanation is that different TA proteins have different residence times in the ER or distinct regions of the ER. Of particular interest is the observation that the TA sequences from Bik- and Cytb5-localized EGFP-TA result in different distributions of the proteins (Fig. 1 B and Fig. 2 A). Consistent with the CTS sequence FPCVN being a representative of a bona fide localization motif for a novel Bik distribution within the ER, the sequence resulted in the same localization for EGFP-TA-Bcl-2-ΔSHK (Fig. 4). Unexpectedly, inverting the first two amino acids of FPCVN to generate PFCVN resulted in export of EGFP-TA from the ER and localization at Golgi and VAMP5 destinations. In contrast, the sequence RRCVN resulted in 48% assignment of EGFP-TA localization as most similar to Calr-KDEL with negligible protein assigned to Bik localization, indicating that the different sequences result in images that are morphologically distinct. The differences in assigned localizations are most likely due to differences in the distributions of the protein within the ER. Whether the distributions reflect differences in residence times and/or functionally distinct subdomains remains to be determined, but the differing assignments are clearly not the results of different expression levels of the proteins (Fig. S5 C). Data from other mutants reinforce the concept that proteins ostensibly targeted to the same organelle have different distributions, potentially the result of specific ER subdomains. The sequences of two designed mutants, RWCVN and GSCVN, resulted in EGFP-TA assignment to Calr-KDEL, while the CTS sequence WTNFK resulted in assignment of both EGFP-TA and EGFP-TA-Bcl-2-ΔSHK to the Bik landmark. These results also demonstrate that despite our not identifying by random mutagenesis (i.e., within our 995 random mutants) a CTS sequence for localization of EGFP-TA to the Calr-KDEL landmark, such sequences clearly exist. Mutants of EGFP-TA with a CVN sequence and a single R residue that lost assignment preference to a specific ER landmark provided further insight into localization requirements: those with an R in the first position (RWCVN, RCCVN, and RFCVN) were all retained in the ER (40–53%), but less than a third of the protein was assigned to any one of the ER landmarks (Fig. S3). In contrast, mutants with an R at the second position (WRCVN, LRCVN, and VRCVN) were all assigned to the Cytb5 landmark and to locations further along the secretory pathway, including endosomes and the VAMP5 compartments. To our surprise, only mutations that resulted in a distribution similar to the Cytb5 landmark also had partial localization at endosomes. These sequences (WRCVN, RLCVN, LRCVN, VRCVN, SLCVN, SVCVN, MVCVN, and VMCVN) share no obvious similarity and vary in hydrophobicity across the entire scale. However, export of EGFP-TA to the secretory pathway is not a defining feature of the distribution characteristic of the Cytb5 landmark, as other TA CTS sequences including DEPGH, PEHVS, PKWVT, PSNHQ, and RVRPG resulted in localizations assigned as the Cytb5 landmark without significant endosome or plasma membrane localization (Fig. S5). Thus, in this case, if the localization is defined by residence times, it must reflect residences with distinct distributions within the ER. Furthermore, the extent to which the different mutant proteins were exported from the Cytb5 compartment was variable. The mutant MVCVN was equally distributed between the Cytb5, endosome, and VAMP5 compartments, while VMCVN was assigned primarily to VAMP5. In contrast, the six TA proteins assigned to either the Bik or Calr-KDEL landmarks were not exported from the ER (Fig. S5). The bias toward export of TA proteins assigned to the Cytb5 ER landmark suggests the intriguing possibility that the distribution of this landmark represents a subcompartment that may play an important role in sorting TA proteins. Future experimentation is needed to determine if this is unique to TA proteins or if other proteins are sorted from a different ER subcompartments. For example, the Calr-KDEL distribution may represent a subcompartment involved in transport of secretory proteins since that landmark is an ER luminal protein localized to ER in part because it is efficiently recycled from the Golgi. Conclusions We present a database of optically validated images for both human and murine epithelia together with a set of quantitative features that can be used to study protein localization in individual living cells. It combines (1) a large number of images to provide examples of most if not all organelle morphologies including those throughout the secretory pathway that result from dynamic movement, (2) use of an unbiased classification tool to assess subcellular localization that can be readily extended to include additional landmarks, and (3) use of a PWM approach to predict and verify subcellular localization motifs. With these tools, we identified a new motif for targeting TA proteins to mitochondria and multiple morphologically and likely functionally distinct protein distributions within the ER. The sparsity of the sequence coverage provided by 995 unique mutants suggests that our dataset contains many more targeting motifs similar to the CVN sequence that can be characterized using these techniques. Ultimately, the utility of the tools described here is limited only by the creativity of the user. Materials and methods Plasmid construction All coding regions were cloned into pQCXIP, a retroviral expression vector (Clontech) encoding the monomeric fluorescent reporter protein EGFP-S65T upstream of the gene of interest, unless otherwise stated. The vector pQCXIP has an incomplete retroviral 3′ long terminal repeat (LTR) to prevent replication. Therefore, the LTR was repaired by excising the missing piece from pBabe-puro and introducing it into pQCXIP. This restored the function of 3′ LTR (pQCXIP-Repaired [R]), enabling rescue of the virus as described below. To generate pQCIXP-R-EGFP-TA, EGFP from pEGFP (Clontech) was cloned into pQCIXP-R by digestion of the pEGFP vector with AgeI-HF(NEB) and BamHI-HF(NEB). The open reading frame from human VAMP1 (Open Biosystem/Dharmacon) was amplified using Phusion High-Fidelity DNA Polymerase (NEB) and forward (Fwd) and reverse (Rev) primers with the following sequences: Fwd-VAMP1: 5′-GCG​TCG​ACA​TGG​AGA​GCA​GTG​CTG​CCA​AGC​TAA​A-3′, Rev-VAMP1: 5′-GCG​GAT​CCT​CAA​GTA​AAA​AAG​TAG​ATT​ACA​ATA​AC-3′. The resulting products were subcloned into pQCXIP-R-EGFP digested with SalI-HF (NEB) and BamHI-HF (NEB). As landmarks, the following coding regions were cloned into pQCXIP: Rab5, Ras-related protein Rab7A (Rab7), and Rab11 (kindly provided by J.C. Simpson, University College Dublin, Dublin, Ireland); Golgin84 and Bik (Open Biosystem/Dharmacon); ERGIC53, VAMP2, and VAMP5 (OriGene Technologies); Bcl-2 (TA sequence residues: 213–239 [isoform 2]; Zhu et al., 1996); PTDSS1 (Stone and Vance, 2000; kindly provided by J. Vance, University of Alberta, Edmonton, Canada); cytochrome b5 (Zhu et al., 1996); harakiri, membrin, Ras-related protein Rab3C, emerin, and ribosome-attached membrane protein 4 (OriGene Technologies); and the peroxisomal targeting sequence. The sequence encoding the TA region of MAO was assembled from oligonucleotides. pBABE-puro-GFP-wt-lamin A was a gift from T. Misteli (National Cancer Institute, National Institutes of Health, Bethesda, MD; Addgene plasmid 17662; Scaffidi and Misteli, 2008), and LAMP-1-mGFP was a gift from E. Dell’Angelica (University of California, Los Angeles, Los Angeles, CA; Addgene plasmid 34831; Falcón-Pérez et al., 2005). The following coding regions were fused to the N-terminus of EGFP-S65T in the pQCXIP vector: OTC (32 amino acids of the N-terminal sequence; Horwich et al., 1986); the mitochondrial targeting sequence from subunit VIII human CCO; a sequence encoding the N-terminal 81 amino acids of GalT; and the N-terminal 20 amino acids of neuromodulin (Skene and Virág, 1989). Calr-KDEL includes the ER targeting sequence of calreticulin fused to the N-terminus of EGFP and the ER retention sequence KDEL at the C-terminus of EGFP. A plasmid encoding the fluorescent protein mLumin was kindly provided by J, Hardy (University of Massachusetts, Amherst, MA). The coding region for Harakiri was fused to the 3′ end of the mLumin coding sequence by using SalI-HF and BamHI-HF. For expression in MCF10A, the coding regions were subcloned into pLVX-EF1a-IRES-Puro (Clontech) by using NotI-HF (NEB) and AvrII (NEB). Random mutagenesis Generation of random mutations in the VAMP1 CTS in EGFP-TA was performed using PCR with degenerate primers. The coding sequence for the fusion protein EGFP-TA was amplified with primers (Fwd-CF: 5′-CCG​CGG​CCG​CAC​CGG​TCG​CCA​CCA​T-3′, Rev-CF: 5′-GCGGAATTCCGGATCCTCAMNNMNNMNNMNNMNNTACAATAACTACCACGATGATGG-3′; Integrated DNA Technologies) containing five degenerate codons NNK at the 3′ end of the coding sequence. To generate the “CVN” mutants, the following reverse primer was used: Rev-CVN: 5′-GAATTCCGGATCCTCAATTAACACAMMNMMNTACAATAACTGCCACGATGATGGC-3′ (Integrated DNA Technologies). Fwd primer is the same that was used above. The pQCXIP-R plasmid was digested with BamHI-HF (NEB) and AgeI-HF (NEB), the PCR products were ligated into the cut plasmid using the Cold Fusion technology (System Biosciences), and the DNA was transformed into Escherichia coli bacteria DH5α. After incubation at 37°C for 12 h to complete ligation and amplify the plasmids while minimizing the number of copies of identical plasmids, the transformants were harvested and pooled, and plasmid DNA was isolated using a Presto DNA Mini Plasmid kit (FroggaBio) and packaged into retroviral particles (Phoenix cell line). Sequence analysis after rescue from cells (see below) revealed 995 unique sequences from 1,220 clones. Cell lines and culture The cell lines used were selected because they are both relatively genomically stable breast epithelia. Our assumption is that genomically stable cell lines would exhibit a relatively normal distribution of localization phenotypes characteristic of the cell type. NMuMG cells (a generous gift of J. Wrana, Lunenfeld-Tanenbaum Research Institute, Toronto, Canada) were cultured in DMEM, containing 10 µg/ml bovine insulin (Sigma), 10% FBS (Gibco), and penicillin/streptomycin (Wisent). The retroviral packaging cell line (Phoenix) and HEK293T were grown in DMEM (Gibco), supplemented with 10% FBS and penicillin/streptomycin. MCF10A were cultured in DMEM/F12 (Gibco) supplemented with 5% horse serum (Gibco), 10 µg/ml insulin (Sigma), 0.02 µg/ml EGF (Preprotech), 0.5 µg/ml hydrocortisone (Sigma), and 0.25 mg/liter isoproterenol (Sigma). All cell lines were maintained in a 5% CO2 atmosphere at 37°C. All cell lines tested mycoplasma-free using a PCR-based detection system (Hopert et al., 1993). MCF10A were analyzed by comparative genomic hybridization (Mills et al., 2015). Both Phoenix and HEK293T were genotyped at the Centre for Applied Genomics, SickKids, Toronto, Canada. Transfection and transduction Retrovirus was derived by transient transfection of pQCXIP-R-EGFP-TA into the Phoenix packaging cell line using Fugene HD (Promega). After 24 h, the virus-containing cell medium was filtered (0.45 µm, PALL) and transferred onto the target cell line. To increase the efficiency of transduction, 8 µg/ml polybrene (Sigma) was added. Stable colonies were selected in 10% FBS/DMEM containing 2 µg/ml puromycin (Sigma) and followed by sorting single cells (BD FACSAria II) expressing EGFP into individual wells of multiwall plates (TC plate 96-well, standard, F, Sarstedt). Once a colony formed, it was grown under puromycin selection as above until further analysis. Rescuing of proviral DNA To retrieve the nucleotide sequences of the mutants, cell clones individually transduced with the replication-repaired pQCXIP-R-EGFP-TA mutants were seeded in a 96-well format plate (TC plate 96-well, standard, F, Sarstedt) and transfected with the pCL-ECO packaging vector (Imgenex) using Fugene HD (Promega). Transfection with pCL-ECO enables virus production from the cells transduced with pQCXIP-R-EGFP-TA mutants. After 48 h, the supernatant was collected for viral RNA extraction (Murdoch et al., 1997). Briefly, Trizol LS (ThermoFisher) was added to the supernatant in a 3:1 ratio along with chloroform and yeast tRNA under RNase-free conditions. After centrifugation and isopropanol precipitation, viral RNA was subjected to reverse transcription by using SuperScript III Reverse transcription (ThermoFisher) to generate cDNA. Sequence analysis was performed at the Centre for Applied Genomics sequencing facility. Lentivirus production and transduction To express all landmarks in the human cell line MCF10A, the coding regions were subcloned into the lentiviral vector pLVX-EF1a-IRES-Puromycin (Clontech). The lentiviral DNA plasmid and both pPAX2 and pMD2.G plasmids were transfected into HEK293T cells at 1:1:0.1 ratios by means of calcium phosphate precipitation. Prior to transfection, the three plasmids were briefly mixed with sterile 1× Hepes buffered saline. 1× Hepes buffered saline is composed of Hepes (5 g/liter), NaCl (8 g/liter), dextrose (1 g/liter), KCl (3.7 g/liter), and Na2HPO4 × 7H2O (0.19 g/liter). This mixture was then supplemented with 2.5 M CaCl2 to a final concentration of 0.14 M and incubated for 20 min at room temperature before it was added to the HEK293T cells. After 48 h, the virus-containing supernatant was collected from the HEK293T cells and filtered through a 0.45-µm filter unit and transferred onto MCF10A cells. To obtain stable cell lines, selection was performed by adding puromycin (2 µg/ml) 48 h after transduction. Live cell imaging NMuMG and MCF10A cells expressing one of the landmarks or EGFP-TA mutants were seeded in separate wells in 384-well microplates (CellCarrier-384 ultra, B128 SRI/160; Perkin Elmer) and allowed to grow for 24 h before staining with the nuclear dye DRAQ5 (5 nM; Biostatus). To account for any technical interference that may affect the classification output, cell clones expressing the landmarks or mutants were imaged on multiple days and positions within the wells. In addition, the positions within the plates where the cell clones were grown and imaged was varied. Where indicated, cells were also stained with either Mitotracker Red (500 µM; ThermoFisher) or BODIPY-thapsigargin (500 µM; Setareh Biotech) according to the manufacturer’s instructions. Plates were imaged (two to eight wells, 20–30 fields of view) on two different spinning disk automated confocal microscopes of the same model (OPERA QHS; PerkinElmer) with 40× water objectives (NA = 0.9) in a defined temperature (37°C) and CO2 (5%) environment by using EvoShell acquisition software. Images were collected using 3-Peltier-cooled 12-bit CCD cameras (Type sensiCam, camera resolution 1.3 megapixels; PCO.imaging) either with a binning of two or unbinned. Unbinned images were binned numerically before segmentation and feature extraction. Imaging the controls and some query lines on multiple microscopes, with and without camera binning, enabled feature counter selection to remove features characteristic of the imaging platforms. Assessment of new query cell images was processed as described above along with a set of established landmarks as controls. Image processing Image segmentation To identify individual cells, as well as nuclear and cytoplasmic areas for each cell in fluorescence micrographs, image segmentation was performed using PerkinElmer Acapella 2.0 software (Nuclei Detection Algorithm A). The nuclear detection algorithm includes following parameters: threshold adjustment (1.5), individual threshold adjustment (0.4), minimum nuclei distance (15), nuclear splitting adjustment (15), minimum nuclear area (300), and minimum nuclear contrast (0.1). Briefly, the nuclear region was identified using an image of DRAQ5 staining. Using the nucleus as a seed, the cell region was identified using a watershed algorithm and the low-level cytoplasmic staining due to DRAQ5. The region of the cell that is not in the nucleus is considered to be the cytoplasm. Additionally, a ROI mask was computed by identifying all the pixels that are 1.5 times the mean pixel intensity of the cell mask in the EGFP channel. Regions with at least 30 contiguous pixels were retained as ROIs. Images of cells touching the micrograph edge were discarded. Removing out-of-focus cells To investigate the steady-state localization of a protein in cells, images of dividing or dying cells were removed as they have been shown to interfere with classification (Huang and Murphy, 2004). Such cells are typically out of focus and exhibit a different texture and morphology of the nucleus in the DRAQ5 channel compared with other cells. These features enabled the generation of an automated image quality control algorithm that effectively removed images of cells that were out of focus, dividing, or dead. To train a classifier to remove these images, we collected cell images of landmarks that were in-focus and 4 µm below the in-focus plane. The cells were identified by image segmentation, and image features were calculated. Since information regarding cell focus can be obtained from the nuclear staining, which should be similar for cell clones independent of the EGFP protein expressed, we used texture and morphology features from the DRAQ5 images to create a binary random forests classifier and identity features necessary to separate cell images that are in focus and out of focus. Using the new feature subset, a new random forests classifier was built to identify within fields of view images of individual cells that are in focus and out of focus. The classifier was then applied to the entire dataset to remove individual cell images that were out of focus. Examination of the rejected images revealed that this classifier also removed many cells that, when visually inspected, appeared to be in focus, but were usually not correctly segmented, suggesting there may have been a minor problem in focus. Our approach differs from previous work on focus quality control in which an entire image field was identified as out of focus rather than images of individual cells within an image field (Bray et al., 2012). The sensitivity for individual class before and after removal of out-of-focus cells is shown in Fig. S1. Removing cells with improper segmentation Since segmentation was performed using images of the DRAQ5 staining, we used the same information to remove incorrectly segmented objects. We computed two parameters from our existing features: R1, a ratio of average pixel intensity inside the nuclear mask to the average pixel intensity inside the cytoplasm mask, and R2, a ratio of pixel intensity SD inside the nuclear mask to the pixel intensity SD inside the cytoplasm mask.R1=Nucleus IntensityCytoplasm Intensity R2=Standard Deviation of Nucleus IntensityStandard Deviation of Cytoplasm Intensity. The first parameter ensures the appropriate level of nuclear stain since DRAQ5 staining in the cytoplasm is typically far lower compared with staining in the nucleus. Similarly, the second parameter captures the changes in the staining patterns between the nucleus and the cytoplasm region. Due to differences in staining of the heterochromatin inside the nucleus, the deviation of nuclear intensity should be much higher when compared with the intensity deviation in the cytoplasm region. A threshold for both ratios was determined empirically to be 3.5. All the cells below the threshold were removed from the dataset. We then computed the ratio of the nucleus area to the cell area and removed the top and bottom 5% of the cells based on this ratio. This last step removed images of cells that were abnormally large (typically senescent or improperly segmented) or small (typically dying). The same quality control steps were also applied to the EGFP-TA CTS mutants. The 10 mutants for which <50 suitable cell images remained after the quality control steps were removed from the analysis, resulting in a total of 995 unique random mutants. Cell image feature extraction Features were measured for nuclear and cytoplasmic (including ROI) areas for DRAQ5 staining and EGFP expression, respectively, using a custom PerkinElmer Acapella script (code available on https://github.com/DWALab/Schormann-et-al). For each cell image, 495 morphological and statistical image features were calculated (Collins et al., 2015). Micrographs of individual cells were automatically selected for inclusion in the reference library based on numerical assessment of image and segmentation quality (R1 and R2; Fig. S1). Feature selection For classification of localization, a reduced set of features was selected to minimize (1) the impact of intensity variations due to changes in the intensity of the illumination or efficiency of collection, and (2) features that were sensitive to differences in imaging instrument or data collection scheme. The final feature list comprises 160 features, i.e., TAS (n = 140), morphology (n = 18; Boland and Murphy, 2001), and texture (radial moment and angular second moment; Haralick et al., 1973), that were derived exclusively from the EGFP channel. A spot detection script implemented in PerkinElmer Acapella software was used. All feature data tables and a description of feature calculation is available on GitHub (https://github.com/DWALab/Schormann-et-al). Using the images of NMuMG cells expressing the landmarks, the following steps were taken to select features most relevant for the final classification. We first retained only features pertaining to the EGFP channel to eliminate influence of nuclei-derived features on protein localization. Further, only texture TAS and shape features were retained to minimize the effects of protein expression on the final classification. To minimize variation due to different microscopes, a random forests classifier was created to output features that can separate individual landmark cell images based on which microscope they were imaged on. This was done for each landmark separately. We then computed the frequency for each feature, i.e., the number of times the same feature was important in separating the microscopes across landmarks. Features that were important to separate microscopes for at least five landmarks were removed. Correlations between all features and cell intensity were computed. Features with correlation values between −0.5 to 0.5 were retained. A minimum set was identified from the remaining features using a “leave one feature out” test of classification accuracy. For each feature, a classifier was built in which that feature was omitted. Features were retained if the classification accuracy decreased when the feature was removed. Intensity preprocessing For further data processing and final classification, only features from the EGFP channel were used. Cells with <100 intensity units of EGFP expression were eliminated from the analyses. The threshold value of 100 intensity units was obtained empirically by manually thresholding 20 randomly selected EGFP channel images into foreground and background regions. All the features were scaled to have zero mean and unit SD. Further, for each landmark and mutant dataset, cells with integrated total intensity in the top or bottom fifth percentile were removed. The remaining 789,011 and 523,319 validated landmark cell images of NMuMG and MCF10A cells, respectively, constitute the reference libraries uploaded to the Image Data Resource (idr0072; https://idr.openmicroscopy.org) and have been made available as a resource for general use. Quantify effects of protein expression To determine whether differences in the expression levels of the landmark proteins contributed to the localizations assigned by the random forests classifier, landmark images were divided into four different classes based on intensity quartiles (0–25, 25–50, 50–75, and 75–100 quartiles). Cross-validation was performed by training on the landmarks from a quartile bin and using the other quartile bins as queries (Fig. 3, quartile 1–quartile 4). When assessed for all quartiles, the landmarks are classified with high accuracy irrespective of the quartile fraction used for training, thus independent of protein expression levels. To further verify that assignment of localization was not due to differences in intensity, especially for proteins ostensibly targeted to the same organelle, images were randomly selected for training and classification in which the query and landmark had similar distributions of intensities (see below). For each mutant, thousands of cells were imaged from independent experiments performed on multiple days. Further cells were deliberately imaged from different plate locations. Pooling training set cell images from independent experiments and from different automated microscopes reduces the impact of minor fluctuations between experiments. After filtering cell images for focus, segmentation, and expression artifacts, the random forests classifier created using the reference library was then used to classify images of cells not used in training to individual landmarks (Breiman, 2001). Classification of the cell images Random forest classifiers were used to separate the landmarks. Code scripting was performed in the MATLAB environment (R2012b, MathWorks,) including following the RF package (https://github.com/ajaiantilal/randomforest-matlab/tree/master/RF_Class_C). All codes are downloadable from GitHub (https://github.com/DWALab/Schormann-et-al). The classifier was generated setting the “mtry” variable to 12 (approximate square root of the total number of features). The “mtry” variable specifies the number of features that are available for the classifier at each split point when building decision trees. The number of trees was set to 500. Due to differences in the number of cell images for each landmark, we randomly selected 3,200 (70% of Calr-KDEL, the landmark with the smallest number of cells) cell images per landmark for training. The rest of the data were used for testing. To avoid bias due to random sampling, five different random forest classifiers were used, each with randomly sampled data from the landmarks. Every unknown cell image was classified using all five classifiers. Each classifier assigns a cell to a specific landmark. For any cell, the final class was decided using the mode of the predicted class among the five different classifiers. To establish the extent of variation between classifications, 20 classification runs were performed on mitochondrial and CVN mutants. The result is presented as mean values (percentage classified) including SD (see Table S1). In general, the variation in percentage of cells assigned to a particular location was <1. Classification of cell images within intensity range For control experiments using cells with a defined range of intensities, cells were automatically selected from the image dataset based on total cellular EGFP intensity. The intensity ranges used were 300–500 and 200–1,000 arbitrary units for mitochondrial and CVN tail-anchored mutants, respectively. To display the intensity profiles as continuous lines, the cell intensities were fit to a density curve (Fig. 7). Comparison of the classification results confirmed that intensity differences did not contribute significantly (Fig. 7 and Fig. S5). Visualization of the spatial relationships of the landmark cell images in 160-dimensional space We used the t-SNE algorithm (Van der Maaten and Hinton, 2008) to reduce the multidimensional data into a two-dimensional plot using code obtained from the author’s website (Van der Maaten and Hinton, 2008). We set the perplexity value to 10 and used 1,000 iterations to generate the two-dimensional representation. Due to the limitations of the algorithm regarding memory and computation cost, we did not use all the data for visualization. For each landmark only the 200 cells nearest to the median of the landmark were used for visualization. This resulted in a total of 3,400 (murine landmark library) and 4,000 (human landmark library) data points for visualization. Naming system When organelle names are used, it is to designate multiple landmarks in aggregate. The following labeling was used: ER corresponds to the landmarks Cytb5, Calr-KDEL, and Bik; Golgi apparatus corresponds to GalT and Golgin84; endosome corresponds to Rab5 and Rab7; MAM is represented by PTDSS1; mitochondria correspond to MAO and CCO; and ERGIC is represented by ERGIC53. Motif visualization Visualization of amino acid motifs was performed using pLogo plots (O’Shea et al., 2013). The foreground sequences were all the sequences that belonged to the cluster, while the background sequences were all the rest of the mutants generated by random mutagenesis with a CTS five amino acids long. PWM As an alternative approach to predict the localization of an unknown sequence of amino acids, a PWM was computed for the amino acid sequences of the CTS for mutants assigned to a specific localization. A foreground-normalized frequency matrix was first computed by calculating for each position the frequency of each amino acid at that position. If an amino acid was not present at a given position, its probability would be zero and would bring the likelihood computation to zero. To avoid this, a pseudocount value of 0.25 was added to the raw frequency for all amino acids for all positions. Thus, the frequency matrix was divided by the number of sequences in the cluster plus a value of 5 (0.25 × 20), resulting in a normalized frequency matrix. The background-normalized frequency matrix was calculated for all the sequences that were generated using random mutagenesis with a five–amino acids CTS. In this case, we did not use pseudocounts as every amino acid was represented at least once at every position. The foreground-normalized frequency matrix was then divided by the background frequency matrix. The log of the resultant gives the PWM. The simple rules and PWMs for predicting mitochondrial localization were tested on the database of sequences having five amino acids in the CTS. The list of predicted localizations based on the simple rule or PWM was compared with the “true assignments” based on image classification. The PPV, sensitivity, and specificity were calculated from the predicted and true assignments. Statistical enrichment All enrichments were computed by fitting a hypergeometric distribution. For all calculations (added hydrophobic core, positive charges, and negative charges) except enrichment of actual length of the CTS, only sequences with five amino acids in the CTS were used. All of the sequences were used to calculate the enrichment for the lengths of the CTS. For pLogo, all the sequences were fed through their website (O’Shea et al., 2013). Background sequences were computed using all the sequences of length five that were generated. Sequences classified as individual landmark were used as foreground sequences. The following equation was used to determine the height of any amino acid at a specific position:Residue Height K,N,p=-logPr(k ∀k ≥K | N,p)Pr(k ∀k ≤K | N,p), where K is the number of residues of a given type at specific position, N is total number of residues at specific position, and p is the probability of a residue at a given residue computed from background sequences. The probabilities are defined as below: Prk⋅∀k·≥K | N,p=∑k=KNbinomial(k,N,p) Prk⋅∀k·≤K | N,p=∑k=0Kbinomial(k,N,p). Online supplemental material Fig. S1 illustrates the image processing pipeline used and sample images. Fig. S2 provides Euclidean distances between centroids of the landmarks in NMuMG cells as a colored heat map and as a table of values. Fig. S3 provides heatmaps of the amino acid enrichments associated with different localizations for EGFP-TA mutants. Fig. S4 provides sample images at high magnification demonstrating that images of different mitochondrial localized proteins correctly assigned by classification care not reliably distinguished visually. Fig. S5 provides heatmaps of the localization assignments for the XXCVN mutants to different distributions within the ER. Table S1 shows mean values (including SD) of 20 classification runs of CVN and mitochondrial mutants. Table 1. Protein localization information used for landmarks Protein Abbreviation Number of cells Protein sequence UniProt Localization Reference Cytochrome b5 Cytb5 Mu: 10,428 99-134 P00167 ER D’Arrigo et al., 1993 Mu: 4,452 Signal sequence of calreticulin, ER retention sequence (KDEL) Calr-KDEL Mu: 5,367 N/A P27797 ER Fliegel et al., 1989; Munro and Pelham, 1987 Hu: 24,059 Bcl-2 interacting killer Bik Mu: 148,030 111-160 Q13323 ER Germain et al., 2002 Hu: 11,046 Ribosome-attached membrane protein 4 RAMP4 Hu: 2,269 ­1–166 Q9Y6X1 ER Schröder et al., 1999 ER-Golgi intermediate compartment 53 kD ERGIC53 Mu: 39,178 1–517 Q9D0F3 ERGIC Scheel et al., 1997 Hu: 73,059 β1,4-galactosyl-transferase GalT Mu: 60,260 1–81 P15291 Trans-Golgi Roth and Berger, 1982 Hu: 19,797 Golgin84 Golgin84 Mu: 147,916 674-731 Q8TBA6 Cis-Golgi Diao et al., 2003 Hu: 5,632 Golgi SNAP receptor complex member 2 Membrin Hu: 19,288 1–212 O14653 Cis-Golgi Hong, 2005 Monoamine oxidase A MAO Mu: 34,521 489-527 Q49A63 OMM de Champlain et al., 1969 Hu: 22,935 Cytochrome c oxidase, subunit VIII CCO Mu: 11,023 1–29 Q53XN1 IMM Rizzuto et al., 1995 Hu: 92,159 Phosphatidylserine synthase 1 PTDSS1 Mu: 137,908 1–473 Q99LH2 MAM, ER Stone and Vance, 2000 Ras-related protein Rab3C Rab3 Hu: 27,545 1–227 Q96E17 Secretory vesicles Fischer von Mollard et al., 1994 Ras-related protein Rab5A Rab5 Mu: 18,098 1–205 A0A024R2K1 Early endosome Chavrier et al., 1990 Hu: 27,137 Ras-related protein Rab7A Rab7 Mu: 45,878 1–207 A0A158RFU6 Late endosome Bucci et al., 2000 Hu: 42,792 Ras-related protein Rab11B Rab11B Hu: 21,961 1–218 Q15907 Recycling endosome Schlierf et al., 2000 Vesicle-associated membrane protein 2 VAMP2 Mu: 19,411 1–116 P63027 Secretory vesicles Grote et al., 1995; Chen and Scheller, 2001 Hu: 12,239 Vesicle-associated membrane protein 5 VAMP5 Mu: 43,352 1–116 O95183 Plasma membrane Hong, 2005 Hu: 39,946 Neuromodulin Neuromodulin Hu: 17,852 1–20 P17677 Plasma membrane Skene and Virág, 1989 Vesicle-associated membrane protein 1, TMD deleted ΔTMD-VAMP1 Mu: 19,074 1–99 P23763 Cytoplasm, nucleus This study Lamin A Lamin A Mu: 15,197 1–690 P02545 Nuclear envelope Scaffidi and Misteli, 2008 Peroxisome targeting signal 1 PTS-1 Mu: 18,396 SKL N/A Peroxisome Gould et al., 1989 Hu: 27,712 Lysosomal-associated membrane protein 1 LAMP-1 Mu: 14,974 1–417 P11279 Lysosomes Falcón-Pérez et al., 2005 Hu: 17,162 Emerin Emerin Hu: 14,277 1–254 P50402 Nuclear envelope Pfaff et al., 2016 The common name, abbreviation, and amino acid sequence used to construct the landmarks is indicated, with the UniProt ID for each of the coding regions. The reference provided includes the data for the assignment of localization and identification of the responsible targeting sequence. The number of cells refers to the number of cell images used in the analysis here. Hu, human image library; IMM, inner mitochondrial membrane; Mu, mouse image library; OMM, outer mitochondrial membrane. Supplementary Material Table S1 shows mean values (including SD) of 20 classification runs of CVN and mitochondrial mutants. Click here for additional data file. Acknowledgments The authors thank Robert Mullen for critical reading of the manuscript and Jarkko Ylanko for technical support with the high-content imaging. Funding for this research was provided by a foundation grant from the Canadian Institutes of Health Research (FDN 143312) to D.W. Andrews. D.W. Andrews holds the Tier 1 Canada Research Chair in Membrane Biogenesis. The authors declare no competing financial interests. Author contributions: All authors participated in experimental design and data interpretation. W. Schormann and S. Hariharan performed experiments, imaged the cells, and analyzed the data. W. Schormann generated all of the mutants and cell lines. S. Hariharan carried out programming for and automated image analysis and informatics analysis. D.W. Andrews supervised the project. All authors participated in writing, reviewing, and editing the manuscript. ==== Refs References Boland, M.V., and R.F. Murphy. 2001. A neural network classifier capable of recognizing the patterns of all major subcellular structures in fluorescence microscope images of HeLa cells. Bioinformatics. 17 :1213–1223. 10.1093/bioinformatics/17.12.1213 11751230 Bray, M.-A., A.N. Fraser, T.P. Hasaka, and A.E. Carpenter. 2012. Workflow and metrics for image quality control in large-scale high-content screens. J. Biomol. Screen. 17 :266–274. 10.1177/1087057111420292 21956170 Breiman, L. 2001. Random Forests. Mach. Learn. 45 :5–32. 10.1023/A:1010933404324 Bucci, C., P. Thomsen, P. Nicoziani, J. McCarthy, and B. van Deurs. 2000. Rab7: a key to lysosome biogenesis. Mol. Biol. Cell. 11 :467–480. 10.1091/mbc.11.2.467 10679007 Chavrier, P., R.G. Parton, H.P. Hauri, K. Simons, and M. Zerial. 1990. Localization of low molecular weight GTP binding proteins to exocytic and endocytic compartments. Cell. 62 :317–329. 10.1016/0092-8674(90)90369-P 2115402 Chen, Y.A., and R.H. Scheller. 2001. SNARE-mediated membrane fusion. Nat. Rev. Mol. Cell Biol. 2 :98–106. 10.1038/35052017 11252968 Chong, Y.T., J.L.Y. Koh, H. Friesen, S.K. Duffy, M.J. Cox, A. Moses, J. Moffat, C. Boone, and B.J. Andrews. 2015. Yeast Proteome Dynamics from Single Cell Imaging and Automated Analysis. Cell. 161 :1413–1424. 10.1016/j.cell.2015.04.051 26046442 Collins, T.J., J. Ylanko, F. Geng, and D.W. Andrews. 2015. A Versatile Cell Death Screening Assay Using Dye-Stained Cells and Multivariate Image Analysis. Assay Drug Dev. Technol. 13 :547–557. 10.1089/adt.2015.661 26422066 Conrad, C., H. Erfle, P. Warnat, N. Daigle, T. Lörch, J. Ellenberg, R. Pepperkok, and R. Eils. 2004. Automatic identification of subcellular phenotypes on human cell arrays. Genome Res. 14 :1130–1136. 10.1101/gr.2383804 15173118 Cory, S., and J.M. Adams. 2002. The Bcl2 family: regulators of the cellular life-or-death switch. Nat. Rev. Cancer. 2 :647–656. 10.1038/nrc883 12209154 Costello, J.L., I.G. Castro, F. Camões, T.A. Schrader, D. McNeall, J. Yang, E.-A. Giannopoulou, S. Gomes, V. Pogenberg, N.A. Bonekamp, . 2017. Predicting the targeting of tail-anchored proteins to subcellular compartments in mammalian cells. J. Cell Sci. 130 :1675–1687. 10.1242/jcs.200204 28325759 D’Arrigo, A., E. Manera, R. Longhi, and N. Borgese. 1993. The specific subcellular localization of two isoforms of cytochrome b5 suggests novel targeting pathways. J. Biol. Chem. 268 :2802–2808.8428954 de Champlain, J., R.A. Mueller, and J. Axelrod. 1969. Subcellular localization of monoamine oxidase in rat tissues. J. Pharmacol. Exp. Ther. 166 :339–345.4304915 Diao, A., D. Rahman, D.J.C. Pappin, J. Lucocq, and M. Lowe. 2003. The coiled-coil membrane protein golgin-84 is a novel rab effector required for Golgi ribbon formation. J. Cell Biol. 160 :201–212. 10.1083/jcb.200207045 12538640 Falcón-Pérez, J.M., R. Nazarian, C. Sabatti, and E.C. Dell’Angelica. 2005. Distribution and dynamics of Lamp1-containing endocytic organelles in fibroblasts deficient in BLOC-3. J. Cell Sci. 118 :5243–5255. 10.1242/jcs.02633 16249233 Fischer von Mollard, G., B. Stahl, A. Khokhlatchev, T.C. Südhof, and R. Jahn. 1994. Rab3C is a synaptic vesicle protein that dissociates from synaptic vesicles after stimulation of exocytosis. J. Biol. Chem. 269 :10971–10974.8157621 Fliegel, L., K. Burns, D.H. MacLennan, R.A. Reithmeier, and M. Michalak. 1989. Molecular cloning of the high affinity calcium-binding protein (calreticulin) of skeletal muscle sarcoplasmic reticulum. J. Biol. Chem. 264 :21522–21528.2600080 Germain, M., J.P. Mathai, and G.C. Shore. 2002. BH-3-only BIK functions at the endoplasmic reticulum to stimulate cytochrome c release from mitochondria. J. Biol. Chem. 277 :18053–18060. 10.1074/jbc.M201235200 11884414 Gould, S.J., G.A. Keller, N. Hosken, J. Wilkinson, and S. Subramani. 1989. A conserved tripeptide sorts proteins to peroxisomes. J. Cell Biol. 108 :1657–1664. 10.1083/jcb.108.5.1657 2654139 Grote, E., J.C. Hao, M.K. Bennett, and R.B. Kelly. 1995. A Targeting Signal in VAMP Regulating Transport to Synaptic Vesicles. Cell. 81 :581–589. 10.1016/0092-8674(95)90079-9 7758112 Hamilton, N.A., R.S. Pantelic, K. Hanson, and R.D. Teasdale. 2007. Fast automated cell phenotype image classification. BMC Bioinformatics. 8 :110. 10.1186/1471-2105-8-110 17394669 Haralick, R., K. Shanmugam, and I. Dinstein. 1973. Textural features for image classification. IEEE Trans. Syst. Man Cybern. SMC-3 :610–621. 10.1109/TSMC.1973.4309314 Henderson, M.P., Y.T. Hwang, J.M. Dyer, R.T. Mullen, and D.W. Andrews. 2007. The C-terminus of cytochrome b5 confers endoplasmic reticulum specificity by preventing spontaneous insertion into membranes. Biochem. J. 401 :701–709. 10.1042/BJ20060990 16984229 Hong, W. 2005. SNAREs and traffic. Biochim. Biophys. Acta. 1744 :120–144. 10.1016/j.bbamcr.2005.03.014 15893389 Hopert, A., C.C. Uphoff, M. Wirth, H. Hauser, and H.G. Drexler. 1993. Specificity and sensitivity of polymerase chain reaction (PCR) in comparison with other methods for the detection of mycoplasma contamination in cell lines. J. Immunol. Methods. 164 :91–100. 10.1016/0022-1759(93)90279-G 8360512 Horie, C., H. Suzuki, M. Sakaguchi, and K. Mihara. 2002. Characterization of signal that directs C-tail-anchored proteins to mammalian mitochondrial outer membrane. Mol. Biol. Cell. 13 :1615–1625. 10.1091/mbc.01-12-0570 12006657 Horwich, A.L., F. Kalousek, W.A. Fenton, R.A. Pollock, and L.E. Rosenberg. 1986. Targeting of pre-ornithine transcarbamylase to mitochondria: definition of critical regions and residues in the leader peptide. Cell. 44 :451–459. 10.1016/0092-8674(86)90466-6 3943133 Huang, K., and R.F. Murphy. 2004. Boosting accuracy of automated classification of fluorescence microscope images for location proteomics. BMC Bioinformatics. 5 :78. 10.1186/1471-2105-5-78 15207009 Kyte, J., and R.F. Doolittle. 1982. A simple method for displaying the hydropathic character of a protein. J. Mol. Biol. 157 :105–132. 10.1016/0022-2836(82)90515-0 7108955 Levine, J.H., E.F. Simonds, S.C. Bendall, K.L. Davis, A.D. Amir, M.D. Tadmor, O. Litvin, H.G. Fienberg, A. Jager, E.R. Zunder, . 2015. Data-Driven Phenotypic Dissection of AML Reveals Progenitor-like Cells that Correlate with Prognosis. Cell. 162 :184–197. 10.1016/j.cell.2015.05.047 26095251 Li, J., L. Xiong, J. Schneider, and R.F. Murphy. 2012. Protein subcellular location pattern classification in cellular images using latent discriminative models. Bioinformatics. 28 :i32–i39. 10.1093/bioinformatics/bts230 22689776 Marty, N.J., H.J. Teresinski, Y.T. Hwang, E.A. Clendening, S.K. Gidda, E. Sliwinska, D. Zhang, J.A. Miernyk, G.C. Brito, D.W. Andrews, . 2014. New insights into the targeting of a subset of tail-anchored proteins to the outer mitochondrial membrane. Front. Plant Sci. 5 :426. 10.3389/fpls.2014.00426 25237314 Mills, C.E., C. Thome, D. Koff, D.W. Andrews, and D.R. Boreham. 2015. The relative biological effectiveness of low-dose mammography quality X rays in the human breast MCF-10A cell line. Radiat. Res. 183 :42–51. 10.1667/RR13821.1 25536231 Munro, S., and H.R. Pelham. 1987. A C-terminal signal prevents secretion of luminal ER proteins. Cell. 48 :899–907. 10.1016/0092-8674(87)90086-9 3545499 Murdoch, B., D.S. Pereira, X. Wu, J.E. Dick, and J. Ellis. 1997. A rapid screening procedure for the identification of high-titer retrovirus packaging clones. Gene Ther. 4 :744–749. 10.1038/sj.gt.3300448 9282176 Nanni, L., and A. Lumini. 2008. A reliable method for cell phenotype image classification. Artif. Intell. Med. 43 :87–97. 10.1016/j.artmed.2008.03.005 18440791 O’Shea, J.P., M.F. Chou, S.A. Quader, J.K. Ryan, G.M. Church, and D. Schwartz. 2013. pLogo: a probabilistic approach to visualizing sequence motifs. Nat. Methods. 10 :1211–1212. 10.1038/nmeth.2646 24097270 Pärnamaa, T., and L. Parts. 2017. Accurate Classification of Protein Subcellular Localization from High-Throughput Microscopy Images Using Deep Learning. G3 (Bethesda). 7 :1385–1392. 10.1534/g3.116.033654 28391243 Pfaff, J., J. Rivera Monroy, C. Jamieson, K. Rajanala, F. Vilardi, B. Schwappach, and R.H. Kehlenbach. 2016. Emery-Dreifuss muscular dystrophy mutations impair TRC40-mediated targeting of emerin to the inner nuclear membrane. J. Cell Sci. 129 :502–516. 10.1242/jcs.179333 26675233 Rao, M., V. Okreglak, U.S. Chio, H. Cho, P. Walter, and S. Shan. 2016. Multiple selection filters ensure accurate tail-anchored membrane protein targeting. eLife. 5 :e21301. 10.7554/eLife.21301 27925580 Rapaport, D. 2003. Finding the right organelle. Targeting signals in mitochondrial outer-membrane proteins. EMBO Rep. 4 :948–952. 10.1038/sj.embor.embor937 14528265 Raptis, A., B. Torrejón-Escribano, I. Gómez de Aranda, and J. Blasi. 2005. Distribution of synaptobrevin/VAMP 1 and 2 in rat brain. J. Chem. Neuroanat. 30 :201–211. 10.1016/j.jchemneu.2005.08.002 16169186 Rizzuto, R., M. Brini, P. Pizzo, M. Murgia, and T. Pozzan. 1995. Chimeric green fluorescent protein as a tool for visualizing subcellular organelles in living cells. Curr. Biol. 5 :635–642. 10.1016/S0960-9822(95)00128-X 7552174 Roth, J., and E.G. Berger. 1982. Immunocytochemical localization of galactosyltransferase in HeLa cells: codistribution with thiamine pyrophosphatase in trans-Golgi cisternae. J. Cell Biol. 93 :223–229. 10.1083/jcb.93.1.223 6121819 Scaffidi, P., and T. Misteli. 2008. Lamin A-dependent misregulation of adult stem cells associated with accelerated ageing. Nat. Cell Biol. 10 :452–459. 10.1038/ncb1708 18311132 Scheel, J., R. Pepperkok, M. Lowe, G. Griffiths, and T.E. Kreis. 1997. Dissociation of coatomer from membranes is required for brefeldin A-induced transfer of Golgi enzymes to the endoplasmic reticulum. J. Cell Biol. 137 :319–333. 10.1083/jcb.137.2.319 9128245 Schlierf, B., G.H. Fey, J. Hauber, G.M. Hocke, and O. Rosorius. 2000. Rab11b is essential for recycling of transferrin to the plasma membrane. Exp. Cell Res. 259 :257–265. 10.1006/excr.2000.4947 10942597 Schröder, K., B. Martoglio, M. Hofmann, C. Hölscher, E. Hartmann, S. Prehn, T.A. Rapoport, and B. Dobberstein. 1999. Control of glycosylation of MHC class II-associated invariant chain by translocon-associated RAMP4. EMBO J. 18 :4804–4815. 10.1093/emboj/18.17.4804 10469658 Scott, I., and R.J. Youle. 2010. Mitochondrial fission and fusion. Essays Biochem. 47 :85–98. 10.1042/bse0470085 20533902 Sharpe, H.J., T.J. Stevens, and S. Munro. 2010. A comprehensive comparison of transmembrane domains reveals organelle-specific properties. Cell. 142 :158–169. 10.1016/j.cell.2010.05.037 20603021 Skene, J.H., and I. Virág. 1989. Posttranslational membrane attachment and dynamic fatty acylation of a neuronal growth cone protein, GAP-43. J. Cell Biol. 108 :613–624. 10.1083/jcb.108.2.613 2918027 Stone, S.J., and J.E. Vance. 2000. Phosphatidylserine synthase-1 and -2 are localized to mitochondria-associated membranes. J. Biol. Chem. 275 :34534–34540. 10.1074/jbc.M002865200 10938271 Sullivan, D.P., C.F. Winsnes, L. Åkesson, M. Hjelmare, M. Wiking, R. Schutten, L. Campbell, H. Leifsson, S. Rhodes, A. Nordgren, . 2018. Deep learning is combined with massive-scale citizen science to improve large-scale image classification. Nat. Biotechnol. 36 :820–828. 10.1038/nbt.4225 30125267 Van der Maaten, L.J.P., and G.E. Hinton. 2008. Visualizing high-dimensional data using t-sne. J. Mach. Learn. Res. 9 :2579–2605. 10.1007/s10479-011-0841-3 Xu, Y.-Y., F. Yang, Y. Zhang, and H.-B. Shen. 2015. Bioimaging-based detection of mislocalized proteins in human cancers by semi-supervised learning. Bioinformatics. 31 :1111–1119. 10.1093/bioinformatics/btu772 25414362 Xu, Y.-Y., F. Yang, and H.-B. Shen. 2016. Incorporating organelle correlations into semi-supervised learning for protein subcellular localization prediction. Bioinformatics. 32 :2184–2192. 10.1093/bioinformatics/btw219 27153655 Xu, Y.-Y., H.-B. Shen, and R.F. Murphy. 2019. Learning complex subcellular distribution patterns of proteins via analysis of immunohistochemistry images. Bioinformatics.:btz844. 10.1093/bioinformatics/btz844 Zernike, F. 1934. Beugungstheorie des schneidenverfarhens und seiner verbesserten form, der phasenkontrastmethode. Physica. 1 :689–704. 10.1016/S0031-8914(34)80259-5 Zhu, W., A. Cowie, G.W. Wasfy, L.Z. Penn, B. Leber, and D.W. Andrews. 1996. Bcl-2 mutants with restricted subcellular location reveal spatially distinct pathways for apoptosis in different cell types. EMBO J. 15 :4130–4141. 10.1002/j.1460-2075.1996.tb00788.x 8861942

PMC007xxxxxx/PMC7390989.txt

==== Front Bioinformatics Bioinformatics bioinformatics Bioinformatics 1367-4803 1367-4811 Oxford University Press 32437556 10.1093/bioinformatics/btaa518 btaa518 Applications Notes Gene Expression AcademicSubjects/SCI01060 ThETA: transcriptome-driven efficacy estimates for gene-based TArget discovery Failli Mario Institute of Biomedicine, University of Eastern Finland, Kuopio 70210, Finland Department of Chemical, Materials and Industrial Engineering, University of Naples 'Federico II', Naples 80125, Italy Paananen Jussi Institute of Biomedicine, University of Eastern Finland, Kuopio 70210, Finland Blueprint Genetics Ltd, Finland Fortino Vittorio Institute of Biomedicine, University of Eastern Finland, Kuopio 70210, Finland Blueprint Genetics Ltd, Finland Mathelier Anthony Associate Editor To whom correspondence should be addressed. vittorio.fortino@uef.fi 15 7 2020 21 5 2020 21 5 2020 36 14 42144216 29 11 2019 23 4 2020 16 5 2020 © The Author(s) 2020. Published by Oxford University Press. 2020 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Abstract Summary Estimating efficacy of gene–target-disease associations is a fundamental step in drug discovery. An important data source for this laborious task is RNA expression, which can provide gene–disease associations on the basis of expression fold change and statistical significance. However, the simply use of the log-fold change can lead to numerous false-positive associations. On the other hand, more sophisticated methods that utilize gene co-expression networks do not consider tissue specificity. Here, we introduce Transcriptome-driven Efficacy estimates for gene-based TArget discovery (ThETA), an R package that enables non-expert users to use novel efficacy scoring methods for drug–target discovery. In particular, ThETA allows users to search for gene perturbation (therapeutics) that reverse disease-gene expression and genes that are closely related to disease-genes in tissue-specific networks. ThETA also provides functions to integrate efficacy evaluations obtained with different approaches and to build an overall efficacy score, which can be used to identify and prioritize gene(target)–disease associations. Finally, ThETA implements visualizations to show tissue-specific interconnections between target and disease-genes, and to indicate biological annotations associated with the top selected genes. Availability and implementation ThETA is freely available for academic use at https://github.com/vittoriofortino84/ThETA. Contact vittorio.fortino@uef.fi Supplementary information Supplementary data are available at Bioinformatics online. ==== Body pmc1 Introduction In order to minimize the risk of drug development failures, academic and industrial research has focused on target-based drug discovery approaches. This has led to several computational methods to score gene(target)–disease association upon efficacy estimates calculated from different data sources, ranging from scientific publications to omics databases (Koscielny et al., 2017; Nguyen et al., 2017; Piñero et al., 2017). We have recently proposed two transcriptome-driven approaches to identify and score gene–disease associations (Failli et al., 2019), namely tissue-specific efficacy (TSE) and modulation (MOD) scores. The first method identifies genes that are closely related to disease-genes (genes with genetic variants that associate with disease risk) in tissue-specific gene co-expression networks. The second method estimates the likelihood of a gene perturbation (e.g. knockout or know-down) resulting in specific reversion of disease gene-expression profiles. As we have previously reported, these methods can considerably increase the true positive rate of known target–disease associations (Failli et al., 2019). Here, we introduce ThETA, an R package that easily facilitates performing these efficacy scoring methods. In particular, ThETA provides functions (i) to tailor the workflow of the proposed scoring methods, (ii) to integrate these novel scores with efficacy estimates available on the Open Targets Platform and generate an overall efficacy score that can be used to prioritize target–disease associations. Moreover, ThETA provides visualization tools to depict tissue-specific network paths linking top targets (or genes) and disease-genes, to visualize biological annotations associated to set of selected gene targets. An example of workflow that R-users can implement with the ThETA package is depicted in Figure 1. Fig. 1. An overview of the functions provided by ThETA. (1) ThETA generates target(gene)–disease association scores by using two novel mRNA-based scoring methods. (2) ThETA adds and combines efficacy scores retrieved from alternative drug–target discovery platforms (e.g. Open target platform). The table aligned with the steps 2 and 3 indicates the top-ranked targets for Type 2 Diabetes after using the harmonic sum as prioritization score. (3) ThETA compiles efficacy estimates for all annotated disease–gene pairs, and it (4) provides an R-shiny application to display selected drug targets in tissue-specific networks. The tissue-specific gene networks include three different types of node: known disease-genes (red stars), novel targets (light blue triangles) and bridge genes (blue circles), which connect putative targets to known disease-genes. (Color version of this figure is available at Bioinformatics online.) 2 Methods and features This section describes the main features of the R package ThETA. 2.1 Compiling transcriptome-driven efficacy scores The R package ThETA provides the implementation of two transcriptome-based efficacy scoring methods, namely TSE and MOD scores, respectively. By traversing existing tissue/disease specific networks, the tissue-specific scoring (TSE) method detects gene targets that are closely related to disease-genes in disease-relevant tissues. While, the MOD score estimates the likelihood of a gene perturbation (e.g. knockout and knockdown) to result in specific reversion of disease gene-expression profiles. More details on the TSE and MOD scores can be found in our previous study (Failli et al., 2019). In order to compile the TSE score, the user has to provide the following data inputs: tissue-specific gene-expression data, gene–disease pairs from genome-wide association studies, and human protein–protein interaction (PPI) network. Additionally, ThETA provides three pre-computed datasets for this purpose, including data retrieved from GTEx (Ardlie, 2015), DisGeNET (Piñero et al., 2017) and StringDB (Franceschini et al., 2013). In addition, ThETA includes two datasets representing pre-computed node centrality scores from the human PPI network and disease–tissue association scores. These five datasets allow users to rapidly compile TSE scores. However, users still have the possibility to specify different input data and cut-off values for the selection, e.g. for the most significant disease–tissue associations. The MOD score requires lists of up- and down-regulated genes induced by disease and gene perturbations (e.g. gene knockout, knockdown, etc.). For this task, ThETA included gene lists retrieved from Enrichr (Kuleshov et al., 2016). Details of the input data format are included in the Supplementary Material document called ‘Walkthrough ThETA’. Moreover, known target–disease associations from the DrugBank database (Wishart et al., 2006, 2018), the Therapeutic Target Database (Chen et al., 2002; Wang et al., 2020) and the Comparative Toxicogenomics Database (Davis et al., 2017, 2019) are provided in order to allow users to assess the accuracy of the compiled efficacy estimates. These databases include pairwise associations on drugs, molecular targets and diseases. 2.2 Uploading and combining external efficacy estimates Many different drug–target discovery platforms, such as Open Targets (Koscielny et al., 2017) and DisGeNET, provide efficacy scores for drug–target disease associations. These scores, which are freely available for download from their respective web sites, can be integrated with the efficacy estimates provided by ThETA in order to define more robust efficacy estimates for the prioritization of disease–target associations. Currently, ThETA implements two integration methods: the harmonic sum proposed by the authors of Open Targets (https://docs.targetvalidation.org/getting-started/scoring) and the max function. The max simply considers the maximum value across different efficacy estimates. While, the harmonic sum aggregates individual efficacy scores, sorted by descending score i.e. from higher to lower values. 2.3 Compiling tissue-specific networks and biological annotations for selected gene targets An important novelty presented by ThETA is the use of tissue-specific information for the evaluation of genes as drug targets. Indeed, it is acknowledged that drugs modulating tissue-specific targets are more likely to succeed in phase 3 of clinical trials, and that by targeting tissue specificity there are opportunities to identify drug targets with improved efficacy and safety (Ryaboshapkina and Hammar, 2019). Therefore, given the importance of tissue specificity of drug targets, ThETA includes an interactive visualization tool, based on R-shiny (Chang et al., 2017), to display tissue-specific gene networks highlighting genes and pathways that connect putative targets with the genetic loci that underlie disease susceptibility, or simply disease-genes (see Fig. 1). In these graph structures, the selected genes are distinguished from the disease-genes and the so-called bridge genes, which connects genetic variations associated with diseases and selected targets. Another important feature of the presented R package is the possibility to compile extensive biological annotations. By using enrichplot (Yu, 2018) and clusterProfiler (Yu, 2018) R-packages, ThETA can compile different biological annotations, including KEGG (Kanehisa et al., 2019; Kanehisa and Goto, 2000), GO (Ashburner et al., 2000; The Gene Ontology Consortium, 2019) and REACTOME (Fabregat et al., 2018; Jassal et al., 2020), linked to selected targets. In more detail, given a target, it selects all genes in the shortest pathways connecting that target to known disease-genes, within relevant tissue-specific networks, and compiles corresponding biological annotations. Moreover, ThETA can be used to further explore the genes that are closely related to selected targets by using Random Walk with Restart (Fang and Gough, 2014). 3 Conclusion ThETA offers a user-friendly toolbox in R for the computation of mRNA-driven efficacy estimates of disease–target associations. It allows the user to customize the selection of disease-relevant tissues and the estimation of tissue-specific and MOD scores. Comprehensive datasets are included to facilitate easy adaption of the methods. Moreover, different visualization and biological annotation tools are provided to conduct biological interpretations on putative drug targets. Finally, the R package ThETA provides tutorial vignettes including extensive examples on how to use its functions. Financial Support: none declared. Conflict of Interest: Dr M.F. and Dr J.P. have been working at University of Eastern Finland for Business Finland funded project that explores commercialization of drug–target prioritization technologies. Dr J.P. is an employee of Blueprint Genetics Ltd. Supplementary Material btaa518_Supplementary_Data Click here for additional data file. ==== Refs References Ardlie K.G.  et al ; The GTEx Consortium. (2015) Human genomics. The Genotype-Tissue Expression (GTEx) pilot analysis: multitissue gene regulation in humans. Science, 348 , 648–660.25954001 Ashburner M.  et al (2000) Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat. Genet., 25 , 25–29.10802651 Chang,  W.  et al (2017) Shiny: web application framework for R. R package version, 1(5). Chen X.  et al (2002) TTD: therapeutic target database. Nucleic Acids Res., 30 , 412–415.11752352 Davis A.P.  et al (2017) The comparative toxicogenomics database: update 2017. Nucleic Acids Res., 45 , D972–D978.27651457 Davis A.P.  et al (2019) The comparative toxicogenomics database: update 2019. Nucleic Acids Res., 47 , D948–D954.30247620 Fabregat A.  et al (2018) The reactome pathway knowledgebase. Nucleic Acids Res., 46 , D649–D655.29145629 Failli M.  et al (2019) Prioritizing target-disease associations with novel safety and efficacy scoring methods. Sci. Rep., 9 , 9852.31285471 Fang H. , GoughJ. (2014) The “dnet” approach promotes emerging research on cancer patient survival. Genome Med., 6 , 64.25246945 Franceschini A.  et al (2013) STRING v9.1: protein-protein interaction networks, with increased coverage and integration. Nucleic Acids Res., 41 , D808–D815.23203871 Jassal B.  et al (2020) The reactome pathway knowledgebase. Nucleic Acids Res., 48 , D498–D503.31691815 Kanehisa M. , GotoS. (2000) KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res., 28 , 27–30.10592173 Kanehisa M.  et al (2019) New approach for understanding genome variations in KEGG. Nucleic Acids Res., 47 , D590–D595.30321428 Koscielny G.  et al (2017) Open Targets: a platform for therapeutic target identification and validation. Nucleic Acids Res., 45 , D985–D994.27899665 Kuleshov M.V.  et al (2016) Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic Acids Res., 44 , W90–W97.27141961 Nguyen D.-T.  et al (2017) Pharos: collating protein information to shed light on the druggable genome. Nucleic Acids Res., 45 , D995–D1002.27903890 Piñero J.  et al (2017) DisGeNET: a comprehensive platform integrating information on human disease-associated genes and variants. Nucleic Acids Res., 45 , D833–D839.27924018 Ryaboshapkina M. , HammarM. (2019) Tissue-specific genes as an underutilized resource in drug discovery. Sci. Rep., 9 , 7233.31076736 The Gene Ontology Consortium (2019) The Gene Ontology Resource: 20 years and still GOing strong. Nucleic Acids Res., 47 , D330–D338.30395331 Wang Y.  et al (2020) Therapeutic target database 2020: enriched resource for facilitating research and early development of targeted therapeutics. Nucleic Acids Res., 48 , D1031–D1041.31691823 Wishart D.S.  et al (2006) DrugBank: a comprehensive resource for in silico drug discovery and exploration. Nucleic Acids Res., 34 , D668–D672.16381955 Wishart D.S.  et al (2018) DrugBank 5.0: a major update to the DrugBank database for 2018. Nucleic Acids Res., 46 , D1074–D1082.29126136 Yu, G. (2018). enrichplot: visualization of functional enrichment result. R package version 112.

PMC007xxxxxx/PMC7390992.txt

==== Front Bioinformatics Bioinformatics bioinformatics Bioinformatics 1367-4803 1367-4811 Oxford University Press 32437555 10.1093/bioinformatics/btaa533 btaa533 Applications Notes Structural Bioinformatics AcademicSubjects/SCI01060 ProteinFishing: a protein complex generator within the ModelX toolsuite http://orcid.org/0000-0003-0702-2758 Cianferoni Damiano Systems Biology, Centre for Genomic Regulation (CRG), The Barcelona Institute for Science and Technology, Barcelona 08003, Spain Radusky Leandro G Systems Biology, Centre for Genomic Regulation (CRG), The Barcelona Institute for Science and Technology, Barcelona 08003, Spain Head Sarah A Systems Biology, Centre for Genomic Regulation (CRG), The Barcelona Institute for Science and Technology, Barcelona 08003, Spain Serrano Luis Systems Biology, Centre for Genomic Regulation (CRG), The Barcelona Institute for Science and Technology, Barcelona 08003, Spain Universitat Pompeu Fabra (UPF), Barcelona 08002, Spain ICREA, Barcelona 08010, Spain http://orcid.org/0000-0003-1302-5445 Delgado Javier Systems Biology, Centre for Genomic Regulation (CRG), The Barcelona Institute for Science and Technology, Barcelona 08003, Spain Elofsson Arne Associate Editor To whom correspondence should be addressed. javier.delgado@crg.eu 15 7 2020 21 5 2020 21 5 2020 36 14 42084210 18 3 2020 11 5 2020 18 5 2020 © The Author(s) 2020. Published by Oxford University Press. 2020 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Abstract Summary Accurate 3D modelling of protein–protein interactions (PPI) is essential to compensate for the absence of experimentally determined complex structures. Here, we present a new set of commands within the ModelX toolsuite capable of generating atomic-level protein complexes suitable for interface design. Among these commands, the new tool ProteinFishing proposes known and/or putative alternative 3D PPI for a given protein complex. The algorithm exploits backbone compatibility of protein fragments to generate mutually exclusive protein interfaces that are quickly evaluated with a knowledge-based statistical force field. Using interleukin-10-R2 co-crystalized with interferon-lambda-3, and a database of X-ray structures containing interleukin-10, this algorithm was able to generate interleukin-10-R2/interleukin-10 structural models in agreement with experimental data. Availability and implementation ProteinFishing is a portable command-line tool included in the ModelX toolsuite, written in C++, that makes use of an SQL (tested for MySQL and MariaDB) relational database delivered with a template SQL dump called FishXDB. FishXDB contains the empty tables of ModelX fragments and the data used by the embedded statistical force field. ProteinFishing is compiled for Linux-64bit, MacOS-64bit and Windows-32bit operating systems. This software is a proprietary license and is distributed as an executable with its correspondent database dumps. It can be downloaded publicly at http://modelx.crg.es/. Licenses are freely available for academic users after registration on the website and are available under commercial license for for-profit organizations or companies. Contact javier.delgado@crg.eu or luis.serrano@crg.eu Supplementary information Supplementary data are available at Bioinformatics online. Spanish Ministry of Science and Innovation 10.13039/501100004837 BFU2015-63571-P Spanish Ministry of Science and Innovation 10.13039/501100004837 Centro de Excelencia Severo Ochoa Centres de Recerca de Catalunya (CERCA) Programme ‘la Caixa’ Foundation LCF/BQ/DI19/11730061 ==== Body pmc1 Introduction The ModelX toolsuite (Delgado Blanco et al., 2019) has been developed, among other purposes, for modelling biomolecular interactions. ModelX uses fragment libraries generated by in silico digestion of Protein Data Bank (PDB) structures (Berman et al., 2000) and stored in SQL databases. This strategy has proven successful when applied to the design of DNA–protein and RNA–protein interfaces (Blanco et al., 2018; Delgado Blanco et al., 2019). In the protein–protein interactions (PPI) prediction field, few examples of tools performing fast large-scale docking exist. MEGADOCK 4.0 (Ohue et al., 2014) is one, but it requires sophisticated heterogeneous supercomputing environments equipped with hardware accelerators such as GPUs. Another example is InterPred (Mirabello et al., 2017), which uses homology modelling of binding partners and whole protein superimposition to gather interaction templates. Here, we present ProteinFishing, a tool based on the ModelX philosophy that enables the fast generation of 3D interaction models from observed protein–protein interfaces while fulfilling the requirements for local backbone compatibility. 2 New ModelX tools In addition to ProteinFishing, the latest ModelX release contains two more commands: GeneratePeptides, which is needed to populate the FishXDB database, and FishingLure, an automatized version of ProteinFishing. The three mentioned commands can be used with any type of PDB file containing standard amino acids and/or nucleotides, including X-ray, nuclear magnetic resonance (NMR), homology models or any other PDB model created by users. 2.1 GeneratePeptides command The GeneratePeptides command allows ModelX users to customize their fragment library. It takes PDB structures as input and digests them into protein fragments of user-defined length in an overlapping sliding-window fashion. These fragments are stored in FishXDB and are therefore available for the ProteinFishing algorithm. 2.2 ProteinFishing command ProteinFishing uses protein complex structures as input, and requires the user to select one molecule to be part of the output complex (‘Fisher’, Fig. 1A, light blue) and another molecule to be used as the structural template for the retrieval of new docking partners (‘Hook’, Fig. 1A, red). The algorithm requires the user to define an amino acid window from the ‘Hook’ molecule to query the FishXDB protein fragment database with fragment windows interacting with the ‘Fisher’. When the geometrical backbone compatibility and sequence similarity—according to user-configurable options—matches the peptide window with a FishXDB fragment, the full PDB model (‘Fish’, Fig. 1A) from which the fragment was obtained is placed over the ‘Hook’ fragment by local fitting (Fig. 1B). In this way, complexes containing both the ‘Fisher’ and the ‘Fish’ molecules are built (Fig. 1C). Finally, the generated complexes go through two energy filters: the first filter evaluates the presence of atomic clashes between the backbones of the two molecules, and the second filter uses a customizable threshold for free energy values calculated over the generated models. Free energies (representing backbone compatibility) are obtained using a statistical force field embedded in ModelX. The force field is based on a Boltzmann device (Sippl, 1990) with the Kono modification (Kono et al., 1999) of the Sippl method. The models that pass these filters are later returned as PDB files, together with a summary file showing the number of intermolecular contacts, backbone clashes and energy values. Fig. 1. Algorithm description. (A) The IFN-lambda-R1/IFN-lambda-3/IL-10-R2 complex (PDB: 5T5W) containing the ‘Hook’ (IFN-lambda-3: red), the ‘Fisher’ (IL-10-R2: light blue) and IFN-lambda-R1 (grey); (B) The IFN-lambda-R1/IL-10/IL-10-R2 virtual complex superimposed with the ‘Hook’ window (red); (C) The IL-10/IL-10-R2 or ‘Fish/Fisher’ complex (IL-10: dark blue; IL-10-R2: light blue); (D) A comparison between the reported binding levels (first row) and the ΔΔG of interaction as calculated by FoldX (rows 2–12). A unique colour scale has been used to make energies and percentages comparable. The binding loss (%) numerical scale corresponds to 100%—‘binding levels’ for experimentally measured point mutations (Yoon et al., 2006) and the ΔΔG (kcal/mol) numerical scale corresponds to FoldX interaction energy 2.3 FishingLure command The FishingLure command represents a fully automated, multi-thread version of ProteinFishing in which the algorithm itself determines all possible overlapping sliding windows around the ‘Hook’ residues contacting the ‘Fisher’. The FishingLure command allows the use of ProteinFishing over multiple scanning windows computed in parallel. 3 Demonstration To test the utility of our tool, we focused on the complexes of interleukin-10 (IL-10) with its two receptors (IL-10-R1 and IL-10-R2). While structures are available for the IL-10/IL-10-R1 complex (PDB: 1J7V, 1Y6K), the structure of the IL-10/IL-10-R2 complex has not been elucidated. We chose the crystallographic structure of IL-10-R2 complexed with interferon-lambda-3 and interferon-lambda-receptor-1 (PDB: 5T5W) as input. IL-10-R2 was used as the ‘Fisher’ molecule and interferon-lambda-3 was used as the ‘Hook’ (Fig. 1A, red). Defining the scanning window between residues 89–94 of the ‘Hook’, ProteinFishing yielded 11 models that were then energetically minimized. Next, using the BuildModel command of FoldX (Delgado et al., 2019), five point mutations experimentally reported to modify ‘binding levels’ between IL-10 and IL-10-R2 (Yoon et al., 2006) were modelled. For each mutation, we computed the FoldX free energy variations (ΔΔG [kcal/mol] of interaction) between the ‘Fisher’ and the mutated ‘Fish’. Finally, we compared the variations between the ‘binding levels’ of the five IL-10 mutants with IL-10-R2, as reported in literature, with those predicted by FoldX in each of the 11 models (Fig. 1D). The two best-fitting models, as ranked by the statistical force field of ModelX (Supplementary Table S1; 5T5W_1Y6K_8 and 5T5W_1J7V_7), were found to have the best agreement between FoldX energy values and the experimental results (Yoon et al., 2006) (Fig. 1D and Supplementary Table S3). Complete details of the entire process, including the specific parameters used, can be found in Supplementary Appendix: User Tutorial. 4 Conclusions The tools presented here enable the fast structural modelling of PPI suitable for protein design. The ProteinFishing algorithm can be applied in two types of scenarios. The first scenario, described above, allows the user to model a protein complex for which there is no structure available. Depending on the structures with which the user populates the FishXDB, the possible interactors ‘fished’ can be restricted to specific desired targets, or can be exploratory, using all structures from the PDB. In a second scenario, the tool could be used to model different possible conformations between two members of a complex for which a structure already exists. This second scenario could be useful for performing energetic filtering of different conformations, redesigning interfaces by mutagenesis or identifying putative small-molecule binding pockets in the interface between complex members, for example. Supplementary Material btaa533_Supplementary_Data Click here for additional data file. Acknowledgements The authors would like to thank Professor Jesús Delgado Calvo for the inspiring mathematical discussions that helped us improve the computing efficiency of the algorithm, Tony Ferrar for manuscript revision and language editing, the Centre for Genomic Regulation (CRG) Technology & Business Development Office (TBDO) for support with licensing information, the CRG Tecnologías de Información y Comunicación (TIC) for assistance with web hosting and the Scientific Information Technologies (SIT) for distributed computing. We appreciate all the feedback from the members of the L.S. lab, especially from Samuel Miravet-Verde. Funding This work was supported by funding from the Spanish Ministry of Science and Innovation (Plan Nacional BFU2015-63571-P). The authors also acknowledge the support of the Spanish Ministry of Science and Innovation to the EMBL partnership, the Centro de Excelencia Severo Ochoa and the Centres de Recerca de Catalunya (CERCA) Programme/Generalitat de Catalunya. The project that gave rise to these results was supported by a fellowship from ‘la Caixa’ Foundation (ID 100010434; fellowship code LCF/BQ/DI19/11730061). Conflict of Interest: none declared. ==== Refs References Berman,H. M.  et al (2000) The protein data bank. Nucleic Acids Res, 28 , 235–242. 10592235 Blanco J.D.  et al (2018) FoldX accurate structural protein–DNA binding prediction using PADA1 (Protein Assisted DNA Assembly 1). Nucleic Acids Res., 46 , 3852–3863.29608705 Delgado Blanco J.  et al (2019) Protein-assisted RNA fragment docking (RnaX) for modeling RNA-protein interactions using ModelX. Proc. Natl. Acad. Sci. USA, 116 , 24568–24573.31732673 Delgado J.  et al (2019) FoldX 5.0: working with RNA, small molecules and a new graphical interface. Bioinformatics, 35 , 4168–4169.30874800 Kono H.  et al (1999) Structure-based prediction of DNA target sites by regulatory proteins. Proteins Struct. Funct. Genet., 35 , 114–131.10090291 Mirabello C.  et al (2017) InterPred: a pipeline to identify and model protein–protein interactions. Proteins, 85 , 1159–1170.28263438 Ohue M.  et al (2014) MEGADOCK 4.0: an ultra-high-performance protein–protein docking software for heterogeneous supercomputers. Bioinformatics, 30 , 3281–3283.25100686 Sippl M.J. (1990) Calculation of conformational ensembles from potentials of mena force. J. Mol. Biol., 213 , 859–883.2359125 Yoon S.I.  et al (2006) Conformational changes mediate interleukin-10 receptor 2 (IL-10R2) binding to IL-10 and assembly of the signaling complex. J. Biol. Chem., 281 , 35088–35096.16982608

PMC007xxxxxx/PMC7390993.txt

==== Front Bioinformatics Bioinformatics bioinformatics Bioinformatics 1367-4803 1367-4811 Oxford University Press 32348455 10.1093/bioinformatics/btaa280 btaa280 Original Papers Systems Biology AcademicSubjects/SCI01060 Boosting the extraction of elementary flux modes in genome-scale metabolic networks using the linear programming approach Guil Francisco Departamento de Ingeniería y Tecnología de Computadores, Universidad de Murcia, Murcia 30080, Spain Hidalgo José F Departamento de Ingeniería y Tecnología de Computadores, Universidad de Murcia, Murcia 30080, Spain García José M Departamento de Ingeniería y Tecnología de Computadores, Universidad de Murcia, Murcia 30080, Spain Luigi Martelli Pier Associate Editor To whom correspondence should be addressed. fguil@um.es 15 7 2020 10 7 2020 10 7 2020 36 14 41634170 19 12 2019 16 4 2020 22 4 2020 © The Author(s) 2020. Published by Oxford University Press. 2020 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Abstract Motivation Elementary flux modes (EFMs) are a key tool for analyzing genome-scale metabolic networks, and several methods have been proposed to compute them. Among them, those based on solving linear programming (LP) problems are known to be very efficient if the main interest lies in computing large enough sets of EFMs. Results Here, we propose a new method called EFM-Ta that boosts the efficiency rate by analyzing the information provided by the LP solver. We base our method on a further study of the final tableau of the simplex method. By performing additional elementary steps and avoiding trivial solutions consisting of two cycles, we obtain many more EFMs for each LP problem posed, improving the efficiency rate of previously proposed methods by more than one order of magnitude. Availability and implementation Software is freely available at https://github.com/biogacop/Boost_LP_EFM. Contact fguil@um.es Supplementary information Supplementary data are available at Bioinformatics online. AEI 10.13039/100005910 ERDF 10.13039/501100008530 European Regional Development Fund 10.13039/501100008530 RTI2018-098156-B-C53 ==== Body pmc1 Introduction Elementary flux modes (EFMs) are one widely known tool in computational systems biology for studying genome-scale metabolic networks (GSMNs) reconstruction. EFMs represent a finite set of possible states that can generate all the possible states of the network by using convex combinations (see, e.g. De Figueiredo et al., 2009; Gagneur and Klamt, 2004; Klamt et al., 2005; Rezola et al., 2011; Röhl et al., 2019; Schuster et al., 1999, 2000; Schuster and Hilgetag, 1994). Unfortunately, the cardinality of this set is typically very large so it can only be completely computed in a few cases (Hunt et al., 2014). Methods to compute sets of EFMs can be roughly divided into two families (Planes and Beasley, 2008), according to if they rely on properties of the associated graph (path-finding methods) (Arabzadeh et al., 2018; Hidalgo et al., 2015) or just on the study of the stoichiometric matrix of the network. Regarding the second approach, different methods have been proposed to solve the system of linear equations proposed by the stoichiometric matrix. The most used ones are the double description method (DDM) (Fukuda and Prodon, 1995), the mixed integer linear programming (MILP) method (De Figueiredo et al., 2009; Rezola et al., 2011; Röhl and Bockmayr, 2017) and the linear programming (LP) method (Machado et al., 2012; Tabe-Bordbar and Marashi, 2013). The main advantage of the methods based on DDM and MILP is that they can, theoretically, produce the full set of EFMs. On the other hand, LP methods are faster and can more efficiently produce big sets of EFMs, but they cannot assure when all of the sets of EFMs have been computed. Methods based on LP techniques are capable of producing sets of EFMs at a better efficiency rate, both in time and computer resources. Several efforts have been made to propose efficient algorithms that can produce large enough sets of EFMs in GSMNs (Kaleta et al., 2009; Quek and Nielsen, 2014; Pey et al., 2015). The critical point of these techniques is to use different additional constraints and objective functions to transform the stoichiometric and thermodynamic feasibility constraints into an optimization problem. In this way, for any such LP problem, a solution can be an EFM under certain hypothesis (Pey and Planes, 2014). Different strategies try to minimize the associated issues that can appear (usually, infeasible problems or repeated solutions). The efficiency rate of the LP method is defined by how many LPs are needed to run to find a new EFM. It is known that the ‘ideal’ rate is to find a new EFM for each LP problem run. As far as we know, the best efficiency rate was obtained by Pey et al. (2015), where they achieved an efficiency rate of 1.3 (i.e. a new EFM was obtained for each 1.3 LP problems solved). Perhaps due to the fact that it is quite close to the ideal rate, the interest in this topic has slowly decreased. This article presents a new LP approach in which this ideal rate is overcome. We have developed our proposal based on two ideas extracted from the LP problem-solving procedure. First, the objective function (and so the optimal solution obtained) is just a tool to produce a vertex of the (restricted) cone of solutions given by our constraints. As previously noted (Pey and Planes, 2014) just using one additional positive constraint (in a sense explained below) gives that all the vertices of the restricted cone correspond to the subset of EFMs of the network that fulfills the added constraint. Second, the simplex algorithm (Taha, 2016) is based on a series of steps, and each of them can be viewed, after obtaining a first feasible solution, as a way to pass from one vertex to another. Putting these two ideas together, our approach (named EFM-Ta, which stands for EFMs using the tableau) uses the LP simplex algorithm to produce an initial optimal solution that is an EFM and, after that, some simple steps are performed to obtain new vertices, that is, new EFMs. To limit the possible rounding problems arising from the use of floating point numbers, EFM-Ta performs just one possible step from the initial vertex obtained. However, having plenty of possible ways of doing this step leads to the possibility of obtaining lots of new vertices from the initial one. In the article, we call these vertices the adjacent vertices of the first computed solution. As observed in Gerstl et al. (2015), it is desirable to avoid the appearance of two cycles. In the current context, this is especially important for two reasons. First, these solutions are sometimes false EFMs, in the sense of consisting of pairs of irreversible reactions coming from the splitting of a reversible one (Klamt et al., 2005). In general, the output returned by the simplex method cannot be of this unwanted type, but the use of an additional constraint can produce its appearance. In the experiments performed, we have seen that this is a common behavior that dramatically decreases the efficiency of the methods used. Moreover, these two-length vertices (in the sense of corresponding to EFMs with support of cardinality two) tend to have adjacent vertices that also are two length. So, EFM-Ta has been extended to avoid this kind of undesired solution. EFM-Ta obtains efficiency rates that go far beyond those achieved by previous methods by a factor of 10× and so breaking the ideal rate of one EFM by LP problem solved. Our approach has been tested in previously studied GSMNs models and compared with well-known tools (as EFMEvolver or treeEFM), obtaining an increase in the efficiency rate of more than one order of magnitude. The main contributions presented in this article are the following: A further study of the final tableau of the simplex method. For each LP problem posed, new vertices (EFMs) from the original one are calculated, invoking additional elementary simplex steps. The importance of avoiding two-cycle EFMs as solutions. To do that, reactions that come from reversible ones are avoided, and also those that appear in previous EFMs of length 2. The significant improvement by a factor of 10× of the efficiency rate in the extraction of sets of EFMs. We have not used any heuristic rule for this increment (setting apart the avoiding of two cycles), so we expect that this efficiency rate can be greatly improved in subsequent works. We also want to point out that this method of obtaining new vertices from the initial one can also be easily added to other approaches based on LP methods. The associated matrix to any metabolic network is called its stoichiometric matrix, and it represents the processes that can take place on the network. Rows and columns of S represent the internal metabolites and reactions, respectively. So, if the network has m metabolites and n reactions, the associated matrix S has m rows and n columns. Each value of the matrix is the stoichiometric coefficient of the corresponding column (reaction) in the metabolic equation represented by its row. Any given state of the network is characterized by a vector of variables of length equal to the number of reactions. The corresponding values of the variables give the rate at which each reaction is performing in this state. This vector is called a flux rate. If the time interval considered is small enough, it is normal to assume that the concentrations (internal metabolites) are stable. This leads to the so-called steady-state constraint: (1) S⋅v=0. Let R and Irr be the sets of reactions and irreversible reactions of our network. Irreversible reactions are those that can only occur in one direction, while reversible ones are those that are not irreversible. The usual method to deal with a reversible reactions is to replace it with two irreversible ones, representing its two possible directions. For Irr reactions, the flux rates must be non-negative, what is known as the thermodynamic constraint: (2) v[j]≥0, ∀j∈Irr. A flux vector is called a mode if it fulfills both the steady-state and thermodynamic constraints. If v is a mode, its support supp(v) is defined as the set of those reactions r that appear in v with a non-zero rate. A mode v is called an elementary mode, or EFM, if its support is minimal (i.e. there is no other non-zero mode v′ with supp(v′)⊊supp(v)) (see Schuster and Hilgetag, 1994). For any non-trivial network, its set of modes is infinite. However, the set of EFMs of the network is finite and any non-elementary mode can be written (in a non-unique way) as a linear combination of EFMs using non-negative coefficients (see, e.g. Klamt and Stelling, 2003). In this way, the problem of getting all the modes is translated to the problem of computing the set of all EFMs. To tackle the problem of whether a given mode is elementary or not, the well-known characterization of EFMs in terms of the stoichiometric matrix is used. For any mode v, the submatrix Sv of S is constructed by taking just those columns corresponding to reactions in supp(v). If all the reactions of the network are irreversible, linear algebra methods can prove that a mode v is an EFM if and only if rank(Sv)=k−1 for k=|supp(v)| (Klamt et al., 2005; Terzer and Stelling, 2008). Linear programming techniques As stated in Section 1, one approach to find EFMs is the use of linear optimization techniques. In this way, it is common to use any libraries that are publicly available to handle this kind of problems. To do so, the procedure starts from the stoichiometric and thermodynamic constraints that define a (polyhedral) cone of solutions and introduce an ad hoc objective function to convert them into a linear program. To guarantee that this problem is bounded, the objective function is minimized along with the use of non-negative coefficients. The direct translation of a stoichiometric matrix into a linear program defines a clean linear program [Equation (3)]. (3) Minimize ∑i=1nai⋅v[i]subject to S⋅v→=0→ v[i]≥0 ∀ri∈Irr. But, in this case, one minimal solution is obtained by setting all variables to zero. So, the problem must be additionally restricted to get non-trivial solutions. Different conditions must be used to modify this clean linear program in order to constrain it. It can be imposed that certain set of reactions must appear with non-zero flux (at least one of them). This kind of constraints are called positive constraints (see Acuña et al., 2009 or Hidalgo et al., 2017). Other times, the imposition is that a set of reactions must be absent in the final solution (negative constraints). If the set of reactions of interest is J={ri1,…,rik}, the constraints can be written as (4) ∑Jv[ij]=1 (for positive constraints),  (5) ∑Jv[ij]=0 (for negative constraints). So, different modes (solutions) can be obtained by choosing different sets of positive and negative constraints. However, in order to compute EFMs, only certain types of constraints can be used. To avoid possible infeasible LP problems, just one positive constraint is used. The LP problem posed with just one positive constraint and any number of negative ones gives a solution that is an EFM, whenever all reactions are irreversible and the simplex algorithm is used to solve the LP problem (Pey and Planes, 2014). 2 Materials and methods 2.1 Posing the LP problem: the influence of reversible reactions on the positive constraint In order to avoid possible infeasible LP problems, in addition to just use one positive constraint, the blocked reactions are removed from the network (Hidalgo et al., 2017). As stated before, the conventional way to obtain EFMs using the simplex method is by taking into account the stoichiometric and thermodynamic constraints, and additionally constraining the problem with one positive constraint. Then, the following LP problem has to be posed: (6) Minimize ∑i=1nai⋅v[i]subject to S⋅v=0 ∑Jv[ij]=1 v[i]≥0 ∀ri∈Irr, where the objective function includes a random number of reactions with non-negative coefficients, and only one positive constraint is chosen from a set of reactions J={ri1⋯,rik} using Equation (4). Then, every solution returned by the solver is always an EFM. However, we have found that the combination of one positive constraint together with reversible reactions can produce undesired artifacts in the LP problem, leading to the solution found is not always an EFM. Remind that reversible reactions are replaced with two irreversible ones, representing its two possible directions. This anomalous behavior can be observed in the following example. Consider the metabolic network given by the following graph:in which the reaction r1 is reversible and all stoichiometric values are set to 1. This network has three EFMs with supports {r1,r2} (in this case r1 acts in the direction from m1 to m2), {r1,r3} (r1 in the opposite direction) and {r2,r3}. If the reaction r1 is replaced with two irreversible reactions r1+ and r1−, the following graph is derived: For the following order of variables (r1+,r1−,r2,r3) (only considering internal metabolites), this stoichiometric matrix is derived S=(−11001−1−11). Then, the restricted LP problem obtained by adding the constraint r1=1 yields as solution the tuple (1,1,0,0) that does not correspond to any EFM. The problem in the previous example is due to a solution is not allowed to include together the two reactions obtained from the same reversible one. But, as shown, this cannot be assured by just posing LP problems and one positive reaction. Our next result shows how to avoid this undesired behavior:Theorem 1. Let M be a metabolic network. Without loss of generality, the reactions can be reordered so that {r1,…,rk}  are irreversible and rk+1,…,rn  are reversible. For each reversible reaction ri, a pair of irreversible reactions {ri+,ri−}  representing the two possible signs for the flux of the reaction ri are introduced. Let us take a list of non-negative numbers {a1,…,ak}  such that at least one of the ai is non-zero. Consider the subset V⊂ℝn  defined by the following constraints  (7) S⋅v=0v≥0∑i=1kai⋅v[i]=1. If V≠∅  then any extreme point of V corresponds to an EFM of the network. Observe that, in this result, the positive constraint only includes reactions that does not come from reversible ones. This theorem improves the known result from Pey and Planes (2014) in order to avoid situations as in our previous example. A proof of this result can be found in Supplementary Material of this article. Remark: It is easy to prove that the only situation in which a vertex can include both reactions, coming from a reversible one, is in the trivial case that only includes those two reactions with the same flux value. Even so, in practical situations, it is better to avoid them (see Supplementary Material). To conclude this section, LP problems with a positive constraint must only include reactions that do not come from reversible ones to assure the obtention of EFMs. 2.2 Getting more information from the solution The main drawback of the LP methods is that, for every posed LP problem, a unique EFM is obtained, even though the proposed minimization problem can have a non-unique solution, which is a very common case. Moreover, some of the obtained EFMs are repeated, so the efficiency rate is not quite good. EFM-Ta focuses on obtaining, for any LP problem posed, as many solutions as possible, that is, as many vertices as possible from the cone of solutions. To do so, we leverage the tableau information that the simplex method handles along the process of finding the LP problem solution. Let us start by re-examining the steps performed by the simplex method. As commented in Taha (2016), the simplex algorithm is developed in two phases. The first phase is devoted to obtain a first feasible solution that is a vertex of the polyhedron, introducing some artificial variables. Once a vertex is found, all artificial variables should have zero value and be removed so the process continues in the second phase using just the original ones (and slack variables if needed). The second phase uses the found vertex and the gradient of the function to try to obtain another vertex with a lower function value (if it is minimizing the function). Let us briefly examine how this is done. Associated to the function f(x1,…,xn)=∑aixi the stoichiometric matrix S=(s11⋯s1ns21⋯s2n⋯⋯⋯sm1⋯smn) and the positive constraint ∑djxj=1 the following tableau is started: −a1−a2⋯−an0s11s12⋯s1n0s21s22⋯s2n0⋯⋯⋯⋯0sm1sm2⋯smn0d1d2⋯dn1 Suppose that, in one step of the second phase, a vertex P=(x1,…xn) is set. At this point, the simplex method splits the set of reactions into two disjoint subsets. The reactions of these two subsets are called basic and non-basic variables (respectively). The columns of the tableau (putting aside the first row) in the original one corresponding to the basic variables form an invertible matrix B. In this step, another tableau can be obtained from the original one by suitably multiplying by the matrix B−1 and, in this new tableau, the columns corresponding to the basic variables form the identity matrix (except for the order). z1−c1z2−c2⋯zn−cncs′11s′12⋯s′1nb′1s′21s′22⋯s′2nb′2⋯⋯⋯⋯⋯s′m1s′m2⋯s′mnb′md′1d′2⋯d′nb′m+1. The numbers b′1,…,b′m+1 are non-negative numbers. The vertex P from this tableau is calculated by assigning a zero value to the non-basic variables and solving the trivial associated system. The process of trying to obtain a new vertex is started by selecting a non-basic variable xj. The first element of the associated column (i.e. zj−cj) tells how the proposed change is going to affect the value of the function. If this number is 0, then the function will have the same value in the new vertex that it had in P. If the number is negative (or positive), the function value will decrease (or increase) in the new vertex. To obtain the new vertex, the values assigned to the variables must be carefully chosen so that they are all non-negative and fulfill all the constraints of the problem. To do so, a column j is chosen such that there are positive values s′ij with the corresponding b′i also positive. To proceed, all the values of the jth column with this property are taken and the minimum of the corresponding quotients b′is′ij is calculated. Then, the row i, such that b′is′ij>0 is minimal is selected and a pivoting step is done that converts the j column into the column vector ei=(0,…,0,1,…,0,0)T (having the 1 in the ith position). This process affects the values of the variables in the following way: The variable xj is now considered basic and its value is set to b′is′ij>0. Let xi0 be the basic variable such that its column in the previous step is ei. This variable is considered non-basic (its value is 0). The remaining variables remain basic or non-basic. For the basic ones, their values under the pivoting step must be updated. Suppose a basic variable xi0 such that the only non-zero entry in its column appears in the row i1. Then, the following tableau is formed, in which only the columns corresponding to the reaction xi0 and the new basic variable xj and the rows i1 and i are shown. ⋯zi0−ci0⋯zj−cj⋯c⋯0⋯s′1j⋯b′1⋯⋯⋯⋯⋯⋯⋯1⋯s′i1j⋯b′i1⋯⋯⋯⋯⋯⋯⋯0⋯s′ij⋯b′i⋯⋯⋯⋯⋯⋯ In order to transform the i1 entry of the column j into zero, the ith row multiplied by s′i11js′ij must be added to the i1th row. Therefore, the value of this variable changes from b′i1 to b′i1−b′i·s′i11js′ij Most solvers use the revised simplex method in which, in order to speed up the calculations, the tableau is not calculated in intermediate steps. Instead, only the original tableau is kept in memory and only the matrix B−1 is updated in each step, so the tableau is only available in the final step of the process. Therefore, EFM-Ta only calculates the vertices that can be easily computed from the final tableau. Our approach consists of the following steps: Get the final solution (vertex) and final tableau from the solver Obtain the list of non-basic variables xj For each one, check if there are positive values s′ij with the corresponding b′i value also positive If this is the case, obtain a new vertex (EFM) by manually doing the pivoting step as previously explained 2.3 Tuning the computing of the adjacent vertices in the EFM extraction Our proposed approach has still some drawbacks and requires a fine-tuning optimization. First, in some commercial optimization packages, for example in the IBM package CPLEX used in this article (http://www-01.ibm.com/software/integration/optimization/cplex-optim), not all artificial variables are removed at the end of the first phase of the simplex method if the problem is degenerated (as in this case). Instead, the remaining artificial variables are constrained to have zero value during the rest of the optimization process. When the final tableau is retrieved, a list of basic variables is obtained, but not all the variables that are not in that list are non-basic ones. Second, pivoting steps without having the full list of non-basic variables can produce false extreme points (modes that are not EFMs), preventing some other adjacent vertices from being reached by another pivoting step. Finally, the pivoting steps require the use of the S matrix in dense form, which produces a slowdown of the computation process. Therefore, we have elaborated more on our approach to overcoming these drawbacks. We have tuned our proposal for steps 2–4, replacing the pivoting step by invoking another restricted LP problem. Then, EFM-Ta searches for new vertices invoking another LP problem with the same constraints but starting with the previously obtained solution, changing the objective function and limiting to one the number of steps performed by the solver. EFM-Ta uses as new objective functions those defined from subsets of reactions contained in the support of the solution, and do so while the number of repeated adjacent vertices computed is below a certain threshold. Algorithm 1 summarizes our EFM-Ta proposal.Algorithm 1 Computing extra EFMs from the final simplex tableau Data: Matrix S Data: Integer nEFMs Result: set EFMs Function(runExperiment(S)) EFMs←∅; while (|EFMs|<nEFMs) do  f←randomFunction(R);  cons←randomReaction(Irr);  lp←poseLinearProgram(f,S,cons);  {sol}←simplex(lp);  if isNew(sol) then   EFMs←EFMs∪{sol};   sop=supp(sol);   repeated = 0;   while (repeated < threshold) do    f←randomFunction(sop);    lp←poseRestrictedLinearProgram(f,S,cons,sol);    {newSol}←simplex(lp);    if isNew(newSol) then     EFMs←EFMs∪{newSol};     repeated = 0;    end    else     repeated←repeated+1;    end   end  end end return EFMs; In the Algorithm, the randomFunction builds a function based on randomly chosen reactions from the given set; the randomReaction selects a random reaction from the given set; the poseLinearProgram builds the LP problem with objective function f, stoichiometric matrix S, and positive additional constraint cons; and finally, the poseRestrictedLinearProg builds the LP problem similarly to poseLinearProg but using sol as a starting vertex and limiting the number of steps to one. 2.4 Avoiding repeated EFMs As previously stated, one of the main problems of LP methods is the EFM found in every LP call is repeated. This is a common behavior that significantly decreases the efficiency of the methods used. This problem is exacerbated when using EFMs of length 2 (in the sense of corresponding to EFMs with a support of cardinality two), as due to their reduced size is highly probably to be repeated, and also they tend to have adjacent vertices that also are two length. Therefore, we decided not to include a solution in our set of computed EFMs unless it has a length >2 (this excludes such solutions from the set of computed EFMs, but they can be easily recovered after the process is over). EFM-Ta favors this behavior using two measures: (i) only includes in the positive constraint those reactions that had not previously appeared in any computed solution of length 2 and, (ii) reactions coming from reversible ones are also put in the objective function, so the minimizing process would also try to exclude them. Algorithm 2 describes our final EFM-Ta proposal.Algorithm 2 Improvement of efficiency by discarding two-cycle EFMs Data: Matrix S Data: Integer nEFMs Result: set EFMs Function(2cycleEFM(sol)) if  (length(supp(sol))<3)  then  return TRUE; end return FALSE; Function(runExperiment(S)) EFMs←∅; forbidden←∅; while (|EFMs|<nEFMs) do  f←randomFunction(R);  cons←randomReaction(Irr∖forbidden);  lp←poseLinearProg(f,S,cons);  {sol}←simplex(lp);  if  2cycleEFM(sol)  then   forbidden←forbidden∪supp(sol);  end  else   if isNew(sol) then    EFMs←EFMs∪{sol};    sop=supp(sol);    repeated = 0;    while (repeated < threshold) do     f←randomFunction(sop);     lp←poseRestrictedLinearProg(f,S,cons,sol);     {newSol}←simplex(lp);     if isNew(newSol) then      EFMs←EFMs∪{newSol};      repeated = 0;     end     else      repeated←repeated+1;     end    end   end  end end return EFMs; From Algorithm 2, EFM-Ta works in the following manner: Use the stoichiometric matrix in which all blocked reactions are removed. Start with an initially empty set of forbidden reactions. Choose a positive constraint to extract EFMs from this matrix, by randomly choosing a non-forbidden reaction and imposing that this must be equal to 1 (i.e. the obtained solution must include this reaction from this set). Use a similar method to construct the objective function by randomly choosing a set of not forbidden reactions. Additionally, all the forbidden reactions have to be included in this function so if possible, the LP method will try to push these reactions out of the solution. To increase its randomness, also multiply each coefficient of this objective function by a random number (currently set to between 0 and 1). Solve the LP problem associated with this constraint and objective function. Use the techniques explained in Section 3.2 to find adjacent vertices to this solution. Save to a list those that are different from previously obtained ones. This process will stop if the number of consecutive repeated vertices obtained is greater than a given threshold. Each time a solution of length 2 is obtained, store the reactions involved in the set of forbidden reactions and discard this solution. Remind that our approach can be used in any previously proposed method to obtain sets of EFMs by posing LP problems. 3 Results This section shows the evaluation results of EFM-Ta using a specific case study. 3.1 Experimental framework As suggested in Pey et al. (2015), a good unit to measure the efficiency of LP methods is the number of LP problems solved for one EFM obtained. This makes efficiency as independent as possible of the software, hardware and model chosen. However, as EFM-Ta does more things than just solving LP problems (it has to run the Algorithm 2), this section also shows the number of LPs, restricted LPs (RLPs), and the total time (s) of every experiment. To be able to compare our proposal with previous approaches, we define the efficiency rate as the number of solutions obtained by a unit of time defined as the (mean) time required to solve an LP problem. Regarding our evaluation platform, this is equipped with a double socket Cascade Lake Xeon Gold 6238 (44 cores) @ 2.2 GHz with 384 GB of RAM. The system runs on a CentOS Linux 7.5, running CPLEX 12.10 version from IBM and Python 3.6.8 version from Intel. As a case study, we have chosen three different network models available from BIGG models (Schellenberger et al., 2010), ranging from small to medium-large sizes. The main model used is the reconstruction model for Escherichia coli iAF1260 (Feist et al., 2007) with 2382 reactions and 1668 metabolites, being this model previously used in Pey et al. (2015). The two other models are the model for Cricetulus griseus iCHOv1 (Hefzi et al., 2016) with 6663 reactions and 4456 metabolites and the Homo sapiens Recon3D (Brunk et al., 2018) with 10 600 reactions and 5385 metabolites. Finally, in the following experiments blocked reactions have been removed from the stoichiometric matrix. 3.2 Characterization of EFM-Ta First, this section gives the results for our three models of the case study while extracting 100 000 different EFMs. Table 1 shows the number of LPs, restricted LPs (RLPs), total time (s) and efficiency rate (in number of LP per EFM) needed to compute 100 000 different EFMs in our three network models. Table 1. Characterization of EFM-Ta for computing 100 000 different EFMs in three network models Model used LPs RLPs Time (s) Ef. rate (LP/EFM) iAF1260 68 101 825 355 0.099 iCHOv1 84 103 462 1220 0.062 Recon3D 384 106 979 3879 0.067 Next, we give a detailed information for the iAF1260 network model while extracting 1 000 000 different EFMs. Table 2 shows the total number of LPs, RLPs, total time (s) and efficiency rate (LP/EFM) at every 200 000 EFMs found, up to a 1 000 000 different EFMs. Table 2. EFM-Ta characterization for iAF1260 network model for 1 000 000 different EFMs No. of EFMs LPs RLPs Time (s) Eff. rate (LP/EFM) 200 000 100 205 781 672 0.098 400 000 537 414 107 1360 0.096 600 000 981 621 460 2045 0.095 800 000 1962 831 779 2763 0.095 1 000 000 2535 1 038 022 3445 0.094 For each iteration with a starting solution of length >2, EFM-Ta found between no new EFMs and a maximum of 24 859 (in just one iteration). The mean number of new EFMs obtained for iteration is 853.71 (with standard deviation of 3632.74). Regarding the times reported, we have run EFM-Ta in a sequential way, that is, only one copy of the code is run. We expect that a parallel version of EFM-Ta would be faster and could take advantage of the number of cores that our testbed computer has, but this is out of the goal of this article and is postponed as future work. From Tables 1 and 2, we can conclude that most of the computing effort needed to compute those sets of EFMs rely on solving RLPs (i.e. in computing adjacent vertices), and not in full LPs (just 0.24% of the problems solved). Note that solving an RLP is quite faster that doing so for a full LP. One of the main problems when using random sampling as an extraction method is the number of repeated solutions obtained (Pey et al., 2015). In this case, the rate of repeated solutions is <4.1%, as the number of LP and RLP problems needed to obtain 1 000 000 EFMs is of 1 040 557. This leads to a very effective efficiency rate for EFM-Ta that is also very stable in time (as can be seen in Fig. 1). Fig. 1. Evolution of the efficiency rate during the whole experiment Different experiments with this model show that this efficiency rate tends to remain stable in a value between 0.094 and 0.095 with mean of 0.948 and standard deviation of 0.01. Similar behavior has been observed in the other models used. 3.3 Comparing the efficiency with other tools As pointed out in Pey et al. (2015), an exciting problem consists of obtaining EFMs that contains a target reaction in its support. As other tools have reported the number of EFMs obtained including target reactions, in this section we also do in this way to compare EFM-Ta with them. First, EFM-Ta has to be relaxed to compute EFMs containing a specific target reaction. To do so, we fix the positive constraint to be the flux through the desired reaction and use just the objective function to avoid undesired solutions. We have followed this approach to compute 2000 EFMs in the network model iAF1260 passing through reactions with lysine, threonine and arginine. Table 3 shows the final efficiency rate for different reactions in the network and compare them with those obtained with EFMEvolver (Kaleta et al., 2009) and treeEFM (Pey et al., 2015). EFM-Ta obtained slightly worse efficiency rates than in the general case, but still much higher than in previous approaches. Table 3. Comparison of efficiency rates (LP/EFM) for extracting 2000 EFMs including different target reactions Eff. rate EFMEvolver treeEFM EFM-Ta Lysine 2.23 1.38 0.19 Threonine 1.90 1.64 0.16 Arginine 1.80 1.67 0.16 The efficiency rate highly depends on the reaction chosen. In all the studied cases, the efficiency rate tends to grow, stabilized at a specific rate that differs depending on the reaction. Graphs showing the evolution of efficiency while computing 20 000 EFMs passing through reactions with lysine, threonine and arginine can be found in Supplementary Material. 4 Conclusions and future work We have presented a new method to obtain sets of EFMs called EFM-Ta. Its main difference with previous LP-based algorithms is found in the analysis of the final tableau, which enables us to obtain several solutions for each LP problem by performing additional simplex steps. EFM-Ta searches for new vertices invoking additional restricted LP problems with the same constraints but starting with the previously obtained solution, changing the objective function and limiting to one of the number of steps performed by the solver. We have also analyzed the importance of two-cycle EFMs (that usually represent false EFMs and are caused by the imposition of positive constraints), and the negative impact in our method. This impact can be mitigated by using very simple heuristics combining both the positive constraint and objective function. By extending EFM-Ta with the heuristics to avoid two cycles, we have implemented an algorithm that greatly breaks the ideal efficiency rate of 1 by a factor of more than 10×. We have also shown that this highly improved rate is very stable along the time. Finally, we have compared EFM-Ta with other previously used tools when obtaining the sets of EFMs passing through a given target reaction. EFM-Ta manages it by using just the objective function and using the positive constraint to get the desired target. As expected, this produces a slight decrease in the efficiency rate (at least for some reactions), but the obtained rates remain high, and in all the studied cases, EFM-Ta obtains efficiency rates that are better than the previous ones in one order of magnitude. As future work, we plan to explore the changes in the efficiency of EFM-Ta when applied to bigger models. The experiments performed indicate that the efficiency rates increase with the size of the model considered. We think that this can be a consequence of several characteristics of the model such as having a large number of reactions or the difference between this number and the rank of the stoichiometric matrix. It would be interesting to get a better understanding of this behavior so we can have an initial estimation of the efficiency of our method in different networks. In the end, as for any extraction method based on random sampling, it is usually recommended a statistic analysis of the solutions obtained in order to avoid any bias produced by that method (see, e.g. Hidalgo et al., 2018; Tabe-Bordbar and Marashi, 2013). Therefore, another interesting future work is to carry out that statistic analysis to check the diversity and quality of the EFMs found by EFM-Ta. Funding This work was partially funded by the AEI (State Research Agency, Spain) and the ERDF (European Regional Development Fund, EU) [RTI2018-098156-B-C53]. Conflict of Interest: none declared. Supplementary Material btaa280_Supplementary_Data Click here for additional data file. ==== Refs References Acuña V.  et al (2009) Modes and cuts in metabolic networks: complexity and algorithms. Biosystems, 95 , 51–60.18722501 Arabzadeh M.  et al (2018) A graph-based approach to analyze flux-balanced pathways in metabolic networks. Biosystems, 165 , 40–51.29337084 Brunk E.  et al (2018) Recon3d enables a three-dimensional view of gene variation in human metabolism. Nat. Biotechnol., 36 , 272–281.29457794 De Figueiredo L.F.  et al (2009) Computing the shortest elementary flux modes in genome-scale metabolic networks. Bioinformatics, 25 , 3158–3165.19793869 Feist A.M.  et al (2007) A genome-scale metabolic reconstruction for Escherichia coli k-12 mg1655 that accounts for 1260 ORFs and thermodynamic information. Mol. Syst. Biol., vol. 3 , no. 121, pp. 1-18 Fukuda K. , ProdonA. (1995) Double description method revisited. In: Deza,M. et al. (eds) Combinatorics and Computer Science, Volume 1120 of Lecture Notes in Computer Science. Springer, Berlin, Heidelberg, pp. 91–111. Gagneur J. , KlamtS. (2004) Computation of elementary modes: a unifying framework and the new binary approach. BMC Bioinformatics, 5 : 175. Gerstl M.  et al (2015) Metabolomics integrated elementary flux mode analysis in large metabolic networks. Sci. Rep, 5 , 8930.25754258 Hefzi H.  et al (2016) A consensus genome-scale reconstruction of Chinese hamster ovary cell metabolism. Cell Syst., 23 , 434–443. Hidalgo J.F.  et al (2015) A new approach to obtain EFMs using graph methods based on the shortest path between end nodes. In: OrtuñoF. and RojasI. (eds.) Bioinformatics and Biomedical Engineering, Vol. 9043 of lnbi. Springer International Publishing, Granada, Spain, pp. 641–649. Hidalgo J.F.  et al (2017) Representativeness of a set of metabolic pathways. In: Rojas, I. and Ortuño, F. (eds) Bioinformatics and Biomedical Engineering, Vol. 10208 . Springer International Publishing, Granada, Spain, pp. 659–667. Hidalgo J.F.  et al (2018) Improving the EFMs quality by augmenting their representativeness in LP methods. BMC Syst. Biol., 12 (Suppl. 5 ), 101.30458791 Hunt K.  et al (2014) Complete enumeration of elementary flux modes through scalable demand-based subnetwork definition. Bioinformatics, 30 , 1569–1578.24497502 Kaleta C.  et al (2009) Efmevolver: Computing elementary flux modes in genome-scale metabolic networks. Lect. Notes Inf.  P-157. Klamt S. , StellingJ. (2003) Two approaches for metabolic pathway analysis?  Trends Biotechnol, 21 , 64–69.12573854 Klamt S.  et al (2005) Algorithmic approaches for computing elementary modes in large biochemical reaction networks. IEE Proc. Syst. Biol., 152 , 249–255. Machado D.  et al (2012) Random sampling of elementary flux modes in large-scale metabolic networks. Bioinformatics, 28 , i515–i521.22962475 Pey J. , PlanesF. (2014) Direct calculation of elementary flux modes satisfying several biological constraints in genome-scale metabolic networks. Bioinformatics, 30 , 2197.24728852 Pey J.  et al (2015) Treeefm: calculating elementary flux modes using linear optimization in a tree-based algorithm. Bioinformatics, 31 , 897–904.25380956 Planes F. , BeasleyF. (2008) A critical examination of stoichiometric and path-finding approaches to metabolic pathways. Brief. Bioinform., 9 , 422–436.18436574 Quek L.-E. , NielsenL.K. (2014) A depth-first search algorithm to compute elementary flux modes by linear programming. BMC Syst. Biol.,8, 94 . Rezola A.  et al (2011) Exploring metabolic pathways in genome-scale networks via generating flux modes. Bioinformatics, 27 , 534–540.21149278 Röhl A. , BockmayrA. (2017) A mixed-integer linear programming approach to the reduction of genome-scale metabolic networks. BMC Bioinformatics, 18 , 2.28049424 Röhl A.  et al (2019) Computing irreversible minimal cut sets in genome-scale metabolic networks via flux cone projection. Bioinformatics, 35 , 2618–2625.30590390 Schellenberger J.  et al (2010) Bigg: a biochemical genetic and genomic knowledgebase of large scale metabolic reconstructions. BMC Bioinformatics, 11 , 213.20426874 Schuster S. , HilgetagC. (1994) On elementary flux modes in biochemical reaction systems at steady state. J. Biol. Syst., 2 , 165–182. Schuster S.  et al (1999) Detection of elementary flux modes in biochemical networks: a promising tool for pathway analysis and metabolic engineering. Trends Biotechnol., 17 , 53–60.10087604 Schuster S.  et al (2000) A general definition of metabolic pathways useful for systematic organization and analysis of complex metabolic networks. Nat. Biotechnol., 18 , 326–332.10700151 Tabe-Bordbar S. , MarashiS. (2013) Finding elementary flux modes in metabolic networks based on flux balance analysis and flux coupling analysis: application to the analysis of Escherichia coli metabolism. Biotechnol. Lett., 35 , 2039–2044.24078125 Taha H. (2016) Operations Research: An Introduction, 10th edn.   Prentice Hall, Upper Saddle River, NJ Terzer M. , StellingJ. (2008) Large-scale computation of elementary flux modes with bit pattern trees. Bioinformatics, 24 , 2229–2235.18676417

PMC007xxxxxx/PMC7454006.txt

==== Front NAR Cancer NAR Cancer narcancer NAR Cancer 2632-8674 Oxford University Press 32885168 10.1093/narcan/zcaa018 zcaa018 AcademicSubjects/SCI00030 AcademicSubjects/SCI00980 AcademicSubjects/SCI01060 AcademicSubjects/SCI01140 AcademicSubjects/SCI01180 Survey and Summary ACK1–AR and AR–HOXB13 signaling axes: epigenetic regulation of lethal prostate cancers Kim Eric H Division of Urologic Surgery, Washington University in St. Louis, St. Louis, MO 63110, USA Department of Surgery, Washington University in St. Louis, St. Louis, MO 63110, USA Siteman Cancer Center, Washington University in St. Louis, St. Louis, MO 63110, USA Cao Dengfeng Siteman Cancer Center, Washington University in St. Louis, St. Louis, MO 63110, USA Department of Pathology and Immunology, Washington University in St. Louis, St. Louis, MO 63110, USA Mahajan Nupam P Division of Urologic Surgery, Washington University in St. Louis, St. Louis, MO 63110, USA Department of Surgery, Washington University in St. Louis, St. Louis, MO 63110, USA Siteman Cancer Center, Washington University in St. Louis, St. Louis, MO 63110, USA Andriole Gerald L Division of Urologic Surgery, Washington University in St. Louis, St. Louis, MO 63110, USA Department of Surgery, Washington University in St. Louis, St. Louis, MO 63110, USA Siteman Cancer Center, Washington University in St. Louis, St. Louis, MO 63110, USA Mahajan Kiran Division of Urologic Surgery, Washington University in St. Louis, St. Louis, MO 63110, USA Department of Surgery, Washington University in St. Louis, St. Louis, MO 63110, USA Siteman Cancer Center, Washington University in St. Louis, St. Louis, MO 63110, USA To whom correspondence should be addressed. Tel: +1 314 273 7728; Fax: +1 314 272 7771; Email: kiranm@wustl.edu 9 2020 27 8 2020 27 8 2020 2 3 zcaa01813 8 2020 22 7 2020 04 4 2020 © The Author(s) 2020. Published by Oxford University Press on behalf of NAR Cancer. 2020 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Abstract The androgen receptor (AR) is a critical transcription factor in prostate cancer (PC) pathogenesis. Its activity in malignant cells is dependent on interactions with a diverse set of co-regulators. These interactions fluctuate depending on androgen availability. For example, the androgen depletion increases the dependence of castration-resistant PCs (CRPCs) on the ACK1 and HOXB13 cell survival pathways. Activated ACK1, an oncogenic tyrosine kinase, phosphorylates cytosolic and nuclear proteins, thereby avoiding the inhibitory growth consequences of androgen depletion. Notably, ACK1-mediated phosphorylation of histone H4, which leads to epigenetic upregulation of AR expression, has emerged as a critical mechanism of CRPC resistance to anti-androgens. This resistance can be targeted using the ACK1-selective small-molecule kinase inhibitor (R)-9b. CRPCs also deploy the bromodomain and extra-terminal domain protein BRD4 to epigenetically increase HOXB13 gene expression, which in turn activates the MYC target genes AURKA/AURKB. HOXB13 also facilitates ligand-independent recruitment of the AR splice variant AR-V7 to chromatin, compensating for the loss of the chromatin remodeling protein, CHD1, and restricting expression of the mitosis control gene HSPB8. These studies highlight the crosstalk between AR–ACK1 and AR–HOXB13 pathways as key mediators of CRPC recurrence. Phi Beta Psi Sorority Department of Surgery, Washington University National Institutes of Health 10.13039/100000002 1R01CA208258 5R01CA227025 Prostate Cancer Foundation 10.13039/100000892 17CHAL06 ==== Body pmcPROSTATE CANCER: GENERAL PRE-CRPC TREATMENT STRATEGIES Prostate cancer (PC), a common cancer in men worldwide, is a major cause of cancer-related deaths, with an incidence rate of 7.1% (1). Despite its prevalence, PC is not life-threatening in most patients, because it grows slowly at the organ-confined stage. Surgery, chemotherapy, radiation and androgen deprivation therapy (ADT) are some of the treatment options for men with advanced or high-risk PC (Figure 1, left panel). The 10-year progression-free survival for men with high-risk localized PC varies slightly depending on the treatment but is generally ∼81–85% (2). For men with metastatic PC, treatments include chemotherapy, ADT or immunotherapy. Although treatment for metastatic PC has continually improved, the 5-year survival rate for American men with metastatic PC has remained at ∼28%. Bone is a frequent site for metastasis, which is present in 10–12% of men at initial diagnosis (3,4). Comorbidities for men with active growth of metastatic bone PC may include skeletal fractures and spinal cord compression at the metastatic sites, and are often associated with substantial pain. Radiotherapy is mostly palliative at this stage. Moreover, a characteristic feature of castration-resistant PC (CRPC) is a high degree of heterogeneity and tumor plasticity (5,6). Recent advancements in molecular diagnosis, pathology and imaging support the discovery of new targets for clinically significant PC (Figure 1, right panel). Figure 1. A multipronged approach for the treatment of advanced PCs. Analysis of PC is integrated at the molecular, cellular, histological and clinical levels to improve diagnoses and to tailor treatments for superior outcomes. Gleason scoring system: Gleason score 6, low risk; Gleason scores 3 + 4 or 4 + 3, medium risk; Gleason scores 8–10, high risk. Left panel: Organ-confined (very low risk or low-risk) PC is treated with active surveillance, including follow-up for prostate-specific antigen at 3–6 months and a digital rectal exam/biopsy once per year. Organ-confined (intermediate-risk) PC is treated with surgery or focal therapy. Organ-confined (high-risk) PC is treated with surgery (robotic or laparoscopic prostatectomy, bilateral orchiectomy or transurethral resection of the prostate), radiotherapy [external beam radiation therapy, conformal (computer-assisted) radiation therapy, hypofractionated radiation therapy, brachytherapy or internal radiation therapy], intensity-modulated radiation therapy, proton therapy, chemotherapy (docetaxel, cabazitaxel and paclitaxel), luteinizing hormone-releasing hormone agonists, ADT (enzalutamide, apalutamide, darolutamide, abiraterone acetate and ketoconazole) and immunotherapy (Provenge). Right panel: Elucidating the genetic and epigenetic alterations promoting plasticity of metastatic CRPC (mCRPC) is critical to identifying vulnerabilities and novel therapeutic opportunities for intervention. Understanding the heterogeneous nature of PC with advancements in pathology and prostate magnetic resonance imaging should aid in the optimization of the pre-biopsy risk discrimination for clinically significant PC. CRPC Owing to the preponderance of androgen receptor (AR) expression and function at various stages of PC progression, AR axis-targeted therapies have emerged as critical to combat high-risk disease (7,8). These therapies, namely the ADTs, delay the disease progression (9–12). However, patients invariably develop resistance to ADT and progress to a lethal stage referred to as CRPC (11,13). Approximately 10–20% of men with PC develop CRPC within 5 years of initial diagnosis. The risk of death in men with non-metastatic CRPC correlates with age and comorbidities, and biochemically with a <9-month doubling time of prostate-specific antigen levels (14). CRPCs not only are resistant to first-line ADT but also, within ∼18–24 months, develop resistance to second-generation therapies, including enzalutamide, a potent second-generation AR antagonist that prevents AR nuclear translocation and chromatin binding (15,16), and abiraterone, an androgen synthesis inhibitor (17). A treatment-induced resistance mechanism that emerges in CRPCs is the expression of AR splice variants, such as AR-V7 (18–21). AR-V7 lacks the ligand-binding domain (LBD) and thus is insensitive to the levels of anti-androgens in the tumor milieu (22). Thus, elucidating the molecular alterations promoting mCRPC is critical to identify vulnerabilities and novel therapeutic opportunities for intervention. ACK1–AR–HOXB13 AXIS The AR engages a diverse group of proteins (transcription factors and kinases) and consequently promotes PC growth in androgen-dependent and androgen-independent conditions (Figure 2). Recent molecular, cellular and tumor studies have revealed that the ACK1 tyrosine kinase and the HOXB13 transcription factor are two critical regulators of recurrent CRPC growth, particularly in response to second-generation anti-androgens (23–30). These proteins form a self-sustaining co-regulatory protein network axis in PC in response to ADT. Mechanistically, ACK1 modulates AR gene expression through epigenetic regulation (27), and its androgen-independent activity through tyrosine phosphorylation (23,31,32) and interaction with co-regulators (33). In contrast, HOXB13 regulates AR expression and function, as well as AR and AR-V7 chromatin binding in mCRPCs (29,34). Thus, targeting this axis is critical to suppress CRPC recurrence. The current review discusses these mechanisms in depth. Figure 2. Oncogenic and tumor-suppressive pathways regulated by the ACK1–AR and AR–HOXB13 axes in PCs. AR is a transcription factor whose activity is regulated by the binding of its ligand, the androgen testosterone (T) to the C-terminal LBD. AR engages a diverse group of proteins (chromatin regulators, transcription factors and kinases), thus promoting PC growth in the absence of androgen. Specific chromatin regulators such as the tumor suppressor gene CHD1, which has ATP-dependent chromatin remodeling activity, is required for maintaining genome integrity as well as AR transcription. Most CHD1-enriched sites (80%) are promoter independent and colocalize with AR/HOXB13/FOXA1-enriched enhancers. In the androgen-depleted state, the acetyl (ac)-lysine reader bromodomain proteins (BRDs) have emerged as key mediators of castration resistance. An emerging AR interactor is the non-receptor tyrosine kinase ACK1, which regulates AR gene expression through epigenetic regulation and its androgen-independent activity by tyrosine phosphorylation (pY). HOXB13’s interaction with AR regulates its transcriptional activity by influencing AR/AR-V7 chromatin binding in mCRPCs. HOXB13 regulates a proliferative program in CRPC through overexpression of mitotic kinases (AURKA/B and CIT) and repression of a putative tumor suppressor, HSPB8. Autoregulatory BRD4–HOXB13, BRD4–AR and ACK1–AR circuits perpetuate mCRPCs in response to ADT. Figure was created with www.biorender.com. AR IN CRPC PATHOPHYSIOLOGY AR, a transcription factor, is expressed predominantly by the glandular epithelial cells lining the lumen of the prostatic acini and is essential for maintaining the normal prostate gland differentiated state and secretory functions (35,36). AR has three distinct domains: an N-terminal transactivation domain, a middle zinc-finger DNA-binding domain, a hinge region, and a C-terminal LBD (37). The androgen ligands (testosterone or the higher affinity dihydrotestosterone) bind the ligand-binding pocket in the LBD and subsequently induce LBD dimerization and regulation of AR transactivation (37). Some first- and second-generation anti-androgens may block this dimerization of the LBD and thus inhibit AR activity (37). Androgen-bound AR translocates from the cytosol to the nucleus, binds the androgen response elements (AREs) and subsequently drives expression of target genes such as KLK3/PSA (36,38–40). In normal cells, AR activity is exquisitely regulated through limited expression, cytosolic localization and association with protein complexes (40). In contrast, AR is overexpressed (41–43) or functionally activated in most PCs through a variety of mechanisms (24), such as gene body amplification (44–46), AR distal enhancer amplification (44,45,47), increased histone acetylation/phosphorylation at AR enhancers (27,48,49), overexpression of its co-regulators (29,30,43) and protein-stabilizing post-translational modifications (23,50–52). AR deregulation ultimately leads to increased pathologically active AR at tumor-specific AR-binding sites, an outcome correlating with PC progression (27,53,54). The interaction of the AR with its co-regulators and the chromatin state arguably play vital roles in determining which survival pathways are activated after androgen deprivation. AR epigenetic regulators, such as the BRDs, which regulate lineage-specific programs, have emerged as critical mediators of castration resistance (Figure 2). CRPCs overexpress a subset of the bromodomain proteins BRD2, BRD4 and ATAD2 (55). Mechanistic studies have suggested that BRDs directly interact with AR via their N-terminal regions (56), and this interaction regulates chromatin recruitment at AR target genes (55,56). Interaction between AR and BRD4 facilitates increased chromatin accessibility in CRPCs compared with benign prostatic hyperplasias. In agreement with this finding, the chromatin displays maximal internucleosomal spacing, increased transcriptionally permissive histone acetylation and decreased repressive methylation, thus underscoring the structural changes promoting resistance to AR-targeted therapies (57). AR interaction with other epigenetic modifiers, such as histone/lysine acetyltransferases and histone/lysine demethylases, increases chromatin accessibility at AR target genes, some of which are also overexpressed in PC; however, epigenetic alterations occur in a site-specific and context-dependent manner (50,58–60). These epigenetic alterations include removal of repressive histone H3 lysine 9 me2/3 methylation (H3K9me2/3) by the AR interacting histone demethylases KDM1A and KDM3A (61–64) or increased histone H3K27 acetylation by CBP/p300, thus promoting expression of a subset of AR target genes (50,58–60). Whereas the BRD inhibitor JQ1 (29,55,56) and the p300/CBP inhibitors A485 and GNE-049 (65,66) suppress AR target gene expression and inhibit prostate xenograft tumor growth, their broad epigenetic substrate activity has limited their potential clinical use. Consequently, the identification of more targeted approaches and therapeutic vulnerabilities in CRPC is being pursued. ACK1/TNK2: A NOVEL REGULATOR OF AR EPIGENETIC EXPRESSION AND FUNCTION IN CRPCs An emerging AR interactor is the non-receptor tyrosine kinase ACK1, which contributes to PC pathophysiology by rapidly integrating growth signals from receptor tyrosine kinases, thus supporting the growth of PC cells in an androgen-independent manner (67,68). ACK1 phosphorylates AR and AKT, and subsequently initiates an intracellular response associated with cell survival and metabolism (23,69) (Figure 2). The AKT serine/threonine kinase is frequently activated in metastatic PCs and is a well-studied effector of PI3K signaling (70). However, inhibition of PI3K signaling is insufficient to completely block tumor growth in Pten-null PCs, despite inhibition of activated AKT. Increased Ar gene transcription, in a Her2/Her3-dependent manner, is frequently observed in the presence of Pten ablation (71). Moreover, significant increases in phosphorylated ACK1 (pY284-ACK1), AR (pY267-AR) and AKT (pY176-AKT), as well as total AR expression, are observed during human PC progression and in metastasis (23,72). Stimulation of the CRPC-derived cell line CWR-R1 with heregulin induces transactivation of AR, in agreement with HER2’s modulation of AR protein stability and function in PC (73,74). HER2 promotes ACK1 kinase autoactivation (23,72), which is followed by phosphorylation of AR and AKT at Y267 and Y176, respectively, in PTEN-deficient human PC models (23,69,72). Physiologically, probasin-caACK1 (catalytically activated human ACK1) transgenic mice with elevated levels of pY284-ACK1 and pY176-AKT develop prostatic intraepithelial neoplasia as well as rare prostate adenocarcinomas. At the molecular level, the ACK1/pY267-AR complex enters the nucleus and binds the AR promoters/enhancers as well as AR target genes (KLK3), DNA damage response genes (ATM and TP53) and genes associated with lipid metabolism (SREBF1) (23,27,32,69) (Figure 2). In agreement with this finding, human prostate tissue microarray profiling with AR and activated ACK1 antibodies has revealed upregulation of both the AR and activated ACK1 in CRPC (27,72). In an androgen-independent state, the ACK1/AR complex is recruited to the ARE III enhancer of the KLK3/PSA gene (23) as well as to the AREM enhancers present ∼100 kb upstream of the AR gene (27). ACK1 phosphorylates histone H4 at tyrosine 88 (forming H4-pY88) in the context of the chromatin at the AREM enhancers. Treatment with the anti-androgen enzalutamide does not affect ACK1-mediated histone H4-Y88 phosphorylation (27). Subsequently, these pY88-H4 epigenetic marks act as docking sites for the WDR5/MLL2 complex, thus promoting AR and AR-V7 transcription through deposition of transcription-activating H3K4 trimethyl marks and recruitment of the transcription machinery (27). This feed-forward epigenetic mechanism drives CRPC progression and is targetable with ACK1 small-molecule inhibitors (Figure 3). Figure 3. Epigenetic regulation of AR by ACK1 in mCRPCs. The ACK1–AR complex is recruited to the AR enhancer, where it phosphorylates histone H4 at residue Y88, thus driving AR transcription. The ACK1 catalytic inhibitor (R)-9b blocks tyrosine phosphorylation and autoactivation of ACK1 as well as tyrosine phosphorylation of its substrates, thereby inhibiting the androgen-independent AR program in CRPCs. PC cells invariably develop resistance to AR antagonists, and intriguingly, perturbations in AR are common in patients with CRPC. These resistance-causing perturbations include structural alterations in AR enhancers, AR gene amplifications or mutations and AR splice variants that lack the LBD, such as AR-V7. Post-translational modifications that facilitate androgen-independent AR recruitment to chromatin regions distinct from those targeted by androgen-bound AR, as well as the interaction of AR-V7 with HOXB13, are among the distinct mechanisms of transcriptional regulation in CRPCs. Figure was created with www.biorender.com. AR ENHANCERS AS MEDIATORS OF THERAPEUTIC RESISTANCE IN CRPCs Another epigenetic mechanism of therapeutic resistance observed in most mCRPCs is the amplification of a 9-kb genomic region ∼620–700 kb centromeric to the AR gene body, in a manner possibly mediated by the selective pressure of exposure to anti-androgens (44,45,47). The chromatin at this distal AR enhancer contains H3K27 acetylation marks. An intriguing feature of this distal AR enhancer is DNA hypomethylation and the absence of histone marks in adult tissues (45). Moreover, cistrome analysis has revealed the binding of pioneer factors, including HOXB13, FOXA1 and GATA2, at this AR enhancer in LNCaP cells as well as in benign and primary localized prostate tumors (45). However, this AR enhancer lacks H3K27ac, H3K27me3 and H3K4me2 epigenetic marks in primary tumors, thus defining it as a vestigial enhancer that is reactivated specifically in mCRPCs (45). These results suggest that a prerequisite step in CRPC resistance to anti-androgens may be the overexpression and recruitment of these pioneer transcription factors to de novo genomic loci, before transcriptionally activating acetylation is deposited at H3K27 by histone acetyltransferases. The mechanistic details of the enhancer amplification are unclear but may depend on the activity of pioneer factors and chromatin remodeling enzymes. The pioneer factors HOXB13, FOXA1 and GATA2 regulate androgen-dependent and androgen-independent AR function in PCs (75,76). In early-stage PCs, FOXA1 co-regulates the luminal AR transcriptional program through specific gain-of-function mutations in its DNA-binding domain (77). However, the mutational spectrum of FOXA1 differs in metastatic PCs, in which it activates the Wnt signaling program or undergoes structural rearrangements within its locus that drive autoexpression (77). Moreover, 10–20% of CRPC tumors overcome AR dependence and consequently evade AR-targeted therapy (78). One such mechanism is the aberrant N-MYC expression in CRPCs, which enables the switch from luminal epithelial to neuroendocrine lineage subtypes. However, despite the dominance of N-Myc, the pioneer transcription factors HOXB13 and FOXA1 are recruited to specific genomic loci in an AR-independent manner in NEPCs (79). Collectively, these studies underscore a gain of function for the pioneer factors in lineage plasticity. EPIGENETIC REGULATION OF HOXB13, A PC RISK GENE HOX gene family members are evolutionarily conserved and comprise a large group of transcription factors (80). This conservation is evident in their structural organization and the temporal and spatial expression of their individual members in tissues, cells and lineage-specific manners (80). HOX genes regulate development during early embryogenesis through a highly coordinated collinear expression program along the anterior–posterior axis, thus establishing the master body plan (81). Mutation within or deregulated expression of individual HOX genes is associated with malignancies of different cell types as well as developmental defects (82). Hoxb13 is essential for the proper development of the ventral prostate gland in mice (83). In humans, HOXB13 is expressed predominantly in the luminal epithelial cells of the prostate and, to a lesser extent, in cells that compose the basal layer. This gene encodes a 34-kDa sequence-specific DNA-binding protein with a proximal N-terminal HOXA homology domain and a distal 60-amino acid DNA-binding homeodomain. Although its expression is independently regulated and is not under the control of AR during normal prostate development, AR- and FOXA1-binding sites are present near the HOXB13 promoter in PC, thus suggesting a role for chromatin in this de novo recruitment of transcription factors in malignant cells (Figure 4). Moreover, in men with PC, a rare germline mutation in HOXB13 at rs138213197, corresponding to the amino acid change G84E (glycine to glutamate), in addition to mutations at other sites, has been reported (84–86). Specifically, HOXB13 G84E is associated with an increased risk of additional cancers (87). However, the mechanism through which these mutations contribute to prostate tumor development is unclear, thus suggesting the existence of other mechanisms underlying HOXB13 pathogenesis. Figure 4. Pioneer transcription factors are recruited to the HOXB13 regulatory region. Meta-analysis of publicly available chromatin immunoprecipitation sequencing (GSE56288) to probe AR, FOXA1 and HOXB13 binding has revealed occupancy of these factors at the HOXB13 transcription start site and 3′ exon 2 in human prostate tumors and in prostate cell lines expressing all three factors. Two BRD4-binding sites, BRAH1 and BRAH2 (BRD4 recruitment site at HOXB13), are present at nucleotides 268 and 799 upstream of the HOXB13 transcription start site. In mice, Hoxb13 expressed in prostatic lineage cells is developmentally regulated, and its expression is maintained in a poised state through combinatorial expression of transcriptionally activating H3K4me3 and repressive H3K27me3 epigenetic marks (88,89). These bivalent domains are a characteristic feature of the promoters of developmentally regulated protein-coding genes, including Hoxb13 (89). Loss of H3K27me3 at the HOXB13 gene locus also occurs in primary colon tumors, as compared with matching normal mucosa (90). A dynamic swapping of repressive for activating acetylation marks at H3K27 facilitates rapid activation and deactivation of gene expression patterns in response to intracellular and external stimuli (91). However, another layer of epigenetic reinforcement instituted by DNA methylation maintains developmentally regulated genes in a repressed state in differentiated adult cells. DNA hypermethylation at cytosine 5 in the CpG dinucleotide at existing or new loci is targeted for gene silencing through the action of the EZH2-containing polycomb repressor complex, which deposits H3K27me3 marks at the chromatin (92–94); an example is colorectal cancers, in which the DNA methyltransferase DNMT3B targets HOXB13 for repression (95). Therefore, to activate developmental genes such as HOXB13 with hypermethylated CpG islands, a three-step process is required: (i) DNA demethylation; (ii) removal of the repressive H3K27me3 marks (96); and (iii) their replacement with activating histone acetylation marks. Recent studies have revealed enrichment in H3K27 acetylation as well as H4K5 and K8 acetylation at the HOXB13 promoter-proximal region in CRPC (29). Moreover, treatment with GSK126, a PRC2 catalytic inhibitor of H3K27me3, does not affect HOXB13 expression in CRPC cell line C4-2B indicating a state of constitutive activation (29). Increased HOXB13 expression may also open chromatin in CRPCs, thus permitting transcription and compensating for the loss of activity of chromatin remodeling enzymes, such as the tumor suppressor gene CHD1, which encodes a protein with ATP-dependent chromatin remodeling activity (97–99) (Figures 2 and 5). In the non-malignant state, the chromatin remodeling activity of CHD1 maintains genome integrity and AR transcriptional functions (100). In addition to regulating AR chromatin-binding activity, CHD1 regulates homologous recombination (HR) double-strand break repair, and its decreased expression is associated with increased sensitivity to ionizing radiation, and sensitivity to poly(ADP-ribose) polymerase (PARP) inhibitors (101,102). Approximately 10–15% of primary human PCs exhibit recurrent deletions of CHD1 (103,104). However, in mice, prostate-specific genetic ablation of Chd1 does not directly lead to PC, although combining Chd1 ablation with Pten deletion leads to invasive carcinoma, thus indicating cooperativity between these two tumor suppressors (100). Furthermore, silencing of CHD1 expression in high AR-expressing human prostate xenograft tumors (LNCaP/AR) confers resistance to AR blockade, specifically to enzalutamide, owing to alterations in chromatin and enrichment in transcription factors that direct cells away from the luminal lineage (105). Mechanistically, CHD1 regulates the binding of AR and thereby differentiation through promoting the expression of the canonical AR transcriptome. Its deficiency in tumor cells is associated with a diversion of AR binding toward a malignant program (100). In agreement with this finding, CHD1 deletion results in enrichment in HOXB13 binding at most AR enhancers, thus suggesting that the ability of pioneer factors to gain access to inaccessible chromatin is likely to compensate for the loss of CHD1 chromatin remodeling activity (100) (Figure 5). Figure 5. HOXB13 may compensate for CHD1 deficiency, thus promoting chromatin accessibility. CHD1 regulates the binding of AR and consequently differentiation through promoting the expression of the canonical AR transcriptome. In CHD1-deficient conditions, AR enhancers are enriched in HOXB13 binding and HOX motifs. CHD1 deficiency in tumor cells is associated with a diversion of AR binding toward a malignant program likely mediated by the activation of HOXB13 transcriptional network. Figure was created with www.biorender.com. BRD4-MEDIATED EPIGENETIC REGULATION OF HOXB13 IN CRPCs Recent studies have revealed the epigenetic regulation of HOXB13 gene expression by BRD4 as an important mechanism of transcriptional upregulation in PCs. The BET bromodomain protein binds two HOXB13 promoter-proximal sites, BRAH1 and BRAH2 (29). The chromatin at the BRAH1 and BRAH2 sites is characterized by the presence of H3K27 acetylation and enrichment of RNA polymerase II but not repressive H3K27 methylation, a signature indicative of actively transcribed regions (29). In support of this epigenetic regulation, different classes of BET bromodomain inhibitors—such as the prototype compound JQ1 or the dual activity bromodomain kinase inhibitors MA4-022-1, MA4-022-2 and SG3179—effectively suppress HOXB13 expression and inhibit CRPC proliferation and xenograft tumor growth (29,30,56) (Figure 6). Although JQ1 also suppresses expression of c-MYC, another important oncogenic transcription factor, it is not a major effector of BRD4-mediated survival in HOXB13-positive and AR-positive CRPCs (29,30,56). Exogenous expression of HOXB13 rescues JQ1-mediated inhibition of cell proliferation, thus revealing that HOXB13 is a critical downstream effector of BRD4 in CRPCs (29). This finding may be attributed to the active BRD4–HOXB13 regulated network of genes that promote the cell cycle, DNA metabolism, cell division and survival in a MYC-independent manner (29,30). An example is the expression of the mitotic kinase genes AURKA/B in highly proliferative AR-positive luminal epithelial PCs and in circulating tumor cells from patients with CRPC (106). AURKA/B is a target of the MYC transcription factors in neuroendocrine cancers (107) and other cancer types (6,108), but it is regulated by HOXB13 in a subset of CRPCs (29,30). BRD4 deploys HOXB13 and MYC in aggressive PCs, creates a transcription factor dependence in CRPCs and NEPCs (neuroendocrine PCs), respectively, and consequently sensitizes them to BRD4 inhibitors. Collectively, these studies also highlight a means to target the BRD4–HOXB13–AR epigenetic axis in CRPCs and to restrict the pathogenicity of CRPC progression (Figure 6) (29). The contextual dependence of CHD1 normal and deficient human prostate tumors on AR and HOXB13 signaling suggests that ACK1 inhibitors, as well as BET bromodomain inhibitors, may be potential therapeutic targeting strategies either individually or in combination. Figure 6. Targeting AR/HOXB13 pathways in lethal PCs. Among the several known AR co-regulators, HOXB13 has a unique role in defining the AR transcriptome in a context- and stoichiometry-dependent manner. BRD4 epigenetically regulates HOXB13 expression in PCs. BRD4–HOXB13 co-regulated transcriptional networks promote androgen independence and cell proliferation. In addition, AR-V7, a splice variant of AR upregulated in CRPCs, lacks the ability to bind androgen and relies on HOXB13 for chromatin recruitment. HOXB13 downregulation induces HSPB8 expression in mCRPCs, and inhibits cell migration and proliferation. Consequently, HOXB13-positive PCs are sensitive to BET bromodomain inhibitors. HOXB13 REPRESSES HSPB8, A MITOSIS REGULATOR IN PC HOXB13 acquires a neomorphic oncogenic function in human cancers, and its expression is associated with recurrence after radical prostatectomy and aggressive PCs (109–111). The role of HOXB13 in CRPC recurrence is supported by its ability to recruit ligand-independent AR and AR-V7 to genomic sites (29,34,112). In addition to transcriptional activation, HOXB13 may promote prostate tumor progression through direct transcriptional repression of tumor suppressors. A genetic screen for HOXB13 effectors has revealed that the heat shock protein gene HSPB8 is repressed in PCs (30). Accordingly, analysis of The Cancer Genome Atlas (TCGA) prostate adenocarcinoma dataset has revealed that whereas HOXB13 gene expression is increased (Figure 7A), HSPB8 expression is significantly downregulated during disease progression and this decrease is independent of the presence of known frequent genetic alterations observed in PC (Figures 7B-C). Silencing of HOXB13 reverses HSPB8 gene expression in multiple PC cell lines (30). Metastatic PCs with low HSPB8 levels are hyperproliferative, as evidenced by their high expression of mitotic kinases. In contrast, overexpression of HSPB8 restrains proliferation and migration, thus suggesting that it restricts mitosis and motility (30). Consistently, HSPB8 gene expression shows significant differences between patients with prostate adenocarcinoma with metastasis to lymph nodes (LN) LN-N1 compared to LN-N0 with no evidence of metastasis (Figure 7D). Physiologically, autosomal dominant mutations in HSPB8 are associated with myopathy and benign prostatic hyperplasia (113,114). Molecularly, a role of HSPB8 in conjunction with BAG3 in clearing AR polyQ mutant proteins is proposed in a recent study, which, if not cleared, are associated with spinal and bulbar muscular atrophy, an X-linked motoneuron disease (115). Whether specific subsets of high-HOXB13/low-HSPB8 prostate cells are channeled away from differentiation and redirected toward proliferation remains unclear. It will be critical to determine whether a high-HOXB13/low-HSPB8 state defines the subset of men with rapidly growing and potentially fatal cancers, such as the high-risk ductal carcinomas that exhibit high HOXB13 expression (116). Together, these studies suggest that HSPB8 is a mitotic checkpoint protein in normal prostate cells and is one of the first defenses eliminated by PC cells to boost proliferation. Figure 7. HSPB8 expression is downregulated during PC progression (TCGA dataset). (A) Box plots reveal an increasing trend in HOXB13 expression with increasing tumor grade, normal versus cancer; *P = 6.4E−05, **P = 4.4E−08, ***P = 1.81E−08, ****P = 6.27E−09, ns = not significant. (B) Box plots reveal a decreasing trend in HSPB8 expression with increasing tumor grade; normal versus cancer; *P = 8.3E−07, **P = 4.94E−09, ***P = 8.84E−10, ****P = 7.7E−10; Gleason score 6 versus 7, 8 and 9, respectively: *P = 0.013, **P = 0.0015, ***P = 0.0007. (C) Box plots reveal decreased HSPB8 mRNA expression, independently of genetic alteration; normal versus genetic alteration; **P = 6.0E−05, ****P> 1.3E−07, ns = not significant, n = number of cases. (D) HSPB8 mRNA expression is significantly downregulated in LN-positive patients; normal versus LN(N0) P=3.63E-09; normal versus LN(N1) P = 4.77E-10; N0 versus N1 **P = 3.0E−02. The t-test was performed using a PERL script with the Comprehensive Perl Archive Network module ‘Statistics: T-Test’ as described by (139). Collectively, these new studies have uncovered the importance of epigenetic deregulation of AR and HOXB13 as critical mechanisms underlying the lethality of metastatic PCs. Importantly, these studies have revealed the requirement of agents other than anti-androgens that target the ACK1–AR and AR–HOXB13 axes. Because HOXB13 facilitates ligand-independent chromatin recruitment of the AR splice variant AR-V7 in CRPCs, and ACK1 regulates AR/AR-V7 expression, targeting the ACK1 kinase with selective inhibitors may be beneficial. TARGETING CRPCS WITH THE NOVEL ACK1 INHIBITOR (R)-9b The small-molecule inhibitor (R)-9b has emerged as the most promising candidate, owing to its favorable drug-like properties and limited off-target activity (27,117). Nearly all patients with PC treated with abiraterone and enzalutamide acquire resistance via increased expression of AR-V7 splice variants (22); thus, a therapeutic strategy that not only inhibits full-length AR gene expression but also suppresses AR-V7 variant expression is warranted (27). (R)-9b blocks ACK1-mediated H4-Y88 phosphorylation and suppresses AR/AR-V7 transcription, thus overcoming ENZ-resistant CRPC growth (27). Currently, no available data indicate that ACK1 or pY88-H4 epigenetic marks have roles in promoting splicing dysfunction causing AR-V7 expression. ACK1/pY88-H4 signaling is likely to be primarily involved in keeping the AR locus transcriptionally active, even in the presence of AR antagonists. Eventually, cancer cells may use distinct mechanisms to switch from AR to AR-V7 transcription, or low AR-V7 expression due to splicing errors in an AR antagonist-rich environment may select for a subset of resistant CRPCs. ADT resistance may also be mediated by rare CD44 populations (∼0.1%) present in the luminal epithelial-derived PC population (28). ACK1 kinase is active in these CD44 stem-like cells, and in preclinical models, ACK1 inhibitors have been found to effectively suppress CD44 stem-like cell tumor growth (28). Thus, (R)-9b may exemplify a ‘third-generation’ AR antagonist that can overcome CRPC and stem-like cell renewal. Pathogenic ACK1 activation has also been reported in cancers of the lung, pancreas and breast, and AKT kinase is a major target activated by ACK1. Thus, (R)-9b has the potential for targeting additional cancers. Other ACK1 inhibitors are in various stages of developmental pipelines, and, together with (R)-9b, hold promise for addressing not only anti-androgen resistance but also radio-resistance of CRPCs (27,31,72,117–120). The tyrosine kinase Src is active in a subset of AR-positive PCs and feeds into the AR axis; Src phosphorylates AR and regulates AR target genes essential for CRPC survival (26,31). However, AR/Src-regulated genes appear to be distinct from those regulated by ACK1: ACK1 performs a specific function in PC stem-like cell renewal (28), whereas the subset of Src-regulated genes is associated with metastasis and poor prognosis (26). Together, kinase inhibitors targeting these non-receptor tyrosine kinases Src and ACK1 have the potential to overcome multiple mechanisms of anti-androgen resistance in hormone refractory PCs. ACK1/AR-pY267 SIGNALING IN THE DNA DAMAGE RESPONSE AND RADIO-RESISTANCE OF CRPCs PC has a heritability of 57% and exhibits a high degree of genetic mosaicism (Figure 8). Consequently, genetic alterations affect macromolecular interactomes involved in chromatin regulation, replication, transcription, DNA repair, RNA splicing and protein turnover (Figure 8) (121,122). Recurrent genetic alterations found in a subset of early-onset PCs are germline or somatic mutations in genes involved in DNA repair comprising the HR-dependent repair pathways (BRCA1, BRCA2, ATM and FANCA); this percentage doubles in men with metastatic PC (122–124). In contrast, mCRPCs display both frequent amplification and overexpression of the AR gene as well as frequent mutations in the PTEN and TP53 tumor suppressor genes, thus suggesting that selection of cells with AKT activation and loss of the G1 checkpoint may result in a survival advantage due to the avoidance of apoptosis and activation of metabolic programs (123). In agreement with these findings, mouse modeling studies with Pten and Trp53 deletions have recapitulated the genomic alterations observed in humans (125). Activated oncogenic kinases further establish a hierarchy in this regulation by modifying several key proteins and promoting PTEN-independent AKT activation (67). Prostates from probasin-caACK1 transgenic mice expressing activated human ACK1 as well as its substrates pY176-AKT and pY267-AR display higher expression of the DNA damage response protein ATM in cancerous tissue than in the normal prostate (32,69). Androgen-independent pY267-AR recruitment to the ATM enhancer facilitates an increase in ATM levels in CRPCs (32). Moreover, human CRPCs exhibit a significant increase in ATM gene expression that correlates with activated ACK1/pY267-AR expression. Similarly to the participation of androgen-bound AR in the cellular response to DNA damage (126–129), androgen-independent ACK1/AR signaling is active under castration conditions and is associated with radio-resistance of CRPCs (32). Notably, this radio-resistance is targetable with novel ACK1 inhibitors. Figure 8. Heterogeneity of PCs. Mutual exclusivity as well as collaboration between genetic and epigenetic alterations within individual subsets provides a framework of genetic redundancy, tumor heterogeneity and tumor evolution. Master regulatory networks directed by pioneer transcription factors respond to cues from the intracellular and extracellular environment and consequently switch cell dependences on and off and enable transition to different cell states. Oncogenes, tumor suppressors and collaborator genes that contribute to the heterogeneity of PC include BRCA2, DNA repair-associated (breast cancer gene 2); BRCA1 (breast cancer gene 1); PARP [poly(ADP-ribose) polymerase 1]; ATM (ataxia telangiectasia mutated); MLL4/KMT2B (lysine methyltransferase 2B); KDMs (histone lysine demethylases); CBP/CREBBP (CREB-binding protein, histone lysine acetyltransferase); EP300 (E1A-binding protein P300); EZH2/KMT6 (enhancer of zeste 2 polycomb repressive complex 2 subunit); ACK1/TNK2 (tyrosine kinase non-receptor 2); SRC (SRC proto-oncogene); WEE1 (WEE1 G2 checkpoint kinase); AR (androgen receptor); HOXB13 (homeobox B13); FOXA1 (forkhead box A1); MYC (myelocytomatosis viral oncogene); ERG (erythroblast transformation-specific transcription factor); SPOP (speckle-type BTB/POZ protein); DNMT (DNA methyltransferase); PCAT1 (prostate cancer-associated transcript 1); ARLNC1 (AR-regulated long noncoding RNA 1); BRD4 (bromodomain containing 4); MLL2/KMT2D (mixed-lineage leukemia protein 2); WDR5 (WD40 repeat protein); AKT1 (V-Akt murine thymoma viral oncogene-like protein 1); CDK7 (cyclin-dependent kinase 7); CDK12 (cyclin-dependent kinase 12); AURKB (Aurora kinase B); SREBF1 (sterol regulatory element-binding transcription factor 1); PTEN (phosphatase and tensin homolog); TP53 (tumor suppressor protein P53); CHD1 (chromodomain helicase DNA-binding protein 1); NKX3.1 (NKX3 homeobox 1); and RB1 (retinoblastoma 1). Whether ACK1 inhibition can synergize with PARP inhibitors in mitigating mCRPC growth is currently unknown. PARP functions in the base excision repair pathway, which repairs single-strand DNA breaks. Importantly, mutations in genes comprising the HR repair pathway sensitize cells to PARP inhibition. Clinically, chemotherapy- and ADT-treated patients with mCRPC with HR defects rather than HR proficiency show a higher response rate to PARP inhibition (88% in HR deficient versus 33% HR proficient) and longer progression-free survival and overall survival (130). However, PARP may negatively regulate the non-homologous end joining pathway. Consequently, in HR-deficient cells treated with a PARP inhibitor, non-homologous end joining-mediated repair is stimulated, thereby contributing to cytotoxicity and genome instability (131). Whether ACK1 inhibitors synergize with PARP inhibitors in preventing the emergence of resistant clones and might extend treatment benefits remains to be determined. ACK1 inhibitors may also have treatment potential through their indirect participation in mechanisms leading to tumor development, such as the rare but recurrent somatic missense mutations in the tumor suppressor gene SPOP (CULLIN3‐based E3 ubiquitin ligase substrate‐binding adaptor gene, speckle-type POZ protein) found in 10% of clinically localized PCs (124,132). SPOP regulates transcription through multiple processes, including BRD4 degradation, splicing, DNA repair, proliferation and regulation of the AR and AKT survival pathways, two key targets of ACK1 (69,132–134). Most tumors with SPOP mutations display intrachromosomal modifications (i.e. deletions, inversions and translocations), which precede mutations in CHD1 (103,104,135). Mechanistically, stabilization of the bromodomain protein BRD4 is observed in some hotspot SPOP mutants with loss-of-function mutations, such as F133V, owing to decreased interaction of SPOP with BRD4. However, the SPOP Q165P mutant protein retains partial binding to BRD4. Although SPOP F133V-expressing cells are resistant to the prototype BET inhibitor JQ1, they remain sensitive to dual activity BET-CBP/p300 inhibitors (136–138). The effects of SPOP mutations such as F133V on BRD4-mediated HOXB13 epigenetic regulation are unknown but may favor tumor growth, particularly under conditions of androgen deprivation due to reduced BRD4 degradation. Together, these data suggest that ACK1/AR signaling in association with the pioneer transcription factor HOXB13 plays a critical role in maintaining AR/AR-V7 mRNA expression and functionality in CRPC recurrence. Consequently, inhibition of activated ACK1 may be a highly effective strategy for overcoming CRPC resistance to anti-androgens. LIMITATIONS AND FUTURE DIRECTIONS What is the mechanism by which CRPCs upregulate ACK1 expression under the stress of androgen deprivation? If overexpression/activation of ACK1 or HOXB13 confers resistance to anti-androgens such as enzalutamide, a convergence to critical nodes is suggested. Is there a hierarchy in how the ACK1–AR–HOXB13 axis is regulated? Which axis takes precedence: the epigenetic regulation of HOXB13 or ACK1-mediated AR regulation in CRPC? How do pioneer factors identify ‘vestigial enhancers’ for targeting and activation in CRPCs? Further mechanistic studies are required to understand the crosstalk between the AR–ACK1 and AR–HOXB13 networks and specific activation during CRPC recurrence. SUMMARY Overcoming resistance to second-line therapies has emerged as a major challenge for PC researchers and clinicians. The heterogeneity and mosaic patterns emerging from treatment-induced changes in cancer cells underscore the roles of genetic and epigenetic alterations in tumor plasticity. Pharmacological profiling of patient-derived organoids ex vivo will help address the challenge presented by tumor cell heterogeneity and enable the evaluation of specific targetable signaling pathways active in tumors. One limitation that we foresee is that further mechanistic studies will be necessary to understand the crosstalk between the AR–ACK1 and AR–HOXB13 networks and specific activation in CRPC progression. The redundancy and potential compensatory roles between AR–ACK1 and AR–HOXB13 must also be established. Collectively, these studies underscore the need to preempt tumor plasticity by targeting aberrantly activated tyrosine kinases or epigenetic regulators that promote androgen independence, either alone or in conjunction with PARP inhibitors. ACKNOWLEDGEMENTS This manuscript was edited by the Scientific Editing Service supported by the Institute of Clinical and Translational Sciences at Washington University. FUNDING Phi Beta Psi Sorority [to K.M.]; Department of Surgery, Washington University [to K.M.]; National Institutes of Health [1R01CA208258 and 5R01CA227025 to N.P.M.]; Prostate Cancer Foundation [17CHAL06 to N.P.M.]. Conflict of interest statement. K.M. and N.P.M. are co-founders of Technogenesys, a startup company that controls the intellectual property and patents on the ACK1 inhibitor (R)-9b. ==== Refs REFERENCES 1. Bray  F., FerlayJ., SoerjomataramI., SiegelR.L., TorreL.A., JemalA.  Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J. Clin. 2018; 68 :394–424.30207593 2. Boorjian  S.A., KarnesR.J., ViterboR., RangelL.J., BergstralhE.J., HorwitzE.M., BluteM.L., BuyyounouskiM.K.  Long-term survival after radical prostatectomy versus external-beam radiotherapy for patients with high-risk prostate cancer. Cancer. 2011; 117 :2883–2891.21692049 3. Berish  R.B., AliA.N., TelmerP.G., RonaldJ.A., LeongH.S.  Translational models of prostate cancer bone metastasis. Nat. Rev. Urol. 2018; 15 :403–421.29769644 4. Sathiakumar  N., DelzellE., MorriseyM.A., FalksonC., YongM., ChiaV., BlackburnJ., AroraT., KilgoreM.L.  Mortality following bone metastasis and skeletal-related events among men with prostate cancer: a population-based analysis of US Medicare beneficiaries, 1999–2006. Prostate Cancer Prostatic Dis. 2011; 14 :177–183.21403668 5. Chandran  U.R., MaC., DhirR., BiscegliaM., Lyons-WeilerM., LiangW., MichalopoulosG., BecichM., MonzonF.A.  Gene expression profiles of prostate cancer reveal involvement of multiple molecular pathways in the metastatic process. BMC Cancer. 2007; 7 :64.17430594 6. Beltran  H., PrandiD., MosqueraJ.M., BenelliM., PucaL., CyrtaJ., MarotzC., GiannopoulouE., ChakravarthiB.V., VaramballyS.et al .  Divergent clonal evolution of castration-resistant neuroendocrine prostate cancer. Nat. Med. 2016; 22 :298–305.26855148 7. Drake  C.G., SharmaP., GerritsenW.  Metastatic castration-resistant prostate cancer: new therapies, novel combination strategies and implications for immunotherapy. Oncogene. 2014; 33 :5053–5064.24276248 8. Watson  P.A., AroraV.K., SawyersC.L.  Emerging mechanisms of resistance to androgen receptor inhibitors in prostate cancer. Nat. Rev. Cancer. 2015; 15 :701–711.26563462 9. Feldman  B.J., FeldmanD.  The development of androgen-independent prostate cancer. Nat. Rev. 2001; 1 :34–45. 10. Edwards  J., BartlettJ.M.  The androgen receptor and signal-transduction pathways in hormone-refractory prostate cancer. Part 1: modifications to the androgen receptor. BJU Int. 2005; 95 :1320–1326.15892825 11. Burnstein  K.L.  Regulation of androgen receptor levels: implications for prostate cancer progression and therapy. J. Cell. Biochem. 2005; 95 :657–669.15861399 12. Dai  C., HeemersH., SharifiN.  Androgen signaling in prostate cancer. Cold Spring Harb. Perspect. Med. 2017; 7 :a030452.28389515 13. Chen  C.D., WelsbieD.S., TranC., BaekS.H., ChenR., VessellaR., RosenfeldM.G., SawyersC.L.  Molecular determinants of resistance to antiandrogen therapy. Nat. Med. 2004; 10 :33–39.14702632 14. Whitney  C.A., HowardL.E., FreedlandS.J., DeHoedtA.M., AmlingC.L., AronsonW.J., CooperbergM.R., KaneC.J., TerrisM.K., DaskivichT.J.  Impact of age, comorbidity, and PSA doubling time on long-term competing risks for mortality among men with non-metastatic castration-resistant prostate cancer. Prostate Cancer Prostatic Dis. 2019; 22 :252–260.30279582 15. Tran  C., OukS., CleggN.J., ChenY., WatsonP.A., AroraV., WongvipatJ., Smith-JonesP.M., YooD., KwonA.et al .  Development of a second-generation antiandrogen for treatment of advanced prostate cancer. Science. 2009; 324 :787–790.19359544 16. Arora  V.K., SchenkeinE., MuraliR., SubudhiS.K., WongvipatJ., BalbasM.D., ShahN., CaiL., EfstathiouE., LogothetisC.et al .  Glucocorticoid receptor confers resistance to antiandrogens by bypassing androgen receptor blockade. Cell. 2013; 155 :1309–1322.24315100 17. Ryan  C.J., SmithM.R., de BonoJ.S., MolinaA., LogothetisC.J., de SouzaP., FizaziK., MainwaringP., PiulatsJ.M., NgS.et al .  Abiraterone in metastatic prostate cancer without previous chemotherapy. N. Engl. J. Med. 2013; 368 :138–148.23228172 18. Watson  P.A., ChenY.F., BalbasM.D., WongvipatJ., SocciN.D., VialeA., KimK., SawyersC.L.  Constitutively active androgen receptor splice variants expressed in castration-resistant prostate cancer require full-length androgen receptor. Proc. Natl Acad. Sci. U.S.A. 2010; 107 :16759–16765.20823238 19. Dehm  S.M., SchmidtL.J., HeemersH.V., VessellaR.L., TindallD.J.  Splicing of a novel androgen receptor exon generates a constitutively active androgen receptor that mediates prostate cancer therapy resistance. Cancer Res. 2008; 68 :5469–5477.18593950 20. Hu  R., DunnT.A., WeiS., IsharwalS., VeltriR.W., HumphreysE., HanM., PartinA.W., VessellaR.L., IsaacsW.B.et al .  Ligand-independent androgen receptor variants derived from splicing of cryptic exons signify hormone-refractory prostate cancer. Cancer Res. 2009; 69 :16–22.19117982 21. Guo  Z., YangX., SunF., JiangR., LinnD.E., ChenH., ChenH., KongX., MelamedJ., TepperC.G.et al .  A novel androgen receptor splice variant is up-regulated during prostate cancer progression and promotes androgen depletion-resistant growth. Cancer Res. 2009; 69 :2305–2313.19244107 22. Antonarakis  E.S., LuC., WangH., LuberB., NakazawaM., RoeserJ.C., ChenY., MohammadT.A., ChenY., FedorH.L.et al .  AR-V7 and resistance to enzalutamide and abiraterone in prostate cancer. N. Engl. J. Med. 2014; 371 :1028–1038.25184630 23. Mahajan  N.P., LiuY., MajumderS., WarrenM.R., ParkerC.E., MohlerJ.L., EarpH.S., WhangY.E.  Activated Cdc42-associated kinase Ack1 promotes prostate cancer progression via androgen receptor tyrosine phosphorylation. Proc. Natl Acad. Sci. U.S.A. 2007; 104 :8438–8443.17494760 24. Gelman  I.H.  Androgen receptor activation in castration-recurrent prostate cancer: the role of Src-family and Ack1 tyrosine kinases. Int. J. Biol. Sci. 2014; 10 :620–626.24948875 25. Mahajan  K., MahajanN.P.  Cross talk of tyrosine kinases with the DNA damage signaling pathways. Nucleic Acids Res. 2015; 43 :10588–10601.26546517 26. Chattopadhyay  I., WangJ., QinM., GaoL., HoltzR., VessellaR.L., LeachR.W., GelmanI.H.  Src promotes castration-recurrent prostate cancer through androgen receptor-dependent canonical and non-canonical transcriptional signatures. Oncotarget. 2017; 8 :10324–10347.28055971 27. Mahajan  K., MallaP., LawrenceH.R., ChenZ., Kumar-SinhaC., MalikR., ShuklaS., KimJ., CoppolaD., LawrenceN.J.et al .  ACK1/TNK2 regulates histone H4 Tyr88-phosphorylation and AR gene expression in castration-resistant prostate cancer. Cancer Cell. 2017; 31 :790–803.28609657 28. Mahajan  N.P., CoppolaD., KimJ., LawrenceH.R., LawrenceN.J., MahajanK.  Blockade of ACK1/TNK2 to squelch the survival of prostate cancer stem-like cells. Sci. Rep. 2018; 8 :1954.29386546 29. Nerlakanti  N., YaoJ., NguyenD.T., PatelA.K., EroshkinA.M., LawrenceH.R., AyazM., KuenziB.M., AgarwalN., ChenY.et al .  Targeting the BRD4–HOXB13 coregulated transcriptional networks with bromodomain-kinase inhibitors to suppress metastatic castration-resistant prostate cancer. Mol. Cancer Ther. 2018; 17 :2796–2810.30242092 30. Yao  J., ChenY., NguyenD.T., ThompsonZ.J., EroshkinA.M., NerlakantiN., PatelA.K., AgarwalN., TeerJ.K., DhillonJ.et al .  The homeobox gene, HOXB13, regulates a mitotic protein-kinase interaction network in metastatic prostate cancers. Sci. Rep. 2019; 9 :9715.31273254 31. Liu  Y., KaracaM., ZhangZ., GioeliD., EarpH.S., WhangY.E.  Dasatinib inhibits site-specific tyrosine phosphorylation of androgen receptor by Ack1 and Src kinases. Oncogene. 2010; 29 :3208–3216.20383201 32. Mahajan  K., CoppolaD., RawalB., ChenY.A., LawrenceH.R., EngelmanR.W., LawrenceN.J., MahajanN.P.  Ack1-mediated androgen receptor phosphorylation modulates radiation resistance in castration-resistant prostate cancer. J. Biol. Chem. 2012; 287 :22112–22122.22566699 33. De Silva  D., ZhangZ., LiuY., ParkerJ.S., XuC., CaiL., WangG.G., EarpH.S., WhangY.E.  Interaction between androgen receptor and coregulator SLIRP is regulated by Ack1 tyrosine kinase and androgen. Sci. Rep. 2019; 9 :18637.31819114 34. Chen  Z., WuD., Thomas-AhnerJ.M., LuC., ZhaoP., ZhangQ., GeraghtyC., YanP.S., HankeyW., SunkelB.et al .  Diverse AR-V7 cistromes in castration-resistant prostate cancer are governed by HoxB13. Proc. Natl Acad. Sci. U.S.A. 2018; 115 :6810–6815.29844167 35. Cinar  B., KoenemanK.S., EdlundM., PrinsG.S., ZhauH.E., ChungL.W.  Androgen receptor mediates the reduced tumor growth, enhanced androgen responsiveness, and selected target gene transactivation in a human prostate cancer cell line. Cancer Res. 2001; 61 :7310–7317.11585771 36. Taplin  M.E., BalkS.P.  Androgen receptor: a key molecule in the progression of prostate cancer to hormone independence. J. Cell. Biochem. 2004; 91 :483–490.14755679 37. Nadal  M., PrekovicS., GallasteguiN., HelsenC., AbellaM., ZielinskaK., GayM., VilasecaM., TaulesM., HoutsmullerA.B.et al .  Structure of the homodimeric androgen receptor ligand-binding domain. Nat. Commun. 2017; 8 :14388.28165461 38. Wong  C.I., ZhouZ.X., SarM., WilsonE.M.  Steroid requirement for androgen receptor dimerization and DNA binding: modulation by intramolecular interactions between the NH2-terminal and steroid-binding domains. J. Biol. Chem. 1993; 268 :19004–19012.8360187 39. Huang  W., ShostakY., TarrP., SawyersC., CareyM.  Cooperative assembly of androgen receptor into a nucleoprotein complex that regulates the prostate-specific antigen enhancer. J. Biol. Chem. 1999; 274 :25756–25768.10464314 40. Norris  J.D., ChangC.Y., WittmannB.M., KunderR.S., CuiH., FanD., JosephJ.D., McDonnellD.P.  The homeodomain protein HOXB13 regulates the cellular response to androgens. Mol. Cell. 2009; 36 :405–416.19917249 41. Urbanucci  A., SahuB., SeppalaJ., LarjoA., LatonenL.M., WalteringK.K., TammelaT.L., VessellaR.L., LahdesmakiH., JanneO.A.et al .  Overexpression of androgen receptor enhances the binding of the receptor to the chromatin in prostate cancer. Oncogene. 2012; 31 :2153–2163.21909140 42. Clegg  N.J., WongvipatJ., JosephJ.D., TranC., OukS., DilhasA., ChenY., GrillotK., BischoffE.D., CaiL.et al .  ARN-509: a novel antiandrogen for prostate cancer treatment. Cancer Res. 2012; 72 :1494–1503.22266222 43. Ylitalo  E.B., ThysellE., JernbergE., LundholmM., CrnalicS., EgevadL., StattinP., WidmarkA., BerghA., WikstromP.  Subgroups of castration-resistant prostate cancer bone metastases defined through an inverse relationship between androgen receptor activity and immune response. Eur. Urol. 2017; 71 :776–787.27497761 44. Quigley  D.A., DangH.X., ZhaoS.G., LloydP., AggarwalR., AlumkalJ.J., FoyeA., KothariV., PerryM.D., BaileyA.M.et al .  Genomic hallmarks and structural variation in metastatic prostate cancer. Cell. 2018; 174 :758–769.30033370 45. Takeda  D.Y., SpisakS., SeoJ.H., BellC., O’ConnorE., KorthauerK., RibliD., CsabaiI., SolymosiN., SzallasiZ.et al .  A somatically acquired enhancer of the androgen receptor is a noncoding driver in advanced prostate cancer. Cell. 2018; 174 :422–432.29909987 46. Koivisto  P., KononenJ., PalmbergC., TammelaT., HyytinenE., IsolaJ., TrapmanJ., CleutjensK., NoordzijA., VisakorpiT.et al .  Androgen receptor gene amplification: a possible molecular mechanism for androgen deprivation therapy failure in prostate cancer. Cancer Res. 1997; 57 :314–319.9000575 47. Viswanathan  S.R., HaG., HoffA.M., WalaJ.A., Carrot-ZhangJ., WhelanC.W., HaradhvalaN.J., FreemanS.S., ReedS.C., RhoadesJ.et al .  Structural alterations driving castration-resistant prostate cancer revealed by linked-read genome sequencing. Cell. 2018; 174 :433–447.29909985 48. Berman  B.P., FrenkelB., CoetzeeG.A., JiaL.  Androgen receptor responsive enhancers are flanked by consistently-positioned H3-acetylated nucleosomes. Cell Cycle. 2010; 9 :2249–2250.20495358 49. Jia  L., KimJ., ShenH., ClarkP.E., TilleyW.D., CoetzeeG.A.  Androgen receptor activity at the prostate specific antigen locus: steroidal and non-steroidal mechanisms. Mol. Cancer Res. 2003; 1 :385–392.12651911 50. Fu  M., RaoM., WangC., SakamakiT., WangJ., Di VizioD., ZhangX., AlbaneseC., BalkS., ChangC.et al .  Acetylation of androgen receptor enhances coactivator binding and promotes prostate cancer cell growth. Mol. Cell. Biol. 2003; 23 :8563–8575.14612401 51. Fu  M., WangC., WangJ., ZhangX., SakamakiT., YeungY.G., ChangC., HoppT., FuquaS.A., JaffrayE.et al .  Androgen receptor acetylation governs trans activation and MEKK1-induced apoptosis without affecting in vitro sumoylation and trans-repression function. Mol. Cell. Biol. 2002; 22 :3373–3388.11971970 52. Gaughan  L., StockleyJ., WangN., McCrackenS.R., TreumannA., ArmstrongK., ShaheenF., WattK., McEwanI.J., WangC.et al .  Regulation of the androgen receptor by SET9-mediated methylation. Nucleic Acids Res. 2011; 39 :1266–1279.20959290 53. Lupien  M., BrownM.  Cistromics of hormone-dependent cancer. Endocr. Relat. Cancer. 2009; 16 :381–389.19369485 54. Sharma  N.L., MassieC.E., Ramos-MontoyaA., ZecchiniV., ScottH.E., LambA.D., MacArthurS., StarkR., WarrenA.Y., MillsI.G.et al .  The androgen receptor induces a distinct transcriptional program in castration-resistant prostate cancer in man. Cancer Cell. 2013; 23 :35–47.23260764 55. Urbanucci  A., BarfeldS.J., KytolaV., ItkonenH.M., ColemanI.M., VodakD., SjoblomL., ShengX., TolonenT., MinnerS.et al .  Androgen receptor deregulation drives bromodomain-mediated chromatin alterations in prostate cancer. Cell Rep. 2017; 19 :2045–2059.28591577 56. Asangani  I.A., DommetiV.L., WangX., MalikR., CieslikM., YangR., Escara-WilkeJ., Wilder-RomansK., DhanireddyS., EngelkeC.et al .  Therapeutic targeting of BET bromodomain proteins in castration-resistant prostate cancer. Nature. 2014; 510 :278–282.24759320 57. Braadland  P.R., UrbanucciA.  Chromatin reprogramming as an adaptation mechanism in advanced prostate cancer. Endocr. Relat. Cancer. 2019; 26 :R211–R235.30844748 58. Debes  J.D., SeboT.J., LohseC.M., MurphyL.M., HaugenD.A., TindallD.J.  p300 in prostate cancer proliferation and progression. Cancer Res. 2003; 63 :7638–7640.14633682 59. Fu  M., WangC., ReutensA.T., WangJ., AngelettiR.H., Siconolfi-BaezL., OgryzkoV., AvantaggiatiM.L., PestellR.G.  p300 and p300/cAMP-response element-binding protein-associated factor acetylate the androgen receptor at sites governing hormone-dependent transactivation. J. Biol. Chem. 2000; 275 :20853–20860.10779504 60. Gong  J., ZhuJ., GoodmanO.B.Jr, PestellR.G., SchlegelP.N., NanusD.M., ShenR.  Activation of p300 histone acetyltransferase activity and acetylation of the androgen receptor by bombesin in prostate cancer cells. Oncogene. 2006; 25 :2011–2021.16434977 61. Tateishi  K., OkadaY., KallinE.M., ZhangY.  Role of Jhdm2a in regulating metabolic gene expression and obesity resistance. Nature. 2009; 458 :757–761.19194461 62. Yamane  K., ToumazouC., TsukadaY., Erdjument-BromageH., TempstP., WongJ., ZhangY.  JHDM2A, a JmjC-containing H3K9 demethylase, facilitates transcription activation by androgen receptor. Cell. 2006; 125 :483–495.16603237 63. Li  X., LiT., ChenD., ZhangP., SongY., ZhuH., XiaoY., XingY.  Overexpression of lysine-specific demethylase 1 promotes androgen-independent transition of human prostate cancer LNCaP cells through activation of the AR signaling pathway and suppression of the p53 signaling pathway. Oncol. Rep. 2016; 35 :584–592.26534764 64. Kahl  P., GullottiL., HeukampL.C., WolfS., FriedrichsN., VorreutherR., SollederG., BastianP.J., EllingerJ., MetzgerE.et al .  Androgen receptor coactivators lysine-specific histone demethylase 1 and four and a half LIM domain protein 2 predict risk of prostate cancer recurrence. Cancer Res. 2006; 66 :11341–11347.17145880 65. Jin  L., GarciaJ., ChanE., de la CruzC., SegalE., MerchantM., KharbandaS., RaisnerR., HavertyP.M., ModrusanZ.et al .  Therapeutic targeting of the CBP/p300 bromodomain blocks the growth of castration-resistant prostate cancer. Cancer Res. 2017; 77 :5564–5575.28819026 66. Lasko  L.M., JakobC.G., EdaljiR.P., QiuW., MontgomeryD., DigiammarinoE.L., HansenT.M., RisiR.M., FreyR., ManavesV.et al .  Discovery of a selective catalytic p300/CBP inhibitor that targets lineage-specific tumours. Nature. 2017; 550 :128–132.28953875 67. Mahajan  K., MahajanN.P.  PI3K-independent AKT activation in cancers: a treasure trove for novel therapeutics. J. Cell. Physiol. 2012; 227 :3178–3184.22307544 68. Mahajan  K., MahajanN.P.  ACK1/TNK2 tyrosine kinase: molecular signaling and evolving role in cancers. Oncogene. 2015; 34 :4162–4167.25347744 69. Mahajan  K., CoppolaD., ChallaS., FangB., ChenY.A., ZhuW., LopezA.S., KoomenJ., EngelmanR.W., RiveraC.et al .  Ack1 mediated AKT/PKB tyrosine 176 phosphorylation regulates its activation. PLoS One. 2010; 5 :e9646.20333297 70. Drake  J.M., GrahamN.A., LeeJ.K., StoyanovaT., FaltermeierC.M., SudS., TitzB., HuangJ., PientaK.J., GraeberT.G.et al .  Metastatic castration-resistant prostate cancer reveals intrapatient similarity and interpatient heterogeneity of therapeutic kinase targets. Proc. Natl Acad. Sci. U.S.A. 2013; 110 :E4762–E4769.24248375 71. Carver  B.S., ChapinskiC., WongvipatJ., HieronymusH., ChenY., ChandarlapatyS., AroraV.K., LeC., KoutcherJ., ScherH.et al .  Reciprocal feedback regulation of PI3K and androgen receptor signaling in PTEN-deficient prostate cancer. Cancer Cell. 2011; 19 :575–586.21575859 72. Mahajan  K., ChallaS., CoppolaD., LawrenceH., LuoY., GevariyaH., ZhuW., ChenY.A., LawrenceN.J., MahajanN.P.  Effect of Ack1 tyrosine kinase inhibitor on ligand-independent androgen receptor activity. Prostate. 2010; 70 :1274–1285.20623637 73. Mellinghoff  I.K., VivancoI., KwonA., TranC., WongvipatJ., SawyersC.L.  HER2/neu kinase-dependent modulation of androgen receptor function through effects on DNA binding and stability. Cancer Cell. 2004; 6 :517–527.15542435 74. Gregory  C.W., WhangY.E., McCallW., FeiX., LiuY., PongutaL.A., FrenchF.S., WilsonE.M., EarpH.S.3rd  Heregulin-induced activation of HER2 and HER3 increases androgen receptor transactivation and CWR-R1 human recurrent prostate cancer cell growth. Clin. Cancer Res. 2005; 11 :1704–1712.15755991 75. Hankey  W., ChenZ., WangQ.  Shaping chromatin states in prostate cancer by pioneer transcription factors. Cancer Res. 2020; 80 :2427–2436.32094298 76. Wu  D., SunkelB., ChenZ., LiuX., YeZ., LiQ., GrenadeC., KeJ., ZhangC., ChenH.et al .  Three-tiered role of the pioneer factor GATA2 in promoting androgen-dependent gene expression in prostate cancer. Nucleic Acids Res. 2014; 42 :3607–3622.24423874 77. Parolia  A., CieslikM., ChuS.C., XiaoL., OuchiT., ZhangY., WangX., VatsP., CaoX., PitchiayaS.et al .  Distinct structural classes of activating FOXA1 alterations in advanced prostate cancer. Nature. 2019; 571 :413–418.31243372 78. Bluemn  E.G., ColemanI.M., LucasJ.M., ColemanR.T., Hernandez-LopezS., TharakanR., Bianchi-FriasD., DumpitR.F., KaipainenA., CorellaA.N.et al .  Androgen receptor pathway-independent prostate cancer is sustained through FGF signaling. Cancer Cell. 2017; 32 :474–489.29017058 79. Berger  A., BradyN.J., BarejaR., RobinsonB., ConteducaV., AugelloM.A., PucaL., AhmedA., DardenneE., LuX.et al .  N-Myc-mediated epigenetic reprogramming drives lineage plasticity in advanced prostate cancer. J. Clin. Invest. 2019; 130 :3924–3940. 80. Krumlauf  R.  Evolution of the vertebrate Hox homeobox genes. Bioessays. 1992; 14 :245–252.1350721 81. Mallo  M., AlonsoC.R.  The regulation of Hox gene expression during animal development. Development. 2013; 140 :3951–3963.24046316 82. Shah  N., JinK., CruzL.A., ParkS., SadikH., ChoS., GoswamiC.P., NakshatriH., GuptaR., ChangH.Y.et al .  HOXB13 mediates tamoxifen resistance and invasiveness in human breast cancer by suppressing ERα and inducing IL-6 expression. Cancer Res. 2013; 73 :5449–5458.23832664 83. Economides  K.D., CapecchiM.R.  Hoxb13 is required for normal differentiation and secretory function of the ventral prostate. Development. 2003; 130 :2061–2069.12668621 84. Ewing  C.M., RayA.M., LangeE.M., ZuhlkeK.A., RobbinsC.M., TembeW.D., WileyK.E., IsaacsS.D., JohngD., WangY.et al .  Germline mutations in HOXB13 and prostate-cancer risk. N. Engl. J. Med. 2012; 366 :141–149.22236224 85. Akbari  M.R., TrachtenbergJ., LeeJ., TamS., BristowR., LoblawA., NarodS.A., NamR.K.  Association between germline HOXB13 G84E mutation and risk of prostate cancer. J. Natl Cancer Inst. 2012; 104 :1260–1262.22781434 86. Cardoso  M., MaiaS., PauloP., TeixeiraM.R.  Oncogenic mechanisms of HOXB13 missense mutations in prostate carcinogenesis. Oncoscience. 2016; 3 :288–296.28050579 87. Beebe-Dimmer  J.L., HathcockM., YeeC., OkothL.A., EwingC.M., IsaacsW.B., CooneyK.A., ThibodeauS.N.  The HOXB13 G84E mutation is associated with an increased risk for prostate cancer and other malignancies. Cancer Epidemiol. Biomarkers Prev. 2015; 24 :1366–1372.26108461 88. Bernhart  S.H., KretzmerH., HoldtL.M., JuhlingF., AmmerpohlO., BergmannA.K., NorthoffB.H., DooseG., SiebertR., StadlerP.F.et al .  Changes of bivalent chromatin coincide with increased expression of developmental genes in cancer. Sci. Rep. 2016; 6 :37393.27876760 89. Bernstein  B.E., MikkelsenT.S., XieX., KamalM., HuebertD.J., CuffJ., FryB., MeissnerA., WernigM., PlathK.et al .  A bivalent chromatin structure marks key developmental genes in embryonic stem cells. Cell. 2006; 125 :315–326.16630819 90. Hahn  M.A., LiA.X., WuX., YangR., DrewD.A., RosenbergD.W., PfeiferG.P.  Loss of the polycomb mark from bivalent promoters leads to activation of cancer-promoting genes in colorectal tumors. Cancer Res. 2014; 74 :3617–3629.24786786 91. Calo  E., WysockaJ.  Modification of enhancer chromatin: what, how, and why?. Mol. Cell. 2013; 49 :825–837.23473601 92. Ohm  J.E., McGarveyK.M., YuX., ChengL., SchuebelK.E., CopeL., MohammadH.P., ChenW., DanielV.C., YuWet al .  A stem cell-like chromatin pattern may predispose tumor suppressor genes to DNA hypermethylation and heritable silencing. Nat. Genet. 2007; 39 :237–242.17211412 93. Schlesinger  Y., StraussmanR., KeshetI., FarkashS., HechtM., ZimmermanJ., EdenE., YakhiniZ., Ben-ShushanE., ReubinoffB.E.et al .  Polycomb-mediated methylation on Lys27 of histone H3 pre-marks genes for de novo methylation in cancer. Nat. Genet. 2007; 39 :232–236.17200670 94. Widschwendter  M., FieglH., EgleD., Mueller-HolznerE., SpizzoG., MarthC., WeisenbergerD.J., CampanM., YoungJ., JacobsI.et al .  Epigenetic stem cell signature in cancer. Nat. Genet. 2007; 39 :157–158.17200673 95. Ghoshal  K., MotiwalaT., ClausR., YanP., KutayH., DattaJ., MajumderS., BaiS., MajumderA., HuangT.et al .  HOXB13, a target of DNMT3B, is methylated at an upstream CpG island, and functions as a tumor suppressor in primary colorectal tumors. PLoS One. 2010; 5 :e10338.20454457 96. McGarvey  K.M., Van NesteL., CopeL., OhmJ.E., HermanJ.G., Van CriekingeW., SchuebelK.E., BaylinS.B.  Defining a chromatin pattern that characterizes DNA-hypermethylated genes in colon cancer cells. Cancer Res. 2008; 68 :5753–5759.18632628 97. Zhao  D., LuX., WangG., LanZ., LiaoW., LiJ., LiangX., ChenJ.R., ShahS., ShangX.et al .  Synthetic essentiality of chromatin remodelling factor CHD1 in PTEN-deficient cancer. Nature. 2017; 542 :484–488.28166537 98. Shenoy  T.R., BoysenG., WangM.Y., XuQ.Z., GuoW., KohF.M., WangC., ZhangL.Z., WangY., GilV.et al .  CHD1 loss sensitizes prostate cancer to DNA damaging therapy by promoting error-prone double-strand break repair. Ann. Oncol. 2017; 28 :1495–1507.28383660 99. Metzger  E., WillmannD., McMillanJ., ForneI., MetzgerP., GerhardtS., PetrollK., von MaessenhausenA., UrbanS., SchottA.K.et al .  Assembly of methylated KDM1A and CHD1 drives androgen receptor-dependent transcription and translocation. Nat. Struct. Mol. Biol. 2016; 23 :132–139.26751641 100. Augello  M.A., LiuD., DeonarineL.D., RobinsonB.D., HuangD., StellooS., BlattnerM., DoaneA.S., WongE.W.P., ChenY.et al .  CHD1 loss alters AR binding at lineage-specific enhancers and modulates distinct transcriptional programs to drive prostate tumorigenesis. Cancer Cell. 2019; 35 :603–617.30930119 101. Zhou  J., LiJ., SerafimR.B., KetchumS., FerreiraC.G., LiuJ.C., CoeK.A., PriceB.D., YusufzaiT.  Human CHD1 is required for early DNA-damage signaling and is uniquely regulated by its N terminus. Nucleic Acids Res. 2018; 46 :3891–3905.29529298 102. Kari  V., MansourW.Y., RaulS.K., BaumgartS.J., MundA., GradeM., SirmaH., SimonR., WillH., DobbelsteinM.et al .  Loss of CHD1 causes DNA repair defects and enhances prostate cancer therapeutic responsiveness. EMBO Rep. 2016; 17 :1609–1623.27596623 103. Burkhardt  L., FuchsS., KrohnA., MasserS., MaderM., KluthM., BachmannF., HulandH., SteuberT., GraefenM.et al .  CHD1 is a 5q21 tumor suppressor required for ERG rearrangement in prostate cancer. Cancer Res. 2013; 73 :2795–2805.23492366 104. Grasso  C.S., WuY.M., RobinsonD.R., CaoX., DhanasekaranS.M., KhanA.P., QuistM.J., JingX., LonigroR.J., BrennerJ.C.et al .  The mutational landscape of lethal castration-resistant prostate cancer. Nature. 2012; 487 :239–243.22722839 105. Zhang  Z., ZhouC., LiX., BarnesS.D., DengS., HooverE., ChenC.C., LeeY.S., ZhangY., WangC.et al .  Loss of CHD1 promotes heterogeneous mechanisms of resistance to AR-targeted therapy via chromatin dysregulation. Cancer Cell. 2020; 37 :584–598.32220301 106. Singhal  U., WangY., HendersonJ., NiknafsY.S., QiaoY., GurskyA., ZaslavskyA., ChungJ.S., SmithD.C., KarnesR.J.et al .  Multigene profiling of CTCs in mCRPC identifies a clinically relevant prognostic signature. Mol. Cancer Res. 2018; 16 :643–654.29453313 107. Quintanal-Villalonga  A., ChanJ.M., YuH.A., Pe’erD., SawyersC.L., SenT., RudinC.M.  Lineage plasticity in cancer: a shared pathway of therapeutic resistance. Nat. Rev. Clin. Oncol. 2020; 17 :382.32203275 108. Park  J.W., LeeJ.K., SheuK.M., WangL., BalanisN.G., NguyenK., SmithB.A., ChengC., TsaiB.L., ChengD.et al .  Reprogramming normal human epithelial tissues to a common, lethal neuroendocrine cancer lineage. Science. 2018; 362 :91–95.30287662 109. Zabalza  C.V., AdamM., BurdelskiC., WilczakW., WittmerC., KraftS., KrechT., SteurerS., KoopC., Hube-MaggC.et al .  HOXB13 overexpression is an independent predictor of early PSA recurrence in prostate cancer treated by radical prostatectomy. Oncotarget. 2015; 6 :12822–12834.25825985 110. Varinot  J., FurudoiA., DrouinS., PheV., PennaR.R., RoupretM., BitkerM.O., CussenotO., ComperatE.  HOXB13 protein expression in metastatic lesions is a promising marker for prostate origin. Virchows Arch. 2016; 468 :619–622.26931741 111. Beebe-Dimmer  J.L., IsaacsW.B., ZuhlkeK.A., YeeC., WalshP.C., IsaacsS.D., JohnsonA.M., EwingC.E., HumphreysE.B., ChowdhuryW.H.et al .  Prevalence of the HOXB13 G84E prostate cancer risk allele in men treated with radical prostatectomy. BJU Int. 2014; 113 :830–835.24148311 112. Sharp  A., ColemanI., YuanW., SprengerC., DollingD., Nava RodriguesD., RussoJ.W., FigueiredoI., BertanC., SeedG.et al .  Androgen receptor splice variant-7 expression emerges with castration resistance in prostate cancer. J. Clin. Invest. 2018; 129 :192–208.30334814 113. Al-Tahan  S., WeissL., YuH., TangS., SaportaM., ViholaA., MozaffarT., UddB., KimonisV.  New family with HSPB8-associated autosomal dominant rimmed vacuolar myopathy. Neurol. Genet. 2019; 5 :e349.31403083 114. Cristofani  R., RusminiP., GalbiatiM., CicardiM.E., FerrariV., TedescoB., CasarottoE., ChierichettiM., MessiE., PiccolellaM.et al .  The regulation of the small heat shock protein B8 in misfolding protein diseases causing motoneuronal and muscle cell death. Front. Neurosci. 2019; 13 :796.31427919 115. Cicardi  M.E., CristofaniR., CrippaV., FerrariV., TedescoB., CasarottoE., ChierichettiM., GalbiatiM., PiccolellaM., MessiE.et al .  Autophagic and proteasomal mediated removal of mutant androgen receptor in muscle models of spinal and bulbar muscular atrophy. Front. Endocrinol. (Lausanne). 2019; 10 :569.31481932 116. Park  C.K., ShinS.J., ChoY.A., JooJ.W., ChoN.H.  HoxB13 expression in ductal type adenocarcinoma of prostate: clinicopathologic characteristics and its utility as potential diagnostic marker. Sci. Rep. 2019; 9 :20205.31882852 117. Lawrence  H.R., MahajanK., LuoY., ZhangD., TindallN., HuseyinM., GevariyaH., KaziS., OzcanS., MahajanN.P.et al .  Development of novel ACK1/TNK2 inhibitors using a fragment-based approach. J. Med. Chem. 2015; 58 :2746–2763.25699576 118. Maxson  J.E., AbelM.L., WangJ., DengX., ReckelS., LutyS.B., SunH., GorensteinJ., HughesS.B., BottomlyDet al .  Identification and characterization of tyrosine kinase nonreceptor 2 mutations in leukemia through integration of kinase inhibitor screening and genomic analysis. Cancer Res. 2016; 76 :127–138.26677978 119. Groendyke  B.J., PowellC.E., FeruF., GeroT.W., LiZ., SzaboH., PangK., FeutrillJ., ChenB., LiB.et al .  Benzopyrimidodiazepinone inhibitors of TNK2. Bioorg. Med. Chem. Lett. 2020; 30 :126948.31928839 120. Kopecky  D.J., HaoX., ChenY., FuJ., JiaoX., JaenJ.C., CardozoM.G., LiuJ., WangZ., WalkerN.P.et al .  Identification and optimization of N3,N6-diaryl-1H-pyrazolo[3,4-d]pyrimidine-3,6-diamines as a novel class of ACK1 inhibitors. Bioorg. Med. Chem. Lett. 2008; 18 :6352–6356.18993068 121. Cooney  K.A.  Inherited predisposition to prostate cancer: from gene discovery to clinical impact. Trans. Am. Clin. Climatol. Assoc. 2017; 128 :14–23.28790484 122. Na  R., ZhengS.L., HanM., YuH., JiangD., ShahS., EwingC.M., ZhangL., NovakovicK., PetkewiczJ.et al .  Germline mutations in ATM and BRCA1/2 distinguish risk for lethal and indolent prostate cancer and are associated with early age at death. Eur. Urol. 2017; 71 :740–747.27989354 123. Robinson  D., Van AllenE.M., WuY.M., SchultzN., LonigroR.J., MosqueraJ.M., MontgomeryB., TaplinM.E., PritchardC.C., AttardG.et al .  Integrative clinical genomics of advanced prostate cancer. Cell. 2015; 161 :1215–1228.26000489 124. Romanel  A., GarritanoS., StringaB., BlattnerM., DalfovoD., ChakravartyD., SoongD., CotterK.A., PetrisG., DhingraP.et al .  Inherited determinants of early recurrent somatic mutations in prostate cancer. Nat. Commun. 2017; 8 :48.28663546 125. Zou  M., ToivanenR., MitrofanovaA., FlochN., HayatiS., SunY., Le MagnenC., ChesterD., MostaghelE.A., CalifanoA.et al .  Transdifferentiation as a mechanism of treatment resistance in a mouse model of castration-resistant prostate cancer. Cancer Discov. 2017; 7 :736–749.28411207 126. Jividen  K., KedzierskaK.Z., YangC.S., SzlachtaK., RatanA., PaschalB.M.  Genomic analysis of DNA repair genes and androgen signaling in prostate cancer. BMC Cancer. 2018; 18 :960.30305041 127. Mantoni  T.S., ReidG., GarrettM.D.  Androgen receptor activity is inhibited in response to genotoxic agents in a p53-independent manner. Oncogene. 2006; 25 :3139–3149.16434973 128. Schiewer  M.J., KnudsenK.E.  DNA damage response in prostate cancer. Cold Spring Harb. Perspect. Med. 2019; 9 :a030486.29530944 129. Yin  Y., LiR., XuK., DingS., LiJ., BaekG., RamanandS.G., DingS., LiuZ., GaoY.et al .  Androgen receptor variants mediate DNA repair after prostate cancer irradiation. Cancer Res. 2017; 77 :4745–4754.28754673 130. Mateo  J., CarreiraS., SandhuS., MirandaS., MossopH., Perez-LopezR., Nava RodriguesD., RobinsonD., OmlinA., TunariuN.et al .  DNA-repair defects and olaparib in metastatic prostate cancer. N. Engl. J. Med. 2015; 373 :1697–1708.26510020 131. Patel  A.G., SarkariaJ.N., KaufmannS.H.  Nonhomologous end joining drives poly(ADP-ribose) polymerase (PARP) inhibitor lethality in homologous recombination-deficient cells. Proc. Natl Acad. Sci. U.S.A. 2011; 108 :3406–3411.21300883 132. Barbieri  C.E., BacaS.C., LawrenceM.S., DemichelisF., BlattnerM., TheurillatJ.P., WhiteT.A., StojanovP., Van AllenE., StranskyN.et al .  Exome sequencing identifies recurrent SPOP, FOXA1 and MED12 mutations in prostate cancer. Nat. Genet. 2012; 44 :685–689.22610119 133. Geng  C., KaocharS., LiM., RajapaksheK., FiskusW., DongJ., FoleyC., DongB., ZhangL., KwonO.J.et al .  SPOP regulates prostate epithelial cell proliferation and promotes ubiquitination and turnover of c-MYC oncoprotein. Oncogene. 2017; 36 :4767–4777.28414305 134. Blattner  M., LiuD., RobinsonB.D., HuangD., PoliakovA., GaoD., NatarajS., DeonarineL.D., AugelloM.A., SailerV.et al .  SPOP mutation drives prostate tumorigenesis in vivo through coordinate regulation of PI3K/mTOR and AR signaling. Cancer Cell. 2017; 31 :436–451.28292441 135. Huang  S., GulzarZ.G., SalariK., LapointeJ., BrooksJ.D., PollackJ.R.  Recurrent deletion of CHD1 in prostate cancer with relevance to cell invasiveness. Oncogene. 2012; 31 :4164–4170.22179824 136. Zhang  P., WangD., ZhaoY., RenS., GaoK., YeZ., WangS., PanC.W., ZhuY., YanY.et al .  Intrinsic BET inhibitor resistance in SPOP-mutated prostate cancer is mediated by BET protein stabilization and AKT-mTORC1 activation. Nat. Med. 2017; 23 :1055–1062.28805822 137. Dai  X., GanW., LiX., WangS., ZhangW., HuangL., LiuS., ZhongQ., GuoJ., ZhangJ.et al .  Prostate cancer-associated SPOP mutations confer resistance to BET inhibitors through stabilization of BRD4. Nat. Med. 2017; 23 :1063–1071.28805820 138. Yan  Y., MaJ., WangD., LinD., PangX., WangS., ZhaoY., ShiL., XueH., PanY.et al .  The novel BET-CBP/p300 dual inhibitor NEO2734 is active in SPOP mutant and wild-type prostate cancer. EMBO Mol. Med. 2019; 11 :e10659.31559706 139. Chandrashekar  D.S., BashelB., BalasubramanyaS.A.H., CreightonC.J., Ponce-RodriguezI., ChakravarthiB.V.S.K., VaramballyS.  UALCAN: A Portal for Facilitating Tumor Subgroup Gene Expression and Survival Analyses. Neoplasia. 2017; 19 :649–658.28732212

PMC007xxxxxx/PMC7520847.txt

==== Front NAR Cancer NAR Cancer narcancer NAR Cancer 2632-8674 Oxford University Press 33015625 10.1093/narcan/zcaa025 zcaa025 AcademicSubjects/SCI00030 AcademicSubjects/SCI00980 AcademicSubjects/SCI01060 AcademicSubjects/SCI01140 AcademicSubjects/SCI01180 Survey and Summary Addressing cancer signal transduction pathways with antisense and siRNA oligonucleotides http://orcid.org/0000-0002-9107-6603 Juliano Rudolph L Initos Pharmaceuticals LLC, Chapel Hill, NC 27514-7631, USA To whom correspondence should be addressed. Email: rudyatinitos@gmail.com 9 2020 25 9 2020 25 9 2020 2 3 zcaa02507 9 2020 24 8 2020 20 7 2020 © The Author(s) 2020. Published by Oxford University Press on behalf of NAR Cancer. 2020 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Abstract Signal transduction pathways play key roles in the initiation, progression and dissemination of cancer. Thus, signaling molecules are attractive targets for cancer therapeutics and enormous efforts have gone into the development of small molecule inhibitors of these pathways. However, regrettably, there has been only moderate progress to date, primarily in connection with the RAS signaling pathway. Oligonucleotide-based drugs potentially offer several advantages for addressing signaling pathways, including their exquisite selectivity and their ability to exploit both enzymatic and nonenzymatic targets. Nonetheless, there are problems inherent in the oligonucleotide approach, not the least being the challenge of effectively delivering these complex molecules to intracellular sites within tumors. This survey article will provide a selective review of recent studies where oligonucleotides were used to address cancer signaling and will discuss both positive aspects and limitations of those studies. This will be set in the context of an overview of various cancer signaling pathways and small molecule approaches to regulate those pathways. The survey will also evaluate the challenges and opportunities implicit in the oligonucleotide-based approach to cancer signaling and will point out several possibilities for future research. National Center for Advancing Translational Sciences 10.13039/100006108 1R41TR002692-01 ==== Body pmcINTRODUCTION Aberrations in signaling pathways frequently underlie the initiation and progression of cancer. Activation or overexpression of oncogenes or loss or inhibition of tumor suppressor genes connected to signal transduction can lead to the multiple manifestations of cancer (1). This includes both changes to the tumor cells themselves and alterations of the tumor microenvironment. Hallmark features of cancer cells include enhanced cell proliferation, resistance to programmed cell death, altered metabolism and changes in cell fate and differentiation such as the epithelial–mesenchymal transition often seen in carcinomas (2,3). Tumors can also influence their local and distant microenvironments through signaling processes that affect angiogenesis, inflammation and modulation of potential metastatic sites (4,5). In considering therapeutic intervention in cancer signal transduction, it is important to recall that these pathways are highly convoluted, with multiple interconnections between pathways as well as numerous feedback and feed-forward control mechanisms (6). Thus, perturbation of one pathway can have unintended consequences for other pathways; this reality has often impeded the development of therapies directed toward signaling. A powerful concept for cancer therapeutics has been the idea of ‘oncogene addiction’ whereby tumor cells become highly dependent on the activated state of a particular oncogene (7). However, there are limitations on this approach, including the emergence of resistance to oncogene-directed therapy (8). Another complexity relates to the existence in many tumors of cancer stem cells whose properties are markedly different from the bulk cell population. These differences often include modifications of signaling pathways and altered responses to therapy (9). Cancer signaling pathways offer many potential opportunities for oligonucleotide-based therapeutics. Indeed, there has been interest in this possibility since the earliest days of research on antisense oligonucleotides (ASOs) (10). However, progress has been slow due to many factors, not the least being the complexity of cancer-related signaling. This survey will focus on recent developments in the application of siRNA oligonucleotides, ASOs and splice-switching oligonucleotides (SSOs) to the regulation of signal transduction in cancer. It will deal solely with efforts directed at core cytosolic signaling pathways and will not discuss upstream ligands and receptors nor downstream mechanisms of cell cycle control, cell death or cell differentiation. An initial overview of individual pathways and small molecule inhibitors of those pathways will precede discussion of oligonucleotide-based therapeutic approaches. CANCER SIGNALING PATHWAYS The following sections provide simplified descriptions of some of the key signaling pathways involved in cancer and explore how they have been addressed with small molecule drugs. This will provide context for the subsequent discussion of oligonucleotide-based approaches. RAS-related signaling RAS GTPases are molecular switches that play a critical role in many cancers (11). In normal cells, RAS is activated by receptor tyrosine kinases whose autophosphorylation recruits guanine nucleotide exchange factors such as SOS (Son of Sevenless) to the plasma membrane where they can interact with membrane-bound RAS converting it to its active GTP-bound state. Conversely, GTPase-activating proteins such as neurofibromin 1 return RAS to its inactive GDP-bound state. Activated RAS interacts with multiple downstream effectors setting up signaling cascades that regulate many cellular activities, including proliferation, survival, metabolism and cytoskeletal organization. Initial RAS effectors contain weakly homologous RAS-binding domains (RBDs) that interact with RAS and trigger conformational changes that lead to activation of the effector. The two RAS signaling pathways most prominently associated with cancer are the MAP kinase pathway regulating cell proliferation and the phosphoinositide 3-kinase (PI3K) pathway that regulates cell metabolism and survival (Figure 1). Figure 1. The RAS signaling pathway. This figure and Figures 2–4 present simplified versions of complex pathways. There are multiple additional connections within each pathway that are not depicted as well as interconnections between pathways. RAS signaling in cancer has two major aspects: the MAP kinase pathway and the PI3K pathway. These regulate cell cycle control, apoptosis, cell metabolism and protein synthesis. Green arrows indicate activation, while red lines indicate inhibition. Abbreviations: RTK, receptor tyrosine kinase; Grb2, growth factor receptor-bound protein, an adaptor protein; SOS, Son of Sevenless, a guanine nucleotide exchange factor for RAS; RAS, a small GTPase; Raf, a serine–threonine kinase; Mek, Map kinase/Erk kinase, a dual-specificity kinase; Erk, extracellular signal-regulated kinase, a serine–threonine kinase; PTEN, phosphatase and tensin homolog deleted on chromosome 10, a lipid phosphatase; PIP2, phosphatidylinositol 4,5-bisphosphate; PIP3, phosphatidylinositol 3,4,5-trisphosphate; PDK1, phosphoinositide-dependent protein kinase 1, a serine–threonine kinase activated by PIP3; AKT, a serine–threonine kinase; TSC1,2, tuberous sclerosis proteins, GTPase-activating proteins for Rheb; Rheb, RAS homolog enriched in brain, a small GTPase; mTORC1, mechanistic target of rapamycin, a multiprotein complex regulating metabolism; FOXO, forkhead box O, transcription factors involved in control of apoptosis. In the MAP kinase pathway (12), activated RAS binds to the RBD of a member of the RAF family of serine–threonine kinases, thus relieving an auto-inhibition and activating the kinase. RAF then activates MEK, a dual-specificity kinase, which in turn activates the ERK MAP kinase. ERK can enter the nucleus and phosphorylate several transcription factors, including ELK-1 and MYC leading to transcriptional activation of genes that positively regulate the cell cycle. PI3Ks catalyze the phosphorylation of PIP2 to PIP3 that serves as a second messenger for numerous downstream processes (2,13). There are multiple PI3Ks in mammals, but several of them are activated via the interaction of activated RAS with an RBD in the p110 kinase subunit. The activity of PI3Ks is countered by inositol lipid phosphatases, particularly PTEN, which has an important role as a tumor suppressor. A key downstream effector of PI3K is the AKT serine/threonine kinase that influences both cell survival and cell metabolism. Thus, AKT-mediated phosphorylation of FOXO transcription factors (14) prevents their activation of pro-apoptotic genes and thus enhances cell survival. AKT also modulates the mTORC1 complex (15) that senses nutrient levels and coordinates metabolism and protein synthesis. Thus, AKT phosphorylates and inhibits the tumor suppressor proteins TSC1 and TSC2 that in turn inhibit Rheb, a GTPase that is an important regulatory component of the mTOR complex. Activating mutations in RAS and its downstream effectors are present in many types of cancer. Mutations in RAS, particularly the KRAS and NRAS isoforms, are prevalent in pancreatic, colorectal and lung cancers (11,12), whereas mutations in B-RAF are common in melanoma and colorectal cancer (12,16). Activating mutations have also been identified in PI3K, particularly in the p110α subunit, and are associated with cancers of the breast, colon, stomach, cervix, prostate and lung (13,17). However, a major aspect of the role of PI3K in cancer involves the loss or inactivation of the PTEN tumor suppressor that leads to dysregulation of the PI3K pathway (18). Small molecule modulation of Ras and its downstream effectors RAS inhibitors Enormous efforts have gone into the search for drugs that control cancer by affecting signaling pathways. Obviously, RAS itself has been a major target in this effort, but thus far only modest success has been attained (19,20). RAS normally associates with the plasma membrane through a farnesyl lipid linked to a cysteine residue in its COOH terminus. A number of farnesyl transferase inhibitors were developed, but these led to disappointing results in clinical trials. One reason for this is the existence of an alternative lipidation pathway involving geranylgeranyl transferases. A more recent approach has been to directly address activating mutations in RAS. Thus, compounds were developed that cause irreversible allosteric inhibition of KRAS G12C, thereby trapping RAS in its inactive GDP-bound state. Two phase I clinical trials of this type of compound are underway. However, it is unclear whether this progress can be replicated for other activating mutations, such as the more common G12D mutation, where a reactive cysteine is not available. RAF and MEK inhibitors In contrast to the situation with RAS itself, there has been substantial progress in the development of inhibitors in the MAP kinase pathway. A key finding was discovery of the V600E BRAF mutation in a majority of melanomas (16). In contrast to wild-type (WT) RAFs that require RAS interaction and dimerization for activation, BRAF V600E is constitutively active as a monomer (21). A number of ATP-competitive selective inhibitors of BRAF have been developed and two (vemurafenib and dabrafenib) have been approved by the FDA. These molecules dramatically changed the therapeutic landscape in melanoma (22,23). Vemurafenib and dabrafenib as single agents both produce objective responses in ∼50% of melanoma patients with BRAF V600E mutations and also provide improved survival. However, these drugs are much less effective in colon cancers and non-small cell lung cancers that also have the V600E mutation. Additionally, patients treated with these agents as monotherapy rapidly develop resistance that is likely due to reactivation of downstream elements in the MAP kinase pathway. Another issue is the paradoxical activation of WT BRAF by these drugs leading to activation of the MAP kinase cascade and potentially to tumor formation in tissues that lack V600E BRAF. Inhibitors of the MEK kinase have also been developed. While these molecules have been used as monotherapy, their most important role is in combination with RAF inhibitors (22,24). Two compounds (trametinib and cobimetinib) have been approved by the FDA for use in melanoma, while several other Mek inhibitors are at various stages of clinical evaluation. The use of BRAF and MEK inhibitors in combination has at least partially overcome some of the limitations of monotherapy with V600E targeted drugs, particularly the reactivation of the MAP kinase pathway. Studies are also underway on direct inhibitors of ERK itself, but these are still at an early stage of clinical development. There has also been substantial work on the use of RAF and MEK inhibitor combinations in cancers other than melanoma. Despite these advances, melanoma remains a dire disease. A recent retrospective analysis of the dual use of dabrafenib plus trametinib in melanoma found an overall 5-year survival of 34% (25) demonstrating that new therapies are badly needed in this disease as well as in other cancers. In that vein, two recent papers suggest an exciting new approach to cancer therapy via the RAS pathway (26,27). RAS-driven pancreatic cancers are essentially refractory to RAF and MEK inhibitors. However, combined use of a MEK inhibitor and an inhibitor of autophagy led to a synergistic cytotoxic effect both in cell culture and in animal tumor models. This approach of dual inhibition of two distinct but interconnected pathways mirrors the traditional approach of cytotoxic chemotherapy but with greater precision. It will be very interesting to see whether this approach can be extended to other tumors and whether it can advance to clinical trials. PI3K inhibitors As mentioned earlier, activating mutations of PI3K play a role in multiple cancers and thus there has been a great deal of work on PI3K inhibitors (17,28). Drugs under development include isoform-specific inhibitors, pan-PI3K inhibitors and dual PI3K/mTOR inhibitors. Thus far, two drugs have been approved by the FDA for use in lymphomas and leukemias. Idelalisib is an isoform-selective inhibitor, while duvelisib is a pan-PI3K inhibitor. Overall, PI3K inhibitors have had only modest effects as single agents and their use has been compromised by substantial toxicities, including autoimmune reactions (28). Thus, over the last decade there has been substantial progress in the use of inhibitors of the RAS pathway in treatment of cancer, particularly in melanoma but in other cancers as well. Nonetheless, many challenges remain including heterogeneous efficacy, rapid emergence of resistance and unacceptable toxicities. Thus, innovative approaches to therapy for cancers that involve the RAS pathway are very much needed. WNT pathway signaling The importance of the WNT pathway in cancer has been appreciated ever since the discovery that mutations in adenomatous polyposis coli (APC), a key component of the pathway, are found in 80–90% of colon cancers (29). WNT signaling involves a multimolecular destruction complex that regulates the intracellular levels of the dual function protein β-catenin (30,31) (Figure 2). The formation of cadherin adherens junction in epithelial cells requires the presence of cytosolic β-catenin, while in the nucleus this protein interacts with TCF/LEF transcription factor to activate genes associated with cell cycle traverse. Appropriate levels of β-catenin are controlled by a protein complex that includes the structural proteins Axin 1 and APC, casein kinase 1, glycogen synthase kinase 3 (GSK3) and β-catenin itself. The phosphorylation of β-catenin marks it for recognition by the ubiquitin ligase β-TrCP that then ubiquitinates β-catenin and triggers its destruction by the proteasome. Figure 2. The Wnt signaling pathway. The key element of the Wnt pathway, β-catenin, has dual functions. It is both a transcriptional activator in the nucleus that regulates cell growth and also a key component of cell adhesion junctions. A multiprotein complex in the cytosol regulates the intracellular levels of β-catenin. Abbreviations: cadherins; a family of cell–cell adhesion proteins; Wnt, a family of polypeptide mediators; Fzd, Frizzled, a receptor for Wnt; Lrp5/6, lipoprotein receptor-related protein 5/6, a co-receptor for Wnt; Axin 1, a structural protein; APC, adenomatous polyposis coli, a structural protein; Ck1; casein kinase 1, a serine–threonine kinase; Gsk3, glycogen synthase kinase 3, a serine–threonine kinase; β-TrCP, β-transducin repeat-containing protein, a ubiquitin ligase; Tcf, TCF/LEF, T-cell factor/lymphoid enhancer factor, transcription factors. Binding of WNT to its cell surface receptors FZD and LRP5/6 triggers disassembly of the cytosolic β-catenin destruction complex. The WNT-bound receptors recruit the cytosolic protein DVL allowing membrane docking of AXIN 1 and its associated kinases. This leads to reduced degradation of β-catenin and its accumulation in the nucleus. Recently, a second important Wnt-related regulatory pathway has been identified that involves enzymes termed tankyrases (TNKS) (32). The TNKS bind to AXIN 1 and catalyze the addition of poly(ADP-ribose) moieties. The modified AXIN 1 is then ubiquitinated by the RNF146 E3 ligase leading to its proteosomal degradation and thus disruption of the β-catenin destruction complex. Small molecules affecting the WNT pathway Despite the importance of the WNT pathway in cancer, there has been only modest progress in the development of inhibitors (31,33). A number of TNKS inhibitors have been developed, but these displayed concerning gastrointestinal toxicities (34). Inhibitors of binding between FZD and DVL have been described, as well as GSK3 inhibitors, but are at an early stage of development. Thus, there seem to be opportunities for development of novel approaches to therapeutically modulate the WNT pathway. Notch and Hedgehog signaling Cell-to-cell signaling through the NOTCH pathway is fundamental to cell fate decisions in developmental processes and also plays an important role in cancers (35,36). The interactions of NOTCH ligands on one cell with NOTCH on an adjacent cell lead to proteolytic release of the NOTCH intracellular domain (NICD) that then migrates to the nucleus to interact with CSL transcription factors (Figure 3). NOTCH ligands are transmembrane proteins that in mammals comprise three delta-type ligands (Dll1–3) and two jagged ligands (Jag1 and Jag2). The four mammalian NOTCH receptors are also transmembrane glycoproteins. Ligand–receptor interaction causes conformational changes that allow sequential cleavage of NOTCH first by an ADAM protease and then by γ-secretase leading to release of the NICD. The signaling outputs of this relatively simple system are remarkably complex and context dependent and may reflect different outputs from different ligand–receptor pairs as well as epigenetic distinctions among cells. While NOTCH may play a direct oncogenic role in some cancers such as T-cell lymphomas, more commonly its impact is on the tumor microenvironment (35,37). Thus, NOTCH ligand–receptor interactions can take place between tumor cells and adjacent stromal cells or between different lineages within the tumor cell population. For example, NOTCH signaling, particularly that involving the Jag1 ligand, can play a role in providing a favorable niche for cancer stem cells. Figure 3. The Notch pathway. Notch signaling involves cell-to-cell communication. Notch ligands on an adjacent cell interact with Notch causing it to be cleaved by two proteases. This releases the Notch intracellular domain that then migrates to the nucleus to interact with transcription factors. Abbreviations: Notch ligands, transmembrane proteins of the Delta or Jagged type; ADAM, a disintegrin and metalloproteinase, an extracellular protease; γ-secretase, a membrane-associated protease complex; NICD, Notch intracellular domain; Csl, CBF1, Suppressor of Hairless, Lag-1, a transcription factor. Hedgehog signaling (Figure 4) also plays a role in multiple developmental processes, while abnormalities in this pathway have been linked to several types of cancer, including basal cell carcinoma, medulloblastoma, and breast, lung, prostate and pancreatic cancers (38,39). Hedgehog signaling in mammals is associated with primary cilia that are microtubule-based structures at the cell surface (40,41). Signaling is activated by three hedgehog ligands, the best known being Sonic hedgehog (SHH). In the absence of SHH binding to its transmembrane receptor PTCH, the GPCR-like protein SMO is inhibited by PTCH and signaling is quiescent. SUFU proteins act with cytosolic kinases to keep transcription factors Gli1–3 in their repressor form. In the presence of SHH, inhibition is relieved and Smo migrates to the primary cilium and initiates the downstream signaling cascade. This results in the activation and nuclear translocation of Gli1–3. Figure 4. The Hedgehog pathway. The interaction of hedgehog ligands with their membrane receptor Patched controls the activation state of Gli family transcription factors. Abbreviations: Shh, Sonic hedgehog; PTCH, Patched, the receptor for hedgehog ligands; Smo, Smoothened, an intermediate protein in the pathway; Gli1–3, glioma-associated oncogene homolog, transcriptional regulators; SUFU, Suppressor of fused, a negative regulator of the pathway; MT, microtubules. Small molecules that affect Notch and Hedgehog signaling Small molecule development in the NOTCH pathway has primarily focused on γ-secretase inhibitors (37,42). These molecules have shown promising results in early phase clinical trials. However, since γ-secretase has over 90 substrates including many important transmembrane proteins other than Notch, the potential for off-target effects is high. In patients, the dose-limiting toxicity is diarrhea due to intestinal goblet cell metaplasia. There has been a good deal of drug development activity for the Hedgehog pathway focusing on antagonists of SMO (37–39) but also including inhibitors of Gli. Currently, two SMO antagonists (vismodegib and sonidegib) are FDA approved for use in advanced basal cell cancer. Hedgehog pathway inhibitors are also being clinically evaluated in a variety of other cancers as single agents or in conjunction with standard chemotherapy. While there are many variations, the overview is that these molecules are not very effective in unselected patient cohorts but may be more successful in patients whose disease clearly has a Hedgehog-driven component. OLIGONUCLEOTIDE APPROACHES TO CANCER SIGNALING Basic aspects of oligonucleotide therapeutics Over the last decade, oligonucleotide-based therapeutics has evolved from basic research to clinical reality with FDA approval of seven drugs, including two ASOs, two siRNA oligonucleotides and three SSOs (43,44). Progress has largely been based on advances in oligonucleotide chemistry that have improved stability, increased efficacy and reduced off-target and immunostimulatory effects (45–48). Some notable examples have been the development of methoxyethoxy (49), linked nucleic acid (50) and constrained ethyl (c-Et) (47) modifications that have markedly increased binding affinity, as well as backbone modifications such as the uncharged morpholino structure that has proven useful for SSOs (51). Despite successes, a major remaining challenge for oligonucleotide-based therapeutics concerns the effective delivery of these molecules to their intracellular sites of action (52). This problem has two aspects. The first is to obtain sufficient accumulation of oligonucleotide in the tissue of interest. The second is to overcome nonproductive trapping of oligonucleotide within endosomal compartments. Although systemically administered oligonucleotides distribute broadly to tissues other than the central nervous system, there is great variability in tissue uptake with liver and kidney being predominant (49). Enormous efforts have been devoted to improving the delivery of oligonucleotides, particularly siRNA, primarily involving the use of various cationic lipid or polymer nanoparticles (53,54). However, most nanoparticles can exit from the bloodstream only at sites where the vascular endothelium has gaps of 100 nm or more, thus limiting nanoparticle delivery to the liver, spleen and some rapidly growing tumors, but not to many other tumors (55,56). There is also much interest in ligand–oligonucleotide conjugates that can interact with specific receptors and thus potentially allow tissue-selective targeting (57). Thus, carbohydrate conjugates of siRNAs and ASOs have shown dramatically increased uptake into the liver via the hepatic asialoglycoprotein receptor (58). The extent to which this approach will work with other receptors in other tissues, particularly tumors, is not entirely clear at this point but is a key area to explore. The second major delivery issue concerns the intracellular trafficking of oligonucleotides (59–61). Whether administered as ‘free’ molecules or associated with nanoparticles, oligonucleotides are taken up via endocytosis and then traffic to intracellular membrane-bound compartments including early and late endosomes and lysosomes. Within these compartments, oligonucleotides are pharmacologically inert since they cannot access their molecular targets in the cytosol or nucleus. Certain types of nanoparticle carriers, particularly cationic lipoplexes (53), can facilitate oligonucleotide escape from endosomes. Additionally, another approach has evolved recently that utilizes endosome-destabilizing small molecules to promote oligonucleotide release and thus increased effectiveness (62,63). In summary, while great progress has been made, there still remain important challenges to therapeutic use of oligonucleotides. These include off-target effects at the nucleic acid level, toxicities due to interactions with proteins (64) and, most importantly, the efficient intracellular delivery of these molecules to their intracellular sites of action. Oligonucleotides in cancer There is a very large literature describing use of various types of oligonucleotides in cancer and a number of clinical trials are currently underway. However, there has been more limited work done regarding oligonucleotide modulation of cancer signaling pathways. Several recent reviews address various aspects of the broad potential role of oligonucleotides in cancer therapy. Thus, an article by Chen et al. provides an overview of siRNA-based approaches and lists a number of recent clinical trials of siRNA in cancer (65). A review by Yamakawa et al. focuses on pancreatic cancer but describes several types of potential oligonucleotide therapies (66). Lee et al. discuss both potential siRNA targets in cancer and delivery strategies (67). A review by Martinez-Montiel et al. explores the role of alternative RNA splicing in cancer, discusses potential use of SSOs and provides information on current clinical trials (68). In the context of these broader views of oligonucleotides in cancer, this survey will focus on oligonucleotides that address cancer signaling pathways. Oligonucleotide modulation of the RAS pathway As discussed earlier, there has been limited progress in developing small molecule inhibitors of RAS and its effectors. In a sense, this presents an opportunity for deploying oligonucleotide approaches to this challenging problem. While there has been more success with small molecules for the enzymes of the MAP kinase and PI3K pathways, there still remain difficulties. Table 1 summarizes a selection of recent investigations of the use of ASOs or siRNA to address the overall RAS pathway. A few of these will be discussed in more detail below. Table 1. Oligonucleotides targeting Ras and its downstream effectors: selected examples illustrating the molecular target, cancer type, type of oligonucleotide used and approach to in vivo delivery Molecular target Cancer type Oligo type In vivo delivery modality Cell studies Animal model Other Reference KRAS NSCLC c-Et ASO None WT and KRAS mutant cell lines KRAS mutant xenografts Clinical trial NCT03101839 (69) KRAS G12D Pancreatic Plasmid for bi- shRNA Lipid nanoparticle WT and KRAS mutant reporters KRAS mutant xenografts (70) KRAS Ovarian Poly-siRNA Chitosan nanoparticle Ovarian cancer cells Allograft tumor model Co-use of PI3K inhibitor (71) KRAS Pancreatic siRNA Peptide nanoparticle Pancreatic and colon cancer cells Murine pancreatic tumor tumor (72) KRAS G12S NSCLC siRNA IgG and Poloxamer nanoparticle A549 cells A549 xenograft Co-use of erlotinib (73) KRAS G12D Pancreatic siRNA Exosomes PANC-1 cells PANC-1 xenografts, genetically engineered mouse tumor Clinical trial NCT03608631 (74) KRAS Lung siRNA Novel nanoparticles Lung adenocarcinoma cells Genetically engineered mouse tumor (75) KRAS Lung, colon siRNA Neutral liposomes A549 cells Lung, colon xenografts (76) KRAS Colon siRNA Anti-EGFR antibody/protamine Various Colon xenografts (77) KRAS G12D Pancreatic siRNA Sustained release capsule Xenografts and syngeneic pancreatic tumors Clinical trial NCT01188785 (78) BRAF Thyroid siRNA Nanoparticle with near-infrared (NIR) fluorescence BRaf mutated cell lines BRaf mutated thyroid tumor xenografts (79) BRAF Melanoma siRNA Liposomes topical BRaf mutated melanoma cells (80) BRAF Melanoma siRNA Gold nanorods BRaf mutated melanoma cells BRaf mutated xenografts (81) PI3K Colon siRNA PI3K and Ras mutated colon cancer cells (82) As indicated in Table 1, during the last few years there have been many publications concerning effects of oligonucleotides, especially siRNA, on RAS-driven cancers. However, progress has been limited as only a few of these studies have matured to clinical investigations. Many studies have relied on use of commercially available unmodified siRNAs that are known to be quite unstable as opposed to the highly modified siRNAs that have successfully progressed through clinical trials (48). Another concern is that while several studies have specifically targeted mutant forms of RAS, in most cases there was not definitive proof that the siRNA could discriminate the mRNAs of mutant versus WT RAS. Designing siRNAs to discriminate a single base change is very challenging. While this has been accomplished in highly controlled in vitro studies (83,84), whether it has been attained in a complex in vivo situation is less clear. In these contexts, it is interesting to delve into a few studies selected from Table 1. Investigators from Ionis Pharmaceuticals have described AZD4785, a cEt-modified ASO that targets the 3′ untranslated region of human KRAS (69). Thus, the ASO equally affects mutant and WT forms. When tested in cell lines with mutant or WT Ras, the ASO inhibited the MAP kinase pathway and cell growth in KRAS mutant cells but not WT cells. Apparently, the cell lines chosen displayed ‘oncogene addiction’ in that they were highly dependent on the mutant form of KRAS. Further, while use of a Mek inhibitor triggered positive feedback phosphorylation of Mek on activating residues, this was not observed for the ASO. AZD4785 was administered systemically, without any delivery agents, in KRAS mutant lung cancer xenograft models. This resulted in substantial reduction of intratumoral KRAS and an inhibition of tumor growth. Similar observations were made in a patient-derived xenograft model. However, both of these studies used quite high doses of ASO (50 mg/kg) as well as extended dosing intervals (5× weekly, 4 weeks). In this study, no observations were made concerning the possibility of additional positive feedback loops induced by RAS depletion, for example the well-known activation of EGFR leading to PI3K stimulation via relief of a negative feedback loop from ERK (85). Another issue is whether this ASO would be effective in tumors that have an activated RAS but where oncogene addiction is not strongly manifested. A phase I study of AZD4785 has been completed (NCT03101839), but it is not clear whether further development is contemplated. A group from MD Anderson Cancer Center published a rather remarkable study involving the use of exosomes to deliver KRAS G12D siRNA to pancreatic tumors (74). Exosomes were prepared from fibroblasts and loaded with the siRNA via electroporation. The exosomes contained CD47, a membrane protein that discourages phagocytosis by monocytes and macrophages, thus leading to long circulation times and potentially allowing increased tumor uptake. The siRNA exosomes reduced KRAS G12D mRNA and downstream signaling in mutant PANC-1 cells but not in tumor cells with WT RAS. In PANC-1 xenografts, daily treatment with siRNA exosomes almost entirely reduced tumor growth and greatly extended survival; this was paralleled by a reduction in KRAS G12D mRNA and in ERK activation. Similar, if slightly less impressive, observations were made in a genetically engineered pancreatic tumor model. Although observations were made regarding differential effects on mutant and WT KRAS RNA during cell studies, these were not confirmed at the protein level. The siRNA used in these studies was apparently unmodified raising questions about stability. A concern with all studies involving the potential therapeutic use of exosomes relates to the difficulty of reproducibly scaling up production of these complex entities (86,87), although this group has made a strong effort in that direction (88). A phase I clinical trial of this approach has been initiated (NCT03608631). It will be interesting to see whether these impressive studies in model systems hold up in the clinic, noting that other initial reports of exciting results with exosome delivery have apparently not progressed very far (89). Investigators at Harvard have explored therapy of aggressive anaplastic thyroid cancer using nanoparticle delivery of siRNA (79). They developed a hybrid nanoparticle that includes BRAF siRNA, a cationic lipid, an NIR emitting polymer and polyethylene glycol. The NIR characteristics provided the ability to noninvasively image uptake of the nanoparticles into tissues. Nanoparticles with BRAF siRNA were used to treat BRAF V600E thyroid tumor cells and were found to reduce cell growth and inhibit ERK activation. Treatment of thyroid cancer xenografts and orthotopic grafts showed reduced tumor growth and the reduction of intratumoral BRAF. In this study, the effects on BRAF reduction and tumor growth inhibition were somewhat modest. Additionally, there was no attempt to explore the feedback mechanisms often seen with other inhibitors of RAF kinases. Thus, while there have been some very encouraging reports on the use of oligonucleotides to inhibit the Ras pathway in cancer, only a few of those have matured to the level of clinical trials. Currently, there are 71 clinical trials involving siRNA and 186 involving ASOs listed on the ClinicalTrials.gov website, but very few involve RAS pathway signaling. Additionally, perusal of the websites of several leading antisense and siRNA companies did not identify clinical trials involving the RAS pathway. Thus, despite numerous academic publications on this topic, there has been only modest progress in translating these studies to the clinical environment. Nonetheless, it should be noted that most of the studies to date have not fully benefited from recent advances in the chemistry of siRNAs nor from newer approaches to selective delivery involving conjugation with targeting ligands. Oligonucleotide modulation of the Wnt, Notch and Hedgehog pathways Focus on these pathways for cancer therapy has come later than for the RAS pathway. Thus, the literature on oligonucleotide effects on the Wnt, Notch and Hedgehog pathways is more limited. Some of the recent literature is summarized in Table 2. Table 2. Oligonucleotides targeting the Wnt, Notch and Hedgehog pathways: selected examples illustrating the molecular target, cancer type, type of oligonucleotide used and approach to in vivo delivery Molecular target Cancer type Oligo type In vivo delivery modality Cell studies Animal model Other Reference β-Catenin Several Dicer substrate siRNA Lipid nanoparticles Several cancer cell types Xenografts of colon, hepatic cancers also metastatic models (90) β-Catenin Melanoma Dicer substrate siRNA Lipid nanoparticles Syngeneic mouse tumors Used with checkpoint blockade antibodies (91) β-Catenin Morpholino and PNA SSOs HEK293 Produced dominant negative β-catenin (92) HOTAIR lncRNA Cervical siRNA HeLa SiHOTAIR modulates Wnt signaling (93) Frizzled Esophageal siRNA Esophageal carcinoma cells (94) NOTCH-1, β- catenin, Stat3 Breast siRNA MCF7 Stronger cell growth inhibition with siRNA combinations (95) NOTCH-3 Ovarian siRNA SKOV3 siNOTCH reduces paclitaxel resistance (96) NOTCH-1 Breast shRNA MCF7, MDA-MB231 (97) Notch-2 Bladder shRNA Bladder cancer cell lines Orthotopic xenograft Stable expression of shNOTCH reduces tumor growth (98) Sonic hedgehog Breast siRNA MCF-7, SKBR-3 siSHH and small molecule inhibitor of Gli1 synergize (99) Gli1 Lung cancer siRNA SBC-5 Combined siGli1 and kinase inhibition block cell growth (100) Many of the studies using oligonucleotides in the WNT, NOTCH and Hedgehog pathways, while interesting, were at a very early stage of development. Perhaps the most complete report is from investigators at Dicerna Pharmaceuticals who targeted β-catenin (90). Using siRNAs that are Dicer substrates and a cationic lipid nanoparticle, they examined delivery of the siRNA to tumors, reduction of β-catenin mRNA levels and inhibition of tumor growth. Several colon and liver cancer xenograft and metastatic models were used. This report also included extensive analysis of the design of the nanoparticles used. Interestingly, significant tumor inhibition was observed in WNT-dependent colon and hepatic tumors but not those that were WNT independent. A second report using this technology explored the interaction between siRNA-mediated β-catenin inhibition and the effectiveness of checkpoint inhibitors in syngeneic mouse tumors (91). One weakness of these reports was a limited amount of information regarding possible toxicities and off-target effects. Currently, no studies using this approach are listed in ClinicalTrials.gov. Despite the limited progress thus far, it would seem that the WNT, NOTCH and Hedgehog pathways offer many possibilities for the development of therapeutic oligonucleotides. Within these pathways are several key targets that are nonenzymatic proteins and thus difficult to address with small molecules but easily addressed with antisense or siRNA. Especially in the NOTCH and Hedgehog pathways there are distinct isoforms of signaling proteins whose mRNAs can be selectively modulated by oligonucleotides, whereas selective modulation at the protein level would be challenging. Thus, hopefully the near future will see additional progress in this area. SUMMARY AND ANALYSIS Perturbations of intracellular signaling pathways are intimately involved in the initiation and progression of many types of cancer. Accordingly, the ability to modulate these pathways is an obvious goal for cancer therapeutics. Unfortunately, there has been only limited progress toward this goal. The development of small molecule inhibitors has often been frustrated by the complexity of cancer signaling pathways. Inhibition at one locus can sometimes trigger compensating feedback processes leading to resistance, such as that seen with BRAF inhibitors. Blocking key multifunctional proteins such as RAS or β-catenin can lead to undesired effects on normal cells and tissues. Small molecule drugs can often be designed to inhibit enzymes, but more rarely to affect nonenzymatic signaling pathway proteins (although there are exceptions such as the inhibitors of SMO in the Hedgehog pathway). Conceptually, oligonucleotide-based drugs, because of their exquisite specificity, should be able to overcome some of the limitations of small molecule therapeutics in cancer. For example, ASOs and siRNAs targeting KRAS rather than NRAS or HRAS have been described in several publications and it may even be possible, although difficult, to discriminate mutant from WT KRAS (Table 2). Likewise, it is possible to target individual NOTCH isoforms in that pathway or individual Gli transcription factors in the Hedgehog pathway. The ability to target either enzymatic or nonenzymatic pathway components is clearly an asset for oligonucleotide-based approaches. Despite these advantages, oligonucleotide modulation of cancer signaling pathways has progressed only slowly. A major impediment concerns the ability to effectively deliver oligonucleotides to intracellular targets in tumors. Much of the work on siRNA in cancer has involved use of nanoparticles (53,65). This can be an effective strategy in rapidly growing tumors in mice where a highly abnormal intratumoral vasculature allows escape of the nanoparticles to the tumor parenchyma. However, it may be less effective in more slowly growing tumors, including many human tumors, where the vascular abnormalities are not as extreme (56,101). ASOs are often administered as ‘free’ molecules but here tumor uptake may be low as compared to hepatic or kidney uptake (102). An emerging approach to this problem is the use of ASOs or chemically stabilized siRNAs conjugated to ligands that bind specific cell surface receptors (57,58,103). In some instances, there are well-known cell surface markers for particular tumor types that could be exploited for tumor-selective delivery, although this is not always the case (104,105). Some types of nanoparticles have the ability to overcome endosomal trapping of oligonucleotides by causing endosome membrane destabilization (53), but this is not true of ‘free’ oligonucleotides or ligand–oligonucleotide conjugates that lack that ability. Thus, new strategies are needed to overcome the limited tumor uptake of oligonucleotides while still dealing with the endosome entrapment issue. Based on these considerations, there are some hypothetical strategies that may advance the use of oligonucleotides in modulating cancer signaling pathways. First, focus on nonenzymatic pathway components. There is vast medicinal chemistry expertise on the design of, for example, kinase inhibitors. It is not clear that oligonucleotides will offer advantages over small molecules in that context. Second, use conjugates to selectively deliver oligonucleotides to cells where a particular pathway is active. For example, in the Hedgehog pathway ligand-bound Patched is internalized via endocytosis (39). Thus, an oligonucleotide with a ligand for Patched may be taken up more effectively in cells where Hedgehog components are strongly expressed. The oligonucleotide could be designed to downregulate one of the intracellular Hedgehog pathway components. Third, consider simultaneously addressing targets in two interconnected pathways. Recent work has shown that dual inhibition of the MAP kinase and autophagy pathways can lead to synergistic effects in pancreatic cancer cells (26,27). Similar dual inhibition approaches could be used with oligonucleotides, especially for pathways where good small molecule inhibitors are lacking. This is also consistent with traditional strategies in cancer chemotherapy where several drugs with different molecular mechanisms are used to attain therapeutic effects while distributing toxicity to different tissues. An approach that has not been substantially pursued to date is to use oligonucleotides to manipulate levels of tumor suppressor proteins. This might be done via ASO or siRNA inhibition of expression of proteins that negatively modulate tumor suppressors. Alternatively, in some cases it may be possible to use SSOs to increase expression of tumor suppressors as has been done for other types of proteins (106,107). Finally, as suggested by the history of tyrosine kinase inhibitors in cancer (108), it will be important to develop therapeutic oligonucleotides that are appropriate for specific cohorts of cancer patients based on the molecular pathology of their disease, rather than expecting to find oligonucleotide drugs that work for large, unselected populations. FUNDING National Center for Advancing Translational Sciences [1R41TR002692-01]. Conflict of interest statement. None declared. ==== Refs REFERENCES 1. Sever  R., BruggeJ.S.  Signal transduction in cancer. Cold Spring Harb. Perspect. Med. 2015; 5 :a006098.25833940 2. Shaw  R.J., CantleyL.C.  Ras, PI(3)K and mTOR signalling controls tumour cell growth. Nature. 2006; 441 :424–430.16724053 3. Wong  R.S.  Apoptosis in cancer: from pathogenesis to treatment. J. Exp. Clin. Cancer Res. 2011; 30 :87.21943236 4. Lambert  A.W., PattabiramanD.R., WeinbergR.A.  Emerging biological principles of metastasis. Cell. 2017; 168 :670–691.28187288 5. Viallard  C., LarriveeB.  Tumor angiogenesis and vascular normalization: alternative therapeutic targets. Angiogenesis. 2017; 20 :409–426.28660302 6. Csermely  P., KorcsmarosT., NussinovR.  Intracellular and intercellular signaling networks in cancer initiation, development and precision anti-cancer therapy: RAS acts as contextual signaling hub. Semin. Cell Dev. Biol. 2016; 58 :55–59.27395026 7. Weinstein  I.B.  Cancer. Addiction to oncogenes—the Achilles heal of cancer. Science. 2002; 297 :63–64.12098689 8. Nagel  R., SemenovaE.A., BernsA.  Drugging the addict: non-oncogene addiction as a target for cancer therapy. EMBO Rep. 2016; 17 :1516–1531.27702988 9. Matsui  W.H.  Cancer stem cell signaling pathways. Medicine (Baltimore). 2016; 95 :S8–S19.27611937 10. Monia  B.P., JohnstonJ.F., EckerD.J., ZounesM.A., LimaW.F., FreierS.M.  Selective inhibition of mutant Ha-ras mRNA expression by antisense oligonucleotides. J. Biol. Chem. 1992; 267 :19954–19962.1400312 11. Simanshu  D.K., NissleyD.V., McCormickF.  RAS proteins and their regulators in human disease. Cell. 2017; 170 :17–33.28666118 12. Braicu  C., BuseM., BusuiocC., DrulaR., GuleiD., RadulyL., RusuA., IrimieA., AtanasovA.G., SlabyO.et al .  A comprehensive review on MAPK: a promising therapeutic target in cancer. Cancers (Basel). 2019; 11 :1618.31652660 13. Fruman  D.A., ChiuH., HopkinsB.D., BagrodiaS., CantleyL.C., AbrahamR.T.  The PI3K pathway in human disease. Cell. 2017; 170 :605–635.28802037 14. Coomans de Brachene  A., DemoulinJ.B.  FOXO transcription factors in cancer development and therapy. Cell. Mol. Life Sci. 2016; 73 :1159–1172.26686861 15. Ben-Sahra  I., ManningB.D.  mTORC1 signaling and the metabolic control of cell growth. Curr. Opin. Cell Biol. 2017; 45 :72–82.28411448 16. Dankner  M., RoseA.A.N., RajkumarS., SiegelP.M., WatsonI.R.  Classifying BRAF alterations in cancer: new rational therapeutic strategies for actionable mutations. Oncogene. 2018; 37 :3183–3199.29540830 17. Yang  J., NieJ., MaX., WeiY., PengY., WeiX.  Targeting PI3K in cancer: mechanisms and advances in clinical trials. Mol. Cancer. 2019; 18 :26.30782187 18. Lee  Y.R., ChenM., PandolfiP.P.  The functions and regulation of the PTEN tumour suppressor: new modes and prospects. Nat. Rev. Mol. Cell Biol. 2018; 19 :547–562.29858604 19. Friedlaender  A., DrilonA., WeissG.J., BannaG.L., AddeoA.  KRAS as a druggable target in NSCLC: rising like a phoenix after decades of development failures. Cancer Treat. Rev. 2020; 85 :101978.32062493 20. Khan  I., RhettJ.M., O’BryanJ.P.  Therapeutic targeting of RAS: new hope for drugging the “undruggable”. Biochim. Biophys. Acta Mol. Cell Res. 2020; 1867 :118570.31678118 21. Brummer  T., McInnesC.  RAF kinase dimerization: implications for drug discovery and clinical outcomes. Oncogene. 2020; 39 :4155–4169.32269299 22. Sanchez  J.N., WangT., CohenM.S.  BRAF and MEK inhibitors: use and resistance in BRAF-mutated cancers. Drugs. 2018; 78 :549–566.29488071 23. Samatar  A.A., PoulikakosP.I.  Targeting RAS–ERK signalling in cancer: promises and challenges. Nat. Rev. Drug Discov. 2014; 13 :928–942.25435214 24. Savoia  P., FavaP., CasoniF., CremonaO.  Targeting the ERK signaling pathway in melanoma. Int. J. Mol. Sci. 2019; 20 :1483.30934534 25. Robert  C., GrobJ.J., StroyakovskiyD., KaraszewskaB., HauschildA., LevchenkoE., Chiarion SileniV., SchachterJ., GarbeC., BondarenkoI.et al .  Five-year outcomes with dabrafenib plus trametinib in metastatic melanoma. N. Engl. J. Med. 2019; 381 :626–636.31166680 26. Bryant  K.L., StalneckerC.A., ZeitouniD., KlompJ.E., PengS., TikunovA.P., GundaV., PierobonM., WatersA.M., GeorgeS.D.et al .  Combination of ERK and autophagy inhibition as a treatment approach for pancreatic cancer. Nat. Med. 2019; 25 :628–640.30833752 27. Kinsey  C.G., CamolottoS.A., BoespflugA.M., GuillenK.P., FothM., TruongA., SchumanS.S., SheaJ.E., SeippM.T., YapJ.T.et al .  Protective autophagy elicited by RAF→MEK→ERK inhibition suggests a treatment strategy for RAS-driven cancers. Nat. Med. 2019; 25 :620–627.30833748 28. Greenwell  I.B., IpA., CohenJ.B.  PI3K inhibitors: understanding toxicity mechanisms and management. Oncology (Williston Park). 2017; 31 :821–828.29179250 29. Kinzler  K.W., VogelsteinB.  Lessons from hereditary colorectal cancer. Cell. 1996; 87 :159–170.8861899 30. van Kappel  E.C., MauriceM.M.  Molecular regulation and pharmacological targeting of the beta-catenin destruction complex. Br. J. Pharmacol. 2017; 174 :4575–4588.28634996 31. Nusse  R., CleversH.  Wnt/beta-catenin signaling, disease, and emerging therapeutic modalities. Cell. 2017; 169 :985–999.28575679 32. Mariotti  L., PollockK., GuettlerS.  Regulation of Wnt/beta-catenin signalling by tankyrase-dependent poly(ADP-ribosyl)ation and scaffolding. Br. J. Pharmacol. 2017; 174 :4611–4636.28910490 33. Zhong  Z., VirshupD.M.  Wnt signaling and drug resistance in cancer. Mol. Pharmacol. 2020; 97 :72–89.31787618 34. Krishnamurthy  N., KurzrockR.  Targeting the Wnt/beta-catenin pathway in cancer: update on effectors and inhibitors. Cancer Treat. Rev. 2018; 62 :50–60.29169144 35. Meurette  O., MehlenP.  Notch signaling in the tumor microenvironment. Cancer Cell. 2018; 34 :536–548.30146333 36. Siebel  C., LendahlU.  Notch signaling in development, tissue homeostasis, and disease. Physiol. Rev. 2017; 97 :1235–1294.28794168 37. Takebe  N., MieleL., HarrisP.J., JeongW., BandoH., KahnM., YangS.X., IvyS.P.  Targeting Notch, Hedgehog, and Wnt pathways in cancer stem cells: clinical update. Nat. Rev. Clin. Oncol. 2015; 12 :445–464.25850553 38. Cortes  J.E., GutzmerR., KieranM.W., SolomonJ.A.  Hedgehog signaling inhibitors in solid and hematological cancers. Cancer Treat. Rev. 2019; 76 :41–50.31125907 39. Peer  E., TesanovicS., AbergerF.  Next-generation Hedgehog/GLI pathway inhibitors for cancer therapy. Cancers (Basel). 2019; 11 :538.30991683 40. Carballo  G.B., HonoratoJ.R., de LopesG.P.F., SpohrT.  A highlight on Sonic hedgehog pathway. Cell Commun. Signal. 2018; 16 :11.29558958 41. Liu  A.  Proteostasis in the Hedgehog signaling pathway. Semin. Cell Dev. Biol. 2019; 93 :153–163.31429406 42. Pine  S.R.  Rethinking gamma-secretase inhibitors for treatment of non-small-cell lung cancer: is notch the target?. Clin. Cancer Res. 2018; 24 :6136–6141.30104200 43. Levin  A.A.  Treating disease at the RNA level with oligonucleotides. N. Engl. J. Med. 2019; 380 :57–70.30601736 44. Bennett  C.F.  Therapeutic antisense oligonucleotides are coming of age. Annu. Rev. Med. 2019; 70 :307–321.30691367 45. Shen  X., CoreyD.R.  Chemistry, mechanism and clinical status of antisense oligonucleotides and duplex RNAs. Nucleic Acids Res. 2018; 46 :1584–1600.29240946 46. Egli  M., ManoharanM.  Re-engineering RNA molecules into therapeutic agents. Acc. Chem. Res. 2019; 52 :1036–1047.30912917 47. Shen  W., De HoyosC.L., MigawaM.T., VickersT.A., SunH., LowA., BellT.A.3rd, RahdarM., MukhopadhyayS., HartC.E.et al .  Chemical modification of PS-ASO therapeutics reduces cellular protein-binding and improves the therapeutic index. Nat. Biotechnol. 2019; 37 :640–650.31036929 48. Smith  C.I.E., ZainR.  Therapeutic oligonucleotides: state of the art. Annu. Rev. Pharmacol. Toxicol. 2019; 59 :605–630.30285540 49. Bennett  C.F., BakerB.F., PhamN., SwayzeE., GearyR.S.  Pharmacology of antisense drugs. Annu. Rev. Pharmacol. Toxicol. 2017; 57 :81–105.27732800 50. Hagedorn  P.H., PerssonR., FunderE.D., AlbaekN., DiemerS.L., HansenD.J., MollerM.R., PapargyriN., ChristiansenH., HansenB.R.et al .  Locked nucleic acid: modality, diversity, and drug discovery. Drug Discov. Today. 2018; 23 :101–114.28988994 51. Summerton  J.E.  Invention and early history of morpholinos: from pipe dream to practical products. Methods Mol. Biol. 2017; 1565 :1–15.28364229 52. Juliano  R.L.  The delivery of therapeutic oligonucleotides. Nucleic Acids Res. 2016; 44 :6518–6548.27084936 53. Cullis  P.R., HopeM.J.  Lipid nanoparticle systems for enabling gene therapies. Mol. Ther. 2017; 25 :1467–1475.28412170 54. Dong  Y., SiegwartD.J., AndersonD.G.  Strategies, design, and chemistry in siRNA delivery systems. Adv. Drug Deliv. Rev. 2019; 144 :133–147.31102606 55. Fang  J., NakamuraH., MaedaH.  The EPR effect: unique features of tumor blood vessels for drug delivery, factors involved, and limitations and augmentation of the effect. Adv. Drug Deliv. Rev. 2011; 63 :136–151.20441782 56. Danhier  F.  To exploit the tumor microenvironment: since the EPR effect fails in the clinic, what is the future of nanomedicine?. J. Control. Release. 2016; 244 :108–121.27871992 57. Seth  P.P., TanowitzM., BennettC.F.  Selective tissue targeting of synthetic nucleic acid drugs. J. Clin. Invest. 2019; 129 :915–925.30688661 58. Lee  S.H., KangY.Y., JangH.E., MokH.  Current preclinical small interfering RNA (siRNA)-based conjugate systems for RNA therapeutics. Adv. Drug Deliv. Rev. 2016; 104 :78–92.26514375 59. Crooke  S.T., WangS., VickersT.A., ShenW., LiangX.H.  Cellular uptake and trafficking of antisense oligonucleotides. Nat. Biotechnol. 2017; 35 :230–237.28244996 60. Dowdy  S.F.  Overcoming cellular barriers for RNA therapeutics. Nat. Biotechnol. 2017; 35 :222–229.28244992 61. Juliano  R.L.  Intracellular trafficking and endosomal release of oligonucleotides: what we know and what we don’t. Nucleic Acid Ther. 2018; 28 :166–177.29708838 62. Yang  B., MingX., CaoC., LaingB., YuanA., PorterM.A., Hull-RydeE.A., MaddryJ., SutoM., JanzenW.P.et al .  High-throughput screening identifies small molecules that enhance the pharmacological effects of oligonucleotides. Nucleic Acids Res. 2015; 43 :1987–1996.25662226 63. Juliano  R.L., WangL., TavaresF., BrownE.G., JamesL., AriyarathnaY., MingX., MaoC., SutoM.  Structure–activity relationships and cellular mechanism of action of small molecules that enhance the delivery of oligonucleotides. Nucleic Acids Res. 2018; 46 :1601–1613.29361039 64. Crooke  S.T., VickersT.A., LiangX.H.  Phosphorothioate modified oligonucleotide–protein interactions. Nucleic Acids Res. 2020; 48 :5235–5253.32356888 65. Chen  X., MangalaL.S., Rodriguez-AguayoC., KongX., Lopez-BeresteinG., SoodA.K.  RNA interference-based therapy and its delivery systems. Cancer Metastasis Rev. 2018; 37 :107–124.29243000 66. Yamakawa  K., Nakano-NarusawaY., HashimotoN., YokohiraM., MatsudaY.  Development and clinical trials of nucleic acid medicines for pancreatic cancer treatment. Int. J. Mol. Sci. 2019; 20 :4224.31470511 67. Lee  S.J., KimM.J., KwonI.C., RobertsT.M.  Delivery strategies and potential targets for siRNA in major cancer types. Adv. Drug Deliv. Rev. 2016; 104 :2–15.27259398 68. Martinez-Montiel  N., Rosas-MurrietaN.H., Anaya RuizM., Monjaraz-GuzmanE., Martinez-ContrerasR.  Alternative splicing as a target for cancer treatment. Int. J. Mol. Sci. 2018; 19 :545.29439487 69. Ross  S.J., RevenkoA.S., HansonL.L., EllstonR., StaniszewskaA., WhalleyN., PandeyS.K., RevillM., RooneyC., BuckettL.K.et al .  Targeting KRAS-dependent tumors with AZD4785, a high-affinity therapeutic antisense oligonucleotide inhibitor of KRAS. Sci. Transl. Med. 2017; 9 :eaal5253.28615361 70. Rao  D.D., LuoX., WangZ., JayC.M., BrunicardiF.C., MalteseW., ManningL., SenzerN., NemunaitisJ.  KRAS mutant allele-specific expression knockdown in pancreatic cancer model with systemically delivered bi-shRNA KRAS lipoplex. PLoS One. 2018; 13 :e0193644.29851957 71. Kim  M.J., LeeS.J., RyuJ.H., KimS.H., KwonI.C., RobertsT.M.  Combination of KRAS gene silencing and PI3K inhibition for ovarian cancer treatment. J. Control. Release. 2020; 318 :98–108.31838203 72. Strand  M.S., KrasnickB.A., PanH., ZhangX., BiY., BrooksC., WetzelC., SankpalN., FlemingT., GoedegebuureS.P.et al .  Precision delivery of RAS-inhibiting siRNA to KRAS driven cancer via peptide-based nanoparticles. Oncotarget. 2019; 10 :4761–4775.31413817 73. Perepelyuk  M., ShoyeleO., BirbeR., ThangavelC., LiuY., DenR.B., SnookA.E., LuB., ShoyeleS.A.  siRNA-encapsulated hybrid nanoparticles target mutant K-ras and inhibit metastatic tumor burden in a mouse model of lung cancer. Mol. Ther. Nucleic Acids. 2017; 6 :259–268.28325292 74. Kamerkar  S., LeBleuV.S., SugimotoH., YangS., RuivoC.F., MeloS.A., LeeJ.J., KalluriR.  Exosomes facilitate therapeutic targeting of oncogenic KRAS in pancreatic cancer. Nature. 2017; 546 :498–503.28607485 75. Xue  W., DahlmanJ.E., TammelaT., KhanO.F., SoodS., DaveA., CaiW., ChirinoL.M., YangG.R., BronsonR.et al .  Small RNA combination therapy for lung cancer. Proc. Natl Acad. Sci. U.S.A. 2014; 111 :E3553–E3561.25114235 76. Pecot  C.V., WuS.Y., BellisterS., FilantJ., RupaimooleR., HisamatsuT., BhattacharyaR., MaharajA., AzamS., Rodriguez-AguayoC.et al .  Therapeutic silencing of KRAS using systemically delivered siRNAs. Mol. Cancer Ther. 2014; 13 :2876–2885.25281617 77. Baumer  S., BaumerN., AppelN., TerheydenL., FremereyJ., SchelhaasS., WardelmannE., BuchholzF., BerdelW.E., Muller-TidowC.  Antibody-mediated delivery of anti-KRAS-siRNA in vivo overcomes therapy resistance in colon cancer. Clin. Cancer Res. 2015; 21 :1383–1394.25589625 78. Zorde Khvalevsky  E., GabaiR., RachmutI.H., HorwitzE., BrunschwigZ., OrbachA., ShemiA., GolanT., DombA.J., YavinE.et al .  Mutant KRAS is a druggable target for pancreatic cancer. Proc. Natl Acad. Sci. U.S.A. 2013; 110 :20723–20728.24297898 79. Liu  Y., GundaV., ZhuX., XuX., WuJ., AskhatovaD., FarokhzadO.C., ParangiS., ShiJ.  Theranostic near-infrared fluorescent nanoplatform for imaging and systemic siRNA delivery to metastatic anaplastic thyroid cancer. Proc. Natl Acad. Sci. U.S.A. 2016; 113 :7750–7755.27342857 80. Dorrani  M., GarbuzenkoO.B., MinkoT., Michniak-KohnB.  Development of edge-activated liposomes for siRNA delivery to human basal epidermis for melanoma therapy. J. Control. Release. 2016; 228 :150–158.26965957 81. Zhang  Y., ZhanX., PengS., CaiY., ZhangY.S., LiuY., WangZ., YuY., WangY., ShiQ.et al .  Targeted-gene silencing of BRAF to interrupt BRAF/MEK/ERK pathway synergized photothermal therapeutics for melanoma using a novel FA-GNR-siBRAF nanosystem. Nanomedicine. 2018; 14 :1679–1693.29684526 82. Fernandes  M.S., MeloS., VelhoS., CarneiroP., CarneiroF., SerucaR.  Specific inhibition of p110alpha subunit of PI3K: putative therapeutic strategy for KRAS mutant colorectal cancers. Oncotarget. 2016; 7 :68546–68558.27602501 83. Dahlgren  C., ZhangH.Y., DuQ., GrahnM., NorstedtG., WahlestedtC., LiangZ.  Analysis of siRNA specificity on targets with double-nucleotide mismatches. Nucleic Acids Res. 2008; 36 :e53.18420656 84. Schwarz  D.S., DingH., KenningtonL., MooreJ.T., SchelterJ., BurchardJ., LinsleyP.S., AroninN., XuZ., ZamoreP.D.  Designing siRNA that distinguish between genes that differ by a single nucleotide. PLoS Genet. 2006; 2 :e140.16965178 85. Gysin  S., SaltM., YoungA., McCormickF.  Therapeutic strategies for targeting ras proteins. Genes Cancer. 2011; 2 :359–372.21779505 86. Syn  N.L., WangL., ChowE.K., LimC.T., GohB.C.  Exosomes in cancer nanomedicine and immunotherapy: prospects and challenges. Trends Biotechnol. 2017; 35 :665–676.28365132 87. Arrighetti  N., CorboC., EvangelopoulosM., PastoA., ZucoV., TasciottiE.  Exosome-like nanovectors for drug delivery in cancer. Curr. Med. Chem. 2019; 26 :6132–6148.30182846 88. Mendt  M., KamerkarS., SugimotoH., McAndrewsK.M., WuC.C., GageaM., YangS., BlankoE.V.R., PengQ., MaX.et al .  Generation and testing of clinical-grade exosomes for pancreatic cancer. JCI Insight. 2018; 3 :e99263.29669940 89. Alvarez-Erviti  L., SeowY., YinH., BettsC., LakhalS., WoodM.J.  Delivery of siRNA to the mouse brain by systemic injection of targeted exosomes. Nat. Biotechnol. 2011; 29 :341–345.21423189 90. Ganesh  S., KoserM.L., CyrW.A., ChopdaG.R., TaoJ., ShuiX., YingB., ChenD., PandyaP., ChipumuroEet al .  Direct pharmacological inhibition of beta-catenin by RNA interference in tumors of diverse origin. Mol. Cancer Ther. 2016; 15 :2143–2154.27390343 91. Ganesh  S., ShuiX., CraigK.P., ParkJ., WangW., BrownB.D., AbramsM.T.  RNAi-mediated beta-catenin inhibition promotes T cell infiltration and antitumor activity in combination with immune checkpoint blockade. Mol. Ther. 2018; 26 :2567–2579.30274786 92. Vonbrull  M., RiegelE., HalterC., AignerM., BockH., WernerB., LindhorstT., CzernyT.  A dominant negative antisense approach targeting beta-catenin. Mol. Biotechnol. 2018; 60 :339–349.29524201 93. Guo  X., XiaoH., GuoS., LiJ., WangY., ChenJ., LouG.  Long noncoding RNA HOTAIR knockdown inhibits autophagy and epithelial–mesenchymal transition through the Wnt signaling pathway in radioresistant human cervical cancer HeLa cells. J. Cell. Physiol. 2019; 234 :3478–3489.30367473 94. Liu  X., YanY., MaW., WuS.  Knockdown of frizzled-7 inhibits cell growth and metastasis and promotes chemosensitivity of esophageal squamous cell carcinoma cells by inhibiting Wnt signaling. Biochem. Biophys. Res. Commun. 2017; 490 :1112–1118.28669726 95. Alshaer  W., AlqudahD.A., WehaibiS., AbuarqoubD., ZihlifM., HatmalM.M., AwidiA.  Downregulation of STAT3, beta-catenin, and Notch-1 by single and combinations of siRNA treatment enhance chemosensitivity of wild type and doxorubicin resistant MCF7 breast cancer cells to doxorubicin. Int. J. Mol. Sci. 2019; 20 :3696.31357721 96. Kang  H., JeongJ.Y., SongJ.Y., KimT.H., KimG., HuhJ.H., KwonA.Y., JungS.G., AnH.J.  Notch3-specific inhibition using siRNA knockdown or GSI sensitizes paclitaxel-resistant ovarian cancer cells. Mol. Carcinog. 2016; 55 :1196–1209.26207830 97. Lai  X.X., LiG., LinB., YangH.  Interference of Notch 1 inhibits the proliferation and invasion of breast cancer cells: involvement of the beta-catenin signaling pathway. Mol. Med. Rep. 2018; 17 :2472–2478.29207146 98. Hayashi  T., GustK.M., WyattA.W., GorikiA., JagerW., AwreyS., LiN., OoH.Z., Altamirano-DimasM., ButtyanR.et al .  Not all NOTCH is created equal: the oncogenic role of NOTCH2 in bladder cancer and its implications for targeted therapy. Clin. Cancer Res. 2016; 22 :2981–2992.26769750 99. Wang  X., WeiS., ZhaoY., ShiC., LiuP., ZhangC., LeiY., ZhangB., BaiB., HuangY.et al .  Anti-proliferation of breast cancer cells with itraconazole: Hedgehog pathway inhibition induces apoptosis and autophagic cell death. Cancer Lett. 2017; 385 :128–136.27810405 100. Onishi  H., NakamuraK., NagaiS., YanaiK., YamasakiA., KawamotoM., ImaizumiA., MorisakiT.  Hedgehog inhibition upregulates TRK expression to antagonize tumor suppression in small cell lung cancer cells. Anticancer Res. 2017; 37 :4987–4992.28870922 101. He  H., LiuL., MorinE.E., LiuM., SchwendemanA.  Survey of clinical translation of cancer nanomedicines—lessons learned from successes and failures. Acc. Chem. Res. 2019; 52 :2445–2461.31424909 102. Geary  R.S., NorrisD., YuR., BennettC.F.  Pharmacokinetics, biodistribution and cell uptake of antisense oligonucleotides. Adv. Drug Deliv. Rev. 2015; 87 :46–51.25666165 103. Springer  A.D., DowdyS.F.  GalNAc–siRNA conjugates: leading the way for delivery of RNAi therapeutics. Nucleic Acid Ther. 2018; 28 :109–118.29792572 104. Leon  G., MacDonaghL., FinnS.P., CuffeS., BarrM.P.  Cancer stem cells in drug resistant lung cancer: targeting cell surface markers and signaling pathways. Pharmacol. Ther. 2016; 158 :71–90.26706243 105. Najafi  M., FarhoodB., MortezaeeK.  Cancer stem cells (CSCs) in cancer progression and therapy. J. Cell. Physiol. 2019; 234 :8381–8395.30417375 106. Dzierlega  K., YokotaT.  Optimization of antisense-mediated exon skipping for Duchenne muscular dystrophy. Gene Ther. 2020; doi:10.1038/s41434-020-0156-6. 107. Fay  A.J., KnoxR., NeilE.E., StroberJ.  Targeted treatments for inherited neuromuscular diseases of childhood. Semin. Neurol. 2020; 40 :335–341.32294764 108. Martinez-Marti  A., NavarroA., FelipE.  Epidermal growth factor receptor first generation tyrosine-kinase inhibitors. Transl. Lung Cancer Res. 2019; 8 :S235–S246.31857948

PMC007xxxxxx/PMC7575049.txt

==== Front Bioinformatics Bioinformatics bioinformatics Bioinformatics 1367-4803 1367-4811 Oxford University Press 32579220 10.1093/bioinformatics/btaa587 btaa587 Applications Notes Structural Bioinformatics AcademicSubjects/SCI01060 HaDeX: an R package and web-server for analysis of data from hydrogen–deuterium exchange mass spectrometry experiments Puchała Weronika Institute of Biochemistry and Biophysics, Polish Academy of Sciences, Warsaw 02-106, Poland http://orcid.org/0000-0001-8926-582X Burdukiewicz Michał Faculty of Mathematics and Information Science, Warsaw University of Technology, Warsaw 00-662, Poland Kistowski Michał Institute of Biochemistry and Biophysics, Polish Academy of Sciences, Warsaw 02-106, Poland Dąbrowska Katarzyna A Institute of Biochemistry and Biophysics, Polish Academy of Sciences, Warsaw 02-106, Poland http://orcid.org/0000-0003-1832-4364 Badaczewska-Dawid Aleksandra E Faculty of Chemistry, Biological and Chemical Research Center, University of Warsaw, Warsaw 02-093, Poland http://orcid.org/0000-0001-6206-0672 Cysewski Dominik Institute of Biochemistry and Biophysics, Polish Academy of Sciences, Warsaw 02-106, Poland Dadlez Michał Institute of Biochemistry and Biophysics, Polish Academy of Sciences, Warsaw 02-106, Poland Ponty Yann Associate Editor † The authors wish it to be known that, in their opinion, Weronika Puchała and Michał Burdukiewicz should be regarded as Joint First Authors. To whom correspondence should be addressed. Email: michald@ibb.waw.pl 15 8 2020 24 6 2020 24 6 2020 36 16 45164518 29 8 2019 08 5 2020 16 6 2020 © The Author(s) 2020. Published by Oxford University Press. 2020 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Abstract Motivation Hydrogen–deuterium mass spectrometry (HDX-MS) is a rapidly developing technique for monitoring dynamics and interactions of proteins. The development of new devices has to be followed with new software suites addressing emerging standards in data analysis. Results We propose HaDeX, a novel tool for processing, analysis and visualization of HDX-MS experiments. HaDeX supports a reproducible analytical process, including data exploration, quality control and generation of publication-quality figures. Availability and implementation HaDeX is available primarily as a web-server (http://mslab-ibb.pl/shiny/HaDeX/), but its all functionalities are also accessible as the R package (https://CRAN.R-project.org/package=HaDeX) and standalone software (https://sourceforge.net/projects/HaDeX/). Supplementary information Supplementary data are available at Bioinformatics online. Foundation of Polish Science TEAM TECH CORE FACILITY/2016-2/2 Mass Spectrometry of Biopharmaceuticals ==== Body pmc1 Introduction The understanding of interactions between proteins and other molecules is crucial for studying complex biological systems. Among the methods for characterization of conformational dynamics of proteins and their complexes, hydrogen–deuterium mass spectrometry (HDX-MS) has proven to be both rapid and sensitive (Konermann et al., 2011). This technique is especially important in the case of proteins that are difficult to study with other methods such as membrane proteins, oligomerizing proteins or intrinsically disordered proteins (Goswami et al., 2013). Hydrogen–deuterium exchange monitors an exchange of amide hydrogens in peptide bonds. Protein incubation in D2O leads to the exchange of hydrogen to deuterium atoms in amides. The speed of such exchange depends mostly on the stability of hydrogen bonds formed by the hydrogen amides and also by the accessibility to the solvent. The importance of the influence of these factors is discussed, but it is considered that the dominant component is the stability of hydrogen bonds and not the availability of solvent. Thus, this process may be slowed due to structural factors: stability, flexibility and accessibility. Therefore the method, by measuring the level of protection of amides at different times of incubation in D2O, recognizes the stability of hydrogen bonding networks and regions of protein with limited solvent accessibility. HDX probes the dynamic nature of proteins systems, as opposed to static structures, offered by X-ray crystallography. It also allows mapping the regions affected by the interaction between the proteins concerned. The main scheme of local (continuous-labeling, bottom-up) HDX-MS experiments consist of: the incubation of a protein in a D2O solution, buffered to native or native-like conditions, followed by exchange quench, proteolytic digestion and mass spectrometry measurement of resulting peptide’s masses. Results generated by HDX-MS are complex and demanding in terms of analysis, interpretation and visual presentation. While there are many open-source and free to use software packages addressing the challenges of HDX-MS data analysis (Hourdel et al., 2016; Kavan and Man, 2011; Lau et al., 2019; Lumpkin and Komives, 2019), HaDeX aims to cover post-processing workflow, where results of the experiment are analyzed and presented in a publication-friendly format (for comparison of mentioned tools see Supplementary Information SI1). It forces HDX-MS users to rely on several pieces of software, thus making the already laborious process even more time-consuming. Yet, another challenge of the HDX-MS data analysis is a proper visualization of results on 3D protein structure which is supported by HDX-Viewer (Bouyssié et al., 2019). As the field HDX-MS is still growing, researchers introduce new standards, including data analysis and reporting (Masson et al., 2019). The available software does not allow automatic generation of reports according to the new guidelines, which adds work for experimentalists. To alleviate these issues, we propose HaDeX, a comprehensive software suite for analysis of HDX-MS data. The aim of HaDeX is not only to provide a comprehensive way to study results of HDX-MS experiments but also to report their results in a reproducible way by including all parameters relevant to data analysis as the size of confidence intervals. 2 Materials and methods HaDeX dissects work into three steps: (i) general properties of a sequence reconstructed from measured peptides, (ii) uncertainty of measurements and their significance and (iii) visualization of results (Fig. 1). Fig. 1. The core functionalities of the HaDeX GUI. (A) Multistate discriminative analysis of peptides; (B) Woods plots; (C) coverage of the protein sequence by peptides measured with mass spectrometry and (D) kinetics of hydrogen–deuterium exchange The only input necessary to start work with HaDeX is an exported data as a.csv datafile (in the Cluster format) produced by the DynamX™ 2.0 or 3.0 (Waters Corp.) Our software does not require any external preprocessing, which not only streamlines the whole workflow but also increases its reproducibility. HaDeX uses a well-established method to compute confidence intervals for measured peptides (Houde et al., 2011) (see Supplementary Information SI2.5). Additionally, we enhanced this functionality by providing uncertainties derived by error propagation (Joint Committee for Guides in Metrology, 2008) (see Supplementary Information SI2.2). Known in the literature as Woods charts (Woods and Hamuro, 2001), these types of plots are used to visually inspect results of HDX-MS studies (Kupniewska-Kozak et al., 2010). A user can further enhance these charts by indicating specified confidence levels (Fig. 1B). All figures are exportable in vector formats. HaDeX provides a highly customizable report generation module, which increases the reproducibility of its analytic workflow. The report not only contains partial results of the analysis, but also the additional input provided by the user (e.g. altered significance levels). 3 Conclusion and availability HaDeX supports a large part of the analytic workflow of HDX-MS data (Claesen and Burzykowski, 2017), from the quality control of the data to publication-ready visualizations. However, our tool does not provide any high-resolution output (as 3D visualizations or deuteration heatmaps based on single residues) thus we refer the user to methods addressing this problem (Gessner et al., 2017). Thanks to the input of HDX-MS users, our software is not only a convenient re-implementation of already existing methods but also provides unique functionalities unavailable elsewhere: novel, ISO-based uncertainty computations, multistate analysis and downloadable reports produced according to the novel guidelines (Masson et al., 2019). As our tool is targeted at experimentalists, it is available as a web server and a standalone GUI (Windows only). However, bioinformaticians can access HaDeX functions programatically, as it is also available as an R package. We hope that thanks to its comprehensiveness and reproducibility-oriented features HaDeX can satisfy requirements of users from both academia and industry. Supplementary Material btaa587_supplementary_data Click here for additional data file. Acknowledgements The authors would like to thank Julien Marcoux (IPBS-Toulouse) and members of the Mass Spectrometry Core Facility (Institute of Biochemistry and Biophysics Polish Academy of Sciences) for suggestions with software features. Funding Foundation of Polish Science TEAM TECH CORE FACILITY/2016-2/2 Mass Spectrometry of Biopharmaceuticals—improved methodologies for qualitative, quantitative and structural characterization of drugs, proteinaceous drug targets and diagnostic molecules. Conflict of Interest: none declared. ==== Refs References Bouyssié D.  et al (2019) HDX-Viewer: interactive 3D visualization of hydrogen–deuterium exchange data. Bioinformatics, 35 , 5331–5333.31287496 Claesen J. , BurzykowskiT. (2017) Computational methods and challenges in hydrogen/deuterium exchange mass spectrometry. Mass Spectr. Rev., 36 , 649–667. Gessner C.  et al (2017) Computational method allowing hydrogen–deuterium exchange mass spectrometry at single amide resolution. Sci. Rep., 7 , 3789.28630467 Goswami D.  et al (2013) Time window expansion for HDX analysis of an intrinsically disordered protein. J. Am. Soc. Mass Spectr., 24 , 1584–1592. Houde D.  et al (2011) The utility of hydrogen/deuterium exchange mass spectrometry in biopharmaceutical comparability studies. J. Pharm. Sci., 100 , 2071–2086.21491437 Hourdel V.  et al (2016) MEMHDX: an interactive tool to expedite the statistical validation and visualization of large HDX-MS datasets. Bioinformatics, 32 , 3413–3419.27412089 Joint Committee for Guides in Metrology. (2008) JCGM 100: evaluation of measurement data – guide to the expression of uncertainty in measurement. Technical report, JCGM. Kavan D. , ManP. (2011) MSTools—web based application for visualization and presentation of HXMS data. Int. J. Mass Spectr., 302 , 53–58. Konermann L.  et al (2011) Hydrogen exchange mass spectrometry for studying protein structure and dynamics. Chem. Soc. Rev., 40 , 1224–1234.21173980 Kupniewska-Kozak A.  et al (2010) Intertwined structured and unstructured regions of exrage identified by monitoring hydrogen–deuterium exchange. J. Mol. Biol., 403 , 52–65.20732329 Lau A.M.C.  et al (2019) Deuteros: software for rapid analysis and visualization of data from differential hydrogen deuterium exchange-mass spectrometry. Bioinformatics, 35 , 3171–3173. Lumpkin R. , KomivesE.A. (2019) DECA, a comprehensive, automatic post-processing program for HDX-MS data. Mol. Cell. Proteomics., 18 , 2516–2523.31594786 Masson G.R.  et al (2019) Recommendations for performing, interpreting and reporting hydrogen deuterium exchange mass spectrometry (HDX-MS) experiments. Nat. Methods, 16 , 595–602.31249422 Woods V.L. , HamuroY. (2001) High resolution, high-throughput amide deuterium exchange-mass spectrometry (DXMS) determination of protein binding site structure and dynamics: utility in pharmaceutical design. J. Cell. Biochem., 37 , 89–98.

PMC007xxxxxx/PMC7594209.txt

==== Front J Adv Nurs J Adv Nurs 10.1111/(ISSN)1365-2648 JAN Journal of Advanced Nursing 0309-2402 1365-2648 John Wiley and Sons Inc. Hoboken 30644123 10.1111/jan.13809 JAN13809 Research Methodology: Empirical Research ‐ Methodology Research Papers Research Methodology: Empirical Research ‐ Methodology Improving reporting of meta‐ethnography: The eMERGe reporting guidance 改进元人种志的报告:新兴报告指南 FRANCE et al. France Emma F. 1 emma.france@stir.ac.uk @Emma_France Cunningham Maggie 1 Ring Nicola 2 Uny Isabelle 1 Duncan Edward AS 1 Jepson Ruth G 3 Maxwell Margaret 1 Roberts Rachel J. 1 Turley Ruth L. 4 Booth Andrew 6 Britten Nicky 7 Flemming Kate 8 Gallagher Ian 9 Garside Ruth 7 Hannes Karin 10 Lewin Simon 11 12 Noblit George W. 13 Pope Catherine 14 Thomas James 15 Vanstone Meredith 16 Higginbottom Gina M. A. 17 Noyes Jane 5 1 University of Stirling Stirling UK 2 Edinburgh Napier University Edinburgh UK 3 University of Edinburgh Edinburgh UK 4 Cardiff University Cardiff UK 5 Bangor University Bangor UK 6 University of Sheffield Sheffield UK 7 University of Exeter Medical School Exeter UK 8 Department of Health Sciences University of York York UK 9 eMERGe project Stirling UK 10 University of Leuven Leuven Belgium 11 Global Health Unit Norwegian Institute of Public Health and Health Systems Research Unit Oslo Norway 12 South African Medical Research Council Capetown South Africa 13 University of North Carolina at Chapel Hill USA 14 University of Southampton Southampton UK 15 EPPI‐Centre Institute of Education London UK 16 McMaster University Hamilton ON Canada 17 School of Health Sciences & Centre for Evidence Based Health Care The University of Nottingham Nottingham UK * Correspondence Emma France, NMAHP Research Unit, Unit 13 Scion House, Stirling University Innovation Park, Stirling, FK9 4NF, UK. Email: emma.france@stir.ac.uk 15 1 2019 5 2019 75 5 10.1111/jan.2019.75.issue-5 11261139 22 6 2018 13 6 2017 03 7 2018 © 2019 The Authors. Journal of Advanced Nursing Published by John Wiley & Sons Ltd. https://creativecommons.org/licenses/by-nc/4.0/ This is an open access article under the terms of the http://creativecommons.org/licenses/by-nc/4.0/ License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. Abstract Aims The aim of this study was to provide guidance to improve the completeness and clarity of meta‐ethnography reporting. Background Evidence‐based policy and practice require robust evidence syntheses which can further understanding of people's experiences and associated social processes. Meta‐ethnography is a rigorous seven‐phase qualitative evidence synthesis methodology, developed by Noblit and Hare. Meta‐ethnography is used widely in health research, but reporting is often poor quality and this discourages trust in and use of its findings. Meta‐ethnography reporting guidance is needed to improve reporting quality. Design The eMERGe study used a rigorous mixed‐methods design and evidence‐based methods to develop the novel reporting guidance and explanatory notes. Methods The study, conducted from 2015 ‐ 2017, comprised of: (1) a methodological systematic review of guidance for meta‐ethnography conduct and reporting; (2) a review and audit of published meta‐ethnographies to identify good practice principles; (3) international, multidisciplinary consensus‐building processes to agree guidance content; (4) innovative development of the guidance and explanatory notes. Findings Recommendations and good practice for all seven phases of meta‐ethnography conduct and reporting were newly identified leading to 19 reporting criteria and accompanying detailed guidance. Conclusion The bespoke eMERGe Reporting Guidance, which incorporates new methodological developments and advances the methodology, can help researchers to report the important aspects of meta‐ethnography. Use of the guidance should raise reporting quality. Better reporting could make assessments of confidence in the findings more robust and increase use of meta‐ethnography outputs to improve practice, policy, and service user outcomes in health and other fields. This is the first tailored reporting guideline for meta‐ethnography. This article is being simultaneously published in the following journals: Journal of Advanced Nursing, Psycho‐oncology, Review of Education, and BMC Medical Research Methodology. 目的 本研究的目的是为提高元人种志报告的完整性和清晰度提供指导。 背景 基于证据的政策和实践需要强有力的证据合成,以进一步了解人们的经验和相关的社会过程。 种志是由诺伯特和黑尔开发的一种严谨的七‐相定性证据综合方法。元人种志在健康研究中被广泛使用,但报告质量往往很差,这阻碍了对研究结果的信任和使用。元人种志报告指南是提高报告质量所必需的。 设计 本研究采用了严格的混合方法设计和基于证据的方法来开发新的报告指南和注释。 方法 这项研究从2015年到2017年进行,包括:(1)对元人种志行为和报告指南的方法系统审查;(2)审查和审计已出版的元人种志,以确定良好的实践原则;(3)国际、多学科的共识——建立过程以达成指导内容;(4)创新发展的指导和说明。 研究发现 新确定的所有七个元人种志实施和报告阶段的建议和良好做法提供了19项报告准则,并附有详细的指南。 结论 定制的新兴报告指南,包含了新的方法论的发展和进步,可以帮助研究人员报告元人种志的重要方面。使用指南应提高报告质量。更好的报告可以使对调查结果的信心评估更加可靠,并增加对元人种学输出的使用,以改进卫生和其他领域的实践、政策和服务用户结果。这是第一个为元人种学量身定制的报告指南。本文同时发表于《高级护理杂志》、《精神肿瘤学》、《教育评论》和《英国医学委员会医学研究方法论》等期刊。 guideline meta‐ethnography nursing publication standards qualitative evidence synthesis qualitative research reporting research design systematic review National Institute of Health Research (NIHR) 10.13039/501100000272 13/114/60 Economic and Social Research Council 10.13039/501100000269 RES‐590‐28‐0005 Welsh GovernmentUK Clinical Research Collaboration 10.13039/100011417 source-schema-version-number2.0 cover-dateMay 2019 details-of-publishers-convertorConverter:WILEY_ML3GV2_TO_JATSPMC version:6.2.9 mode:remove_FC converted:22.06.2023 France EF , Cunningham M , Ring N , et al. Improving reporting of meta‐ethnography: The eMERGe reporting guidance. J Adv Nurs. 2019;75 :1126–1139. 10.1111/jan.13809 30644123 Funding information This study was funded by an NIHR Health Service and Delivery Research HS&DR grant (13/114/60). The views and opinions expressed therein are those of the authors and do not necessarily reflect those of the HS&DR Programme, NIHR, NHS, or the Department of Health. This article is being simultaneously published in the Journal of Advanced Nursing, Psycho‐oncology, Review of Education, and BMC Medical Research Methodology. The article followed a double‐blind peer‐review model managed by the Journal of Advanced Nursing, and the editors from each of the journals in question consolidated on the decision process. [Correction added after first online publication on 19 August 2020: Information about DECIPHer has been moved from the Funding section to the Acknowledgements per the copyright license requirements.] ==== Body pmc Why is this research or review needed? No bespoke reporting guidance exists for meta‐ethnography, one of the most commonly used yet often poorly reported, methodologies for qualitative evidence synthesis which could contribute robust evidence for policy and practice. Existing generic guidance for reporting qualitative evidence syntheses pays insufficient attention to reporting the complex synthesis processes of meta‐ethnography—tailored guidance should improve reporting and could improve quality of conduct. Better reporting of meta‐ethnographies will likely have greater impact on understanding of specific phenomena of interest which will subsequently inform intervention development and changes in policy and practice. What are the key findings? Recommendations, guidance, and good practice for conducting and/or reporting all seven phases of a meta‐ethnography were identified for the first time, along with uncertainties and evidence gaps regarding good practices. Nineteen reporting criteria were developed including detailed guidance on Phases 3–6: approach to reading/extracting data; processes for/ outcome of relating studies; processes for/ outcome of translation and synthesizing translations. The analysis and interpretation of methodological evidence and novel development work underpinning this new tailored reporting guidance advances meta‐ethnography methodology, for example, to incorporate good practice in translation and synthesis. How should the findings be used to influence policy/practice/research/education? Use of the guidance by researchers, peer‐reviewers, and journal editors to ensure complete and transparent reporting of meta‐ethnographies will ensure their findings are optimized for use in policy and practice. The guidance can be used to inform the design and conduct of meta‐ethnographies because of the underpinning rigorous, comprehensive analysis, interpretation, and synthesis of the latest methodological evidence. 1 INTRODUCTION Evidence‐based decision‐making for health services, policies, and programmes requires qualitative and quantitative research; this is recognized by leading evidence‐producing organisations including Cochrane, the Campbell Collaboration, and the World Health Organization (Noyes et al., 2018; Uny, France, & Noblit, 2017). To make sense of large volumes of research, robust syntheses of all types of research are needed (Noyes et al., 2018). Syntheses of qualitative studies, such as meta‐ethnographies, can be used to develop theory about how a service, policy, strategy, or intervention works and how people experience these (Noyes & Lewin, 2011); provide evidence of the acceptability, feasibility, and appropriateness of interventions or services (Booth, Carroll, Ilott, Low, & Cooper, 2013; Glenton & Lewin, 2016b; Glenton, Lewin, & Gulmezoglu, 2016a; Gulmezoglu, Chandler, Shepperd, & Pantoja, 2013; Pearson, Wiechula, Court, & Lockwood, 2005); convey people's experiences of, for example, illness (Campbell et al., 2011; Pound et al., 2005); and inform the development, implementation, and evaluation of complex interventions (Carroll, 2017; Rycroft‐Malone & Burton, 2015). 1.1 What is meta‐ethnography? Meta‐ethnography is a seven phase, theory‐based (Turner, 1980) and potentially theory‐generating, interpretive methodology for qualitative evidence synthesis developed by sociologists Noblit and Hare (1988) in the field of education. Meta‐ethnography aims to produce novel interpretations that transcend individual study findings, rather than aggregate findings (Thorne, 2015). Meta‐ethnography involves systematically comparing conceptual data from primary qualitative studies to identify and develop new overarching concepts, theories, and models. It was designed to preserve the original meanings and contexts of study concepts (Campbell et al., 2011; Noblit & Hare, 1988). The originators of meta‐ethnography developed a distinctive analytic synthesis process of “translation” and “synthesis of translations” (Noblit & Hare, 1988), underpinned by the theory of social comparison (Turner, 1980), which involves analysing the conceptual data, for example, concepts, themes, developed by authors of primary studies. 1.2 Why is reporting guidance needed Meta‐ethnography is a distinct, complex and increasingly common and influential qualitative methodology. It is the most widely used qualitative evidence synthesis methodology in health and social care research (Dixon‐Woods, Booth, & Sutton, 2007; Hannes & Macaitis, 2012; Ring, Jepson, & Ritchie, 2011b) and is increasingly used by other academic disciplines (Uny et al., 2017). Many other qualitative evidence synthesis methodologies and methods are based on or influenced by it (Dixon‐Woods et al., 2006; Paterson, 2011; Uny et al., 2017). A methodological evaluation of the effectiveness of meta‐ethnography for synthesizing qualitative studies in health and health care concluded that meta‐ethnography can lead to important new conceptual understandings of health care issues (Campbell et al., 2011) and high quality meta‐ethnographies have informed clinical guidelines (Nunes et al., 2009; Ring et al., 2011a). However, the quality of reporting in published meta‐ethnographies varies and is often poor despite methodological advances (Britten et al., 2002; Campbell et al., 2003, 2011; France et al., 2014; Hannes & Macaitis, 2012). Adequate quality in reporting is one of several prerequisites to assessing confidence in meta‐ethnography findings that could inform evidence‐based policy and practice, for instance, in health and social care (Lewin et al., 2015). Reporting guidance is commonly used in health and social care research and can raise publication standards (Plint et al., 2006). For systematic reviews and meta‐analyses of quantitative studies, the most commonly used guidance is Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) (Moher, Liberati, Tetzlaff, Altman, & Group, 2009). For reviews of qualitative studies, the most commonly used one is the generic 2012 ENTREQ (Enhancing transparency in reporting the synthesis of qualitative research) statement (Tong, Flemming, McInnes, Oliver, & Craig, 2012). Qualitative evidence synthesis methodologies differ greatly; therefore, unique reporting guidance for metanarrative reviews was recently developed (Wong, Greenhalgh, Westhorp, Buckingham, & Pawson, 2013). There is currently no guidance on reporting the complex synthesis process of meta‐ethnography. Such guidance should improve the transparency and completeness of reporting and thus maximize the ability of meta‐ethnographies to contribute robust evidence to health, social care, and other disciplines, such as education. Although meta‐ethnography continues to evolve, reporting guidance is needed currently for this complex methodology. 2 METHODS The methods used to develop the eMERGe meta‐ethnography reporting guidance followed a rigorous approach consistent with, but exceeding, good practice recommendations (Moher, Schulz, Simera, & Altman, 2010) and were published in a protocol (France et al., 2015). The research questions were: What are the existing recommendations and guidance for conducting and reporting each process in a meta‐ethnography and why? (Stage 1) What good practice principles can we identify in meta‐ethnography conduct and reporting to inform recommendations and guidance? (Stage 2) From the good practice principles, what standards can we develop in meta‐ethnography conduct and reporting to inform recommendations and guidance? (Stage 2) What is the consensus of experts and other stakeholders on key standards and domains for reporting meta‐ethnography in an abstract and main report/publication? (Stages 3 & 4). Details of the methods are given in supplementary File S1. Guidance development was conducted by the grant project team (the first 10 authors), in consultation with the one of the two originators of meta‐ethnography, George Noblit and supported by a multidisciplinary project advisory group of national and international academics, policy experts, nonacademic users of syntheses such as clinical guideline developers and lay advisors, who had an active role in the development of the guidance and whose contributions were central throughout the project (the 11 authors from A. B. onwards were advisory group members). Guidance development took place over a 2‐year period from 2015 to 2017 and comprised four stages, outlined in Figure 1: Identification of potential reporting standards to include in the guidance; Development and application of potential standards to published meta‐ethnographies; Consensus on guidance content; Development of reporting criteria for the guidance and explanatory notes. Figure 1 Guidance development flowchart 2.1 Stage 1. Identification of standards Stage 1 was conducted by the grant project team who undertook a systematic review (PROSPERO CRD42015024709) of relevant methodological and reporting guidance on meta‐ethnographies to identify potential reporting standards (France et al., 2015). From this review, we identified 138 recommendations for meta‐ethnography standards on reporting from 57 included publications (see supplementary File S2). 2.2 Stage 2. Development and application of the standards The grant project team reviewed 29 published meta‐ethnographies (see supplementary File S3) from various academic disciplines and interviewed nonacademic end users of meta‐ethnographies to identify good practice principles and recommendations which we then developed into an audit tool of 109 measurable provisional standards. The 29 meta‐ethnographies were chosen by academic experts who were asked to justify why they considered them seminal (i.e., they had influenced or significantly advanced thinking and/or were of central importance in the field of meta‐ethnography) or relatively poorly reported, or meta‐ethnographies were identified as poorly reported from published reviews. The team applied the provisional standards to a purposive sample of 40 published health and social care‐related meta‐ethnographies (selected from 571 identified through comprehensive systematic searches to give variation in, for example, journal, academic discipline, topic, number of included studies and of authors—supplementary File S1 gives full sampling details) in a retrospective audit to determine the extent to which the standards were met (“not at all”, “in part” or “in full”) and to identify ways the standards could be refined. 2.3 Stage 3. Consensus on guidance content From the results of Stage 2, the project team reviewed and refined the 109 provisional standards by clarifying ambiguous wording, merging duplicative standards, and combining standards on similar processes to create 53 items which were discussed in an online workshop and tested in Delphi consensus studies (Linstone & Turoff, 2002) with academic and nonacademic potential end users. Two parallel, online Delphi consensus studies with identical questions were conducted: one Delphi for international experts in qualitative methods (comprising editors or researchers with prior meta‐ethnography/qualitative evidence synthesis experience) and one for professional/academic and lay people (potential end‐users of meta‐ethnographies). Sixty‐two people (39 experts and 23 professional/lay people) completed all three rounds of the Delphi. Four items failed to reach consensus in both Delphi studies and so were excluded from the final guidance (these were the abstract should ideally differentiate between reported findings of the primary studies and of the synthesis; state the qualitative research expertise of reviewers; state in which order primary study accounts had data extracted from them; state the order in which studies were translated/synthesized). Participants reached consensus that 49 of 53 items should be included in the guidance, too many for usable reporting guidance; therefore, further steps were undertaken to condense these items into fewer reporting criteria. 2.4 Stage 4. Development of the guidance To develop the final reporting criteria for the guidance, a project advisory group meeting was convened which had 26 attendees including expert academics, other professionals, and lay members. The group discussed and agreed the structure of the guidance and the accompanying explanatory notes. Following this meeting, the grant project team agreed which Delphi items should be merged to create usable guidance. The project advisory group then commented on the readability and usability of the guidance. Members of the grant project team then further refined the guidance and explanatory notes. The final guidance and explanatory notes were checked against the Delphi items to ensure content and meaning had been preserved throughout this iterative process. Members of the project advisory group and project team reviewed and agreed the final guidance table and explanatory notes. Supplementary File S1 gives details of the methods which also appear in a published protocol (France et al., 2015) and funder's report (Cunningham et al., 2018). 3 HOW TO USE THE GUIDANCE The eMERGe reporting guidance is designed for use by researchers conducting a meta‐ethnography (referred to throughout as “reviewers”: the term “reviewers” for people who conduct and report meta‐ethnographies was the preferred term identified from the eMERGe Delphi studies in line with the increasing use of systematic review methodology for qualitative evidence syntheses), peer reviewers, journal editors, and end‐users of meta‐ethnographies including policy makers and practitioners. The eMERGe guidance also provides a helpful structure for anyone contemplating or conducting a meta‐ethnography. While the guidance was developed for meta‐ethnography, some of the reporting criteria, such as those relating to stating a review question and reporting literature search and selection strategies, might also be applicable to other forms of qualitative evidence synthesis and thus overlap with the generic ENTREQ guidance for reporting a wide range of qualitative evidence syntheses (Tong et al., 2012). In contrast to eMERGe, ENTREQ does not provide guidance regarding reporting of the complex analytic synthesis processes (Phases 4–6) in a meta‐ethnography and did not follow good practice guidance for developing a reporting guideline (Moher et al., 2010), for example, it was not designed with the consensus of a wider community of experts (Cunningham et al., 2018; Flemming, Booth, Hannes, Cargo, & Noyes, 2018). The eMERGe guidance consists of three parts: Part 1: Table of reporting criteria that are common to all meta‐ethnographies, Part 2: Detailed explanatory notes on how to apply the common reporting criteria including supplementary detail of findings for phases 3–6 (see supplementary information Table S4), Part 3: Extensions for reporting steps and processes which are not common to every meta‐ethnography. Readers should refer to and use all three parts of the guidance. Parts 1 and 2 of the eMERGe reporting guidance are organized by the seven phases of meta‐ethnography. Suggestions are provided in the grey cells of the table in Part 1 for where specific reporting criteria could be reported under journal article section headings. Where appropriate, reviewers should also consider additional relevant guidance for reporting other common qualitative evidence synthesis steps and processes, such as searches for evidence. See for example, the “STARLITE” guidance (Booth, 2006) and PRISMA (Moher et al., 2009) for reporting literature searches (refer to the EQUATOR Network for a comprehensive database of up‐to‐date reporting guidance https://www.equator-network.org/). Part 3 covers eMERGe extensions for format and content of the meta‐ethnography output (for example, of an abstract); assessment of methodological strengths and limitations of included primary studies; and using the GRADE CERQual approach to assess confidence in findings from qualitative evidence syntheses (Lewin et al., 2015; Noyes et al., 2018). Users of this guidance should note that meta‐ethnography is an iterative process and although the guidance is presented by meta‐ethnography phases, we are not advocating a linear approach to meta‐ethnography conduct. Furthermore, those conducting meta‐ethnographies may need to be creative and adapt the methodology to their specific research/review question (Noblit, 2016). 3.1 Part 1: Guidance table (see Table 1) 3.2 Part 2: Explanatory notes 3.2.1 PHASE 1—Selecting meta‐ethnography and getting started Reporting criterion 1—Rationale and context for the meta‐ethnography Consider whether a meta‐ethnography of this topic is needed (Finlayson & Dixon, 2008; Kangasniemi, Lansimies‐Antikainen, Halkoaho, & Pietila, 2012; Toye et al., 2014), for example, is there an existing meta‐ethnography on the topic and if so, provide a reason for updating it (France, Wells, Lang, & Williams, 2016) and describe the gap in research or knowledge to be filled by the meta‐ethnography. This should include reviewers describing the availability of qualitative data which potentially could be synthesized and the context of the meta‐ethnography, for instance, the political, cultural, social, policy, or other relevant contexts; any funding sources for the meta‐ethnography; and the timescales for the meta‐ethnography conduct. Reviewers should consider referring to frameworks which provide guidance on how to specify context, such as Noyes et al. (2018). Reporting criterion 2—Aim(s) of the meta‐ethnography The intention of meta‐ethnography is to produce a new configuration/interpretation, a new model, conceptual framework, or theory, although ultimately this might not be possible, for instance, if no conceptual innovation had occurred since an early, conceptually rich primary study account (Atkins et al., 2008; Campbell et al., 2011; Malpass et al., 2009). The aim(s) of the meta‐ethnography should be explicitly stated and should be compatible with such intentions. The aim may be refined after reading the literature and examining the available data (Booth et al., 2016; Campbell et al., 2003, 2011; Finfgeld‐Connett, 2014; Finfgeld‐Connett & Johnson, 2013). If the initial aim(s) is (are) changed during Phases 1 and 2, give details of any refinements made. Reporting criterion 3—Focus of the meta‐ethnography The review question(s) should be explicitly stated and be congruent with the intention of meta‐ethnography. If, during later phases, the initial review question(s) or objective(s) needed to be refined, give details of any refinements. A well‐defined review question, specifying a precise focus, can lead to a more efficient synthesis and more useful output (Atkins et al., 2008; Finfgeld‐Connett, 2014; Finfgeld‐Connett & Johnson, 2013), for instance, by contributing to clear study inclusion criteria for Phase 2. Reporting criterion 4—Rationale for using meta‐ethnography Many qualitative evidence synthesis methodologies and methods exist (Booth et al., 2016). Unlike meta‐ethnography, some of these are aggregative (e.g., thematic analysis, Joanna Briggs Institute methods), combine qualitative and quantitative data (e.g., critical interpretive synthesis, metanarrative, metastudy, metasummary, realist synthesis), or have a realist epistemology (e.g., thematic synthesis, framework synthesis) (Booth et al., 2016; Noyes & Lewin, 2011; Paterson, 2011). The rationale should be given for why meta‐ethnography was chosen as the most appropriate metet al.hodology for conducting an interpretive synthesis (Toye et al., 2014). If reviewers made adaptations or modifications to Noblit and Hare's (1988) methodology or methods, state why meta‐ethnography was still considered the most appropriate methodology and describe all adaptations and modifications made. 3.2.2 PHASE 2—Deciding what is relevant Reporting criterion 5—Search strategy Explain how the search strategy was informed by the research aim(s), question, or objectives and the meta‐ethnography's purpose (Booth, 2013; Finfgeld‐Connett & Johnson, 2013). Reviewers should provide a rationale for whether the approach to searching was comprehensive (search strategies sought all available studies), purposeful (e.g., searching sought all available concepts until theoretical saturation was achieved), or a combination of approaches. Purposeful searches may be suited for theory‐generating syntheses (Booth, 2013; Finfgeld‐Connett & Johnson, 2013). In addition, provide a rationale for the selection of bibliographic databases and other sources of literature; when searching was stopped if purposeful searches were used; and any search limiters (restrictions to the searches) such as the years covered, geography, language, and so on. Reporting criterion 6—Search processes Describe and provide a rationale for how the literature searching was conducted, following appropriate guidance for reporting qualitative literature searches, for example, STARLITE (Booth, 2006), some journals may also require use of PRISMA (Moher et al., 2009). Reporting criterion 7—Selecting primary studies Describe the screening method, such as by title, abstract, and/or full text review and identify who was involved in study selection. Specify the inclusion and exclusion criteria for study selection, for example, in terms of population, language, year limits, type of publication, study type, methodology, epistemology, country, setting, type of qualitative data, methods, conceptual richness of data, and so on. Also, describe any sampling decisions for study selection—were all relevant studies included or a purposive or theoretical sample of studies (Finfgeld‐Connett & Johnson, 2013; Suri & Clarke, 2009)? Reporting criterion 8—Outcome of study selection Provide details on the number of primary studies assessed for eligibility and included in the meta‐ethnography. Give reasons for exclusion, for example, for comprehensive searches provide numbers of studies screened indicated in a figure/flowchart; for purposeful searching describe reasons for study exclusion and inclusion based on modifications to the review question and/or contribution to theory development. Outcome of study selection can be presented as a primary study flow diagram or narrative—reviewers should note publication requirements—many journals require a PRISMA type flow diagram (Moher et al., 2009). If comprehensive literature searches were conducted, reviewers should follow appropriate reporting guidance formats, such as PRISMA (Moher et al., 2009) and STARLITE (Booth, 2006). If publication requirements prevent full reporting, reviewers should state where readers can access these data in full, for example, on a project website, in online files. 3.2.3 PHASE 3—Reading included studies Reporting Criterion 9—Reading and data extraction approach This is the phase where the clearest divergence can start to be seen from other types of qualitative evidence syntheses. As described in the original meta‐ethnography text:“… we think it is best to identify this phase as the repeated reading of the accounts and the noting of interpretative metaphors. Meta‐ethnography is the synthesis of texts; this requires extensive attention to the details in the accounts and what they tell you about your substantive concerns.” (Noblit & Hare, 1988, p. 28) Reviewers should describe: the process and strategy for reading included studies to indicate how close (critical) reading was achieved and who was involved in reading studies. the strategy for extracting or recording data from included studies and state who was involved in this, whether processes were conducted independently by reviewers and whether data were checked for accuracy and if so, how. the process for identifying and recording concepts, themes, and metaphors from the primary studies (France et al., 2014). Indicate whether data were extracted from across the full primary study (desirable), or specific sections only, for example, findings (not recommended because conceptual data may appear throughout the account and the primary study context could be lost (Noblit, 2016; Toye et al., 2014)). Clarify which kind(s) of primary study findings were extracted, such as participant quotes and/or concepts developed by authors of primary studies (sometimes called first‐ and second‐order constructs, respectively; Britten et al., 2002) so that readers can follow reviewers’ concept development. Examples of how data extraction has been done include: create a list of metaphors and themes (Campbell et al., 2011), create a grid or table of concepts (Britten & Pope, 2012; Erasmus, 2014; Malpass et al., 2009), or code concepts in a software programme for the analysis of qualitative data such as QSR NVivo (Toye et al., 2014). Reviewers should state what they mean by the terminology they have used for the units of synthesis, for example, metaphor, concept, theme. Reporting criterion 10—Presenting characteristics of included studies Provide a detailed description in narrative and/or table or other diagrammatic format of included studies and their study characteristics (such as year of publication, population, number of participants, data collection, methodology, analysis, research questions, study funder) (Britten & Pope, 2012; Toye et al., 2014). If publication requirements prevent full reporting, state where readers can access these data in full, for example, a project website, online files. In addition, provide key contextual information about the primary studies and comment on their relevance to the context(s) specified in the meta‐ethnography review question (Atkins et al., 2008; Thorne, Jensen, Kearney, Noblit, & Sandelowski, 2004; Toye et al., 2013). Context of included primary studies can influence the analysis process (Atkins et al., 2008), for example, primary study accounts published after a certain date may reflect a change in health policy/practice such as the introduction of a smoking ban in enclosed public places. If two or more included primary study accounts, for example, papers, were derived from the same primary study, this should be made explicit. Contextual information should include details about the primary study participants (such as their gender, age, socioeconomic status, ethnicity, and so on); the setting such as a geographical setting (a country, region, city) or organisation (hospital, school, company, community); and key political, historical, and cultural factors of relevance, for instance, the introduction of a major international guideline, which affected clinical care, preceded publication of included studies. If such contextual information is not available in the primary study accounts, reviewers should make this clear to readers (Table 1). Table 1 The eMERGe meta‐ethnography reporting guidance No. Criteria Headings Reporting Criteria Phase 1—Selecting meta‐ethnography and getting started Introduction 1 Rationale and context for the meta‐ethnography Describe the gap in research or knowledge to be filled by the meta‐ethnography, and the wider context of the meta‐ethnography 2 Aim(s) of the meta‐ethnography Describe the meta‐ethnography aim(s) 3 Focus of the meta‐ethnography Describe the meta‐ethnography review question(s) (or objectives) 4 Rationale for using meta‐ethnography Explain why meta‐ethnography was considered the most appropriate qualitative synthesis methodology Phase 2—Deciding what is relevant Methods 5 Search strategy Describe the rationale for the literature search strategy 6 Search processes Describe how the literature searching was carried out and by whom 7 Selecting primary studies Describe the process of study screening and selection, and who was involved Findings 8 Outcome of study selection Describe the results of study searches and screening Phase 3—Reading included studies Methods 9 Reading and data extraction approach Describe the reading and data extraction method and processes Findings 10 Presenting characteristics of included studies Describe characteristics of the included studies Phase 4—Determining how studies are related Methods 11 Process for determining how studies are related Describe the methods and processes for determining how the included studies are related: ☐ Which aspects of studies were compared AND ☐ How the studies were compared Findings 12 Outcome of relating studies Describe how studies relate to each other Phase 5—Translating studies into one another Methods 13 Process of translating studies Describe the methods of translation: ☐ Describe steps taken to preserve the context and meaning of the relationships between concepts within and across studies ☐ Describe how the reciprocal and refutational translations were conducted ☐ Describe how potential alternative interpretations or explanations were considered in the translations Findings 14 Outcome of translation Describe the interpretive findings of the translation. Phase 6—Synthesizing translations Methods 15 Synthesis process Describe the methods used to develop overarching concepts (“synthesised translations”) Describe how potential alternative interpretations or explanations were considered in the synthesis Findings 16 Outcome of synthesis process Describe the new theory, conceptual framework, model, configuration, or interpretation of data developed from the synthesis Phase 7—Expressing the synthesis Discussion 17 Summary of findings Summarize the main interpretive findings of the translation and synthesis and compare them to existing literature 18 Strengths, limitations, and reflexivity Reflect on and describe the strengths and limitations of the synthesis: ☐ Methodological aspects—for example, describe how the synthesis findings were influenced by the nature of the included studies and how the meta‐ethnography was conducted. ☐ Reflexivity—for example, the impact of the research team on the synthesis findings 19 Recommendations and conclusions Describe the implications of the synthesis John Wiley & Sons, Ltd 3.2.4 PHASE 4—Determining how studies are related Reporting criterion 11—Process for determining how studies are related Reviewers should describe which aspects of the primary studies were compared and why, to determine how they are related, bearing in mind the aim of their meta‐ethnography. Aspects could include: (i) research design, such as the: study aims; contexts; type of studies; theoretical approach/paradigm; participant characteristics, for example, their gender, ethnicity, culture, or age; study focus, for example, a health or social issue, long‐term conditions, other diseases, or care settings; (ii) findings—the meaning of the concepts, metaphors, and/or themes (Noblit & Hare, 1988); the overarching storyline or explanation of a phenomenon from the primary study accounts (Noblit, 2016) and (iii) other contextual factors, such as the time period, for instance, whether findings of primary study accounts differed because they were conducted in different time contexts. In addition, reviewers should describe how the studies were compared, that is, the methods and process of comparison. There is a wide variety of methods for comparing studies; examples of how Phase 4 has been reported include: Campbell et al. (2003); Atkins et al. (2008); Malpass et al. (2009); Beck (2009); Britten and Pope (2012); Erasmus (2014). Reporting criterion 12—Outcome of relating studies Describe how primary studies relate: (i) to each other; (ii) to the review question; and (iii) to the prespecified aspects of context which were considered important, for example, do they relate reciprocally and/or refutationally, or do they explore different aspects of the topic under study (Atkins et al., 2008; Beck, 2009; Britten & Pope, 2012; Campbell et al., 2011; Erasmus, 2014; France et al., 2014; Malpass et al., 2009; Noblit & Hare, 1988)? When reviewers are reporting how studies are related they should also report “disconfirming cases” (Booth et al., 2013; Thorne et al., 2004) that is, where one or more findings (e.g., metaphors or concepts) from a study differ from those of other studies for reasons that may be explained by differences in participants, settings, or study design. Reviewers can describe how studies were related in narrative, tabular, and/or diagrammatic form. 3.2.5 PHASE 5—Translating studies into one another Reporting criterion 13—Process of translating studies There is a variety of ways to conduct translation; therefore, reviewers should state their understanding and working definitions of reciprocal and refutational translation. Examples of approaches to translation identified by our systematic review are: Atkins et al. (2008), Campbell et al. (2011), Garside (2008), Toye et al. (2014), and Doyle (2003). Examples of refutational translation include Garside (2008) and Wikberg and Bondas (2010). Reviewers should also: state who was involved in translation; describe how meaning was translated from one study into another, for instance, by reporting one or more examples of how this was done; describe how relationships between concepts within and across studies, were preserved in the translation, such as by drawing concept maps to show relationships between concepts (Kinn, Holgersen, Ekeland, & Davidson, 2013; Malpass et al., 2009) (grids, tables, and other visual diagrams could also be used); describe how the contexts of the primary studies were preserved in the process of translation, for example, were subgroups of studies translated according to a common health condition or time‐period (Campbell et al., 2011)? clearly indicate whose interpretation is being presented (France et al., 2014)—that of the research participants, study authors, or reviewers (sometimes called first‐, second‐, and third‐order constructs, respectively) (Britten et al., 2002); describe how potential alternative interpretations or explanations were considered in the translation. Refutational translation is often overlooked (Booth et al., 2013; Thorne et al., 2004); its purpose is to explain differences and to explore and explain exceptions, incongruities, and inconsistencies (Barnett‐Page & Thomas, 2009; Booth, 2013). An entire study could refute another study (Bondas & Hall, 2007; Britten & Pope, 2012) or concepts/metaphors within studies could refute one another (Bondas & Hall, 2007; Britten & Pope, 2012; Finfgeld‐Connett, 2014), in which case it may be possible to do both reciprocal and refutational translation in a meta‐ethnography rather than one or the other. Reviewers should identify disconfirming cases that could inform or have an impact on translation and, subsequently, synthesis. Some argue that synthesizing a large number of studies might result in a superficial synthesis that loses its “groundedness” in the studies (Campbell et al., 2011); too few studies might result in underdeveloped theory/concepts (Finfgeld‐Connett, 2014; Toye et al., 2014). There is no consensus over what constitutes too few or too many studies; perceptions of a “large” number of studies varies from over 40 (Campbell et al., 2011) to over 100 (Thorne et al., 2004). The volume of data will also depend on the richness and length of those accounts and team size will affect the ability to manage the data. If a large volume of data were synthesized, reviewers should explicitly describe how translation was achieved given this volume, for example, did they translate studies in smaller clusters to preserve conceptual richness and/or stay grounded in the data? Reporting criterion 14—Outcome of translation Describe the interpretive findings of the reciprocal translation and refutational translation—including how each primary study contributed to the translation (Booth, 2013) and describe alternative interpretations/explanations. Clearly document from which concepts in primary studies, the reviewers’ concepts are derived (Booth, 2013). Reviewers need to differentiate between concepts derived from the participants of primary study accounts (sometimes called first order constructs) and those derived by the authors of the primary study accounts (sometimes called second‐order constructs). An example of how this has been reported is Britten et al. (2002) and a clear table describing the different levels of constructs can be found in Malpass et al. (2009). Descriptions of the study concepts and reviewers’ concepts and their interrelationships can be provided in table, diagrammatic or narrative form, with additional information in supplementary files. When quotes are used, reviewers should state their origin—primary study participants, primary study authors, or the reviewers’ own analysis notes. If any study was reported in more than one paper/account, describe how this was dealt with. 3.2.6 PHASE 6—Synthesizing translations Reporting criterion 15—Synthesis process There are two aspects of Phase 6: synthesizing translations and line of argument synthesis. The synthesized translations (concepts) represent the reviewers’ interpretation of the translations and are referred to in Britten et al. (2002) as third‐order constructs. A line of argument synthesis aims to provide a fresh interpretation; it goes further than translation and puts any similarities and dissimilarities into a new interpretive context (Noblit & Hare, 1988). George Noblit (2016) has more recently further defined a line of argument as the new “storyline” or overarching explanation of a phenomenon. Reviewers should describe the methods used to develop synthesized translations and how the line of argument synthesis was conducted. If line of argument synthesis was not conducted, state why not. In addition, describe: how many and which studies were synthesized. Sometimes studies are excluded in Phases 5 and 6 (for instance, because they lack conceptual depth), so the number of synthesized studies may differ from the number of studies meeting review inclusion criteria. who was involved in the synthesis and explain how synthesis findings have been considered from alternative perspectives (for example, from different academic disciplines) (Atkins et al., 2008; Bondas & Hall, 2007; Garside, 2008). how reviewers remained grounded with primary study data and avoided losing conceptual richness during synthesis, particularly if a large amount of data were synthesized. (See the discussion on volume of data to be synthesized in Phase 5). Reporting criterion 16—Outcome of synthesis process Describe the interpretive findings of the synthesis of translations, the line of argument synthesis and any new model, conceptual framework or theory developed in a narrative, grid, table and/or visually, for instance, as an illustration, diagram or film. Any of these may be considered to be a synthesis product and a single synthesis may have more than one product. Reviewers should show the inter‐relationships between the data from the primary studies and the reviewers’ new interpretations. If development of a new theory, conceptual framework, or model was not possible, state why not. Describe the context where the new theory, model, or framework applies, or not, based on the characteristics of included primary studies. For example, the new theory may have been based solely on studies of young, white women, or studies conducted in countries with private health care, or the included studies may be older and/or predate a significant development in the field. 3.2.7 PHASE 7—Expressing the synthesis Reporting Criterion 17—Summary of findings Relate the main interpretive findings to the synthesis objective(s), review question(s), focus, and intended audience(s) (Atkins et al., 2008; Bearman & Dawson, 2013; Bondas & Hall, 2007; Campbell et al., 2011; Noblit & Hare, 1988). Compare the concept, model, or theory generated in the synthesis to the existing literature, such as research and policy publications. Reviewers should consider the possible influence of findings from other authors (both from primary study accounts and the wider literature) on their own conclusions (Booth et al., 2013). Reporting criterion 18—Strengths, Limitations, and Reflexivity Consideration of methodological and other strengths and limitations and how they may influence the final interpretation is a key to meta‐ethnography reporting. Reviewers should reflect on and describe the effect of these on the synthesis process and outcomes because they may affect the credibility and trustworthiness (in other fields, this is referred to as validity and reliability) of the synthesis findings. Strengths and limitations of: (i) the included primary studies; and (ii) how the meta‐ethnography was conducted should be described. The latter are infrequently reported in published meta‐ethnographies. Reviewers should comment on how these aspects may have influenced or limited the synthesis findings: the characteristics, content and context of the primary studies, such as the temporal context, type of participant, cultural factors, study design. the conduct of the synthesis. Considerations include, but are not restricted to: the order in which studies were synthesized (France et al., 2014; Garside, 2008), the impact of study selection and sampling, the number of included studies/ volume of data (may affect depth of analysis), the context of the synthesis, and any modifications made to Noblit and Hare's (1988) original methodology. Reflexivity—critically reflecting on the context of knowledge construction, especially the effect of the researcher on the research process—should include comment on how the reviewers influenced the interpretive process and synthesis findings (Walsh & Downe, 2005), for example: the reviewers’ background, perspectives, and experience, such as, but not limited to, epistemological position(s), professional position(s) held, academic discipline, organisation(s), or professional bodies represented (Thorne et al., 2004); if the reviewers have a specific view, stance, or personal interest, for example, the reviewer's viewpoint on access to abortion care for a review about women's reproductive health care services. any influence of the funder of the meta‐ethnography; any conflicts of interests of the reviewers, that is, any factor, for example, financial, political, or organizational, which might influence the judgement of the reviewers when conducting the interpretation and synthesis. how each reviewer was involved and how their contribution to literature searching and screening, reading of studies, data extraction, translation, and synthesis may have influenced the interpretive process (Atkins et al., 2008; Bondas & Hall, 2007; Garside, 2008; Toye et al., 2014). Reporting criterion 19—Recommendations and conclusions Describe the implications of the synthesis findings for policy, practice, and/or theory. Policy and practice implicet al.ations were particularly important to eMERGe nonacademic and lay project advisors. Identify any areas where further primary or secondary research is needed. 3.3 Part 3: Extensions The first three extensions for reporting steps and processes that are not common to every meta‐ethnography are available as supplementary material to this paper. 4 DISCUSSION The eMERGe guidance is intended to increase transparency and completeness of reporting, making it easier for diverse stakeholders to judge the trustworthiness and credibility of meta‐ethnographies and also intended to make the findings more usable and useful to inform services and interventions, such as in health, social care, and education. The development of this guidance used methods following, but exceeding, good practice in developing reporting guidance (Moher et al., 2010) incorporating systematic literature reviews; consensus methods; and consultation with one of the two originators of meta‐ethnography, George Noblit. The team believe that the guidance is unusual among current reporting guidance in the extent to which it has involved lay people in all aspects of the development (France et al., 2015). This guidance is not intended as a detailed guide in how to conduct a meta‐ethnography—some such publications exist (e.g., Atkins et al., 2008; Britten & Pope, 2012; Campbell et al., 2011; France et al., 2016; Malpass et al., 2009) and others from the eMERGe project are in preparation (see http://emergeproject.org/publications/). The guidance is designed to raise the reporting quality of meta‐ethnographies and thus to assist those writing, reviewing, updating, and using meta‐ethnographies in making judgements about quality of meta‐ethnography conduct and output. It might also help users of qualitative evidence syntheses to recognize other forms of qualitative evidence synthesis mislabelled as a meta‐ethnography, a common occurrence (France et al., 2014). The guidance does, however, advance the methodology through its comprehensive analysis, interpretation and synthesis of methodological publications on meta‐ethnography, published since Noblit and Hare's original monograph, which underpin the reporting criteria and explanatory notes. Some might argue that the guidance is overly prescriptive and detracts from the original purposes of meta‐ethnography and, indeed, qualitative research. It is our view and that of others (Thorne, 2017) who conducting a meta‐ethnography involves creative, interpretive, qualitative analysis methods; however, a creative and interpretive approach should not preclude describing clearly how the research was conducted and some guidance is required to avoid misuse or mislabelling of the methods (Thorne, 2015) and poor or misleading reporting. In this guidance, definitions and requirements have not been imposed arbitrarily, unnecessarily, or where consensus is lacking. Meta‐ethnography has been described as an advanced qualitative research methodology (Campbell et al., 2011; Finlayson & Dixon, 2008; Toye et al., 2014) probably reflecting its complexity as a methodology. Training materials to accompany this guidance including video clips and slides (available from http://emergeproject.org/resources) have been developed as part of the eMERGe project. This guidance has been designed to have the flexibility to be applied to diverse reporting formats with differing publication requirements (for example, journal articles, reports, book chapters) and this explains why some standards, which apply only to certain formats, are included as “extensions” to the guidance. Publication requirements can limit manuscript length; therefore, reviewers might need to provide some data in an alternative format, such as online, to achieve full reporting. Methodological developments in meta‐ethnography and in relevant qualitative evidence synthesis methodology generally will continue to occur. This guidance was created with an eye to accommodating these future developments which will be monitored through our discussion list: www.jiscmail.ac.uk/META-ETHNOGRAPHY. Future research will investigate the impact of the eMERGe reporting guidance, for example, by updating our earlier systematic review of meta‐ethnography reporting practices (France et al., 2014), with a view to updating the guidance and we regard this guidance as one baseline from which to track the evolution of meta‐ethnography. 5 CONCLUSION This guidance has been developed following a rigorous approach in line with and exceeding good practice in creating reporting guidance. It is intended to improve the clarity and completeness of reporting of meta‐ethnographies to facilitate use of their findings to inform the design and delivery of services and interventions in health, social care, and other fields. Qualitative data are essential for conveying people's (e.g., patients, carers, clinicians) experiences and understanding social processes and it is important that they contribute to the evidence base. Meta‐ethnography is an evolving qualitative evidence synthesis methodology with huge potential to contribute evidence for policy and practice. In future, changes to the guidance might be required to encompass methodological advances and accommodate changes identified after evaluation of the impact of the guidance. CONFLICT OF INTEREST Catherine Pope is an author of the book Pope C, Mays N, Popay J. Synthesizing qualitative and quantitative health evidence: a guide to methods. Buckingham: Open University Press 2007 which discusses meta‐ethnography; she receives royalties from this. Jane Noyes is a Journal of Advanced Nursing Editor. She was recused from the Journal of Advanced Nursing management of this paper. No conflict of interest has been declared by the remaining author(s). AUTHORS CONTRIBUTIONS All authors have agreed on the final version and meet at least one of the following criteria [recommended by the ICMJE ( http://www.icmje.org/recommendations/)]: substantial contributions to conception and design, acquisition of data, or analysis and interpretation of data; drafting the article or revising it critically for important intellectual content. Supporting information Additional supporting information may be found online at https://osf.io/nmf7v/.   Click here for additional data file.   Click here for additional data file.   Click here for additional data file.   Click here for additional data file. ACKNOWLEDGEMENTS We warmly thank the members of our project advisory group: Geoff Allan, Marjory Barton, Ian Gallagher, Anne Ward, Lorna Neill, Gordon Snedden, Veronica Saunders (lay members), Andrew Booth, Nicky Britten, Rona Campbell, Margaret Cargo, Kate Flemming, Ruth Garside, Claire Glenton, Karin Hannes, Angela Harden, Gina Higginbottom, Christine Johnstone, Simon Lewin, George W. Noblit, Sandy Oliver, Cathy Pope, Karen Ritchie, James Thomas, Meredith Vanstone, and Geoff Wong (academic, other experts, and professional end‐users); and Sheena Blair, independent chair of the advisory group; Kevin Swingler, Steve Boulton, the participants in the Delphi, and our funder the National Institute of Health Research (NIHR). The systematic reviews were undertaken with the support of DECIPHer, a UKCRC Public Health Research: Centre of Excellence. Their funders (the British Heart Foundation, Cancer Research UK, Economic and Social Research Council RES‐590‐28‐0005, Medical Research Council, the Welsh Government and the Wellcome Trust (WT087640MA), under the auspices of the UK Clinical Research Collaboration) are gratefully acknowledged. ==== Refs REFERENCES Atkins, S. , Lewin, S. , Smith, H. , Engel, M. , Fretheim, A. , & Volmink, J. (2008). Conducting a meta‐ethnography of qualitative literature: Lessons learnt. BMC Medical Research Methodology, 8 , 21.18416812 Barnett‐Page, E. , & Thomas, J. (2009). Methods for the synthesis of qualitative research: A critical review. BMC Medical Research Methodology, 9 (1 ), 59.19671152 Bearman, M. , & Dawson, P. (2013). Qualitative synthesis and systematic review in health professions education. Medical Education, 47 (3 ), 252–260.23398011 Beck, C. T. (2009). Metasynthesis: A goldmine for evidence‐based practice. AORN Journal, 90 (5 ), 701‐2–705‐10.19895928 Bondas, T. , & Hall, E. O. (2007). Challenges in approaching metasynthesis research. Qualitative Health Research, 17 (1 ), 113–121.17170249 Booth, A. (2006). “Brimful of STARLITE”: Toward standards for reporting literature searches. J Med Libr Assoc, 94 (4 ), 421–429, e205.17082834 Booth, A. (2013). Acknowledging a dual heritage for qualitative evidence synthesis: Harnessing the qualitative research and systematic review research traditions. Sheffield: PhD University of Sheffield. Booth, A. , Carroll, C. , Ilott, I. , Low, L. L. , & Cooper, K. (2013). Desperately seeking dissonance: Identifying the disconfirming case in qualitative evidence synthesis. Qualitative Health Research, 23 (1 ), 126–141.23166156 Booth, A. , Noyes, J. , Flemming, K. , Gerhardus, A. , Wahlster, P. , Van Der Wilt, G. , … Rehfuess E. (2016). Guidance on choosing qualitative evidence synthesis methods for use in health technology assessments of complex interventions. INTEGRATE‐HTA. Britten, N. , Campbell, R. , Pope, C. , Donovan, J. , Morgan, M. , & Pill, R. (2002). Using meta ethnography to synthesise qualitative research: A worked example. Journal of Health Services & Research Policy, 7 (4 ), 209–215. Britten, N. , & Pope, C. (2012) Medicine taking for asthma: A worked example of meta‐ethnography (Chapter 3). ( K. Hannes & C Lockwood eds.) (pp. 41–58). Chichester: Wiley‐Blackwell BMJ Books. Campbell, R. , Pound, P. , Morgan, M. , Daker‐White, G. , Britten, N. , Pill, R. , … Donovan, J. (2011). Evaluating meta‐ethnography: Systematic analysis and synthesis of qualitative research. Health Technology Assessment, 15 (43 ), 1126–164. Campbell, R. , Pound, P. , Pope, C. , Britten, N. , Pill, R. , Morgan, M. , & Donovan, J. (2003). Evaluating meta‐ethnography: A synthesis of qualitative research on lay experiences of diabetes and diabetes care. Social Science and Medicine, 56 (4 ), 671–684.12560003 Carroll, C. (2017). Qualitative evidence synthesis to improve implementation of clinical guidelines. BMJ, 356 , j80.28093384 Cunningham, M. , France, E.F. , Ring, N. , Uny, I. , Duncan, E.A.S. , Roberts, R.J. , & Noyes, J. (2018) 13/114/60. Developing a reporting guideline to improve meta‐ethnography in health research: The eMERGe mixed‐methods study. Health Services and Delivery Research Journal (in press). Dixon‐Woods, M. , Booth, A. , & Sutton, A. J. (2007). Synthesizing qualitative research: A review of published reports. Qualitative Research, 7 (3 ), 375–422. Dixon‐Woods, M. , Cavers, D. , Agarwal, S. , Annandale, E. , Arthur, A. , Harvey, J. , … Sutton, A. J. (2006). Conducting a critical interpretive synthesis of the literature on access to healthcare by vulnerable groups. BMC Medical Research Methodology, 6 , 35.16872487 Doyle, L. H. (2003). Synthesis through meta‐ethnography: Paradoxes, enhancements and possibilities. Qualitative Research, 3 (3 ), 321–344. Erasmus, E. (2014). The use of street‐level bureaucracy theory in health policy analysis in low‐ and middle‐income countries: A meta‐ethnographic synthesis. Health Policy Plan, 29 (Suppl 3 ), iii70–iii78.25435538 Finfgeld‐Connett, D. (2014). Metasynthesis findings: Potential versus reality. Qualitative Health Research, 24 (11 ), 1581–1591.25192758 Finfgeld‐Connett, D. , & Johnson, E. D. (2013). Literature search strategies for conducting knowledge‐building and theory‐generating qualitative systematic reviews. Journal of Advanced Nursing, 69 (1 ), 194–204.22591030 Finlayson, K. W. , & Dixon, A. (2008). Qualitative meta‐synthesis: A guide for the novice. Nurse Researcher, 15 (2 ), 59–71. Flemming, K. , Booth, A. , Hannes, K. , Cargo, M. , & Noyes, J. (2018). Cochrane Qualitative and Implementation Methods Group guidance series – Paper 6: Reporting guidelines for qualitative, implementation and process evaluation evidence syntheses. Journal of Clinical Epidemiology, 97 , 79–85.29222060 France, E. F. , Ring, N. , Noyes, J. , Maxwell, M. , Jepson, R. , Duncan, E. , … Uny, I. (2015). Protocol‐developing meta‐ethnography reporting guidelines (eMERGe). BMC Medical Research Methodology, 15 (1 ), 103.26606922 France, E. F. , Ring, N. , Thomas, R. , Noyes, J. , Maxwell, M. , & Jepson, R. (2014). A methodological systematic review of what's wrong with meta‐ethnography reporting. BMC Medical Research Methodology, 14 (1 ), 119.25407140 France, E. F. , Wells, M. , Lang, H. , & Williams, B. (2016). Why, when and how to update a meta‐ethnography qualitative synthesis. Systematic Reviews, 5 (1 ), 44.26979748 Garside, R. (2008). A Comparison of methods for the systematic review of qualitative research: Two examples using meta‐ethnography and meta‐study. Exeter: PhD University of Exeter. Glenton, C. , & Lewin, S. (2016b) Using evidence from qualitative research to develop WHO guidelines (Chapter 15). In World Health Organization (Ed.), WHO Handbook for guideline development (2nd ed.). Geneva: WHO. Glenton, C. , Lewin, S. , & Gulmezoglu, A. M. (2016a). Expanding the evidence base for global recommendations on health systems: Strengths and challenges of the OptimizeMNH guidance process. Implementation Science, 11 (1 ), 98.27430879 Gulmezoglu, A. M. , Chandler, J. , Shepperd, S. , & Pantoja, T. (2013). Reviews of qualitative evidence: A new milestone for Cochrane. Cochrane Database Systematic Review, 11 (11 ), ED000073. Hannes, K. , & Macaitis, K. (2012). A move to more systematic and transparent approaches in qualitative evidence synthesis: Update on a review of published papers. Qualitative Research, 12 (4 ), 402–442. Kangasniemi, M. , Lansimies‐Antikainen, H. , Halkoaho, A. , & Pietila, A. M. (2012). Examination of the phases of metasynthesis: A study on patients’ duties as an example. Professioni Infermieristiche, 65 (1 ), 55–60.22463754 Kinn, L. G. , Holgersen, H. , Ekeland, T. J. , & Davidson, L. (2013). Metasynthesis and bricolage: an artistic exercise of creating a collage of meaning. Qualitative Health Research, 23 (9 ), 1285–1292.23964060 Lewin, S. , Glenton, C. , Munthe‐Kaas, H. , Carlsen, B. , Colvin, C. J. , Gulmezoglu, M. , … Rashidian, A. (2015). Using qualitative evidence in decision making for health and social interventions: An approach to assess confidence in findings from qualitative evidence syntheses (GRADE‐CERQual). PLoS Medicine, 12 (10 ), e1001895.26506244 Linstone, H. A. T. , & Turoff, M. (2002). The delphi method: Techniques and applications. Technometrics, 18 , 363. Malpass, A. , Shaw, A. , Sharp, D. , Walter, F. , Feder, G. , Ridd, M. , & Kessler, D. (2009). “Medication career” or “moral career”? The two sides of managing antidepressants: A meta‐ethnography of patients’ experience of antidepressants. Social Science and Medicine, 68 (1 ), 154–168.19013702 Moher, D. , Liberati, A. , Tetzlaff, J. , Altman, D. G. , & Group, P. (2009). Preferred reporting items for systematic reviews and meta‐analyses: The PRISMA statement. PLoS Medicine, 6 (7 ), e1000097.19621072 Moher, D. , Schulz, K. F. , Simera, I. , & Altman, D. G. (2010). Guidance for developers of health research reporting guidelines. PLoS Medicine, 7 (2 ), e1000217.20169112 Noblit, G.W. (2016) How qualitative (or interpretive or critical) is qualitative synthesis and what we can do about this? In A public lecture by Professor George W. Noblit, University of North Carolina at Chapel Hill Vol. 2018. Edinburgh. Retrieved from http://emergeproject.org/wp-content/uploads/2016/09/How-qualitative.pdf. Noblit, G. W. , & Hare, R. D. (1988). Meta‐ethnography: Synthesizing qualitative studies. California: Sage Publications. Noyes, J. , Booth, A. , Flemming, K. , Garside, R. , Harden, A. , Lewin, S. , … Thomas, J. (2018). Cochrane Qualitative and Implementation Methods Group guidance series‐paper 3: Methods for assessing methodological limitations, data extraction and synthesis and confidence in synthesized qualitative findings. Journal of Clinical Epidemiology, 97 , 49–58.29247700 Noyes, J. , & Lewin, S. (2011) Chapter 6: Supplemental guidance on selecting a method of qualitative evidence synthesis and integrating qualitative evidence with cochrane intervention reviews. In J. Noyes , A. Booth , K. Hannes , A. Harden , J. Harris , S. Lewin & C. Lockwood (Eds.), Supplementary guidance for inclusion of qualitative research in Cochrane systematic reviews of interventions. Version 1 (updated August 2011). Cochrane Collaboration Qualitative Methods Group. Available from: http://cqrmg.cochrane.org/supplemental‐handbook‐guidance Nunes, V. , Neilson, J. , O'Flynn, N. , Calvert, N. , Kuntze, S. , Smithson, H. , … Crome, P. (2009) Clinical guidelines and evidence review for medicines adherence: Involving patients in decisions about prescribed medicines and supporting adherence (Vol. CG76 ). London: National Collaborating Centre for Primary Care and Royal College of General Practitioners. Paterson, B.L. (2011) “It Looks Great but How do I know if it Fits?”: An Introduction to Meta‐Synthesis Research. In J. Barroso & M. Sandelowski (Eds.), Synthesizing qualitative research (pp. 1126–20). Hoboken, NJ: John Wiley & Sons, Ltd. Pearson, A. , Wiechula, R. , Court, A. , & Lockwood, C. (2005). The JBI model of evidence‐based healthcare. International Journal of Evidence‐Based Healthcare, 3 (8 ), 207–215.21631749 Plint, A. C. , Moher, D. , Morrison, D. , Schulz, K. , Altman, D.G. , Hill C. , & Gaboury I. (2006). Does the CONSORT checklist improve the quality of reports of randomised controlled trials? A systematic review. Medical Journal of Australia, 185 (5 ), 263.16948622 Pound, P. , Britten, N. , Morgan, M. , Yardley, L. , Pope, C. , Daker‐White, G. , & Campbell, R. (2005). Resisting medicines: A synthesis of qualitative studies of medicine taking. Social Science and Medicine, 61 (1 ), 133–155.15847968 Ring, N. , Jepson, R. , Hoskins, G. , Wilson, C. , Pinnock, H. , Sheikh, A. , & Wyke, S. (2011a). Understanding what helps or hinders asthma action plan use: A systematic review and synthesis of the qualitative literature. Patient Education and Counseling, 85 (2 ), e131–e143.21396793 Ring, N. , Jepson, R. , & Ritchie, K. (2011b). Methods of synthesizing qualitative research studies for health technology assessment. International Journal of Technology Assessment in Health Care, 27 (4 ), 384–390.22004781 Rycroft‐Malone, J. , & Burton, C.R. (2015) The synthesis of qualitative data (Chapter 8). In D. A. Richards & I. R. Hallberg (eds.), Complex interventions in health: An overview of research methods. Abingdon: Routledge. Suri, H. , & Clarke, D. (2009). Advancements in research synthesis methods: From a methodologically inclusive perspective. Review of Educational Research, 79 (1 ), 395–430. Thorne, S. E. (2015). Qualitative metasynthesis: A technical exercise or a source of new knowledge? Psycho‐Oncology, 24 (11 ), 1347–1348.26291310 Thorne, S. (2017). Metasynthetic madness: What kind of monster have we created? Qualitative Health Research, 27 (1 ), 3–7.27956657 Thorne, S. , Jensen, L. , Kearney, M. H. , Noblit, G. , & Sandelowski, M. (2004). Qualitative metasynthesis: Reflections on methodological orientation and ideological agenda. Qualitative Health Research, 14 (10 ), 1342–1365.15538004 Tong, A. , Flemming, K. , McInnes, E. , Oliver, S. , & Craig, J. (2012). Enhancing transparency in reporting the synthesis of qualitative research: ENTREQ. BMC Medical Research Methodology, 12 , 181.23185978 Toye, F. , Seers, K. , Allcock, N. , Briggs, M. , Carr, E. , Andrews, J. , & Barker, K. (2013). ‘Trying to pin down jelly’ – Exploring intuitive processes in quality assessment for meta‐ethnography. BMC Medical Research Methodology, 13 (1 ), 46.23517438 Toye, F. , Seers, K. , Allcock, N. , Briggs, M. , Carr, E. , & Barker, K. (2014). Meta‐ethnography 25 years on: Challenges and insights for synthesising a large number of qualitative studies. BMC Medical Research Methodology, 14 , 80.24951054 Turner, S. P. (1980). Sociological explanation as translation. Cambridge: Cambridge University Press. Uny, I. , France, E. F. , & Noblit, G. W. (2017). Steady and delayed: Explaining the different development of meta‐ethnography in health care and education. Ethnography and Education, 12 (2 ), 243–257. Walsh, D. , & Downe, S. (2005). Meta‐synthesis method for qualitative research: A literature review. Journal of Advanced Nursing, 50 (2 ), 204–211.15788085 Wikberg, A. , & Bondas, T. (2010). A patient perspective in research on intercultural caring in maternity care: A meta‐ethnography. International Journal of Qualitative Studies on Health and Well‐being, 5 (1 ). Wong, G. , Greenhalgh, T. , Westhorp, G. , Buckingham, J. , & Pawson, R. (2013). RAMESES publication standards: Meta‐narrative reviews. BMC Medicine, 11 (1 ), 20.23360661

PMC007xxxxxx/PMC7613183.txt

==== Front ACS Nano ACS Nano nn ancac3 ACS Nano 1936-0851 1936-086X American Chemical Society 35785960 10.1021/acsnano.2c05391 Article Single-File Translocation Dynamics of SDS-Denatured, Whole Proteins through Sub-5 nm Solid-State Nanopores Soni Neeraj †‡ Freundlich Noam † Ohayon Shilo † Huttner Diana † https://orcid.org/0000-0001-7082-0985 Meller Amit *†‡# † Department of Biomedical Engineering, Technion−IIT, Haifa, 3200003 Israel ‡ Russell Berrie Nanotechnology Institute Technion−IIT, Haifa, 3200003 Israel * Email: ameller@technion.ac.il. 03 07 2022 26 07 2022 03 07 2023 16 7 1140511414 © 2022 The Authors. Published by American Chemical Society 2022 The Authors https://creativecommons.org/licenses/by-nc-nd/4.0/ Permits non-commercial access and re-use, provided that author attribution and integrity are maintained; but does not permit creation of adaptations or other derivative works (https://creativecommons.org/licenses/by-nc-nd/4.0/). The ability to routinely identify and quantify the complete proteome from single cells will greatly advance medicine and basic biology research. To meet this challenge of single-cell proteomics, single-molecule technologies are being developed and improved. Most approaches, to date, rely on the analysis of polypeptides, resulting from digested proteins, either in solution or immobilized on a surface. Nanopore biosensing is an emerging single-molecule technique that circumvents surface immobilization and is optimally suited for the analysis of long biopolymers, as has already been shown for DNA sequencing. However, proteins, unlike DNA molecules, are not uniformly charged and harbor complex tertiary structures. Consequently, the ability of nanopores to analyze unfolded full-length proteins has remained elusive. Here, we evaluate the use of heat denaturation and the anionic surfactant sodium dodecyl sulfate (SDS) to facilitate electrokinetic nanopore sensing of unfolded proteins. Specifically, we characterize the voltage dependence translocation dynamics of a wide molecular weight range of proteins (from 14 to 130 kDa) through sub-5 nm solid-state nanopores, using a SDS concentration below the critical micelle concentration. Our results suggest that proteins’ translocation dynamics are significantly slower than expected, presumably due to the smaller nanopore diameters used in our study and the role of the electroosmotic force opposing the translocation direction. This allows us to distinguish among the proteins of different molecular weights based on their dwell time and electrical charge deficit. Given the simplicity of the protein denaturation assay and circumvention of the tailor-made necessities for sensing protein of different folded sizes, shapes, and charges, this approach can facilitate the development of a whole proteome identification technique. solid-state nanopores single-molecule sensing protein translocation electroosmotic force electrical charge deficit SDS−protein complex voltage-driven translocation H2020 European Research Council 10.13039/100010663 833399 Azrieli Foundation 10.13039/501100005155 NA H2020 European Research Council 10.13039/100010663 966824 document-id-old-9nn2c05391 document-id-new-14nn2c05391 ccc-price ==== Body pmcIntroduction The fast-growing demand for on-site and rapid clinical diagnostics, accelerated by the SARS-Cov-2 pandemic, has created a rising need for alternative biomolecular sensing technologies, particularly those that offer single-molecule sensing resolution. Among the various methods, nanopore sensors have carved a unique niche due to their relatively straightforward operation principle, their versatility in terms of the range of detectable analytes, and their potential integration in fully portable low-power devices.1,2 Importantly, in recent years, researchers have been able to shift nanopore research from basic proof of concept demonstration using laboratory-made synthetic analytes toward clinical applications, which often involve complex-to-analyze biofluids. The versatility of nanopore sensors has already been illustrated in multiple applications, such as sensing of DNA cancer mutations or mRNA cancer metastasis biomarkers in clinical samples, RNA of viral infections, epigenetic modifications, and native proteins, to name a few.3−11 In parallel, the development of single-molecule proteomics technologies has gained significant momentum, partly owing to emerging sensing strategies.12 Also here, nanopore sensors have been proposed for protein sequencing or protein identification, using similar principles underlying the successful nanopore DNA sequencer.13−20 Whole proteome analyses introduce significant challenges, which are considerably more complex than genome sequencing. Unlike DNA, proteins cannot be amplified to facilitate their analysis, and human proteins harbor 20 different amino acids as opposed to the DNA’s four canonical nucleotides. Similar to mass spectrometry, some of the nanopore approaches focus on analyzing short peptides produced from sheared or digested proteins,21−23 whereas other approaches are centered on the analysis of full-length proteins.24 An advantage of the latter approach is that each protein is counted once, and the full information regarding the protein identity is maintained. Hence, downstream data analysis is potentially simplified. To date, an antibody-free generic method for full-length protein identification with a single-molecule resolution has remained elusive. The biochemical diversity and complexity of biological proteomes in terms of proteins’ structures, charge, mass, and hydrophobic nature are overwhelming. One way to tackle this challenge is to employ a denaturation protocol, which resolves the proteins’ higher-order structures while stabilizing a random coil state. The unfolded protein structure lends itself to molecule-by-molecule analysis using a single-size nanopore sensor for all the proteins regardless of their length, provided that the unfolded protein can be linearly threaded through the pore.25−29 Particularly, ionic surfactants, such as sodium dodecyl sulfate (SDS), have been extensively used in bulk for stabilization of the proteins’ denatured state. Furthermore, SDS facilitates uniform electromigration of denatured proteins, since the uniform negative charge of the SDS molecules adsorbed to the proteins supersedes by a large margin the proteins’ native net charge.30−32 Recently, it has been shown that the uniformly distributed SDS charge facilitates electrophoretic-based capture and translocation of the proteins through ∼10 nm solid-state nanopores.33 However, the mechanism governing their translocation dynamics, particularly through small nanopores, which can potentially extend the translocation dwell-times, has remained to date unclear. Past studies of dsDNA molecules’ translocations suggested that the DNAs’ dwell-time in a solid-state nanopore is strongly dependent on the nanopore diameter.34 In particular, about an order of magnitude increase in dwell-time was observed when the pore size was reduced from roughly 5 nm to 3 nm (the mean dsDNA cross-section is 2.2 nm). Lengthening the translocation dwell-time is advantageous, as it extends the dynamic range of nanopore analysis, permitting the analysis of smaller molecular weight proteins. Here we hypothesized that substantial slowing down of the protein samples may be achieved using nanopore diameters of roughly 3–5 nm, comparable with the expected cross-section of the SDS-denatured protein complexes (roughly 1.5–2.5 nm).35 This allows us to effectively analyze the translocation dynamics of several proteins of different molecular weights starting from about 14 kDa up to 130 kDa. These proteins were analyzed under different voltages, SDS concentrations, and nanopore sizes, affording insights into the SDS-denatured protein translocation dynamics and the underlying forces governing this process. Interestingly, even when the SDS concentration is dialed below the critical micelle concentration (CMC), we observe a significant slowing down of the SDS-denatured protein complexes translocating through small nanopores. This is attributed to the interplay of the electroosmotic force and the electrophoretic force. According to our results, smaller proteins are clearly distinguishable from larger proteins by their dwell-times and their ion-current amplitudes. Moreover, the multiplications of the events’ dwell-times by their corresponding current amplitudes (the so-called electrical charge deficits, or ECDs) provide a nanopore-specific analogous measurement of the proteins’ molecular weights, hence permitting wide-range discrimination among proteins. Histograms of the ECD values for each protein display two distinct peaks, suggesting the translocations of two different populations: a complete SDS–unfolded protein complex and a partially structured SDS–protein complex. Results and Discussion 1 Nanopore Discrimination between SDS Micelles and Proteins Translocations Due to Debye shielding at high solution ionic strengths, SDS molecules spontaneously form spherical micelles at a significantly lower threshold concentration (the CMC) than in pure water. Specifically, at a near-physiological concentration of 200 mM NaCl, the CMC drops from roughly 6 mM (in pure water) to <1 mM (0.03% w/w).36 High ionic strength solutions are necessary for nanopore sensing assays to support ion-current measurements; hence it was necessary to characterize the translocation pattern of SDS micelles versus the translocation of SDS-denatured proteins. To this end, we performed a set of measurements using a standard “nanopore buffer” (0.4 M NaCl in 1× PBS, pH 7.4) using nanopore sizes in the range of 4–5 nm. These pores were characterized by their conductance (G = iO/V where iO is the measured open-pore current and V is the applied voltage, typically 300 mV). Here we varied the SDS concentration from 700 μM down to 175 μM, while keeping the protein concentration of bovine serum albumin protein (BSA, Mw = 66.4 kDa, and the number of amino acids Naa = 583) at 1–1.5 nM. On the basis of the literature we expected that the actual CMC at 0.4 M NaCl would be <1 mM SDS, but the exact CMC value and the optimal SDS concentration for the nanopore experiments must be determined empirically. Starting from a stable and unperturbed nanopore open pore current (3.6 nA at V = 300 mV, G = 12 nS), upon addition of 700 μM SDS, we observed a fast and uniform ion-blockade pattern with an event amplitude of ΔI = 0.67 nA and FWHM of 0.03 nA (Figure 1a, left panel). We attributed these events to the SDS micelles’ translocations, as the individual SDS molecules are too small to be sensed in the nanopores used. Given that SDS micelles are small (Mw < 18 kDa) and highly charged (NSDS ≈ 60) spherical particles, we expected to observe an extremely fast translocation dwell-time. Our data, however, suggest that the micelles’ event dwell-times are peaked between 100 and 150 μs (Figure 1b-I), comparable to a dsDNA molecule of 300 bp, which is roughly an order of magnitude larger in size. We rationalized that the main factor contributing to the seemingly long translocation time of the SDS micelles is the increased level of electroosmotic counter flow, due to the partial coating of all surfaces, including the nanopore inner walls, by the negatively charged SDS molecules.37 Consistent with previous results, a small fraction of events (<0.5%) appears at even longer time scales (a few ms), perhaps due to the random insertion of nonspherical micelles or micelles that strongly interact with the SiN in the nanopore lumen.33 Figure 1 Characterization of SDS micelles and SDS-denatured BSA proteins translocations using sub-5 nm ssNPs. (a) Ion current traces at different protein and SDS concentrations show two distinct events amplitudes corresponding to SDS micelles and BSA proteins (BSA concentration 1–1.5 nM). (b) Event diagrams, shown as heatmaps, of the event amplitude (ΔI) vs dwell-time (log tD) for different SDS and protein concentrations: (I) SDS only (700 μM), (II) SDS + BSA (350 μM/1 nM), (III) SDS + BSA (175 μM/1.5 nM). (c) Corresponding histograms of event amplitudes show two distinct peaks for SDS micelles (0.5–0.7 nA) and BSA proteins (1.1–1.3 nA). Next, we used a similar pore size (G = 12 nS) to test a sample containing 700 μM SDS and 1 nM BSA protein concentration (Figure 1a). The BSA proteins were heat-denatured in the presence of a higher SDS concentration (see Methods) and diluted at room temperature (RT), 22.0 ± 0.5 °C, to the final SDS and protein concentrations before being introduced to the nanopore. As before, we primarily observe fast and uniform events with a mean amplitude of ∼0.5 nA, but unlike panel I, a second population of events with a larger mean amplitude of 0.86 nA emerged (Figure S1). Further decreasing the SDS concentration to 350 μm while keeping the BSA at 1.0 nM resulted in a clear two-peak pattern in the mean events’ amplitudes (Figure 1b-II and 1c, middle panel): (1) 0.65 ± 0.11 nA and (2) 1.3 ± 0.26 nA, as marked in the heat-map diagram. This time the fractions of the two populations were comparable. Finally, when further decreasing the SDS concentration to 175 μM, we observe primarily events at ΔI = 1.1 nA, whereas the low-amplitude population (ΔI = 0.45 nA) became the minor population of events (Figure 1b-III and 1c, right panel). The set of experiments described in Figure 1 suggests that at high SDS concentration (>350 μM) the micelles’ translocations dominate the overall population of the events. At lower SDS concentrations, and particularly below the NaCl salt-adjusted CMC, the vast majority of events could be attributed to protein/SDS complexes, which exhibit a distinguished events pattern based on their larger ion-current blocked amplitude. Notably, the protein/SDS complex translocations exhibit a broader dwell-time and blockade current amplitude distributions than that of the SDS micelles, as shown in the heat maps (Figure 1 panels b-II and b-III). 2 SDS-Denatured Translocation Patterns at Different Nanopore Sizes Suggest Primarily Single-File Dynamics and Multiple Protein Structures To shed more light on the translocation pattern of the protein/SDS complexes, we selected a smaller protein than BSA, carbonic anhydrase (CA, Mw = 28.9 kDa, Naa = 259), which was denatured in the presence of SDS using the same protocol. We tested the CA translocation dynamics using three different nanopore sizes at the low SDS concentration (175 μM). Starting from a G = 7 nS pore (∼3 nm), we observed long translocation events with a characteristic dwell-time of 650 ± 60 μs (Figure 2a left panel and Figure S3b). When a 12 nS pore (∼4 nm) was used, the characteristic mean dwell-time was reduced to 180 ± 40 μs (Figure 2a middle panel). Finally using a large pore of G = 21 nS (∼7 nm), we observed events with a mean dwell-time of ∼100 μs (Figure 2a right panel). Notably, in all three cases the events’ amplitudes remained invariant with a mean value of ΔI ≈ 0.9 nA and STD = 0.11 nA. This value is also consistent with the mean BSA event amplitude (1.1 and STD = 0.3 nA) shown in Figure 1. Figure 2 Translocation dynamics of SDS-denatured carbonic anhydrase (CA) proteins in three different nanopore sizes. (a) Translocation events diagrams shown as heat-maps for CA proteins using three nanopore sizes (left to right): ∼3 nm, G = 7 ns; ∼4 nm, G = 12 nS; and ∼7 nm, G = 21 nS. In each case, the event amplitudes (ΔI) are shown vs the event dwell-times (log tD). Typical events are displayed in insets, showing a clear shift toward longer events in the smaller nanopores, but similar event amplitudes. (b) Histograms of the electrical charge deficit (ECD) for the three experiments (shown in semilog scale). In all cases, the ECD histogram displays a prominent peak around 106e and secondary peaks at 108.3e, 107e, and 106.5e for the smaller and larger pores, respectively. Minor peaks at much lower ECD, possibly due to collisions, are also visible. Red curves are double Gaussian fits (see text). Drawing an analogy from the translocation dynamics of dsDNA molecules through ssNPs, we can roughly approximate the effective analyte’s cross-section using , where iO is the nanopore’s open-pore current and d is its effective diameter.34 Accordingly, for the three experiments displayed in Figure 2, we obtained a = 1.96 ± 0.22 nm, 2.01 ± 0.20 nm, and 2.83 ± 0.31 nm for the 3, 4, and 7 nm pores, respectively. While these are only crude approximations, they fall within the expected cross-sections of the SDS–polypeptide complexes between 1.5 and 2.5 nm.35 We, therefore, hypothesized that the main portion of the events represents single-file translocations of the unfolded proteins through the nanopore. Further observation of the events diagram heat-maps suggests that the smaller pores give rise to significantly longer protein mean translocation dwell-times. Particularly, the characteristic translocation dwell-time of the CA proteins is 650 μs (using the ∼3 nm ssNP), which is roughly an order of magnitude longer than previous studies using ∼10 nm ssNPs,33 considering similar molecular weight proteins. This finding may permit the detection of a broader range of proteins, but the effect is not as strong as the one observed for dsDNA translocation dynamics through ssNPs.34 Temperature-induced protein unfolding and the stabilization of the denatured state by surfactants, such as SDS, is a complex process that can lead to multiple structural states. In addition to a fully unfolded, random coil state, the protein may maintain some of the most stable secondary structural motifs, which may be resolved upon further application of force on the protein chain. To resolve multiple states of dsDNA using nanopore sensors, previous studies38 calculated the electrical charge deficit, defined as ECD = ΔI × tD/e where e is the elementary charge. For example, the ECD values of folded and unfolded forms of the same DNA were found to be strikingly different.38 Following this idea, we hypothesized that the ECD might be useful in better resolving multiple protein/SDS complexes in our measurements. Figure 2b presents the histograms of the ECD values for the three nanopore size experiments performed with CA/SDS complexes. In all three cases, two main peaks in the ECD histogram emerged: a main peak at around 106e to 106.5e and a secondary peak, which varied between ∼108.5e and ∼107e for the smaller and larger pores, respectively. Interestingly, the intermediate pore size (∼4 nm) displayed the largest fraction of secondary ECD events (∼25%), while smaller and larger pores showed smaller fractions of the secondary peak (∼10%). This observation is consistent with an interpretation in which the secondary ECD events correspond to partially folded complexes. 3 Alpha-lactalbumin Translocation Dynamics as a Function of Voltage Further Reveals the Interplay between Electrophoretic and Electroosmotic Forces The experiments shown in Figure 1 and Figure 2 already exposed some of the additional complexities associated with the translocation dynamics of SDS/protein complexes as compared to a simpler polyelectrolyte, such as dsDNA, and the potential role of electroosmotic flow (EOF). To further isolate the effect of surface charge and EOF, we used a smaller protein, alpha-lactalbumin (α-LA, 14 kDa, Naa = 123), which is readily denatured using heat/SDS treatment. We found that at V = 300 mV the characteristic translocation dwell-time of α-LA through a G = 12 nS pore was too short for detection. Therefore, to ensure that our results are minimally biased by the experimental temporal bandwidth, particularly at larger voltages, we performed this set of experiments in the presence of 10% glycerol (v/v), which is expected to increase the solution viscosity by a factor of 1.63 (at RT).39 Notably, a higher glycerol concentration than 10% could further increase the apparent dwell-times of the proteins in the nanopore. But higher viscosities proportionally reduce the effective protein capture rate by the nanopore, hence limiting its practical use with relatively low glycerol concentrations. In Figure S6, we show a series of SDS-denatured α-LA measurements performed using an 11 nS nanopore at different voltages (450–750 mV) in 10% glycerol (v/v). The data were analyzed as in the previous cases (Figures 1 and Figure 2): first, we extracted the events’ tD and ΔI, and then we calculated the ECD histograms for each data set. Next, we fitted the log of the ECD distributions with a double Gaussian function to obtain the mean values of the two populations. A summary of our results is presented in Figure 3. The left panel shows the ECD values for the main peak (solid black circles) and for the secondary peak (solid red squares) as a function of voltage. Interestingly, the main ECD peak appears to be independent of the voltage, with a mean value of 106.0±0.03 electron units. This result aligns well with the results shown in Figure 2, suggesting that the main ECD peak is also independent of nanopore size. In other words, the increase of ΔI with voltage is offset by a similar decrease in log(td) with voltage. Figure 3 Voltage dependence of translocation dynamics of SDS-denatured alpha-lactalbumin proteins. (A) Electrical charge deficit (ECD) vs voltage. Black circles represent the main peak, and red squares correspond to the secondary peak. The main ECD peak is nearly independent of voltage, whereas the secondary peak shows a mild decrease with voltage. (B) Dwell-time (td) vs voltage. Black circles correspond to events in the main peak, and red squares (inset) correspond to the secondary peak. The solid lines are decaying exponential fits. The experiments were performed in 10% glycerol (v/v), 0.4 M NaCl, 1× PBS, and 175 μM SDS. The right panel displays the events’ dwell-time dependence on the applied voltage for the translocation events of the primary and secondary ECD peaks (black circles and red squares, respectively). In both cases, we see a decrease in dwell-times as the voltage increases, reassuring evidence that these blockade events represent translocations of the proteins from the cis side (at which proteins are introduced) to the trans side of the membrane and not random collisions of the protein complexes with the pore, which would have resulted in a longer dwell-time with increasing voltage bias. Some representative events that correspond to the primary and secondary peaks are displayed in Figure S7. For both the primary and secondary groups, exponential functions can fit our data well, as expected for a voltage-driven translocation process through small nanopores.34 The typical decay constants for the SDS-denatured proteins obtained from the exponential fitting are 10.2 ± 1.0 V–1 and 9.2 ± 3.4 V–1 for the primary and secondary peaks, respectively. Interestingly, these values are about a factor of 5 smaller than the typical decay constants for dsDNA translocations through similar ssNPs.34 The much weaker voltage dependence of the protein/SDS complexes compared with the dsDNA experiments may be attributed to a smaller effective overall charge of the former, which manifests as a smaller electrophoretic driving force. An additional important factor is the possibility of the anionic SDS molecules coating the interior of the nanopore surfaces, resulting in a significant EOF that slows down the overall translocation of proteins. The magnitude of the EOF is expected to scale linearly with voltage, and combined with the apparent decrease in dwell-times with increasing voltage (Figure 3b), we concluded that electrophoretic forces remain the dominant contributors to the protein/SDS translocation dynamics. This interpretation is consistent with the relatively long translocation dwell-times that the SDS-denatured proteins exhibit in the nanopore as compared with ssDNA (or dsDNA) of a comparable chain length: for the 123 aa α-LA we measured a dwell-time of about 250 μs (at 450 mV), whereas a similar length dsDNA would translocate in less than 10 μs using a similar ssNP size and voltage. Additionally, a previous publication reporting folded state (no SDS) translocations of similar size proteins and nanopores reported much shorter dwell-times (∼50 μs) even at a lower bias voltage.40 This again supports the important contribution of EOF to slowing down the SDS-denatured protein molecules’ translocations. 4 Translocation Dwell-Times and ECDs Correlate with Proteins’ Molecular Weights Assuming that the majority of the proteins/SDS complex events represent single-file translocations, we evaluated the capability of solid-state nanopores to discriminate among proteins having different molecular weights. To that end, we accumulated data from six different proteins spanning molecular weights between 14 and roughly 130 kDa. Figure 4a displays an SDS-PAGE (polyacrylamide gel electrophoresis) analysis of the six proteins analyzed (alpha-lactalbumin, carbonic anhydrase, ovalbumin, bovine serum albumin, phosphorylase B, and SARS-Cov-2 spike protein, labeled lanes 1 to 6, respectively), along a protein molecular weight ladder (marked with “L”). For all proteins, we observed a single band, except OVA, which consistently displayed two bands, presumably due to two alternative glycosylation states, as suggested by the vendor (Sigma). The proteins’ molecular weight versus their measured in-gel migration distance showed an exponential dependency (Figure 4b). Figure 4 SDS-PAGE analysis of six proteins used in this study compared with nanopore measurements. (a) PAGE analysis, from right to left: alpha-lactalbumin, carbonic anhydrase, ovalbumin, BSA, phosphorylase B, and spike protein were denatured and separated on a 4–20% Tris-glycine SDS-PAGE. The gel was fixed, stained with Flamingo fluorescent dye overnight, and imaged by the Pharos scanner (Bio-Rad). L denotes lanes with protein ladders (see Methods). (b) Dependence of the SDS-PAGE migration distance of the six proteins on their molecular weights. The solid line is an exponential fit. (c) Dependence of SDS-denatured protein translocation dwell-time on the PAGE migration distance. The solid line is a linear fit. We analyzed the same proteins using the nanopore sensors. As before, we first calculated the ECD values for every event and then separated the events in each case according to their ECD main and secondary groups. We then plotted their characteristic dwell-time from the exponential fitting of the histograms. The characteristic mean dwell-times of the events in the primary ECD group showed a linear relationship with the in-gel bulk migration distance (Figure 4c). These results further support our interpretation of the primary peak events as single-file translocation of the unfolded proteins. Figure 5a displays the dependencies of the primary and secondary peaks’ mean dwell-times (black circles and red squares, respectively) on molecular weight. For the primary peak, we saw a roughly linear increase of tD with molecular weight, but for the secondary peak, we observed a nonlinear increase of the dwell-time with Mw tentatively associated with the larger propensity of the larger proteins to maintain some secondary structures. In Figure 5b we show the dependence of the ECD main and secondary peaks on the proteins’ Mw. For both event populations we could fit our data with growing exponential functions (ECD = AeMw/ρ), where ρ1,2 are nanopore-characteristic molecular weight constants, ρ1 = 59.1 ± 7.1 kDa, and ρ2 = 28.1 ± 2.2 kDa, describing the rate at which the ECD grows with Mw. Accordingly, for nanopore measurements of SDS-denatured proteins, the ECD may serve as a proxy for the separation of different molecular weights or the single-molecule analog of the migration distance in SDS PAGE analysis. Figure 5 Translocation dynamics of SDS-denatured proteins as a function of molecular weight (Mw). (a) Characteristic dwell-times of the six proteins in the nanopore. The black solid circles represent the main peak (a, b), while the secondary peak is represented by red squares (a, b). We observe a linear increase of td on Mw for the main peak (black straight line fit) and a nonlinear dependence for the second group (red line is a guide for the eye). (b) Summary plot of the ECD as a function of Mw. The main ECD peak event (solid circle) shows a linear dependency on Mw, while the secondary peak event (red squares) exhibits a quadratic dependency. Solid lines are fits. Conclusions The ability to thread and analyze full-length denatured proteins using nanopores is a significant step for single-molecule protein identification and quantification and may lead to a whole proteome analysis approach.12,24,41 A crucial step toward this goal is the elucidation of the voltage-induced translocation dynamics through nanopores comparable in size to the SDS/polypeptide cross-section of 1.5–2.5 nm. To this end in this study, we used sub-5 nm solid-state nanopores and analyzed the proteins’ translocation dynamics at an SDS concentration slightly below the CMC level. Unlike simple polyelectrolytes, such as double-stranded DNA molecules, SDS–protein complexes involve additional factors that must be taken into consideration. Native proteins are not uniformly charged and pose elaborate tertiary structures, which in many cases support their biological function. SDS heat denaturation is an effective way to unfold proteins and decorate them with charged groups. However, this involves some subtleties: first, under high ionic strengths, SDS spontaneously forms micelles at relatively low concentrations, which appear as short translocation events and may be misinterpreted as proteins (Figure 1). We showed here that this issue can be addressed if SDS concentration is kept right below its salt-adjusted CMC. Second, SDS molecules coat all surfaces, including the nanopore itself, adding a significant amount of negative surface charge to the pore interfaces. The addition of surfactants increases the nanopore stability, making it more hydrophilic. But at the same time under voltage bias, it produces a stronger electroosmotic flow as compared to uncharged (or less charged) nanopore interfaces (Figure 6b). The EOF is a nanopore size-dependent phenomenon and can be exploited for slowing down or trapping the protein in a thin membrane, allowing label-free analysis of the protein dynamics.12,42,43 As suggested in Figure 3, the effect of the EOF is to slow down the proteins’ translocations, as it is directed in the reverse direction to the electrophoretic force (EP). But the EOF may also reduce the event capture rate into the nanopore;44 hence future studies may need to develop means to control the EOF effect more precisely. Figure 6 Model for elucidation of voltage-induced translocation of SDS/protein complexes through small solid-state nanopores. (a) Electrical charge deficit (ECD) for phosphorylase B, where two peaks can be distinguished clearly. The representative events corresponding to the first peak (represented in black color) show no sublevels in the ionic current trace during translocations. However, the event with the sublevel ionic current traces (red-colored events) corresponds to subtle secondary or tertiary motifs that exist during the protein translocations. (c) Schematic illustration for the two different scenarios. (b) Effect of SDS coating on the pore walls enhancing the EOF, resulting in the longer translocation time in comparison to a similar size dsDNA molecule. As common in nanopore experiments, SDS-denatured proteins’ translocations dynamics were characterized using two main parameters, the events’ dwell-times and the events’ current amplitudes (tD and ΔI, respectively). We find that both parameters may exhibit multiple peaks, complicating a straightforward analysis of the results. However, their multiplication, the events’ electrical charge deficit, is a useful tool for the separation of the events into groups. Particularly, a striking feature that all our measurements suggest is the existence of a primary (where most of the events reside) and a secondary ECD group for all the proteins that were analyzed. Figure 6a shows a typical ECD histogram, measured for phosphorylase B proteins, using a 4 nm pore. The events associated with the primary ECD peak (black annotated) are typically single-level events with an amplitude around 1.2 nA. To a first approximation, this ion-current level is independent of nanopore size. In contrast, events from the secondary group (labeled in red) typically display multi-ion-current levels and are significantly longer than the events in the primary group (Figure 6a right-hand panel). An analysis of multiple proteins ranging from about 14 to 130 kDa shows that the characteristic dwell-time of the primary ECD group events grows linearly with the proteins’ Mw, which is directly proportional to the proteins’ polypeptide chain lengths. It is therefore plausible to assume that this group of events corresponds to single-file translocation of fully linearized proteins through the nanopores, as shown schematically in Figure 6c. An analysis of the events’ current amplitude level, in this case, provides a rough estimate for the mean diameter of the polypeptide/SDS complexes of 2.0 to 2.5 nm. However, one should keep in mind that the events’ amplitude is considerably affected also by an access resistance due to increased repulsion of ions from the pore during the protein’s translocation, particularly for the long proteins. The events of the secondary ECD peak exhibit much longer dwell-times than the primary group and often showed multiple ion-current levels during translocation. We hypothesize that, in this case, proteins are not fully denatured and contain some unresolved structures during their translocation process, as shown schematically in Figure 6c (right-hand panel). These results are supported by bulk time-resolved fluorescence measurements showing that at high SDS concentration the denatured proteins adopt an unfolded coil state, but below the SDS concentration used in our study, some local protein structure emerges (Figures S16 and S17). Further studies will be required to show that additional information on the proteins’ structure can be obtained from a detailed analysis of the ion-current time traces. Full denaturation of the polypeptide can be mediated by the EP force applied to the proteins, hence generating long dwell-times with nonlinear dependency on the proteins’ length (Figure 5a inset). In the future, more specific single-molecule studies, perhaps using electrooptical means or the inclusion of FRET probes, may be proven useful in support of this model.45 However, the fact that also the dwell-time of these secondary group events exponentially decays with voltage (Figure 3) may suggest that these events do not represent proteins that get stuck in the nanopore and retract to the cis side after some time. The latter scenario would entail growing dwell-times with increasing voltage, which is opposite to our observation. In summary, we present a study of voltage-driven translocation dynamics of SDS-denatured proteins through solid-state nanopores with diameters that are just larger than the proteins/SDS cross-section. Small nanopores allow us to extend the dwell-times of proteins in the nanopores by nearly an order of magnitude, compared to previous studies, hence significantly extending the detectable range of molecular weights of unfolded full-length proteins. Calculation of the events’ ECD has revealed two distinct groups of events, which we attribute to the fully denatured proteins and to proteins with partially folded structures. The ECD can be used as a proxy for estimation of each protein’s Mw, in an analogous manner to the traditional analysis of proteins’ migration length in bulk SDS-PAGE. Our results constitute an essential step toward using solid-state nanopores for the identification of linearized full-length proteins, with single-molecule resolution. Furthermore, our study highlights the role of EOF in slowing down SDS-denatured proteins’ translocating through nanopores. Future studies, using fluorescently labeled proteins and computational simulations are required for elucidating its exact mode of action. Methods Sample Preparation for Nanopore Experiments For experiments presented in Figures 2, 3, and 5, the denatured protein samples were prepared as follows: 100 nM of each protein was diluted in 1× PBS pH 7.4 containing 17.5 mM SDS and 5 mM tris(2-carboxyethyl)phosphine. The protein samples were allowed to shake at 300 rpm for 30 min at 25 °C to dissociate the protein’s disulfide bonds, followed by a complete denaturation of the proteins at 90 °C for 5 min. Then, the protein samples were cooled to room temperature before adding 1 μL of the protein sample to 100 μL of the nanopore buffer (400 mM NaCl, 175 μM SDS in 1× PBS, pH 7.4). Experiments in Figure 1 were prepared as described, except that the initial SDS concentration was 17.5, 35, or 70 mM. With the 100-fold dilution, the final SDS concentrations in the nanopore’s cis chamber were 175, 350, or 700 μM, accordingly, with a fixed protein concentration of 1 nM. For the SDS micelle translocation experiment (Figure 1a), the sample was prepared in the same manner without adding a protein. All proteins were purchased from Sigma-Aldrich-Merck, except the spike glycoprotein, which was purchased from ProSpec-Tany Techno-Gene Ltd. Sample Preparation and Procedure for SDS-PAGE For performing the SDS-PAGE experiments of proteins used in this study (Figure 4), the proteins were defrosted on ice, centrifuged at 10000 xg for 2 min at 4 °C, and kept on ice prior to sample preparation. A 20 pmol amount of alpha-lactalbumin, 6 pmol of carbonic anhydrase, 8 pmol of ovalbumin, 6 pmol of bovine serum albumin, 6 pmol of phosphorylase B and 4 pmol of spike were taken and mixed with Laemmli sample buffer containing 50 mM Tris-HCl, pH 6.8, 100 mM dithiothreitol, 2% (w/v) SDS, and 10% (v/v) glycerol. The samples were denatured at 95 °C for 5 min and separated on a 4–20% Tris-glycine gel (Bio-Rad) at a constant voltage of 150 V for 45 min using Laemmli running buffer (25 mM Tris base, 250 mM glycine, and 0.1% (w/v) SDS). The gel was fixed by incubating in 40% ethanol and 10% acetic acid for 3 h with gentle shaking and then stained using 1X Flamingo (Bio-Rad) overnight (16 h). The proteins’ bands were visualized using the 532 nm laser (532 nm excitation and 605 nm emission filters), and the gel image was acquired using the Pharos scanner (Bio-Rad). Nanopore Fabrication and Device Assembly A four-inch silicon wafer coated with 500 nm thermal silicon dioxide and 50 nm low-stress amorphous silicon nitride was used as a substrate for nanopore chips. The SiNx was locally thinned to 8–10 nm (∼2 μm circular wells) by reactive ion etching (RIE), followed by wet etching with buffered hydrofluoric acid (HF) etching to remove the SiO2. Following the procedure described elsewhere, the etched SiNx acted as a hard mask for subsequent anisotropic Si etching in KOH (33% m/v).46 Nanopore devices were cleaned in a hot 2:1 solution of H2SO4/H2O2 to make them hydrophilic and subsequently glued using EcoFlex 5 (smooth-on) onto a custom-made Teflon insert, immersed in buffer (1× PBS containing 400 mM NaCl, pH 7.4), and placed in a Teflon cell. The buffer was filtered using a 20 nm syringe filter before use. Nanopores were drilled in the thinned SiNx regions either using a dielectric breakdown or by laser drilling, as described previously.47,48 After drilling, the cis chamber buffer was changed to the “translocation buffer” (1× PBS containing 400 mM NaCl, 175 μM SDS, pH 7.4) and the open pore current was monitored for 5–10 min. If the pore remained stable, about 1 μL of the analyte was added to the cis chamber in each experiment. Data Acquisition and Analysis Before adding the sample to the nanopore’s cis chamber, a stable open-pore current was obtained by setting the bias voltage to V = 100–200 mV. The next step proceeded once a steady ionic flow and minimum noise were obtained in the nanopore device. An Axon 200B patch-clamp amplifier was used to monitor the translocation events through the nanopore, filtered with a 100 kHz low pass filter, and acquired using a custom LabVIEW program. Another LabVIEW program was used for the initial data analysis. The program identifies each event and provides information on the current blockage (ΔI), the dwell-time (tD), and the time of arrival of the event (ta) according to an electrical threshold. After obtaining ΔI and tD for each event, the ECD was calculated for each event, i.e., (ΔI × tD). The histogram plots of the ECD consist of two clearly distinguishable peaks, which are fitted with the double Gaussian function (Figure S10). Both the peaks have different event densities. The main peak of the ECD is called the primary peak, and the other one is the secondary peak. The peak values of the fitted double Gaussian function are plotted in Figure 5 with the standard deviations. Further, the event under the main peak and the secondary peak were analyzed separately, and corresponding ΔI and tD were fitted using a single Gaussian and an exponential function, respectively, as shown in Figures S11 and S12. The fitting values are plotted in Figure 5. Supporting Information Available The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsnano.2c05391.Supplementary figures of additional protein translocation data, event rate, representative raw translocation data, ECD values for alpha-lactalbumin as a function of nanopore sizes, additional scatter plots and dwell-time histograms, nanopore characterizations, time-resolved fluorescence as a function of SDS concentration and tables of fluorescence lifetime fit values, ECD, and dwell-time mean values (PDF) Supplementary Material nn2c05391_si_001.pdf Author Present Address # The WYSS Institute, Harvard University, Boston, Massachusetts 02115, United States Author Contributions N.S. performed nanopore measurements and analyzed data, N.F. performed bulk analysis and analyzed data, S.O. performed experiments, D.H. prepared protein samples, and A.M. designed and supervised the research. All authors co-wrote the manuscript. The authors declare no competing financial interest. Acknowledgments We thank Dr. Yulia Marom and Dr. Yana Rozevsky for their assistance in the fabrication of nanopore devices and Ms. Karawan Halabi and Mr. Jostine Joby for assistance in nanopore drilling. This project has received funding from the European Research Council (ERC) No. 833399 (NanoProt-ID) and ERC-PoC No. 966824 (OptiPore), both under the European Union’s Horizon 2020 research and innovation program grant agreements. S.O. is supported by the Azrieli Fellowship Foundation. ==== Refs References Deamer D. ; Akeson M. ; Branton D. Three Decades of Nanopore Sequencing. Nat. Biotechnol. 2016, 34 (5 ), 518–524. 10.1038/nbt.3423.27153285 Xue L. ; Yamazaki H. ; Ren R. ; Wanunu M. ; Ivanov A. P. ; Edel J. B. Solid-State Nanopore Sensors. Nat. Rev. Mater. 2020, 5 (12 ), 931–951. 10.1038/s41578-020-0229-6. Nir I. ; Huttner D. ; Meller A. Direct Sensing and Discrimination among Ubiquitin and Ubiquitin Chains Using Solid-State Nanopores. Biophys. J. 2015, 108 (9 ), 2340–2349. 10.1016/j.bpj.2015.03.025.25954891 Schmid S. ; Stömmer P. ; Dietz H. ; Dekker C. Nanopore Electro-Osmotic Trap for the Label-Free Study of Single Proteins and Their Conformations. Nat. Nanotechnol. 2021, 16 (11 ), 1244–1250. 10.1038/s41565-021-00958-5.34462599 Waduge P. ; Hu R. ; Bandarkar P. ; Yamazaki H. ; Cressiot B. ; Zhao Q. ; Whitford P. C. ; Wanunu M. Nanopore-Based Measurements of Protein Size, Fluctuations, and Conformational Changes. ACS Nano 2017, 11 (6 ), 5706–5716. 10.1021/acsnano.7b01212.28471644 Yusko E. C. ; Bruhn B. R. ; Eggenberger O. M. ; Houghtaling J. ; Rollings R. C. ; Walsh N. C. ; Nandivada S. ; Pindrus M. ; Hall A. R. ; Sept D. ; Li J. ; Kalonia D. S. ; Mayer M. Real-Time Shape Approximation and Fingerprinting of Single Proteins Using a Nanopore. Nat. Nanotechnol. 2017, 12 (4 ), 360–367. 10.1038/nnano.2016.267.27992411 He L. ; Tessier D. R. ; Briggs K. ; Tsangaris M. ; Charron M. ; McConnell E. M. ; Lomovtsev D. ; Tabard-Cossa V. Digital Immunoassay for Biomarker Concentration Quantification Using Solid-State Nanopores. Nat. Commun. 2021, 12 (1 ), 1–11. 10.1038/s41467-021-25566-8.33397941 Rozevsky Y. ; Gilboa T. ; Van Kooten X. F. ; Kobelt D. ; Huttner D. ; Stein U. ; Meller A. Quantification of MRNA Expression Using Single-Molecule Nanopore Sensing. ACS Nano 2020, 14 (10 ), 13964–13974. 10.1021/acsnano.0c06375.32930583 Gilboa T. ; Torfstein C. ; Juhasz M. ; Grunwald A. ; Ebenstein Y. ; Weinhold E. ; Meller A. Single-Molecule DNA Methylation Quantification Using Electro-Optical Sensing in Solid-State Nanopores. ACS Nano 2016, 10 (9 ), 8861–8870. 10.1021/acsnano.6b04748.27580095 Burck N. ; Gilboa T. ; Gadi A. ; Patkin Nehrer M. ; Schneider R. J. ; Meller A. Nanopore Identification of Single Nucleotide Mutations in Circulating Tumor DNA by Multiplexed Ligation. Clin. Chem. 2021, 10 , 1–10. 10.1093/clinchem/hvaa328. van Kooten X. F. ; Rozevsky Y. ; Marom Y. ; Ben Sadeh E. ; Meller A. Purely Electrical SARS-CoV-2 Sensing Based on Single-Molecule Counting. Nanoscale 2022, 14 (13 ), 4977–4986. 10.1039/D1NR07787B.35258059 Alfaro J. A. ; Bohländer P. ; Dai M. ; Filius M. ; Howard C. J. ; van Kooten X. F. ; Ohayon S. ; Pomorski A. ; Schmid S. ; Aksimentiev A. ; Anslyn E. V. ; Bedran G. ; Cao C. ; Chinappi M. ; Coyaud E. ; Dekker C. ; Dittmar G. ; Drachman N. ; Eelkema R. ; Goodlett D. ; Hentz S. ; Kalathiya U. ; Kelleher N. L. ; Kelly R. T. ; Kelman Z. ; Kim S. H. ; Kuster B. ; Rodriguez-Larrea D. ; Lindsay S. ; Maglia G. ; Marcotte E. M. ; Marino J. P. ; Masselon C. ; Mayer M. ; Samaras P. ; Sarthak K. ; Sepiashvili L. ; Stein D. ; Wanunu M. ; Wilhelm M. ; Yin P. ; Meller A. ; Joo C. The Emerging Landscape of Single-Molecule Protein Sequencing Technologies. Nat. Methods 2021, 18 (6 ), 604–617. 10.1038/s41592-021-01143-1.34099939 Ouldali H. ; Sarthak K. ; Ensslen T. ; Piguet F. ; Manivet P. ; Pelta J. ; Behrends J. C. ; Aksimentiev A. ; Oukhaled A. Electrical Recognition of the Twenty Proteinogenic Amino Acids Using an Aerolysin Nanopore. Nat. Biotechnol. 2020, 38 (2 ), 176–181. 10.1038/s41587-019-0345-2.31844293 Rodriguez-Larrea D. ; Bayley H. Multistep Protein Unfolding during Nanopore Translocation. Nat. Nanotechnol. 2013, 8 (4 ), 288–295. 10.1038/nnano.2013.22.23474543 Nivala J. ; Marks D. B. ; Akeson M. Unfoldase-Mediated Protein Translocation through an α-Hemolysin Nanopore. Nat. Biotechnol. 2013, 31 (3 ), 247–250. 10.1038/nbt.2503.23376966 Kennedy E. ; Dong Z. ; Tennant C. ; Timp G. Reading the Primary Structure of a Protein with 0.07 Nm 3 Resolution Using a Subnanometre-Diameter Pore. Nat. Nanotechnol. 2016, 11 (11 ), 968–976. 10.1038/nnano.2016.120.27454878 Brinkerhoff H. ; Kang A. S. W. ; Liu J. ; Aksimentiev A. ; Dekker C. Multiple Rereads of Single Proteins at Single-Amino Acid Resolution Using Nanopores. Science (80-.) 2021, 374 (6574 ), 1509–1513. 10.1126/science.abl4381. Zhang S. ; Huang G. ; Versloot R. C. A. ; Bruininks B. M. H. ; de Souza P. C. T. ; Marrink S. J. ; Maglia G. Bottom-up Fabrication of a Proteasome–Nanopore That Unravels and Processes Single Proteins. Nat. Chem. 2021, 13 (12 ), 1192–1199. 10.1038/s41557-021-00824-w.34795436 Hu Z. L. ; Huo M. Z. ; Ying Y. L. ; Long Y. T. Biological Nanopore Approach for Single-Molecule Protein Sequencing. Angew. Chemie - Int. Ed. 2021, 60 (27 ), 14738–14749. 10.1002/anie.202013462. Yan S. ; Zhang J. ; Wang Y. ; Guo W. ; Zhang S. ; Liu Y. ; Cao J. ; Wang Y. ; Wang L. ; Ma F. ; Zhang P. ; Chen H. Y. ; Huang S. Single Molecule Ratcheting Motion of Peptides in a Mycobacterium Smegmatis Porin A (MspA) Nanopore. Nano Lett. 2021, 21 (15 ), 6703–6710. 10.1021/acs.nanolett.1c02371.34319744 Afshar Bakshloo M. ; Kasianowicz J. J. ; Pastoriza-Gallego M. ; Mathé J. ; Daniel R. ; Piguet F. ; Oukhaled A. Nanopore-Based Protein Identification. J. Am. Chem. Soc. 2022, 144 (6 ), 2716–2725. 10.1021/jacs.1c11758.35120294 Oukhaled A. ; Cressiot B. ; Bacri L. ; Pastoriza-Gallego M. ; Betton J. M. ; Bourhis E. ; Jede R. ; Gierak J. ; Auvray L. ; Pelta J. Dynamics of Completely Unfolded and Native Proteins through Solid-State Nanopores as a Function of Electric Driving Force. ACS Nano 2011, 5 (5 ), 3628–3638. 10.1021/nn1034795.21476590 Lucas F. L. R. ; Versloot R. C. A. ; Yakovlieva L. ; Walvoort M. T. C. ; Maglia G. Protein Identification by Nanopore Peptide Profiling. Nat. Commun. 2021, 12 (1 ), 1–9. 10.1038/s41467-021-26046-9.33397941 Ohayon S. ; Girsault A. ; Nasser M. ; Shen-Orr S. ; Meller A. Simulation of Single-Protein Nanopore Sensing Shows Feasibility for Whole-Proteome Identification. PLoS Comput. Biol. 2019, 15 ( (5 ), ),10.1371/journal.pcbi.1007067. Talaga D. S. ; Li J. Single-Molecule Protein Unfolding in Solid State Nanopores. J. Am. Chem. Soc. 2009, 131 (26 ), 9287–9297. 10.1021/ja901088b.19530678 Fologea D. ; Ledden B. ; McNabb D. S. ; Li J. Electrical Characterization of Protein Molecules by a Solid-State Nanopore. Appl. Phys. Lett. 2007, 91 (5 ), 053901 10.1063/1.2767206. Freedman K. J. ; Haq S. R. ; Edel J. B. ; Jemth P. ; Kim M. J. Single Molecule Unfolding and Stretching of Protein Domains inside a Solid-State Nanopore by Electric Field. Sci. Rep. 2013, 3 (1 ), 1–8. 10.1038/srep01638. Cressiot B. ; Oukhaled A. ; Patriarche G. ; Pastoriza-Gallego M. ; Betton J. M. ; Auvray L. ; Muthukumar M. ; Bacri L. ; Pelta J. Protein Transport through a Narrow Solid-State Nanopore at High Voltage: Experiments and Theory. ACS Nano 2012, 6 (7 ), 6236–6243. 10.1021/nn301672g.22670559 Freedman K. J. ; Jürgens M. ; Prabhu A. ; Ahn C. W. ; Jemth P. ; Edel J. B. ; Kim M. J. Chemical, Thermal, and Electric Field Induced Unfolding of Single Protein Molecules Studied Using Nanopores. Anal. Chem. 2011, 83 (13 ), 5137–5144. 10.1021/ac2001725.21598904 Winogradoff D. ; John S. ; Aksimentiev A. Protein Unfolding by SDS: The Microscopic Mechanisms and the Properties of the SDS-Protein Assembly. Nanoscale 2020, 12 (9 ), 5422–5434. 10.1039/C9NR09135A.32080694 Bhuyan A. K. On the Mechanism of SDS-Induced Protein Denaturation. Biopolymers 2010, 93 (2 ), 186–199. 10.1002/bip.21318.19802818 Guo X. H. ; Chen S. H. The Structure and Thermodynamics of Protein-SDS Complexes in Solution and the Mechanism of Their Transports in Gel Electrophoresis Process. Chem. Phys. 1990, 149 (1–2 ), 129–139. 10.1016/0301-0104(90)80134-J. Restrepo-Pérez L. ; John S. ; Aksimentiev A. ; Joo C. ; Dekker C. SDS-Assisted Protein Transport through Solid-State Nanopores. Nanoscale 2017, 9 (32 ), 11685–11693. 10.1039/C7NR02450A.28776058 Wanunu M. ; Sutin J. ; McNally B. ; Chow A. ; Meller A. DNA Translocation Governed by Interactions with Solid-State Nanopores. Biophys. J. 2008, 95 (10 ), 4716–4725. 10.1529/biophysj.108.140475.18708467 Reynolds J. A. ; Tanford C. The Gross Conformation of Protein-Sodium Dodecyl Sulfate Complexes. J. Biol. Chem. 1970, 245 (19 ), 5161–5165. 10.1016/S0021-9258(18)62831-5.5528242 Thongngam M. ; McClements D. J. Influence of PH, Ionic Strength, and Temperature on Self-Association and Interactions of Sodium Dodecyl Sulfate in the Absence and Presence of Chitosan. Langmuir 2005, 21 (1 ), 79–86. 10.1021/la048711o.15620287 Eggenberger O. M. ; Ying C. ; Mayer M. Surface Coatings for Solid-State Nanopores. Nanoscale 2019, 11 (42 ), 19636–19657. 10.1039/C9NR05367K.31603455 Fologea D. ; Gershow M. ; Ledden B. ; McNabb D. S. ; Golovchenko J. A. ; Li J. Detecting Single Stranded DNA with a Solid State Nanopore. Nano Lett. 2005, 5 (10 ), 1905–1909. 10.1021/nl051199m.16218707 Fologea D. ; Uplinger J. ; Thomas B. ; McNabb D. S. ; Li J. Slowing DNA Translocation in a Solid-State Nanopore. Nano Lett. 2005, 5 (9 ), 1734–1737. 10.1021/nl051063o.16159215 Larkin J. ; Henley R. Y. ; Muthukumar M. ; Rosenstein J. K. ; Wanunu M. High-Bandwidth Protein Analysis Using Solid-State Nanopores. Biophys. J. 2014, 106 (3 ), 696–704. 10.1016/j.bpj.2013.12.025.24507610 Restrepo-Pérez L. ; Joo C. ; Dekker C. Paving the Way to Single-Molecule Protein Sequencing. Nat. Nanotechnol. 2018, 13 (9 ), 786–796. 10.1038/s41565-018-0236-6.30190617 Melnikov D. V. ; Hulings Z. K. ; Gracheva M. E. Electro-Osmotic Flow through Nanopores in Thin and Ultrathin Membranes. Phys. Rev. E 2017, 95 (6 ), 063105 10.1103/PhysRevE.95.063105.28709345 Zhang Y. ; Zhao J. ; Si W. ; Kan Y. ; Xu Z. ; Sha J. ; Chen Y. Electroosmotic Facilitated Protein Capture and Transport through Solid-State Nanopores with Diameter Larger than Length. Small Methods 2020, 4 (11 ), 1900893 10.1002/smtd.201900893. Wanunu M. ; Morrison W. ; Rabin Y. ; Grosberg A. Y. ; Meller A. Electrostatic Focusing of Unlabelled DNA into Nanoscale Pores Using a Salt Gradient. Nat. Nanotechnol. 2010, 5 (2 ), 160–165. 10.1038/nnano.2009.379.20023645 Wang R. ; Gilboa T. ; Song J. ; Huttner D. ; Grinstaff M. W. ; Meller A. Single-Molecule Discrimination of Labeled DNAs and Polypeptides Using Photoluminescent-Free TiO2 Nanopores. ACS Nano 2018, 12 (11 ), 11648–11656. 10.1021/acsnano.8b07055.30372037 Squires A. ; Atas E. ; Meller A. Nanopore Sensing of Individual Transcription Factors Bound to DNA. Sci. Rep. 2015, 5 (1 ), 1–11. 10.1038/srep11643. Zrehen A. ; Gilboa T. ; Meller A. Real-Time Visualization and Sub-Diffraction Limit Localization of Nanometer-Scale Pore Formation by Dielectric Breakdown. Nanoscale 2017, 9 (42 ), 16437–16445. 10.1039/C7NR02629C.29058736 Zvuloni E. ; Zrehen A. ; Gilboa T. ; Meller A. Fast and Deterministic Fabrication of Sub-5 Nanometer Solid-State Pores by Feedback-Controlled Laser Processing. ACS Nano 2021, 15 (7 ), 12189–12200. 10.1021/acsnano.1c03773.

PMC007xxxxxx/PMC7681919.txt

==== Front J Gen Physiol J Gen Physiol jgp The Journal of General Physiology 0022-1295 1540-7748 Rockefeller University Press 33211795 jgp.202012667 10.1085/jgp.202012667 Article Computational Biology Cellular Physiology Molecular Physiology Different arrhythmia-associated calmodulin mutations have distinct effects on cardiac SK channel regulation Mechanisms of human mutant calmodulins on SK channel function Ledford Hannah A. 1 Park Seojin 2 https://orcid.org/0000-0003-0170-5937 Muir Duncan 1 Woltz Ryan L. 1 https://orcid.org/0000-0002-9731-8562 Ren Lu 1 Nguyen Phuong T. 3 Sirish Padmini 1 Wang Wenying 2 Sihn Choong-Ryoul 2 https://orcid.org/0000-0002-3993-966X George Alfred L. Jr. 4 https://orcid.org/0000-0003-4956-9735 Knollmann Björn C. 5 Yamoah Ebenezer N. 2 Yarov-Yarovoy Vladimir 3 https://orcid.org/0000-0002-1042-8823 Zhang Xiao-Dong 16 https://orcid.org/0000-0001-9499-8817 Chiamvimonvat Nipavan 16 1 Division of Cardiovascular Medicine, Department of Internal Medicine, School of Medicine, University of California, Davis, Davis, CA 2 Department of Physiology and Cell Biology, University of Nevada, Reno, Reno, NV 3 Department of Physiology and Membrane Biology, School of Medicine, University of California, Davis, Davis, CA 4 Department of Pharmacology, Northwestern University Feinberg School of Medicine, Chicago, IL 5 Vanderbilt Center for Arrhythmia Research and Therapeutics, Department of Medicine, School of Medicine, Vanderbilt University, Nashville, TN 6 Department of Veterans Affairs, Northern California Health Care System, Mather, CA Correspondence to Nipavan Chiamvimonvat: nchiamvimonvat@ucdavis.edu Xiao-Dong Zhang: xdzhang@ucdavis.edu 07 12 2020 19 11 2020 152 12 e20201266706 6 2020 25 8 2020 19 10 2020 © 2020 Ledford et al. 2020 https://creativecommons.org/licenses/by-nc-sa/4.0/ http://www.rupress.org/terms/ This article is distributed under the terms of an Attribution–Noncommercial–Share Alike–No Mirror Sites license for the first six months after the publication date (see http://www.rupress.org/terms/). After six months it is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at https://creativecommons.org/licenses/by-nc-sa/4.0/). Cardiac small-conductance Ca2+-activated K+ (SK) channels are gated solely by beat-to-beat changes in intracellular Ca2+. Ledford et al. show distinct mechanisms by which human calmodulin mutants linked to sudden cardiac death regulate SK channels. Calmodulin (CaM) plays a critical role in intracellular signaling and regulation of Ca2+-dependent proteins and ion channels. Mutations in CaM cause life-threatening cardiac arrhythmias. Among the known CaM targets, small-conductance Ca2+-activated K+ (SK) channels are unique, since they are gated solely by beat-to-beat changes in intracellular Ca2+. However, the molecular mechanisms of how CaM mutations may affect the function of SK channels remain incompletely understood. To address the structural and functional effects of these mutations, we introduced prototypical human CaM mutations in human induced pluripotent stem cell–derived cardiomyocyte-like cells (hiPSC-CMs). Using structural modeling and molecular dynamics simulation, we demonstrate that human calmodulinopathy-associated CaM mutations disrupt cardiac SK channel function via distinct mechanisms. CaMD96V and CaMD130G mutants reduce SK currents through a dominant-negative fashion. By contrast, specific mutations replacing phenylalanine with leucine result in conformational changes that affect helix packing in the C-lobe, which disengage the interactions between apo-CaM and the CaM-binding domain of SK channels. Distinct mutant CaMs may result in a significant reduction in the activation of the SK channels, leading to a decrease in the key Ca2+-dependent repolarization currents these channels mediate. The findings in this study may be generalizable to other interactions of mutant CaMs with Ca2+-dependent proteins within cardiac myocytes. American Heart Association http://dx.doi.org/10.13039/100000968 14BGIA18870087 National Institutes of Health http://dx.doi.org/10.13039/100000002 R56 HL138392 R01 HL085727 R01 HL085844 R01 HL137228 R01 HL083374 R35 HL144980 R01 DC015135 R01 DC015252 P01 AG051443 R01 DC016099 United States Department of Veterans Affairs http://dx.doi.org/10.13039/100000738 I01 BX000576 I01 CX001490 National Heart, Lung, and Blood Institute http://dx.doi.org/10.13039/100000050 T32 HL086350 F31 HL136120 F32 HL151130 California Institute for Regenerative Medicine http://dx.doi.org/10.13039/100000900 NHLBI http://dx.doi.org/10.13039/100000050 T32 HL086350 American Heart Association http://dx.doi.org/10.13039/100000968 Harold S. Geneen Charitable Trust Roger Tatarian Endowed Professorship in Cardiovascular Medicine ==== Body pmcIntroduction Calmodulin (CaM) is a multifunctional Ca2+-binding protein that orchestrates a wide range of intracellular signaling and critical cellular processes (Crivici and Ikura, 1995; Chin and Means, 2000). CaM, a 17-kD protein, is encoded by three distinct genes in humans, namely CALM1, CALM2, and CALM3, each of which encodes an identical CaM protein. CaM is composed of N- and C-terminal lobes linked by a flexible helix (Babu et al., 1985; Heidorn and Trewhella, 1988). Each lobe contains two EF hands, canonical Ca2+-binding motifs, with the N-lobe having slightly lower Ca2+-binding affinity (Watterson et al., 1980; Klevit et al., 1984; Thulin et al., 1984). Ca2+ binding to the EF hands results in structural and functional changes in target molecules (Saimi and Kung, 2002). Recent studies have provided genetic links between human heritable CaM mutations and several types of cardiac arrhythmia syndromes leading to sudden cardiac death, including long QT syndrome (LQTS; Crotti et al., 2013; Reed et al., 2015; Boczek et al., 2016), catecholaminergic polymorphic ventricular tachycardia (CPVT; Nyegaard et al., 2012), and familial idiopathic ventricular fibrillation (IVF; Hwang et al., 2014). N54I and N98S mutations in CaM are associated with cases of CPVT with altered CaM-ryanodine receptor 2 (RYR2) function; F90L has been shown to be linked to IVF; and D96V, D130G, and F142L are associated with LQTS. LQTS mutations of CaM have been shown to reside at or near Ca2+ coordinating sites within the EF hands of the C-lobe of CaM (Limpitikul et al., 2014; Jensen et al., 2018). CALM1-D130G and CALM2-D96V mutations alter highly conserved aspartic acid residues that bind Ca2+ ions in EF-hand domains IV and III, respectively. The CALM1-F142L mutation results in an alteration of the energetic coupling of Ca2+ binding and the conformational change associated with CaM activation (Crotti et al., 2013). All three LQTS-associated mutations show reduced Ca2+ affinity in the C-lobe by 5-, 13-, and 53-fold for F142L, D96V, and D130G, respectively (Crotti et al., 2013). These mutations in CaM result in a significant reduction in Ca2+-dependent inactivation (CDI) of Cav1.2 L-type Ca2+ channel (LTCC; Peterson et al., 1999; Zühlke et al., 1999; Zühlke et al., 2000). In contrast, CPVT mutations in CaM impart little-to-mild reduction of Ca2+-binding affinity (Nyegaard et al., 2012). Clearly, CaM exerts its function on a number of targets in cardiomyocytes, including RYR2 (Meissner, 1986; Fukuda et al., 2014; Hwang et al., 2014; Nomikos et al., 2014; Vassilakopoulou et al., 2015), Kv7.1 (or KCNQ1; Shamgar et al., 2006; Sachyani et al., 2014; Sun and MacKinnon, 2017), LTCCs (Cav1.2, Cav1.3; Liang et al., 2003; Halling et al., 2006; Findeisen and Minor, 2010), and small-conductance Ca2+-activated K+ (SK) channels (Xia et al., 1998; Schumacher et al., 2001). Among the CaM-targets, SK channels are unique because the channels are voltage independent and are gated solely by beat-to-beat changes in intracellular Ca2+ (Ca2+i). SK channels harbor a highly conserved CaM-binding domain (CaMBD); the gating of the channels is controlled by Ca2+ binding and unbinding to CaM associated with SK channels (Xia et al., 1998). Thus, CaM serves as the high affinity Ca2+ sensor for SK channels. Studies on human calmodulinopathy have provided evidence for the critical roles of CaM in the functional regulation of SK channels (Yu et al., 2016; Saljic et al., 2019). However, the mechanistic effects on SK channels by mutations in CaM remain incompletely understood. Ca2+-activated K+ (KCa) channels can be divided into three main subfamilies based on their electrophysiological, pharmacological, and molecular phenotypes. These include large-conductance Ca2+- and voltage-activated K+ channels (BK), the intermediate-conductance KCa channels (IK or SK4), and the small-conductance KCa channels (SK1, 2, and 3; Köhler et al., 1996; Ishii et al., 1997; Joiner et al., 1997; Vergara et al., 1998). SK4 channels have intermediate single-channel conductance and are sensitive to charybdotoxin. In contrast, SK channels have small single-channel conductance and are sensitive to apamin. Recent cryo-EM structure of the SK4 channel–CaM complex (Lee and MacKinnon, 2018) reveals atomic details of the SK channel C-terminal interaction with the C-lobe of CaM, demonstrating that the C-lobe binds to the channel constitutively, while the N-lobe interacts with the S4-S5 linker in a Ca2+-dependent manner. The S4-S5 linker undergoes conformational changes upon CaM binding to open the channel pore (Lee and MacKinnon, 2018). CaM opens SK channels with a median effective intracellular concentration of Ca2+ of ∼100–400 nM (Köhler et al., 1996; Xia et al., 1998). Here, we take advantage of the SK4 channel–CaM complex structure as a template for Rosetta molecular modeling, combined with MD simulation and biochemical and functional analyses to test the effects of the known CaM mutations on cardiac SK2 channel functions. Specifically, we tested the hypothesis that human CaM mutations linked to sudden cardiac death disrupt SK channel function by distinct mechanisms. Mutations in the Ca2+-binding domain of CaM (CALM1-D130G, CALM2-D96V) result in a dominant-negative (DN) effect, while specific mutations with phenylalanine to leucine (CALM1-F90L and -F142L) disrupt the interactions between apo-CaM with the CaMBD of SK channels. SK currents have been shown to be prominently expressed not only in atrial myocytes (Xu et al., 2003; Zhang et al., 2015) but also in cardiac Purkinje cells (Reher et al., 2017). Indeed, Purkinje cells have been shown to be the probable site of origin of cardiac arrhythmias, including in patients with heritable arrhythmia syndromes (Haïssaguerre et al., 2002; Wilde et al., 2019). Additionally, SK channels have been shown to be expressed and play important roles in pacemaking cells, including sinoatrial and atrioventricular nodes (Zhang et al., 2008; Torrente et al., 2017). Distinct mutant CaMs may result in a significant reduction in the activation of the SK channels, leading to a decrease in the key Ca2+-dependent repolarization currents mediated by SK channels, exacerbating the effects of CaM mutations in LQTS or IVF. Materials and methods Plasmid construction Human filamin A (FLNA) in pREP4 vector (Life Technologies) was a kind gift from Dr. Paramita M. Ghosh (University of California Davis, Davis, CA). α-Actinin2 cDNA in pcDNA3 vector was a kind gift from Dr. David Fedida (University of British Columbia, Vancouver, BC, Canada; Cukovic et al., 2001). Construction of SK2 expression plasmids for heterologous expression in human embryonic kidney (HEK) 293 cells was as follows: full-length human cardiac SK2 cDNA was subcloned into pIRES2-EGFP (Takara Bio USA, Inc.) to obtain pSK2-IRES-EGFP plasmid. SK2 channel fusion constructs were generated harboring WT CaM (CaMWT; human SK2 [hSK2]-G4-CaMWT) or mutant CaMs (hSK2-G4 mutant CaM) to specifically occupy the CaMBD of SK2 channels with CaMWT or mutant CaMs using flexible linkers (glycine linkers) as previously described by Mori et al. (2004). To study the subcellular localization of the SK2 channel subunit, modified human influenza hemagglutinin (HA) tag was inserted into the extracellular S1-S2 loop of the channel. Specifically, modified HA epitope was flanked with the ClC-5 chloride channel D1-D2 loop to increase accessibility and inserted in the end of the S1-S2 loop of the SK2 channel subunit as we described previously (Kim et al., 2011). The inserted amino acid sequence was NSEHYPYDVPDYAVTFEERDKCPEWNC. The epitope is shown in bold. Epitope tags were generated by recombination polymerase chain reaction and verified by automated sequencing. HEK 293 cells and plasmid transfection HEK 293 cells were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% FBS. Cell lines were maintained at 37°C in a humidified atmosphere containing 5% CO2. All cell culture reagents were purchased from Life Technologies. HEK 293 cells were transfected using the following plasmid compositions: pSK2-IRES-EGFP, in combination with pREP4-FLNA and pcDNA3–α-actinin2, with WT (CaMWT) or mutant CaM (1 µg for each plasmid), using Lipofectamine 2000 (Life Technologies) according to the manufacturer’s protocol. The 1:1 ratio of the plasmids was determined to be most optimal for SK2 current expression in our prior published studies (Rafizadeh et al., 2014). Patch-clamp recordings Whole-cell Ca2+-activated K+ current (IK,Ca) was recorded from transfected HEK 293 cells and human induced pluripotent stem cell (hiPSC)–derived cardiomyocyte-like cells (hiPSC-CMs) at room temperature using conventional patch-clamp techniques as previously described (Xu et al., 2003). For current recordings, the extracellular solution contained (in mM) 140 N-methylglucamine, 4 KCl, 1 MgCl2, 5 glucose, and 10 HEPES, pH 7.4, with HCl. The internal solution contained (in mM) 144 potassium gluconate, 1.15 MgCl2, 5 EGTA, 10 HEPES, and CaCl2 yielding a free cytosol Ca2+ concentration of 500 nM, using the Maxchelator software by C. Patton of Stanford University as we have previously described (Zhang et al., 2017). The pH was adjusted to 7.25 using KOH. To isolate apamin-sensitive SK currents, extracellular solution containing apamin (10 or 100 nM) was applied during the recordings, and the difference currents between the control and the apamin-containing solution were calculated to be the apamin-sensitive currents. The current was elicited from a holding potential of −55 mV using a voltage ramp protocol ranging from −120 to +60 mV with a 2-s duration. For hiPSC-CM recordings, 0.006 mM CaMWT or mutant CaM peptides was added to the intracellular solution, and the recordings were performed 3 min after forming whole-cell recording for complete dialysis of the peptides. All recordings were performed using 3 M KCl agar bridges. Cell capacitance was calculated as the ratio of total charge (the integrated area under the current transient) to the magnitude of the pulse (20 mV). Currents were normalized to cell capacitance to obtain the current density. Series resistance was compensated electronically. In all experiments, a series resistance compensation of ≥90% was obtained. The currents were recorded using Axopatch 200A amplifier (Molecular Devices), filtered at 1 kHz using a four-pole Bessel filter, and digitized at a sampling frequency of 5 kHz. Data acquisition and analysis were performed using pClamp 10 software (Molecular Devices) and Origin Software (OriginLab). Immunofluorescence confocal laser scanning microscopy of HEK 293 cells HEK 293 cells were cotransfected with human cardiac SK2-HA (SK2 channel with the extracellular HA tag) together with α-actinin, FLNA, and CaMWT or mutant CaM, using Lipofectamine (Life Technologies; catalog #11668-019) as we previously described (Rafizadeh et al., 2014). After blocking with 1% BSA (Sigma; catalog #A7030) under nonpermeabilization (NP) condition, SK2 channels localized on the cell membrane were labeled with monoclonal anti-HA antibody (Covance; catalog #MMS-101P; 1:100 dilution) by incubating overnight in the humidified chamber (4°C), followed by treatment with a chicken anti-mouse Alexa Fluor 555 secondary antibody (Life Technologies; catalog #A-21200; 1:500 dilution) for 1 h at room temperature. Cells were then permeabilized with 0.01% Triton X (Fisher) and blocked with 1% BSA again, and intracellular SK2 channels were labeled with anti-HA antibody (1:250 dilution) at 4°C overnight and a rabbit anti-mouse Alexa Fluor 633 secondary antibody (Life Technologies; catalog #21427; 1:500 dilution). Coverslips were mounted using mounting medium containing 4′,6-diamidino-2-phenylindole (VectaMount; Vector Laboratories, Inc.; catalog #H-5000) and imaged under a Zeiss LSM 700 confocal laser scanning microscope. A fluorescence ratio of 555/633 then represents the ratio of SK2 channel numbers on cell membranes over those inside the cells, as we previously described (Rafizadeh et al., 2014). Cells cotransfected with CaMWT or mutant forms of CaM were immunolabeled in parallel, and all the microscopic settings were kept constant for all groups. Blind analysis was performed on unaltered images using Fiji ImageJ. hiPSC-CMs Feeder-free hiPSCs (iPS-D19-9-T7; WiCell) were cultured with mTeSRTM on hiPSC-qualified Matrigel. hiPSCs were differentiated into hiPSC-CMs following the protocols we previously reported (Sirish et al., 2016; Yamoah et al., 2018). We have observed ∼80–90% cardiomyocytes as shown previously (Yamoah et al., 2018). The hiPSC-CMs at day 40 (maturing cardiomyocytes) were used for experiments. CaMWT and mutant CaM peptides were used to test the functional changes of endogenous SK currents in hiPSC-CMs. The effects of WT compared with mutant CaMs were tested on the apamin-sensitive currents. Generation of recombinant CaM proteins WT and mutant CaM proteins were prepared as previously described (Hwang et al., 2014). Briefly, the recombinant CaM cDNA subcloned into a pET15b vector was mutated using QuikChange site-directed mutagenesis (Agilent Technologies). Proteins were expressed in Escherichia coli BL21 (DE3) cells and purified by hydrophobic chromatography using a phenyl sepharose column. Purified protein was dialyzed overnight at 4°C twice in 50 mM HEPES at pH 7.4, 100 mM KCl, and 5 mM EGTA and twice more with the same buffer except EGTA was lowered to 0.05 mM to remove Ca2+. The molecular mass of all proteins was confirmed using negative electrospray mass spectroscopy. Structural modeling of CaM, CaM mutants, and the hSK2 channel Structural modeling of CaMWT, CaMF90L, CaMF93L, and CaMF142L and the hSK2 channel was performed using Rosetta loop modeling and relax applications (Bonneau et al., 2001; Rohl et al., 2004; Yarov-Yarovoy et al., 2006; Wang et al., 2007; Yarov-Yarovoy et al., 2012; Bender et al., 2016; Alford et al., 2017) based on the hSK4 bound to the apo-CaM cryo-EM structure (PDB accession no. 6CNM; Lee and MacKinnon, 2018) as a template. First, we directly compared the sequence between hSK2 (UniProt accession no. Q9H2S1) with the cryo-EM structure of hSK4 (PDB accession no. 6CNM) that was used as the template for molecular modeling. The sequence identity between the two structures is 46%, while the sequence similarity is 63%. We further compared the sequence for the CaMBD (amino acid residues 412–488 in hSK2, DTQLTKRVKNAAANVLRETWLIYKNTKLVKKIDHAKVRKHQRKFLQAIHQLRSVKMEQRKLNDQANTLVDLAKTQNI). The sequence identity between the two CaMBDs is 47% with a sequence identity of 68%. The hSK2 channel sequence was threaded onto the hSK4 structure. 1,000 models of the hSK2 channel were generated using the Rosetta relax application and then clustered; the top model after clustering was selected as the best hSK2 model. The hSK2 model was used to replace the hSK4 peptide in the CaMBD based on hSK4–CaM complex structure (Lee and MacKinnon, 2018). 1,000 models of the hSK2–CaM complex were generated using the Rosetta relax application and then clustered, and the top model after clustering was selected as the best hSK2-CaM complex model, shown in Fig. 6. To determine the structure of the C-lobe of apo-CaM, CaM was taken from the 6CNM structure, mutated for the mutant models, relaxed, and finally clustered with the top models shown in Fig. 6. The mutant CaM models were created by mutating CaMWT and then using the Rosetta relax application to model potential conformational changes, induced by specific CaM mutations. 10,000 models were generated, the 1,000 lowest energy models were clustered, and the top models after clustering for each CaM mutant were selected as the best model. Molecular graphics and analyses were performed with the University of California San Francisco (UCSF) Chimera package (Pettersen et al., 2004). MD simulations of WT and mutant CaM Structures representing WT and mutant apo-CaM were obtained from Rosetta modeling as described above. The webserver CHARMM-GUI was used to prepare apo-CaM and create the solvent box (Jo et al., 2008; Brooks et al., 2009; Lee et al., 2016). The protein was immersed in a TIP3 water box with dimensions of 9 nm on all sides. Na and Cl ions were added to a final concentration of 150 mM to neutralize charge. In total, the system was composed of ∼15,000 (±1,000) atoms (Jo et al., 2008; Brooks et al., 2009; Lee et al., 2016). Periodic boundary conditions with dimensions equivalent to the water box were used (Phillips et al., 2005). Equilibration and production runs were performed using NAMD 2.13 (Phillips et al., 2005) on a local GPU cluster using CHARMM36m force fields (Vanommeslaeghe et al., 2010; Huang et al., 2017). After 100,000 steps of steepest descent minimization, MD simulations started with a time step of 1 fs, with harmonic restraints initially applied to protein heavy atoms. These restraints were slowly released over 0.5 ns. These systems were equilibrated further for 0.2 ns with a time step of 2 fs. To use a 2-fs time step, all bonds to H atoms were constrained using the SHAKE algorithm. All simulations were performed in isothermal-isobaric ensemble. Temperature was maintained at 303.15 K using a Langevin thermostat, and constant pressure at 1 atm was maintained using a Nosé-Hoover barostat. Electrostatic interactions were computed using the Particle Mesh Ewald method (Darden et al., 1993). Non-bonded pair lists were updated every 10 steps with a list cutoff distance of 16 Å and a real space cutoff of 12 Å, with energy switching starting at 10 Å. After equilibration of 1 ns, the simulation was continued for 1 µs. The full trajectory, including the equilibration and production steps, was used to calculate the RMSD. The 1-µs trajectory of the simulation during the production step was saved every 0.2 ns. This trajectory was then used for analysis. RMSD All measurements and renderings were performed using Visual Molecular Dynamics (Humphrey et al., 1996). RMSD was calculated every 0.2 ns over the equilibration, and production steps aligned to the first frame in the equilibration step. Specifically, for every 0.2 ns of the production simulation, the RMSD was calculated for apo-CaMWT and the mutant CaMs. The structures were clustered, and the cluster with the most decoys was selected as the most probable, with the center of the cluster being used as the representative model. Each mutant CaM structure was aligned individually to the CaMWT representative model using the matchmaker tool in UCSF Chimera, and the distance between the C-α of each residue was calculated. The averaged distance and the SD for each comparison for the full C-terminal domain (residues 81–145) and for Ca2+ binding site 4 (residues 131–138) was quantified (Table S1 and Table S2). Root mean square fluctuation (RMSF) Per residue RMSF was calculated over a sliding average window of 1 ns time and averaged over the 1-µs trajectory. The per-residue RMSF is presented as mean ± SD. Ensembles were generated by clustering the MD run as described in the structural modeling section. The top 10 models were displayed and aligned with the top cluster model. Data analysis Data analysis from patch-clamp recordings was performed by using Origin software (OriginLab). Current density obtained from cells expressing SK2 with CaMWT or mutant CaM was compared by normalizing the currents with cell capacitance. Where appropriate, pooled data are presented as means ± SEM. Statistical comparisons were performed using one-way and two-way ANOVA combined with Tukey’s post hoc analyses. One-way ANOVA was used to determine statistically significant differences among three or more groups in Fig. 1, Fig. 2, Fig. 4, and Fig. 5. Two-way ANOVA was used to determine statistically significant differences with two independent variables among three or more groups in Fig. 3. Post hoc analyses were performed using Tukey’s test. Statistical analyses were performed using GraphPad Prism and Origin Software. Online supplemental material Fig. S1 shows summary data for fluorescence intensity under NP (left bars) and permeabilized (right bars) conditions for HEK 293 cells expressing SK2 fusion protein with HA tag coexpressed with CaMWT or mutant CaMs (N54I, F90L, D96V, N98S, and D130G). Fig. S2 shows RMSD graphs that include the equilibration and production stages. Fig. S3 shows the top 10 clustered models of apo-CaMF90L, apo-CaMF93L, and apo-CaMF142L aligned to represent an ensemble of possible states, illustrating the distribution of helix 3 and Ca2+-binding site 4 regions. Fig. S4 presents graphical comparisons from MD simulations of the C-lobe of apo-CaMs showing hydrogen bond in the apo-CaMWT. The side chain of Asn112 on helix 2 in apo-CaMWT maintains a stable hydrogen bond with the carboxyl oxygen of Ala89 residue on helix 1 during simulations of apo-CaMWT, which is not observed in the simulations of the CaM mutants. Fig. S5 shows comparison of the RMSF of Ca2+-CaMF142L from MD simulations (A) and N-H heteronuclear NOE saturated/unsaturated ratio (B; BMRB: 34262; Wang et al., 2018), with similar flexibility including identification of the most flexible amino acid being Lys116 residue. Video 1, Video 2, and Video 3 show morph videos from the top Rosetta model of the C-lobe of apo-CaMWT to the top Rosetta models of apo-CaMF90L, apo-CaMF93L, and apo-CaMF142L, respectively. Video 4, Video 5, and Video 6 show morph videos from the top MD model of the C-lobe of apo-CaMWT to the top MD models of apo-CaMF90L, apo-CaMF93L, and apo-CaMF142L, respectively. Table S1 shows quantification of the RMSD calculated from the production portion of MD presented in Fig. S2. Table S2 shows quantification of the differences in the RMSD among WT and mutant CaMs, derived from the C-terminal domain (residues 81–145) and Ca2+ binding site 4 (residues 131–138) presented in Fig. 8. Results Inhibitory effects of human CaM mutations on SK channels We first coexpressed human cardiac SK2 channels with different human mutant CaMs compared with CaMWT in HEK 293 cells. Our previous studies showed that cytoskeletal proteins are critical for the proper membrane localization of SK2 channels; therefore, α-actinin2 and FLNA were also used for coexpression in all groups (Lu et al., 2009; Rafizadeh et al., 2014; Zhang et al., 2017). As positive controls, we tested the roles of the N- and C-lobe mutant CaMs in the regulation of SK2 currents using CaM1,2 and CaM3,4, where the two Ca2+-binding sites in the N- and C-lobe are mutated, respectively, as well as a mutant CaM (CaM1,2,3,4) that abolishes the four high affinity Ca2+-binding sites (Xia et al., 1998). Apamin-sensitive IK,Ca is recorded at baseline (before apamin application, black traces) and after apamin application (10 nM, red traces, Fig. 1 A). Both EF hands in the N- and C-lobe of CaM play critical roles in activation of the apamin-sensitive current (Fig. 1 A). Moreover, CaM1,2,3,4 results in the most significant inhibition on the apamin-sensitive current compared with CaM1,2 and CaM3,4 (Fig. 1 A). Summary data for the density of apamin-sensitive currents at the test potentials of −120 and +60 mV are shown in Fig. 1 B. The results are distinct from those previously reported demonstrating the critical roles of Ca2+ binding to the N- but not the C-lobe of CaM in SK channel gating in oocytes (Keen et al., 1999). However, our experiments were performed in a mammalian expression system with coexpression of α-actinin2 and FLNA. Furthermore, our previous study documented binding of α-actinin2 to the CaMBD of SK2 channels (Lu et al., 2007; Lu et al., 2009; Rafizadeh et al., 2014). Figure 1. Mutations in the N-lobe and C-lobe of CaM affect the currents of hSK2 expressed in HEK 293 cells. (A) hSK2 channels were coexpressed with CaMWT or mutant CaMs (CaM1,2, CaM3,4, or CaM1,2,3,4). Current was recorded before (black trace) and after (red trace) application of apamin (10 nM). (B) Summary data at +60 and −120 mV indicating significant reduction of the SK currents by mutant CaMs (n = 12 cells; *, P < 0.05). (C and D) hSK2 channels were coexpressed with CaMWT (black bars), CaMN54I (red bars), CaMF90L (green bars), CaMD96V (blue bars), CaMN98S (cyan bars), and CaMD130G (magenta bars). (C) Current was recorded before (black trace) and after (red trace) application of apamin (10 nM). (D) Summary data at −120 mV and +60 mV (n = 10–13 cells; *, P < 0.05). For B and D, the colors of the asterisks denote comparisons with the corresponding bar graphs of the same colors, using one-way ANOVA combined with Tukey's test (at −120 mV, P = 0.00001 for all pairwise comparisons except P = 0.04 for CaMN54I versus CaMD96V, and P = NS for CaMD96V versus CaMD130G and CaMWT versus CaMF90L; at +60 mV, P = 0.00001 for all pairwise comparisons except P = NS for CaMD96V versus CaMD130G). Data shown represents mean ± SEM. Next, we tested the effects of LQTS (CaMD96V, CaMD130G), CPVT (CaMN54I, CaMN98S), and IVF (CaMF90L) CaM mutants compared with CaMWT. Fig. 1 C shows the most prominent inhibitory effects on the density of apamin-sensitive currents by the two LQTS mutations (CaMD96V and CaMD130G). CaMF90L resulted in the least inhibitory effects. The current densities at −120 mV and +60 mV are shown in Fig. 1 D. No significant alteration in SK2 channel trafficking by human mutant CaMs The effects of CaM mutations on SK2 current may result from a decrease in channel activation or trafficking to the cell membrane. We therefore tested if the reduction in the SK2 current density from coexpression of mutant CaMs results from a decrease in membrane trafficking. Immunofluorescence confocal microscopy was performed using SK2 channels harboring an HA tag in the extracellular S1-S2 loop (SK2-HA) by using NP compared with permeabilized (P) conditions as we previously described (Fig. 2 and Fig. S1; Rafizadeh et al., 2014). Our data suggest that the mutant CaM does not alter SK channel trafficking. Figure 2. Trafficking of hSK2 channels after coexpression with CaMWT or mutant CaMs. (A) HEK 293 cells expressing SK2 fusion protein with HA tag coexpressed with CaMWT or mutant CaMs (N54I, F90L, D96V, N98S, and D130G) using NP and permeabilized (P) conditions. (B) Control represents incubation with secondary antibody alone. Scale bars are 10 µm. (C) Summary data showing fluorescence intensity under NP/P conditions (555/633 fluorescent intensity). Data shown represent mean ± SEM. Statistical analyses were performed using one-way ANOVA combined with Tukey’s test. n = 15–20 cells; P = NS. DAPI, 4′,6-diamidino-2-phenylindole. Figure S1. Trafficking of hSK2 channels after coexpression with CaMWT or mutant CaMs. HEK 293 cells expressing SK2 fusion protein with HA tag coexpressed with CaMWT or mutant CaMs (N54I, F90L, D96V, N98S, and D130G) using NP and permeabilized conditions. Summary data showing fluorescence intensity under NP (left bars) and permeabilized (right bars) conditions (555 and 633 fluorescent intensities in arbitrary units [AU] × 1,000; n = 15–20 cells; P = NS comparing NP with permeabilized for each group). Data shown represents mean ± SEM. Inhibitory effects of mutant CaMs on SK2 channels revealed by SK2 fusion protein harboring CaMWT or mutant CaMs via a flexible glycine (Gn) linker There are some possible caveats inherent in the experiments presented in Fig. 1 since mutations in CaMs may alter the efficiency of CaM expression in HEK 293 cells. To circumvent these possible confounding factors, we generated SK2 channel fusion constructs harboring CaMWT (hSK2-G4-CaMWT) or mutant CaMs (hSK2-G4-CaMMUT) to specifically occupy the CaMBD of SK2 channels with CaMWT or mutant CaMs using flexible linkers (Gn linker; Fig. 3 A), as previously described by Mori et al. (2004). This experimental paradigm enabled us to selectively test the contribution of the mutant CaMs compared with that of CaMWT on SK2 channels (Fig. 3 A). Figure 3. hSK2-CaM fusion proteins exhibit similar inhibitory effects to that of coexpressed hSK2 channels with CaM. (A) Diagram illustrating experimental paradigms using SK2 fusion proteins with flexible Gn linker fused to CaMWT (upper panel) compared with mutant CaMs (lower panel). (B) Current recordings before (black trace) and after (red trace) apamin application (10 nM) for hSK2 expressed with CaMWT versus hSK2-G4-CaMWT versus SK2-G4-CaMWT expressed with DN CaM1,2,3,4. (C) Summary data at −120 mV and +60 mV from B. (D and E) Summary data for apamin-sensitive current density in HEK 293 cells coexpressing hSK2 with CaMWT or mutant CaM constructs (left bars) compared with SK2-G4-CaMWT or SK2-G4-CaMMUT (hashed bars on the right), where CaMMUT refers to CaMN54I (red bars), CaMF90L (green bars), CaMF93L (purple bars), CaMD96V (blue bars), CaMN98S (cyan bars), CaMD130G (magenta bars), and CaMF142L (yellow bars) at −120 mV (D) and +60 mV (E; n = 10–13 cells). Data represent mean ± SEM. Statistical analysis was performed using two-way ANOVA combined with Tukey’s test. The colors of the asterisks denote comparisons with the corresponding bar graphs of the same colors. *, P < 0.05. At −120 mV, P = 0.00001 for all pairwise comparisons except P = 0.04 for CaMN54I versus CaMF93L and CaMN54I versus CaMD96V, and P = NS for CaMWT versus CaMF90L, CaMF93L versus CaMD96V, CaMF93L versus CaMD130G, CaMD96V versus CaMD130G, and CaMN98S versus CaMF142L. At +60 mV, P = 0.00001 for all pairwise comparisons except P = NS for CaMD96V versus CaMD130G and CaMN98S versus CaMF142L. There were no significant differences between hSK2 coexpression with CaM compared with hSK2-G4-CaM for WT or mutant CaMs (P = NS). We first demonstrated that the fusion proteins behave as expected by directly comparing apamin-sensitive IK,Ca recorded from HEK 293 cells expressing hSK2-G4-CaMWT versus hSK2 coexpressed with CaMWT (Fig. 3, B and C). Furthermore, coexpression of DN CaM (CaM1,2,3,4) with hSK2-G4-CaMWT failed to knock down apamin-sensitive current (Fig. 3, B and C) in contrast to hSK2 coexpressed with CaM1,2,3,4, as shown in Fig. 1 A. Having determined that the fusion proteins function as expected, we compared the density of apamin-sensitive currents between hSK2-G4-CaMMUT and coexpression of the SK2 channel and mutant CaMs. The density is comparable, with the most significant inhibitory effects on IK,Ca by the two LQTS mutations tested (CaMD96V and CaMD130G), consistent with a DN effect from these mutations (Fig. 3, D and E). We further tested the effects of two additional mutant CaMs (F93L and F142L). CaMF142L is shown to be linked to LQTS, while the Phe93 residue is predicted from our molecular model to interact with CaMBD of SK2 channels (described in a later section). These two mutations produced intermediate effects on IK,Ca (Fig. 3, D and E). Functional effects of mutant CaMs on SK currents in hiPSC-CMs Thus far, the experiments have relied on a heterologous expression system. We took advantage of hiPSC-CMs. The rationale was to determine the functional effects of mutant CaMs on endogenous SK currents with accompanying ion channel interacting proteins within a human cardiomyocyte-like cell. Even though hiPSC-CMs are relatively immature, we demonstrated the expression of SK2 channels in hiPSC-CMs using immunofluorescence confocal microscopy (Fig. 4 A). Fig. 4 B shows apamin-sensitive current in hiPSC-CMs when the cells were dialyzed by WT or mutant CaM proteins. Dialysis of mutant CaMD96V proteins resulted in the most prominent knockdown of apamin-sensitive currents in hiPSC-CMs, consistent with its DN effects, compared with the effects of CaMF90L (Fig. 4, C and D). Figure 4. Regulation of endogenous SK currents in hiPSC-CMs by intracellular CaM peptides. (A) hiPSC-CMs expressing SK2 channels (green) and α-actinin2 (red). 4′,6-Diamidino-2-phenylindole stain is shown in blue. Scale bar is 10 µm. (B) Effects of mutant CaM peptides (CaMN54I, CaMF90L, and CaMD96V) compared with CaMWT peptide on apamin-sensitive SK currents in hiPSC-CMs. The current was recorded before (black trace) and after (red trace) apamin application (100 nM). (C) Subtracted apamin-sensitive SK currents in the presence of CaMWT and mutant CaM peptides. (D) Comparisons of the apamin-sensitive current densities at −120 mV and +60 mV (n = 8–10 cells). Data represent mean ± SEM. Statistical analyses were performed using one-way ANOVA combined with Tukey’s test; *, P = 0.012 at −120 mV, and *, P = 0.037 at +60 mV between CaMWT and CaMD96V. Altered interaction between the SK2 channel and mutant CaMs Coimmunoprecipitation was performed as we previously described (Lu et al., 2007) between SK2-HA channels (Fig. 5 A) and different mutant CaMs compared with CaMWT. Since native cells contain a large amount of CaM, we constructed a fusion protein of CaM with FLAG-tag (DYKDDDDK; CaMFLAG) to enable identification of transfected mutant CaMs versus endogenous CaMWT (Fig. 5 B). There was a significant decrease in SK2 channels complexed with mutant CaMF90L in the absence of Ca2+ with EGTA (Fig. 5 C). Negative control experiments were performed using nontransfected cells and cells transfected with SK2-HA alone, without CaMFLAG. The data suggest a significant reduction in SK2 channel complexed with apo-CaMF90L compared with other CaM mutants, supporting the lack of DN effects with relatively modest inhibition of CaMF90L on SK2 currents. Figure 5. Interactions of mutant CaMs with SK2 channels revealed by coimmunoprecipitation assay. (A and B) Diagrams depicting the SK2 channel subunit with an HA tag in the S1-S2 extracellular loop and CaM with a FLAG tag in the C terminus under the cytomegalovirus (CMV) promoter. RBS, ribosome-binding site. (C) Coimmunoprecipitation using anti-Flag antibody (α-Flag) to pull down SK2 channels identified using anti–HA antibody (α-HA) in the presence of 2 mM EGTA. The right two lanes show the results from nontransfected HEK 293 cells (NT) and cells transfected with SK2 channels only (SK2) as negative controls. (D) Summary data of the findings in C. The background subtracted intensity for SK2 bands using anti–HA antibody was normalized to CaM bands using anti-Flag antibody (NIH ImageJ). Data shown represent mean ± SEM (n = 3). Statistical analyses were performed using one-way ANOVA combined with Tukey’s test; *, P < 0.05 for CaMWT compared with CaMF90L, CaMD96V, CaMN98S, and CaMD130G, and *, P < 0.05 for CaMF90L compared with CaMD96V, CaMN98S, and CaMD130G (the P values are as follows: CaMWT versus CaMF90L, 0.00003; CaMWT versus CaMD96V, 0.002; CaMWT versus CaMN98S, 0.004; CaMWT versus CaMD130G, 0.004; CaMF90L versus CaMD96V, 0.03; CaMF90L versus CaMN98S, 0.02; and CaMF90L versus CaMD130G, 0.02). IP, immunoprecipitation. Structural modeling of CaMF90L, CaMF93L, and CaMF142L mutants and their interactions with SK2 channels To explore potential structural changes induced by CaM mutations, we generated structural models of the C-lobe of apo-CaMWT, apo-CaMF90L, apo-CaMF93L, and apo-CaMF142L using Rosetta structural modeling. The models of CaM reveal that Phe → Leu mutations resulted in conformational changes that affected helix packing in the C-lobe due to the substitution of a relatively large phenylalanine side chain to a smaller leucine side chain (Fig. 6, B–D). The Phe90 side chain is surrounded by the following residues in apo-CaMWT: Arg87 in helix 1; Ile101 in loop 1–2; Val137 in loop 3–4; and Tyr139 and Phe142 in helix 4. The Phe93 side chain is surrounded by the following residues in apo-CaMWT: Ile101 in loop 1–2; Leu106 and Val109 in helix 2; and Phe142, Met145, and Met146 in helix 4. The Phe142 side chain is surrounded by the following residues in apo-CaMWT: Ile86, Phe90, and Phe93 in helix 1; Ile101 in loop 1–2 and Leu106 in helix 2; Val137 in loop 3–4; and Tyr139, Met145, and Met146 in helix 4. Notably, Phe90, Phe93, and Phe142 are interacting with each other in apo-CaMWT. The conformational effects of the Phe → Leu mutations are explored in more detail using MD simulations. Figure 6. Structural modeling of the interactions between CaM mutants and SK2 CaMBD. (A) Schematic of the four α-helices within the C-lobe of CaM. Location of mutations are shown by labels and colored markers. (B–D) Comparisons of CaMWT (green) and CaMF90L (red; B), CaMF93L (yellow; C), and CaMF142L (purple; D). Side chains of key amino acid residues are shown in stick representation using the color scheme shown in A. Conformational changes due to CaM mutation are indicated by black arrows in each panel. (E) C-lobe of apo-CaMWT (colored in green) bound to the C terminus of hSK2 channel (colored in light brown). Side chains of key amino acids are shown using space-filling representation. (F) Panel E rotated 90° to the left around the y axis. Molecular modeling was performed in Ca2+-free conditions. Additionally, we generated a structural model of apo-CaMWT in complex with CaMBD from the SK2 channel based on the cryo-EM structure of the SK4 channel in complex with apo-CaM (PDB accession no. 6CNM; Lee and MacKinnon, 2018). The three phenylalanine residues (Phe90, Phe93, and Phe142) from the C-lobe of CaM can be seen interacting with hydrophobic residues from the CaMBD of SK2 channels (Fig. 6, E and F). Consistent with our hypothesis and the coimmunoprecipitation study (Fig. 5), we predict that substitution of the phenylalanine side chain may result in a significant decrease in the interaction of apo-CaM and SK2 CaMBD. Even though CaMF90L and CaMF142L were previously described to be linked to human arrhythmias, CaMF93L has not been described in patients. Our molecular modeling predicts that the Phe93 residue intimately interacts with SK2 CaMBD. Indeed, CaMF93L results in a significant decrease in SK2 current, supported by our functional analyses (Fig. 3, D and E). MD simulations provide insights into the conformational changes of CaMF90L, CaMF93L, and CaMF142L To further explore potential conformational changes induced by the Phe → Leu mutations in CaM, we performed MD simulations on the WT and three CaM mutants: F90L, F93L, and F142L. Each simulation was performed for 1 µs after equilibration. The simulation for apo-CaMWT was performed twice. The RMSDs demonstrated that the system equilibrated rapidly and that the RMSD had plateaued for the entire runs (Fig. S2 and Table S1). Figure S2. Graphs of RMSD showing stability throughout the production run. RMSD graphs include the equilibration and prosecution stages. The RMSD from the C-lobe of apo-CaMWT (simulations [sims] 1 and 2), apo-CaMF90L, apo-CaMF93L, and apo-CaMF142L are shown in blue, dark brown, red, yellow, and green, respectively. Runs are slightly different in length due to differences in equilibration time. The total simulation time is the same in all runs (1 µs). Analyses of the RMSF show remarkable findings that the three Phe → Leu mutants affected the per-residue structure in similar manners (Fig. 7), even though the mutants are located in different helices of the C-lobe of CaM (helix 1 contains Phe90 and Phe93, and helix 4 contains Phe142). Compared with apo-CaMWT from the first simulation, all three CaM mutants exhibited more flexibility toward the C-terminal end of helix 2, while the C-terminal end of helix 3 and Ca2+-binding site 4 were less flexible. The largest shifts in RMSF were observed where the side chains of Phe90, Phe93, and Phe142 contact helices 2 and 3. This agrees with the notion that the Phe → Leu mutation creates a void volume, allowing helices 2 and 3 to repack their side chains and readjust their backbone positions (Fig. 8, Fig. S3, Video 1, Video 2, Video 3, Video 4, Video 5, and Video 6). Differences in the RMSD between an equilibrated apo-CaMWT and mutant CaMs shown in Fig. 8 are quantified for the C-terminal domains as well as for Ca2+-binding site 4 (Table S2). Figure 7. Graphs of RMSF showing regional changes in stability. RMSF from MD simulations of the C-lobe of apo-CaMWT is shown in blue (first simulation [sim]) and purple (second simulation), while RMSF from MD simulations of the C-lobe of apo-CaMF90L, apo-CaMF93L, and apo-CaMF142L mutants are shown in red, orange, and green, respectively. Diagrams above the graphs depict helices 1–4 from the C-lobe together with amino acid residues and are scaled to match the residue numbering of the graph. Locations of the three mutations are indicated by their respective colored dot on the graph and bars above the helices. Ca2+-binding sites are indicated with purple dots together with the numbers for the amino acid residues. Data shown are mean ± SD. Figure 8. Graphical comparison from MD simulations of the C-lobe of apo-CaMs showing conformational changes in mutant CaMs with repacking of helices 2 and 3 and highly flexible Ca2+-binding site 4 in apo-CaMWT compared with the mutant CaMs, confirming the RMSF results. (A–C) Top clustered models of the C-lobe of apo-CaMF90L (red), apo-CaMF93L (yellow), and apo-CaMF142L (purple) are aligned with apo-CaMWT (green) in A, B, and C, respectively. Amino acid side chains of Phe90 (pink) and F90L (red) from apo-CaMWT and apo-CaMF90L (A), amino acid side chains of Phe93 (red) and F93L (yellow) from apo-CaMWT and apo-CaMF93L (B), and amino acid side chains of Phe142 (light blue) and F142L (light purple) from apo-CaMWT and apo-CaMF90L (C) are shown in the stick representation. Helix 3 from apo-CaMWT (magenta) resides at a larger angle in reference to the y axis compared with helices 3 from Phe → Leu mutant CaMs (dark blue) in A, B, and C, consistent with repacking of the structures. Quantification of the differences between apo-CaMWT and mutant CaMs shown in A, B, and C are summarized in Table S2. (D) Top 10 clustered models of apo-CaMWT aligned to represent an ensemble of possible states. Amino acid representation and color are conserved from A–C. Helix 3 is in purple. Helix 3 is mostly in a similar angle in reference to the y axis, with a much more extreme model present in the distribution. In addition, Ca2+-binding site 4 has a wide distribution of states, which does not align with those from the mutants (A–C). (E) Panel D rotated 180° to the right around the y axis. Figure S3. Top 10 clustered models of the C-lobe of apo-CaMF90L, apo-CaMF93L, and apo-CaMF142L aligned to represent an ensemble of possible states showing compact distribution of helix 3 and Ca2+-binding site 4 regions. Ensembles of the aligned top 10 clusters of the C-lobe of apo-CaMF90L (red; A), apo-CaMF93L (yellow; B), and apo-CaMF142L (purple; C). Helix 3 is shown in blue, and the Ca2+-binding site 4 is the loop immediately following helix 3 and is labeled. Lower panels are structures from the upper panels rotated 180° to the right around the y axis. 10.1085/jgp.202012667.v1 Video 1. Morph video from the top Rosetta model of the C-lobe of apo-CaMWT to the top Rosetta model of apo-CaMF90L. The video illustrates the conformational changes that are required to accommodate a void volume that is created by the phenylalanine-to-leucine mutation, allowing helices 2 and 3 to repack their side chains and readjust their backbone positions. The coloring is the same as in Fig. 6 and Fig. 8 (residues Phe90, Phe93, and Phe142 are shown in pink, red, and light blue, respectively). Helices 2 and 3 are in the foreground, while helices 1 and 4 are in the background. Helix 3 is shown in magenta. 10.1085/jgp.202012667.v2 Video 2. Morph video from the top Rosetta model of the C-lobe of apo-CaMWT to the top Rosetta model of apo-CaMF93L. The video illustrates the conformational changes that are required to accommodate a void volume that is created by the phenylalanine-to-leucine mutation, allowing helices 2 and 3 to repack their side chains and readjust their backbone positions. The coloring is the same as in Fig. 6 and Fig. 8 (residues Phe90, Phe93, and Phe142 are shown in pink, red, and light blue, respectively). Helices 2 and 3 are in the foreground, while helices 1 and 4 are in the background. Helix 3 is shown in magenta. 10.1085/jgp.202012667.v3 Video 3. Morph video from the top Rosetta model of the C-lobe of apo-CaMWT to the top Rosetta model of apo-CaMF142L. The video illustrates the conformational changes that are required to accommodate a void volume that is created by the phenylalanine-to-leucine mutation, allowing helices 2 and 3 to repack their side chains and readjust their backbone positions. The coloring is the same as in Fig. 6 and Fig. 8 (residues Phe90, Phe93, and Phe142 are shown in pink, red, and light blue, respectively). Helices 2 and 3 are in the foreground, while helices 1 and 4 are in the background. Helix 3 is shown in magenta. 10.1085/jgp.202012667.v4 Video 4. Morph video from the top MD model of the C-lobe of apo-CaMWT to the top MD model of apo-CaMF90L. The video illustrates the conformational changes that are required to accommodate a void volume that is created by the phenylalanine-to-leucine mutation, allowing helices 2 and 3 to repack their side chains and readjust their backbone positions. The coloring is the same as in Fig. 6 and Fig. 8 (residues Phe90, Phe93, and Phe142 are shown in pink, red, and light blue, respectively). Helices 2 and 3 are in the foreground, while helices 1 and 4 are in the background. Helix 3 is shown in magenta. 10.1085/jgp.202012667.v5 Video 5. Morph video from the top MD model of the C-lobe of apo-CaMWT to the top MD model of apo-CaMF93L. The video illustrates the conformational changes that are required to accommodate a void volume that is created by the phenylalanine-to-leucine mutation, allowing helices 2 and 3 to repack their side chains and readjust their backbone positions. The coloring is the same as in Fig. 6 and Fig. 8 (residues Phe90, Phe93, and Phe142 are shown in pink, red, and light blue, respectively). Helices 2 and 3 are in the foreground, while helices 1 and 4 are in the background. Helix 3 is shown in magenta. 10.1085/jgp.202012667.v6 Video 6. Morph video from the top MD model of the C-lobe of apo-CaMWT to the top MD model of apo-CaMF142L. The video illustrates the conformational changes that are required to accommodate a void volume that is created by the phenylalanine-to-leucine mutation, allowing helices 2 and 3 to repack their side chains and readjust their backbone positions. The coloring is the same as in Fig. 6 and Fig. 8 (residues Phe90, Phe93, and Phe142 are shown in pink, red, and light blue, respectively). Helices 2 and 3 are in the foreground, while helices 1 and 4 are in the background. Helix 3 is shown in magenta. The RMSD from Simulation 2 of the apo-CaMWT exhibited a transient increase before returning to baseline toward the end of the simulation (Fig. S2). Closer examinations revealed that the major contributing regions of apo-CaMWT toward the transient increase in the RMSD in Simulation 2 are in helix 3, just proximal to Ca2+-binding site 4, similar to the conformational changes we observed in Simulation 1 but at a larger magnitude. Additionally, Simulation 2 of apo-CaMWT exhibited a similar per-residue RMSF pattern to Simulation 1, albeit lower in magnitude, as demonstrated in Fig. 7. This is consistent with the lower value of RMSD for Simulation 2 compared with Simulation 1 for the majority of the run. Finally, there was a return to the average RMSD after the transient increase, supporting the equilibration of the simulations. Consistent with our observation, the side chain of Asn112 on helix 2 in apo-CaMWT maintained a stable hydrogen bond with the carboxyl oxygen of the Ala89 residue on helix 1 during the simulations of the apo-CaMWT, which was not observed in the simulations of the CaM mutants (Fig. S4). This may also, in part, influence flexibility toward the C-terminal end of helix 2 of the three mutant CaMs, as discussed above. We suggest that these observed changes may result in a decrease in the interaction of apo-CaM and SK2 CaMBD. Finally, to validate the MD simulations, we directly compared RMSF of Ca2+-CaMF142L from the MD simulation with the publicly available N-H heteronuclear NOE saturated/unsaturated ratio from a previously published NMR structure (BMRB accession no. 34262; Wang et al., 2018). Indeed, there is significant similarity in flexibility, including identification of the most flexible amino acid residue as Lys116 (Fig. S5). Figure S4. Graphical comparison from MD simulations of the C-lobe of apo-CaMs showing the hydrogen bond in the apo-CaMWT. (A–D) Top clustered models of the C-lobe of apo-CaMWT (green; A), apo-CaMF90L (red; B), apo-CaMF93L (yellow; C), and apo-CaMF142L (purple; D) are shown. Helix 3 from apo-CaMWT is shown in magenta (A), while helices 3 from Phe → Leu mutant CaMs are shown in dark blue in B–D. The side chain of Asn112 on helix 2 in apo-CaMWT maintains a stable hydrogen bond with the carboxyl oxygen of Ala89 residue on helix 1 during simulations of the apo-CaMWT, which is not observed in the simulations of the CaM mutants. Figure S5. Comparison of the RMSF of Ca2+-CaMF142L from MD simulations and N-H heteronuclear NOE saturated/unsaturated ratio. Related to BMRB accession no. 34262; Wang et al., 2018. (A and B) There is similar flexibility, including identification of the most flexible amino acid being Lys116 residue. Note: flexibility shown on graphs is inversed, with higher values in RMSF and lower values in heteronuclear NOE denoting higher flexibility. Discussion The gating of SK channels relies on Ca2+ binding to CaM. Therefore, mutations of CaM in human calmodulinopathy are predicted to have significant effects on SK channel function. In the current study, we demonstrate a significant reduction in SK currents in the presence of the LQTS CaM mutants CaMD96V and CaMD130G. The marked reduction in SK currents in the presence of LQTS CaM mutants is consistent with their DN effects. Immunofluorescence staining does not show significant changes in membrane-bound SK2 channels in the presence of WT or mutant CaM, suggesting that the effects of these CaM mutants are primarily on the Ca2+-CaM–dependent activation of the channels. The DN effect is predicted to result in a reduction in repolarization reserve. We confirmed these results using tandem SK2-CaM constructs to circumvent potential caveats from endogenous CaM in HEK 293 cells and/or changes in CaM expression from mutant constructs. In contrast to the marked DN effects of CaMD96V and CaMD130G, CaMF90L results in only a modest reduction in IK,Ca, suggesting a distinct mechanism by disrupting the interaction between apo-CaM and the SK channel without the DN effects. Structural modeling and MD simulations reveals the atomistic mechanisms in conformational changes caused by the phenylalanine-to-leucine mutation, which affect helix packing in the C-lobe, disrupting interactions between apo-CaM with CaMBD of SK channels. In Fig. 1, we showed that CaMD96V exhibits the most prominent inhibitory effect, while CaMF90L exerts only a mild inhibitory effect on apamin-sensitive current. Consistently, dialysis of mutant CaMD96V proteins in hiPSC-CMs results in the most prominent knockdown of apamin-sensitive currents in hiPSC-CMs compared with CaMF90L (Fig. 4). As expected, the endogenous SK currents in hiPSC-CMs (Fig. 4) are much smaller than the overexpressed SK currents in HEK 293 cells (Fig. 1). Therefore, the two systems were used as complementary studies to evaluate the effects of CaM mutants on SK currents. There are six CaM alleles, and a mutation in one allele is expected to have only a partial effect on the overall CaM pool within the cell. Therefore, other possible mechanisms may play a role, including cell-specific differential expression of the different alleles. Mechanisms of human calmodulinopathy CaM serves as a Ca2+ sensor for many ion channels and transporters and consequently has been shown to play key roles in the regulation of cardiac excitability; mutations in CaM have been linked to LQTS, CPVT, and IVF (Crotti et al., 2013; George, 2015; Reed et al., 2015; Boczek et al., 2016). Studies on calmodulinopathy have provided evidence of the critical roles of CaM in the functional regulation of Cav1.2, RYR2, Kv7.1, and SK channels, among others (Nyegaard et al., 2012; Crotti et al., 2013; Fukuda et al., 2014; Hwang et al., 2014; George, 2015; Vassilakopoulou et al., 2015; Yu et al., 2016). CaMD96V and CaMD130G occur in highly conserved aspartic acid residues responsible for chelating Ca2+ ions in the C-lobe (Crotti et al., 2013). Consequently, CaMD130G has been shown to have markedly decreased Ca2+-binding affinity (Hwang et al., 2014; Vassilakopoulou et al., 2015). In addition, CaMD130G was reported to have decreased interaction with RYR2, possibly through a DN effect. CaMD96V was previously shown to exhibit an increased Kd for Ca2+ binding compared with CaMWT, although to a lesser extent than CaMD130G (Vassilakopoulou et al., 2015). LQTS mutants (D96V, D130G, and F142L) have been reported to exhibit reduced Ca2+ affinity compared with that of CaMWT and CPVT (N54I and N98S) mutants (Hwang et al., 2014). In addition, the LQTS-associated mutants have been shown to suppress CDI of L-type Ca2+ currents (Limpitikul et al., 2014). Potential roles of cardiac SK channels in calmodulinopathy We previously identified the expression of SK channels (SK1, SK2, and SK3) in the heart and demonstrated the critical roles of the channels in mediating action potential repolarization in human atrial myocytes (Xu et al., 2003; Tuteja et al., 2005). SK2 channels physically coupled to α-actinin2, allowing a close association with LTCCs (Lu et al., 2007, 2009). The subcellular localization of SK channels in close proximity to LTCC and RYR2 provides an immediate Ca2+ source (Zhang et al., 2018), which, through binding of Ca2+ to CaM, triggers conformational changes to activate SK channels. SK channels may also play important roles in the feedback mechanism by regulating the activities of LTCCs and RYR2 to influence local and global Ca2+ signaling. The roles of SK channels in the heart have been extensively studied in the past decade (Zhang et al., 2015). Single nucleotide polymorphisms of KCNN genes encoding for SK channels have been shown to be linked to human atrial fibrillation (AF; Roselli et al., 2018), and SK channels may represent a potential novel therapeutic target against atrial arrhythmias (Diness et al., 2015). Importantly, SK channels participate in the electrical remodeling in AF and heart failure. SK channels are significantly up-regulated in AF animal models (Ozgen et al., 2007; Qi et al., 2014) and failing ventricular myocytes (Chua et al., 2011; Chang et al., 2013). However, the mechanisms underlying SK channel remodeling in the diseased heart remain unclear. SK channels are unique because the channels are solely gated by Ca2+i through a highly conserved CaMBD. Our data demonstrate the functional effects, structural basis, and molecular mechanisms of the disease-causing CaM mutants in the regulation of cardiac SK2 channels, highlighting the potential contribution of cardiac SK channels to human calmodulinopathy. Even though SK currents are more prominently expressed in atrial than in ventricular myocytes (Xu et al., 2003; Zhang et al., 2015), SK currents are known to be up-regulated in failing ventricular myocytes (Chua et al., 2011; Chang et al., 2013) and during hypokalemia (Chan et al., 2015). Recent studies have supported their critical roles in cardiac Purkinje cells (Reher et al., 2017), which are known to be the potential site of origin of cardiac arrhythmias, including in patients with heritable arrhythmia syndrome (Haïssaguerre et al., 2002; Wilde et al., 2019). Moreover, SK channels play critical roles in the pacemaker activity of sinoatrial and atrioventricular nodes (Zhang et al., 2008; Torrente et al., 2017). Therefore, mutations in CaMs may result in a significant reduction in the key Ca2+-dependent repolarization currents mediated by SK channels. Even though no SK channel variants have been reported to be associated with an inherited arrhythmia syndrome to date, a recent study reported a rare variant, c.1509C>G (p.F503L), in one KCNN2 allele in a patient who developed drug-induced LQTS (Ko et al., 2018). Indeed, this is a rapidly expanding area of research. Additionally, SK2 channels colocalized within microdomains of LTCC and RYR2, which serve as the Ca2+ source for the activation of SK channels (Zhang et al., 2018). Activation of SK2 channels provides a direct link between beat-to-beat changes in Ca2+i and membrane potentials, and thus may serve as the feedback mechanisms to regulate the activities of the LTCC and RYR2 channels. Functional effects of CaM mutations on SK channel activation may indirectly disrupt this possible feedback mechanism on LTCC and RYR2 channels. Recent studies have evaluated the effects of arrhythmogenic CaM mutations on SK2 and SK3 channel function in HEK 293 cells (Yu et al., 2016; Saljic et al., 2019). They similarly observed effects independent of SK2 expression and trafficking; however, our study provides new insights by taking advantage of computational analyses as well as native-like cells through our use of hiPSC-CMs. Advancing the atomistic understanding of CaM and SK2 interaction by structural modeling and MD simulations Apo-CaM binding to the CaMBD of SK channels requires the C-lobe, while binding of Ca2+ to the N-lobe is involved in SK channel gating (Keen et al., 1999). Several human CaM mutations affect the activation of SK channels (Yu et al., 2016). However, the exact mechanisms among the different CaM mutations remain incompletely understood. Our study shows a significant reduction of cardiac SK currents by CaMD96V and CaMD130G, not only in the hetero-expression system but also in endogenous SK currents in hiPSC-CMs, consistent with DN effects on channel function. In contrast, phenylalanine mutations in CaM disrupt interaction between apo-CaM and the SK channel without the DN effects. Rosetta molecular modeling is used by taking advantage of the recently published cryo-EM structure of the SK4–apo-CaM complex (Lee and MacKinnon, 2018). We identified the interactions of three key phenylalanine residues (Phe90, Phe93, and Phe142) within the C-lobe of apo-CaM, with hydrophobic amino acids in the CaMBD within the C terminus of the SK2 channel. In addition, previous studies have shown reduced binding affinity to RYR2 by CaMF90L. Circular dichroism experiments in the same study suggest C-lobe destabilization and decreased Ca2+-binding affinity (Nomikos et al., 2014). Our molecular modeling of the CaMF90L mutant shows a significant deviation from CaMWT in the C-lobe structure in the Ca2+-free state (Fig. 6). Taken together with biochemical and functional analyses, the results suggest that the Phe mutations primarily disrupt apo-CaM interactions with the SK2 CaMBD. Indeed, despite significant conformational changes, there were only modest inhibitory effects on apamin-sensitive currents (Fig. 3), consistent with a lack of DN effects; in contrast to CaMD96V and CaMD130G mutants. Structural modeling and MD simulations provide critical insights into the atomistic mechanisms of CaM mutants in human calmodulinopathy. Our findings suggest that all three Phe mutations in CaM result in similar conformational changes, including an increase in flexibility of the C-terminal end of helix 2, a decrease in flexibility of the C-terminal end of helix 3, and a decrease in the flexibility of the Ca2+-binding site 4 (Fig. 7). The findings support our conclusion that the three Phe → Leu mutations exhibit a similar mechanism in altering the regulation of the hSK2 channel, likely through a reduction in the interaction between apo-CaM and CaMBD in the SK2 channel. These critical insights into the distinct mechanisms whereby these CaM mutations exert their effects on Ca2+-dependent ion channels and cardiac action potential repolarization help to pave the way for understanding the genotype–phenotype correlations of human calmodulinopathy. Limitations and future directions RMSD as a metric is circumstantial to derive conclusions on the thermodynamic stability of WT and mutant CaMs. Nonetheless, we have provided additional analyses of RMSF in the current study. Future studies are needed to derive free-energy calculations of the mutations compared with CaMWT. The alteration in flexibility and the possible effects of the mutations on Ca2+-binding site 4 and their cooperativity with other Ca2+-binding sites require further studies as an underlying mechanism affecting CaM mutations. Future studies are needed to test the possible roles of hydrogen bonds between the side chain of Asn112 with the carboxyl oxygen of the Ala89 residue in apo-CaMWT, which is lost in all three Phe → Leu mutations based on MD simulations. Additionally, studies evaluating the effects of CaM mutants in native cells would be ideal; however, the study of mutant proteins within endogenous cells is not without difficulties. To circumvent these issues, the study of CaM mutants in cardiomyocyte-like cells may yet yield valuable information. Consequently, patient-specific hiPSC-CMs may provide additional insights into the effects of CaM mutants on SK channels and action potential profiles. Finally, future studies using CRISPR/Cas9 gene editing would help to further elucidate the effects of CaM mutations on SK channel function. Supplementary Material Table S1 shows quantification of the RMSD calculated from the production portion of MD presented in Fig. S2. Click here for additional data file. Table S2 shows quantification of the differences in the RMSD among WT and mutant CaMs, derived from the C-terminal domain (residues 81–145) and Ca2+ binding site 4 (residues 131–138) presented in Fig. 8. Click here for additional data file. Acknowledgments Jeanne M. Nerbonne served as editor. This work is funded in part by the American Heart Association Beginning Grant-in-Aid 14BGIA18870087 and National Institutes of Health (NIH) grants R56 HL138392 (to X.-D. Zhang); R01 HL085727, R01 HL085844, and R01 HL137228 (to N. Chiamvimonvat); R01 HL083374 (to A.L. George Jr.); R35 HL144980 (to B.C. Knollmann); and R01 DC015135, R01 DC015252, R01 DC016099, and P01 AG051443 (to E.N. Yamoah); and United States Department of Veterans Affairs Merit Review Grants I01 BX000576 and I01 CX001490 (to N. Chiamvimonvat). H.A. Ledford received a Predoctoral Fellowship from NIH/National Heart, Lung, and Blood Institute (NHLBI) Institutional Training Grant in Basic and Translational Cardiovascular Science (T32 HL086350 and F31 HL136120). R.L. Woltz received a Postdoctoral Fellowships from NIH/NHLBI (F32 HL151130). L. Ren received a Predoctoral Fellowship from the American Heart Association. P. Sirish received a Postdoctoral Fellowship from California Institute for Regenerative Medicine Training Grant to University of California Davis and NIH/NHLBI Institutional Training Grant in Basic and Translational Cardiovascular Science (T32 HL086350), an American Heart Association Career Development Award, and the Harold S. Geneen Charitable Trust Award. N. Chiamvimonvat is the holder of the Roger Tatarian Endowed Professorship in Cardiovascular Medicine and a part-time staff physician at VA Northern California Health Care System, Mather, CA. The authors declare no competing financial interests. Author contributions: H.A. Ledford, W. Wang, and X.-D. Zhang performed the electrophysiological experiments and analyzed and interpreted the data. H.A. Ledford, S. Park, L. Ren, and C.-R. Sihn performed the molecular cloning and biochemical experiments. R.L. Woltz, D. Muir, P.T. Nguyen, and V. Yarov-Yarovoy performed the structural modeling and MD simulations. H.A. Ledford performed immunofluorescence imaging. P. Sirish performed stem cell differentiation. H.A. Ledford, X.-D. Zhang, P.T. Nguyen, E.N. Yamoah, V. Yarov-Yarovoy, and N. Chiamvimonvat wrote the manuscript. A.L. George, Jr. and B.C. Knollman provided critical reagents for the study. H.A. Ledford, E.N. Yamoah, X.-D. Zhang, and N. Chiamvimonvat conceived the project, designed the experiments, interpreted the data, and wrote the manuscript. ==== Refs References Alford, R.F., A. Leaver-Fay, J.R. Jeliazkov, M.J. O’Meara, F.P. DiMaio, H. Park, M.V. Shapovalov, P.D. Renfrew, V.K. Mulligan, K. Kappel, . 2017. The Rosetta All-Atom Energy Function for Macromolecular Modeling and Design. J. Chem. Theory Comput. 13 :3031–3048. 10.1021/acs.jctc.7b00125 28430426 Babu, Y.S., J.S. Sack, T.J. Greenhough, C.E. Bugg, A.R. Means, and W.J. Cook. 1985. Three-dimensional structure of calmodulin. Nature. 315 :37–40. 10.1038/315037a0 3990807 Bender, B.J., A. Cisneros III, A.M. Duran, J.A. Finn, D. Fu, A.D. Lokits, B.K. Mueller, A.K. Sangha, M.F. Sauer, A.M. Sevy, . 2016. Protocols for Molecular Modeling with Rosetta3 and RosettaScripts. Biochemistry. 55 :4748–4763. 10.1021/acs.biochem.6b00444 27490953 Boczek, N.J., N. Gomez-Hurtado, D. Ye, M.L. Calvert, D.J. Tester, D. Kryshtal, H.S. Hwang, C.N. Johnson, W.J. Chazin, C.G. Loporcaro, . 2016. Spectrum and Prevalence of CALM1-, CALM2-, and CALM3-Encoded Calmodulin Variants in Long QT Syndrome and Functional Characterization of a Novel Long QT Syndrome-Associated Calmodulin Missense Variant, E141G. Circ. Cardiovasc. Genet. 9 :136–146. 10.1161/CIRCGENETICS.115.001323 26969752 Bonneau, R., C.E. Strauss, and D. Baker. 2001. Improving the performance of Rosetta using multiple sequence alignment information and global measures of hydrophobic core formation. Proteins. 43 :1–11. 10.1002/1097-0134(20010401)43:1<1::AID-PROT1012>3.0.CO;2-A 11170209 Brooks, B.R., C.L. Brooks III, A.D. Mackerell Jr., L. Nilsson, R.J. Petrella, B. Roux, Y. Won, G. Archontis, C. Bartels, S. Boresch, . 2009. CHARMM: the biomolecular simulation program. J. Comput. Chem. 30 :1545–1614. 10.1002/jcc.21287 19444816 Chan, Y.H., W.C. Tsai, J.S. Ko, D. Yin, P.C. Chang, M. Rubart, J.N. Weiss, T.H. Everett IV, S.F. Lin, and P.S. Chen. 2015. Small-Conductance Calcium-Activated Potassium Current Is Activated During Hypokalemia and Masks Short-Term Cardiac Memory Induced by Ventricular Pacing. Circulation. 132 :1377–1386. 10.1161/CIRCULATIONAHA.114.015125 26362634 Chang, P.C., I. Turker, J.C. Lopshire, S. Masroor, B.L. Nguyen, W. Tao, M. Rubart, P.S. Chen, Z. Chen, and T. Ai. 2013. Heterogeneous upregulation of apamin-sensitive potassium currents in failing human ventricles. J. Am. Heart Assoc. 2 :e004713. 10.1161/JAHA.112.004713 23525437 Chin, D., and A.R. Means. 2000. Calmodulin: a prototypical calcium sensor. Trends Cell Biol. 10 :322–328. 10.1016/S0962-8924(00)01800-6 10884684 Chua, S.K., P.C. Chang, M. Maruyama, I. Turker, T. Shinohara, M.J. Shen, Z. Chen, C. Shen, M. Rubart-von der Lohe, J.C. Lopshire, . 2011. Small-conductance calcium-activated potassium channel and recurrent ventricular fibrillation in failing rabbit ventricles. Circ. Res. 108 :971–979. 10.1161/CIRCRESAHA.110.238386 21350217 Crivici, A., and M. Ikura. 1995. Molecular and structural basis of target recognition by calmodulin. Annu. Rev. Biophys. Biomol. Struct. 24 :85–116. 10.1146/annurev.bb.24.060195.000505 7663132 Crotti, L., C.N. Johnson, E. Graf, G.M. De Ferrari, B.F. Cuneo, M. Ovadia, J. Papagiannis, M.D. Feldkamp, S.G. Rathi, J.D. Kunic, . 2013. Calmodulin mutations associated with recurrent cardiac arrest in infants. Circulation. 127 :1009–1017. 10.1161/CIRCULATIONAHA.112.001216 23388215 Cukovic, D., G.W. Lu, B. Wible, D.F. Steele, and D. Fedida. 2001. A discrete amino terminal domain of Kv1.5 and Kv1.4 potassium channels interacts with the spectrin repeats of alpha-actinin-2. FEBS Lett. 498 :87–92. 10.1016/S0014-5793(01)02505-4 11389904 Darden, T., D. York, and L. Pedersen. 1993. Particle mesh Ewald: An N.log(N) method for Ewald sums in large systems. J. Chem. Phys. 98 :10089–10092. 10.1063/1.464397 Diness, J.G., B.H. Bentzen, U.S. Sørensen, and M. Grunnet. 2015. Role of Calcium-activated Potassium Channels in Atrial Fibrillation Pathophysiology and Therapy. J. Cardiovasc. Pharmacol. 66 :441–448. 10.1097/FJC.0000000000000249 25830485 Findeisen, F., and D.L. Minor. 2010. Progress in the structural understanding of voltage-gated calcium channel (CaV) function and modulation. Channels (Austin). 4 :459–474. 10.4161/chan.4.6.12867 21139419 Fukuda, M., T. Yamamoto, S. Nishimura, T. Kato, W. Murakami, A. Hino, M. Ono, H. Tateishi, T. Oda, S. Okuda, . 2014. Enhanced binding of calmodulin to RyR2 corrects arrhythmogenic channel disorder in CPVT-associated myocytes. Biochem. Biophys. Res. Commun. 448 :1–7. 10.1016/j.bbrc.2014.03.152 24755079 George, A.L. Jr. 2015. Calmodulinopathy: a genetic trilogy. Heart Rhythm. 12 :423–424. 10.1016/j.hrthm.2014.11.017 25460856 Haïssaguerre, M., D.C. Shah, P. Jaïs, M. Shoda, J. Kautzner, T. Arentz, D. Kalushe, A. Kadish, M. Griffith, F. Gaïta, . 2002. Role of Purkinje conducting system in triggering of idiopathic ventricular fibrillation. Lancet. 359 :677–678. 10.1016/S0140-6736(02)07807-8 11879868 Halling, D.B., P. Aracena-Parks, and S.L. Hamilton. 2006. Regulation of voltage-gated Ca2+ channels by calmodulin. Sci. STKE. 2006 :er1.16685765 Heidorn, D.B., and J. Trewhella. 1988. Comparison of the crystal and solution structures of calmodulin and troponin C. Biochemistry. 27 :909–915. 10.1021/bi00403a011 3365370 Huang, J., S. Rauscher, G. Nawrocki, T. Ran, M. Feig, B.L. de Groot, H. Grubmüller, and A.D. MacKerell Jr. 2017. CHARMM36m: an improved force field for folded and intrinsically disordered proteins. Nat. Methods. 14 :71–73. 10.1038/nmeth.4067 27819658 Humphrey, W., A. Dalke, and K. Schulten. 1996. VMD: visual molecular dynamics. J. Mol. Graph. 14 :33–38: 27–28. 10.1016/0263-7855(96)00018-5 8744570 Hwang, H.S., F.R. Nitu, Y. Yang, K. Walweel, L. Pereira, C.N. Johnson, M. Faggioni, W.J. Chazin, D. Laver, A.L. George Jr., . 2014. Divergent regulation of ryanodine receptor 2 calcium release channels by arrhythmogenic human calmodulin missense mutants. Circ. Res. 114 :1114–1124. 10.1161/CIRCRESAHA.114.303391 24563457 Ishii, T.M., C. Silvia, B. Hirschberg, C.T. Bond, J.P. Adelman, and J. Maylie. 1997. A human intermediate conductance calcium-activated potassium channel. Proc. Natl. Acad. Sci. USA. 94 :11651–11656. 10.1073/pnas.94.21.11651 9326665 Jensen, H.H., M. Brohus, M. Nyegaard, and M.T. Overgaard. 2018. Human Calmodulin Mutations. Front. Mol. Neurosci. 11 :396. 10.3389/fnmol.2018.00396 30483049 Jo, S., T. Kim, V.G. Iyer, and W. Im. 2008. CHARMM-GUI: a web-based graphical user interface for CHARMM. J. Comput. Chem. 29 :1859–1865. 10.1002/jcc.20945 18351591 Joiner, W.J., L.Y. Wang, M.D. Tang, and L.K. Kaczmarek. 1997. hSK4, a member of a novel subfamily of calcium-activated potassium channels. Proc. Natl. Acad. Sci. USA. 94 :11013–11018. 10.1073/pnas.94.20.11013 9380751 Keen, J.E., R. Khawaled, D.L. Farrens, T. Neelands, A. Rivard, C.T. Bond, A. Janowsky, B. Fakler, J.P. Adelman, and J. Maylie. 1999. Domains responsible for constitutive and Ca(2+)-dependent interactions between calmodulin and small conductance Ca(2+)-activated potassium channels. J. Neurosci. 19 :8830–8838. 10.1523/JNEUROSCI.19-20-08830.1999 10516302 Kim, H.J., P. Lv, C.R. Sihn, and E.N. Yamoah. 2011. Cellular and molecular mechanisms of autosomal dominant form of progressive hearing loss, DFNA2 J. Biol. Chem. 286 :1517–1527. 10.1074/jbc.M110.179010 20966080 Klevit, R.E., D.C. Dalgarno, B.A. Levine, and R.J. Williams. 1984. 1H-NMR studies of calmodulin. The nature of the Ca2+-dependent conformational change. Eur. J. Biochem. 139 :109–114. 10.1111/j.1432-1033.1984.tb07983.x 6697998 Ko, J.S., S. Guo, J. Hassel, P. Celestino-Soper, T.C. Lynnes, J.E. Tisdale, J.J. Zheng, S.E. Taylor, T. Foroud, M.D. Murray, . 2018. Ondansetron blocks wild-type and p.F503L variant small-conductance Ca2+-activated K+ channels. Am. J. Physiol. Heart Circ. Physiol. 315 :H375–H388. 10.1152/ajpheart.00479.2017 29677462 Köhler, M., B. Hirschberg, C.T. Bond, J.M. Kinzie, N.V. Marrion, J. Maylie, and J.P. Adelman. 1996. Small-conductance, calcium-activated potassium channels from mammalian brain. Science. 273 :1709–1714. 10.1126/science.273.5282.1709 8781233 Lee, C.H., and R. MacKinnon. 2018. Activation mechanism of a human SK-calmodulin channel complex elucidated by cryo-EM structures. Science. 360 :508–513. 10.1126/science.aas9466 29724949 Lee, J., X. Cheng, J.M. Swails, M.S. Yeom, P.K. Eastman, J.A. Lemkul, S. Wei, J. Buckner, J.C. Jeong, Y. Qi, . 2016. CHARMM-GUI Input Generator for NAMD, GROMACS, AMBER, OpenMM, and CHARMM/OpenMM Simulations Using the CHARMM36 Additive Force Field. J. Chem. Theory Comput. 12 :405–413. 10.1021/acs.jctc.5b00935 26631602 Liang, H., C.D. DeMaria, M.G. Erickson, M.X. Mori, B.A. Alseikhan, and D.T. Yue. 2003. Unified mechanisms of Ca2+ regulation across the Ca2+ channel family. Neuron. 39 :951–960. 10.1016/S0896-6273(03)00560-9 12971895 Limpitikul, W.B., I.E. Dick, R. Joshi-Mukherjee, M.T. Overgaard, A.L. George Jr., and D.T. Yue. 2014. Calmodulin mutations associated with long QT syndrome prevent inactivation of cardiac L-type Ca(2+) currents and promote proarrhythmic behavior in ventricular myocytes. J. Mol. Cell. Cardiol. 74 :115–124. 10.1016/j.yjmcc.2014.04.022 24816216 Lu, L., Q. Zhang, V. Timofeyev, Z. Zhang, J.N. Young, H.S. Shin, A.A. Knowlton, and N. Chiamvimonvat. 2007. Molecular coupling of a Ca2+-activated K+ channel to L-type Ca2+ channels via alpha-actinin2. Circ. Res. 100 :112–120. 10.1161/01.RES.0000253095.44186.72 17110593 Lu, L., V. Timofeyev, N. Li, S. Rafizadeh, A. Singapuri, T.R. Harris, and N. Chiamvimonvat. 2009. Alpha-actinin2 cytoskeletal protein is required for the functional membrane localization of a Ca2+-activated K+ channel (SK2 channel). Proc. Natl. Acad. Sci. USA. 106 :18402–18407. 10.1073/pnas.0908207106 19815520 Meissner, G. 1986. Ryanodine activation and inhibition of the Ca2+ release channel of sarcoplasmic reticulum. J. Biol. Chem. 261 :6300–6306.2422165 Mori, M.X., M.G. Erickson, and D.T. Yue. 2004. Functional stoichiometry and local enrichment of calmodulin interacting with Ca2+ channels. Science. 304 :432–435. 10.1126/science.1093490 15087548 Nomikos, M., A. Thanassoulas, K. Beck, V. Vassilakopoulou, H. Hu, B.L. Calver, M. Theodoridou, J. Kashir, L. Blayney, E. Livaniou, . 2014. Altered RyR2 regulation by the calmodulin F90L mutation associated with idiopathic ventricular fibrillation and early sudden cardiac death. FEBS Lett. 588 :2898–2902. 10.1016/j.febslet.2014.07.007 25036739 Nyegaard, M., M.T. Overgaard, M.T. Søndergaard, M. Vranas, E.R. Behr, L.L. Hildebrandt, J. Lund, P.L. Hedley, A.J. Camm, G. Wettrell, . 2012. Mutations in calmodulin cause ventricular tachycardia and sudden cardiac death. Am. J. Hum. Genet. 91 :703–712. 10.1016/j.ajhg.2012.08.015 23040497 Ozgen, N., W. Dun, E.A. Sosunov, E.P. Anyukhovsky, M. Hirose, H.S. Duffy, P.A. Boyden, and M.R. Rosen. 2007. Early electrical remodeling in rabbit pulmonary vein results from trafficking of intracellular SK2 channels to membrane sites. Cardiovasc. Res. 75 :758–769. 10.1016/j.cardiores.2007.05.008 17588552 Peterson, B.Z., C.D. DeMaria, J.P. Adelman, and D.T. Yue. 1999. Calmodulin is the Ca2+ sensor for Ca2+-dependent inactivation of L-type calcium channels. Neuron. 22 :549–558. 10.1016/S0896-6273(00)80709-6 10197534 Pettersen, E.F., T.D. Goddard, C.C. Huang, G.S. Couch, D.M. Greenblatt, E.C. Meng, and T.E. Ferrin. 2004. UCSF Chimera--a visualization system for exploratory research and analysis. J. Comput. Chem. 25 :1605–1612. 10.1002/jcc.20084 15264254 Phillips, J.C., R. Braun, W. Wang, J. Gumbart, E. Tajkhorshid, E. Villa, C. Chipot, R.D. Skeel, L. Kalé, and K. Schulten. 2005. Scalable molecular dynamics with NAMD. J. Comput. Chem. 26 :1781–1802. 10.1002/jcc.20289 16222654 Qi, X.Y., J.G. Diness, B.J. Brundel, X.B. Zhou, P. Naud, C.T. Wu, H. Huang, M. Harada, M. Aflaki, D. Dobrev, . 2014. Role of small-conductance calcium-activated potassium channels in atrial electrophysiology and fibrillation in the dog. Circulation. 129 :430–440. 10.1161/CIRCULATIONAHA.113.003019 24190961 Rafizadeh, S., Z. Zhang, R.L. Woltz, H.J. Kim, R.E. Myers, L. Lu, D. Tuteja, A. Singapuri, A.A. Bigdeli, S.B. Harchache, . 2014. Functional interaction with filamin A and intracellular Ca2+ enhance the surface membrane expression of a small-conductance Ca2+-activated K+ (SK2) channel. Proc. Natl. Acad. Sci. USA. 111 :9989–9994. 10.1073/pnas.1323541111 24951510 Reed, G.J., N.J. Boczek, S.P. Etheridge, and M.J. Ackerman. 2015. CALM3 mutation associated with long QT syndrome. Heart Rhythm. 12 :419–422. 10.1016/j.hrthm.2014.10.035 25460178 Reher, T.A., Z. Wang, C.H. Hsueh, P.C. Chang, Z. Pan, M. Kumar, J. Patel, J. Tan, C. Shen, Z. Chen, . 2017. Small-Conductance Calcium-Activated Potassium Current in Normal Rabbit Cardiac Purkinje Cells. J. Am. Heart Assoc. 6 :e005471. 10.1161/JAHA.117.005471 28550095 Rohl, C.A., C.E. Strauss, K.M. Misura, and D. Baker. 2004. Protein structure prediction using Rosetta. Methods Enzymol. 383 :66–93. 10.1016/S0076-6879(04)83004-0 15063647 Roselli, C., M.D. Chaffin, L.C. Weng, S. Aeschbacher, G. Ahlberg, C.M. Albert, P. Almgren, A. Alonso, C.D. Anderson, K.G. Aragam, . 2018. Multi-ethnic genome-wide association study for atrial fibrillation. Nat. Genet. 50 :1225–1233. 10.1038/s41588-018-0133-9 29892015 Sachyani, D., M. Dvir, R. Strulovich, G. Tria, W. Tobelaim, A. Peretz, O. Pongs, D. Svergun, B. Attali, and J.A. Hirsch. 2014. Structural basis of a Kv7.1 potassium channel gating module: studies of the intracellular c-terminal domain in complex with calmodulin. Structure. 22 :1582–1594. 10.1016/j.str.2014.07.016 25441029 Saimi, Y., and C. Kung. 2002. Calmodulin as an ion channel subunit. Annu. Rev. Physiol. 64 :289–311. 10.1146/annurev.physiol.64.100301.111649 11826271 Saljic, A., K.M. Muthukumarasamy, J.M. la Cour, K. Boddum, M. Grunnet, M.W. Berchtold, and T. Jespersen. 2019. Impact of arrhythmogenic calmodulin variants on small conductance Ca2+-activated K+ (SK3) channels. Physiol. Rep. 7 :e14210. 10.14814/phy2.14210 31587513 Schumacher, M.A., A.F. Rivard, H.P. Bächinger, and J.P. Adelman. 2001. Structure of the gating domain of a Ca2+-activated K+ channel complexed with Ca2+/calmodulin. Nature. 410 :1120–1124. 10.1038/35074145 11323678 Shamgar, L., L. Ma, N. Schmitt, Y. Haitin, A. Peretz, R. Wiener, J. Hirsch, O. Pongs, and B. Attali. 2006. Calmodulin is essential for cardiac IKS channel gating and assembly: impaired function in long-QT mutations. Circ. Res. 98 :1055–1063. 10.1161/01.RES.0000218979.40770.69 16556865 Sirish, P., N. Li, V. Timofeyev, X.D. Zhang, L. Wang, J. Yang, K.S. Lee, A. Bettaieb, S.M. Ma, J.H. Lee, . 2016. Molecular Mechanisms and New Treatment Paradigm for Atrial Fibrillation. Circ. Arrhythm. Electrophysiol. 9 :e003721. 10.1161/CIRCEP.115.003721 27162031 Sun, J., and R. MacKinnon. 2017. Cryo-EM Structure of a KCNQ1/CaM Complex Reveals Insights into Congenital Long QT Syndrome. Cell. 169 :1042–1050.e9. 10.1016/j.cell.2017.05.019 28575668 Thulin, E., A. Andersson, T. Drakenberg, S. Forsén, and H.J. Vogel. 1984. Metal ion and drug binding to proteolytic fragments of calmodulin: proteolytic, cadmium-113, and proton nuclear magnetic resonance studies. Biochemistry. 23 :1862–1870. 10.1021/bi00303a043 6722127 Torrente, A.G., R. Zhang, H. Wang, A. Zaini, B. Kim, X. Yue, K.D. Philipson, and J.I. Goldhaber. 2017. Contribution of small conductance K+ channels to sinoatrial node pacemaker activity: insights from atrial-specific Na+/Ca2+ exchange knockout mice. J. Physiol. 595 :3847–3865. 10.1113/JP274249 28346695 Tuteja, D., D. Xu, V. Timofeyev, L. Lu, D. Sharma, Z. Zhang, Y. Xu, L. Nie, A.E. Vázquez, J.N. Young, . 2005. Differential expression of small-conductance Ca2+-activated K+ channels SK1, SK2, and SK3 in mouse atrial and ventricular myocytes. Am J Physiol Heart Circ Physiol. 289 :H2714-23. 10.1152/ajpheart.00534.2005 16055520 Vanommeslaeghe, K., E. Hatcher, C. Acharya, S. Kundu, S. Zhong, J. Shim, E. Darian, O. Guvench, P. Lopes, I. Vorobyov, and A.D. Mackerell Jr. 2010. CHARMM general force field: A force field for drug-like molecules compatible with the CHARMM all-atom additive biological force fields. J. Comput. Chem. 31 :671–690.19575467 Vassilakopoulou, V., B.L. Calver, A. Thanassoulas, K. Beck, H. Hu, L. Buntwal, A. Smith, M. Theodoridou, J. Kashir, L. Blayney, . 2015. Distinctive malfunctions of calmodulin mutations associated with heart RyR2-mediated arrhythmic disease. Biochim. Biophys. Acta. 1850 :2168–2176. 10.1016/j.bbagen.2015.07.001 26164367 Vergara, C., R. Latorre, N.V. Marrion, and J.P. Adelman. 1998. Calcium-activated potassium channels. Curr. Opin. Neurobiol. 8 :321–329. 10.1016/S0959-4388(98)80056-1 9687354 Wang, C., P. Bradley, and D. Baker. 2007. Protein-protein docking with backbone flexibility. J. Mol. Biol. 373 :503–519. 10.1016/j.jmb.2007.07.050 17825317 Wang, K., C. Holt, J. Lu, M. Brohus, K.T. Larsen, M.T. Overgaard, R. Wimmer, and F. Van Petegem. 2018. Arrhythmia mutations in calmodulin cause conformational changes that affect interactions with the cardiac voltage-gated calcium channel. Proc. Natl. Acad. Sci. USA. 115 :E10556–E10565. 10.1073/pnas.1808733115 30348784 Watterson, D.M., F. Sharief, and T.C. Vanaman. 1980. The complete amino acid sequence of the Ca2+-dependent modulator protein (calmodulin) of bovine brain. J. Biol. Chem. 255 :962–975.7356670 Wilde, A.A.M., H. Garan, and P.A. Boyden. 2019. Role of the Purkinje system in heritable arrhythmias. Heart Rhythm. 16 :1121–1126. 10.1016/j.hrthm.2019.01.034 30716412 Xia, X.-M., B. Fakler, A. Rivard, G. Wayman, T. Johnson-Pais, J.E. Keen, T. Ishii, B. Hirschberg, C.T. Bond, S. Lutsenko, . 1998. Mechanism of calcium gating in small-conductance calcium-activated potassium channels. Nature. 395 :503–507. 10.1038/26758 9774106 Xu, Y., D. Tuteja, Z. Zhang, D. Xu, Y. Zhang, J. Rodriguez, L. Nie, H.R. Tuxson, J.N. Young, K.A. Glatter, . 2003. Molecular identification and functional roles of a Ca(2+)-activated K+ channel in human and mouse hearts. J. Biol. Chem. 278 :49085–49094. 10.1074/jbc.M307508200 13679367 Yamoah, M.A., M. Moshref, J. Sharma, W.C. Chen, H.A. Ledford, J.H. Lee, K.S. Chavez, W. Wang, J.E. López, D.K. Lieu, . 2018. Highly efficient transfection of human induced pluripotent stem cells using magnetic nanoparticles. Int. J. Nanomedicine. 13 :6073–6078. 10.2147/IJN.S172254 30323594 Yarov-Yarovoy, V., J. Schonbrun, and D. Baker. 2006. Multipass membrane protein structure prediction using Rosetta. Proteins. 62 :1010–1025. 10.1002/prot.20817 16372357 Yarov-Yarovoy, V., P.G. DeCaen, R.E. Westenbroek, C.Y. Pan, T. Scheuer, D. Baker, and W.A. Catterall. 2012. Structural basis for gating charge movement in the voltage sensor of a sodium channel. Proc. Natl. Acad. Sci. USA. 109 :E93–E102. 10.1073/pnas.1118434109 22160714 Yu, C.C., J.S. Ko, T. Ai, W.C. Tsai, Z. Chen, M. Rubart, M. Vatta, T.H. Everett IV, A.L. George Jr., and P.S. Chen. 2016. Arrhythmogenic calmodulin mutations impede activation of small-conductance calcium-activated potassium current. Heart Rhythm. 13 :1716–1723. 10.1016/j.hrthm.2016.05.009 27165696 Zhang, Q., V. Timofeyev, L. Lu, N. Li, A. Singapuri, M.K. Long, C.T. Bond, J.P. Adelman, and N. Chiamvimonvat. 2008. Functional roles of a Ca2+-activated K+ channel in atrioventricular nodes. Circ. Res. 102 :465–471. 10.1161/CIRCRESAHA.107.161778 18096820 Zhang, X.D., D.K. Lieu, and N. Chiamvimonvat. 2015. Small-conductance Ca2+-activated K+ channels and cardiac arrhythmias. Heart Rhythm. 12 :1845–1851. 10.1016/j.hrthm.2015.04.046 25956967 Zhang, Z., H.A. Ledford, S. Park, W. Wang, S. Rafizadeh, H.J. Kim, W. Xu, L. Lu, V.C. Lau, A.A. Knowlton, . 2017. Distinct subcellular mechanisms for the enhancement of the surface membrane expression of SK2 channel by its interacting proteins, α-actinin2 and filamin A. J. Physiol. 595 :2271–2284. 10.1113/JP272942 27779751 Zhang, X.D., Z.A. Coulibaly, W.C. Chen, H.A. Ledford, J.H. Lee, P. Sirish, G. Dai, Z. Jian, F. Chuang, I. Brust-Mascher, . 2018. Coupling of SK channels, L-type Ca2+ channels, and ryanodine receptors in cardiomyocytes. Sci. Rep. 8 :4670. 10.1038/s41598-018-22843-3 29549309 Zühlke, R.D., G.S. Pitt, K. Deisseroth, R.W. Tsien, and H. Reuter. 1999. Calmodulin supports both inactivation and facilitation of L-type calcium channels. Nature. 399 :159–162. 10.1038/20200 10335846 Zühlke, R.D., G.S. Pitt, R.W. Tsien, and H. Reuter. 2000. Ca2+-sensitive inactivation and facilitation of L-type Ca2+ channels both depend on specific amino acid residues in a consensus calmodulin-binding motif in the(α)1C subunit. J. Biol. Chem. 275 :21121–21129. 10.1074/jbc.M002986200 10779517

PMC007xxxxxx/PMC7695126.txt

==== Front Bioinformatics Bioinformatics bioinformatics Bioinformatics 1367-4803 1367-4811 Oxford University Press 32637989 10.1093/bioinformatics/btaa588 btaa588 Original Papers Sequence Analysis AcademicSubjects/SCI01060 TreeSAPP: the Tree-based Sensitive and Accurate Phylogenetic Profiler http://orcid.org/0000-0002-6714-0898 Morgan-Lang Connor Graduate Program in Bioinformatics, University of British Columbia, Genome Sciences Centre, Vancouver, British Columbia V5Z 4S6, Canada McLaughlin Ryan Graduate Program in Bioinformatics, University of British Columbia, Genome Sciences Centre, Vancouver, British Columbia V5Z 4S6, Canada Armstrong Zachary Genome Science and Technology Program, University of British Columbia, Vancouver, BC V6T 1Z4, Canada Zhang Grace Department of Electrical and Computer Engineering, University of British Columbia, Vancouver BC V6T 1Z4, Canada Chan Kevin Department of Electrical and Computer Engineering, University of British Columbia, Vancouver BC V6T 1Z4, Canada Hallam Steven J Graduate Program in Bioinformatics, University of British Columbia, Genome Sciences Centre, Vancouver, British Columbia V5Z 4S6, Canada Genome Science and Technology Program, University of British Columbia, Vancouver, BC V6T 1Z4, Canada Department of Electrical and Computer Engineering, University of British Columbia, Vancouver BC V6T 1Z4, Canada Department of Microbiology and Immunology, University of British Columbia, 2552-2350 Health Sciences Mall, Vancouver, British Columbia V6T 1Z3, Canada ECOSCOPE Training Program, University of British Columbia, Vancouver, British Columbia V6T 1Z, Canada Elofsson Arne Associate Editor To whom correspondence should be addressed. E-mail: shallam@mail.ubc.ca Present address: Structural Biology Laboratory, Department of Chemistry, The University of York, York YO10 5DD, UK 15 9 2020 08 7 2020 08 7 2020 36 18 47064713 20 1 2020 11 6 2020 30 6 2020 © The Author(s) 2020. Published by Oxford University Press. 2020 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Abstract Motivation Microbial communities drive matter and energy transformations integral to global biogeochemical cycles, yet many taxonomic groups facilitating these processes remain poorly represented in biological sequence databases. Due to this missing information, taxonomic assignment of sequences from environmental genomes remains inaccurate. Results We present the Tree-based Sensitive and Accurate Phylogenetic Profiler (TreeSAPP) software for functionally and taxonomically classifying genes, reactions and pathways from genomes of cultivated and uncultivated microorganisms using reference packages representing coding sequences mediating multiple globally relevant biogeochemical cycles. TreeSAPP uses linear regression of evolutionary distance on taxonomic rank to improve classifications, assigning both closely related and divergent query sequences at the appropriate taxonomic rank. TreeSAPP is able to provide quantitative functional and taxonomic classifications for both assembled and unassembled sequences and files supporting interactive tree of life visualizations. Availability and implementation TreeSAPP was developed in Python 3 as an open-source Python package and is available on GitHub at https://github.com/hallamlab/TreeSAPP. Supplementary information Supplementary data are available at Bioinformatics online. US Department of Energy (DOE) Joint Genome Institute, an Office of Science User Facility Office of Science of the US Department of Energy DE-AC02-05CH11231 Facilities Integrating Collaborations for User Science (FICUS) JGI 10.13039/100015911 NERSC Natural Sciences and Engineering Research Council of Canada 10.13039/501100000038 Genome British Columbia; Genome Canada Compute Canada 10.13039/100013020 Koonkie Inc ==== Body pmc1 Introduction We live in a world dominated by prokaryotic (archaea and bacteria) microorganisms. The collective properties of this unseen majority have an enormous impact on the world, driving matter and energy transformations through networks of metabolite exchange (Canfield et al., 2010; Falkowski et al., 2008; Hurwitz and Sullivan, 2013). Over geological time, these interactions have fundamentally transformed the surface chemistry of the earth and continue to shape elemental fluxes between atmospheric, terrestrial and aquatic compartments of the biosphere. A versatile set of functional genes that evolved early in the history of life are responsible for driving biogeochemical processes through largely defined metabolic pathways (Falkowski et al., 2008). Genes encoding diagnostic steps in these pathways (e.g. functional anchors) can be assembled into a phylogenetic framework that provides information on the distribution, abundance and taxonomic origin of environmental sequences. However, charting the diversity and distribution patterns of functional and taxonomic anchor genes is limited by our inability to quantitatively resolve microbial communities within a standard taxonomic hierarchy. Despite a plethora of published software (Boyd et al., 2018; Buchfink et al., 2015; Darling et al., 2014), taxonomic assignment remains an unsolved problem (Peabody et al., 2015). Many factors must be considered when interpreting taxonomic profiles but perhaps the most vexing are legacy misclassifications in biological sequence databases, which are particularly difficult to predict and fix (Kozlov et al., 2016; Merchant et al., 2014; Nasko et al., 2018). The potential for misclassification is exacerbated by the recent inclusion of composite metagenome-assembled genomes (MAGs) in such databases (Shaiber and Eren, 2019). Moreover, despite enormous progress in high-throughput cultivation (Cross et al., 2019; Nichols et al., 2010) the majority of microorganisms from natural and engineered environments remain to be isolated in a laboratory setting (Rappé and Giovannoni, 2003; Solden et al., 2016; Steen et al., 2019). This ‘cultivation gap’ poses a formidable issue for taxonomic assignment as many sequences derived from the environment are distantly related to representatives in sequence databases that underpin taxonomic classification software. Single-cell amplified genomes (SAGs) have the potential to bridge the gap by linking individual, environmental genotypes to specific taxonomic labels (Rinke et al., 2013). However, their utility in annotation pipelines has yet to be realized. Common methods for taxonomic classification include pairwise sequence alignment (Altschul et al., 1990; Buchfink et al., 2015), k-mer matching (Kim et al., 2016; Ondov et al., 2016) and phylogenetic placement (Barbera et al., 2019; Berger and Stamatakis, 2011; Matsen et al., 2010; Stark et al., 2010). These methods rely on indexed files from either custom, curated sequence datasets or massive repositories. Distance between query and reference sequences can be estimated using sequence similarity, evolutionary distance or derivatives thereof. Current taxonomic assignment methods do not factor these measures into classification, instead using either a lowest common ancestor (LCA), best-hit or ensemble approach to yield a single taxonomic label (Hanson et al., 2016; Huson et al., 2007; Konwar et al., 2013). This frequently leads to over-classification in the presence of unrepresented taxa. Phylogenetic placement methods are well-equipped to handle gene-centric assignment because the branch length distances between query and related reference sequences serve as a coordinate system when estimating taxonomic relationships (Ciccarelli et al., 2006). Applications that calibrate taxonomic rank thresholds to a phylogeny’s branch lengths have been developed (Parks et al., 2018; Wu et al., 2013). As part of the Genome Taxonomy Database (GTDB), nodes in reference phylogenies are calibrated by their relative evolutionary distance (RED) values (Chaumeil et al., 2019). However, the GTDB toolkit is not intended for gene-centric taxonomic assignment but for classifying genomes by placing concatenated single-copy marker genes sequences into pre-computed reference trees. Alternatively GraftM can use pplacer for phylogenetic placement of query sequences into gene trees but stops short of using evolutionary distance to correct for over-classification and annotates query sequences using one reference package at a time (Boyd et al., 2018). Here, we present the Tree-based Sensitive and Accurate Phylogenetic Profiler (TreeSAPP), a gene-centric functional and taxonomic classification tool able to classify proteins and assembled or unassembled nucleotide sequences. Fragments per Kilobase per Million reads values can be calculated and used in all outputs for genomic or transcriptomic data with short-read sequences in FASTQ format. By using RAxML’s evolutionary placement algorithm (EPA), TreeSAPP leverages the evolutionary distance to reference sequences (EDR), a sum of the distal and pendant lengths as well as the mean length from the placement edge to all descendent leaf tips, correlated with taxonomic rank to recommend optimal taxonomic ranks for more accurate classifications. Structural and metabolic feature information can be annotated in the interactive tree of life (iTOL) and exported to allow for improved functional annotation of query sequences (Letunic and Bork, 2016). In addition to a classification table, placement outputs are compatible with iTOL for easy creation of publication-quality figures (Fig. 1). Fig. 1. The workflow of the current study. Sequences for building reference packages were sourced from the NCBI and FunGene databases. Sequences were downloaded from EggNOG for validating reference packages and benchmarking TreeSAPP against GraftM. IMG/M metagenomes were used to explore the global diversity of Mcr 2 Materials and methods TreeSAPP was developed in Python 3. It uses the Environment for Tree Exploration Python package for many tree manipulation operations (Huerta-Cepas et al., 2016b). The multiprocessing Python package is used to run executable processes in parallel unless it is more efficient to use the software’s native parallelization capabilities. BioPython is used for downloading lineage information from Entrez’s taxonomy database for each reference sequence when building new reference packages (Cock et al., 2009). Sequences for the reference packages described in this manuscript were primarily downloaded from the FunGene repository as this resource provides curated sequences for many of the functional anchor genes involved in major biogeochemical cycles (Fish et al., 2013) (Supplementary Table S1). Additionally, to provide the most comprehensive reference trees, newly published sequences that were not included in FunGene 9.6 were included from GenBank and the Joint Genome Institute’s Integrated Microbial Genomes and Microbiomes (IMG/M) (Borrel et al., 2019; Hua et al., 2019; McKay et al., 2019; Seitz et al., 2019; Wang et al., 2019). All benchmarking was performed on a server with a 20 physical core (40 virtual) Intel Xeon CPU (E5-2650 v3), 264 GB of RAM and a 5 TB HDD. The operating system was Red Hat Enterprise Linux Server release 7.5 (Maipo). 2.1 Building reference packages with TreeSAPP create TreeSAPP’s classification workflow requires a multiple sequence alignment (MSA), profile hidden Markov model (HMM), taxonomic lineages and phylogenetic tree for all reference sequences (Supplementary Fig. S1). Together these files constitute a reference package, built using treesapp create, and allow for rapid query sequence filtering, phylogenetic placement and taxonomic classification. Construction began with removing truncated sequences using either a provided HMM or a database’s filter prior to downloading. An empirically determined threshold of 60% profile HMM coverage was required of candidate reference sequences to balance inclusivity and profile quality. NCBI taxonomic lineages were then downloaded for each candidate reference sequence with a valid accession. Lineage information of unaccessioned sequences can be provided via either a table or the FASTA file using a custom header format. Optional taxonomy-based filtering was performed (e.g. to remove viral or eukaryotic sequences). This set can then be clustered using USEARCH at a specified proportional similarity (Edgar, 2010). Non-homologous (i.e. outlier or mis-annotated) sequences were identified by OD-Seq and removed (Jehl et al., 2015). These automatically curated sequences were used to generate the reference MSA with MAFFT’s -‘auto’ algorithm (Katoh and Standley, 2013). The resulting MSA was used by HMMER’s ‘hmmbuild’ module to build a new profile HMM (Eddy, 1998). Before tree construction, the MSA was optionally trimmed using BMGE with the least conserved matrix, BLOSUM30 for proteins or PAM100 otherwise, to decrease the runtime required for phylogeny inference by removing non-conserved positions (Criscuolo and Gribaldo, 2010). By default, RAxML is used for phylogenetic inference, automatically using the optimal substitution model (invoked with the -PROTGAMMAAUTO flag) and the minimum number of bootstraps necessary (using -‘autoMR’) (Pattengale et al., 2010; Stamatakis, 2006). Yet, in light of cursory taxonomic classification performance results indicating little difference between RAxML and FastTree (Supplementary Fig. S6) with a drastic difference in compute time, FastTree is available as well (Price et al., 2010). In this case, a bootstrapped tree will not be generated and the LG amino acid substitution model is used unless the user specifies differently (Le and Gascuel, 2008). To assign taxonomy at the most appropriate rank given a placement’s distance, a linear correlation of taxonomic rank with EDR is estimated for each reference package. Briefly, all sequences from a taxonomic group were removed from the reference tree before clustered sequences from the initial input FASTA file were mapped back to the tree using EPA. The placement distances were calculated and recorded before the next iteration involving a different taxon until all possible taxa have been exhaustively placed. Outliers are then removed from these data and rarefied to improve normality across the ranks before a linear model is fit. 2.2 Testing classification performance with TreeSAPP evaluate TreeSAPP’s classification performance was evaluated using clade exclusion analysis (described below) and the binary classification metric, Matthews’ correlation coefficient (MCC) (Matthews, 1975 (Supplementary Equation S1). TreeSAPP was compared to GraftM and DIAMOND, as implemented in GraftM, using 15 taxonomic and 12 functional anchor reference packages (Supplementary Table S1). Functional anchor reference packages were only used during clade exclusion analysis as their representation in the MCC test data (EggNOG) was too limited and potentially a source of bias. To consistently compare between methods, all performance analyses relied on the NCBI’s taxonomic hierarchy for determining each sequence’s optimal taxonomic assignment: the highest resolution taxonomic classification with respect to the taxonomic composition of the reference package (i.e. the LCA between the query and most closely related reference). Under these conditions, ‘taxonomic distance’ is defined as the number of ranks separating the LCA of the optimal taxonomic assignment and the taxon assigned by the software (Supplementary Fig. S4). Given there are eight conventional taxonomic ranks used in the classification of Bacteria and Archaea, the maximum taxonomic distance is eight while a perfect classification has a taxonomic distance of zero. Classification performance was measured utilizing 15 universal single-copy taxonomic anchor genes and the EggNOG database (v4.5.1) (Supplementary Table S1), with reference sequences from Bacteria and Archaea but excluding Eukaryotes to simplify the hierarchy. In addition to RecA, RadA and RpoB, nearly ubiquitous taxonomic anchor genes used by the GTDB were identified as being present in at least 90% of Archaea and Bacteria (Parks et al., 2018). Out of this set, 12 were selected that also had entries in the PFam database and were therefore easily accessed. Sequences that were either not in the PFam database or did not have an obvious corresponding orthologous group in EggNOG were omitted. For each taxonomic distance between zero and eight, sequence classifications matching their EggNOG annotations within the taxonomic distance threshold were counted as true positives. Sequences that were classified as taxa outside of the threshold or as the wrong gene were counted as false positives. EggNOG sequences that were orthologous to any of the 12 reference packages but were not classified counted as false negatives while all remaining sequences were true negatives. Classification performance was determined using MCC due to its ability to report reasonable values even with very different class sizes (Boughorbel et al., 2017). Taxonomic classification accuracy was estimated by clade exclusion analysis using functional anchor reference packages for dissimilatory sulphite reductase alpha and beta subunits (DsrAB), methyl coenzyme-M reductase alpha, beta and gamma subunits (McrA, McrB and McrG), periplasmic nitrate reductase (NapA), NO-forming nitrite reductase (NirK and NirS), nitrogenase molybdenum–iron protein alpha chain(NifD), nitric oxide reductase subunit B (NorB), nitrite oxidoreductase subunits A and B (NxrA and NxrB) and a combined particulate methane monooxygenase and ammonia monooxygenase (PmoA/AmoA). Clade exclusion analysis measures classification performance in scenarios where the query sequences lack a close relative in the reference set, as is typical during taxonomic classification of metagenomes (Peabody et al., 2015) (Supplementary Fig. S3). The analysis required reference sequences with taxonomic lineages completely resolved to the rank evaluated. Sequences not resolved to the rank being evaluated (e.g. Archaea; Euryarchaeota; environmental samples for Class) were removed as their specific taxonomic relationship to other reference sequences was unknown. For each taxonomic rank tested, representative sequences were selected for each taxon (an arbitrary maximum of five so as to not introduce lineage-specific bias into the final estimate) and sequences belonging to that taxon were removed from the reference package. These representative sequences were then classified and compared to their optimal taxonomic assignment. Distances from their optimal taxonomic assignments were tabulated and reference sequences were returned to the reference package before the next taxon was tested. Only taxa that shared a common ancestor at the rank tested with one or more remaining reference sequences were evaluated. For example, evaluating the ability to classify at the rank of Class using sequences belonging to the order Methanosarcinales would require the reference package to contain other members of the parent taxonomic class Methanomicrobia that are not Methanosarcinales, such as Methanomicrobiales. This allows the software to optimally classify the sequences as Methanomicrobia. 2.3 Classifying query sequences with TreeSAPP assign TreeSAPP begins by predicting open reading frames (ORFs) using Prodigal v2.6.3 if the inputs are nucleotide sequences, otherwise this step is skipped (Supplementary Fig. S2) (Hyatt et al., 2010). Resulting ORFs are conceptually translated into proteins that are then aligned to curated reference sequences using hmmalign, within the HMMER v3.1 package (Eddy, 1998). Homologous sequences are then extracted and mapped onto the reference MSA with hmmalign. Optionally, alignments are trimmed using BMGE to remove non-conserved positions from the alignment file and reduce computation time (Criscuolo and Gribaldo, 2010). BMGE uses the BLOSUM30 substitution matrix for protein sequences, as recommended by Tan et al. (2015). RAxML-EPA places the query sequences in the reference tree by finding the optimal phylogeny likelihood with the sequence inserted (Berger and Stamatakis, 2011). Query sequence placements are filtered by evolutionary distance and likelihood weight ratio prior to predicting taxonomy at the linear-model recommended rank. The complete set of ORF sequences predicted by Prodigal (both nucleotide and amino acid forms), a FASTA file containing only the classified ORFs and a classification table with taxonomy and abundance information are included as outputs. JPlace files, containing reference tree placement coordinates for all queries classified, are additionally provided for each reference package. Along with colour and style information files made for a specific phylogeny, these placements may be visualized in iTOL (Letunic and Bork, 2019). 3 Results Many taxonomic assignment software methods are evaluated by a mock community of known, but severely limited diversity. So while these analyses are useful, their inherent oversimplification prohibits their performance estimates from being extensible to natural and engineered microbial communities. This is especially perplexing for gene-centric annotation software, where the total diversity of query sequences is reduced compared to whole-genome binning and profiling tools. Efforts are being made to provide well-designed mock communities for metagenome analysis, including taxonomic assignment, though they are still limited in their diversity (Sczyrba et al., 2017). Therefore, we decided to compare the taxonomic classification performance of TreeSAPP to GraftM and DIAMOND using the large, but still well curated, EggNOG database (v4.5) (Huerta-Cepas et al., 2016a) comprised of 2031 organisms for universal taxonomic anchor genes. Additionally, we used treesapp evaluate to simulate reference packages that do not represent taxa from Species to Class-level sequence divergence to determine how each tool classifies well and poorly represented query sequences. Finally, TreeSAPP was used to profile all metagenome-derived proteins in the IMG/M database. 3.1 Performance Classification and runtime performance of TreeSAPP were benchmarked against GraftM with the default hmmsearch search strategy and pplacer placement method as well as DIAMOND-mode as a proxy for an alignment-based search and classification strategy (Boyd et al., 2018; Buchfink et al., 2015). Classifying the EggNOG database (version 4.5.0) (Huerta-Cepas et al., 2016a) with 12 single-copy taxonomic anchors revealed TreeSAPP’s overall classification performance was, with little exception, preferable to that of either a pairwise alignment strategy or GraftM (Fig. 2 and Supplementary Table S2). This is due in large part to TreeSAPP’s evolutionary distance-based filtering thresholds that favourably remove false positives without increasing false negatives, thereby increasing precision; GraftM and DIAMOND accrued over 3000 more false positives than TreeSAPP. Notably, this performance was achieved only using RAxML version 8.2.12 as we found pendant length distances were estimated accurately beginning with this version. Based on the performance of TreeSAPP using alignment trimming but assigning taxonomy to queries only by LCA (TreeSAPP-BMGE-Raw), adjusting a query’s assigned taxonomic rank by its EDR is worthwhile. However, relative to DIAMOND and GraftM, TreeSAPP’s recall was not as strong, mostly stemming from the initial HMM search, missing 1775 EggNOG sequences versus 561 and 789 for GraftM and DIAMOND at the most relaxed taxonomic rank distance allowed, respectively (Supplementary Fig. S5 and Table S2). The vast majority of false negatives were accounted for by Eukaryotic sequences and this is sensible given the reference packages were built using only sequences from Bacteria and Archaea. Only 7.1% of unclassified sequences were bacterial or archaeal in origin. Still, 5420 Eukaryotic sequences were correctly classified suggesting TreeSAPP is able to accurately identify very distantly related homologous sequences. False negatives were unevenly distributed across the tested reference packages with over 60% from just 3 (minimum =1 from Ribosomal S3Ae family, maximum =653 from Ribosomal protein L1p/L10e family, median =23). Fig. 2. Classification performance of TreeSAPP, GraftM and DIAMOND as evaluated by the MCC. TreeSAPP was run both with (TreeSAPP-BMGE) and without MSA trimming using BMGE. TreeSAPP-BMGE-Raw represents the classification performance of TreeSAPP with BMGE but without the linear-model-based rank recommendation. Distance from optimal rank is the accepted taxonomic distance in order for a classified sequence to be considered a true positive. Sequences that failed to meet the distance from optimal rank were included in the MCC calculation as false positives Clade exclusion analysis was used to dually determine the accuracy of these methods using functional anchors with less congruent phylogenetic and taxonomic relations as compared to the ribosomal proteins, and how these methods perform when databases lack reference sequences that are closely related to query sequences. Query sequences were downloaded from FunGene version 9.6 and clustered at 99% similarity with USEARCH (Edgar, 2010). EggNOG was not used in this case as its taxonomic breadth of 2031 organisms, many of which do not contain any of the functional markers, was deemed inadequate for reference package construction. Using treesapp evaluate, iterative clade exclusion analyses were performed for every testable taxon and each reference package was independently analysed. Classification performance was variable, but not considerably so, across the reference packages tested. McrA tended to yield the best assignments while PmoA/AmoA and DsrAB consistently performed poorly. GraftM and TreeSAPP both perform much better than DIAMOND when classifying divergent sequences (Class-, Order- and Family-level relations to closest relative in reference set) but comparable when query sequences were similar to the reference set (Fig. 3). Most of TreeSAPP’s classifications were <2 taxonomic ranks away from the optimal assignment on average, regardless of sequence divergence, and sequences were classified with approximately equivalent accuracy across taxonomic ranks. Since false negatives were unavailable for this analysis the F1 score, the harmonic mean of precision and recall, was used to compare across all reference packages (Supplementary Fig. S7). The difference in each of the methods’ ability to handle distantly related sequences is emphasized in these values. All did comparably well when classifying queries closely related to reference sequences (F1 scores between 0.71 and 0.92) but varied significantly for distantly related queries (between 0.04 and 0.8). Fig. 3. Average taxonomic distance across all taxa evaluated for 12 functional anchors. Colours correspond to the taxonomic rank evaluated by clade exclusion analysis and serve as a proxy for sequence divergence. Dashes along the y-axis show the distribution of points on a single plane. P_amoA is a reference package with sequences containing both PmoA and AmoA Clade exclusion analysis was used again to confirm the results observed by classifying the taxonomic anchor genes in EggNOG. The mean distances and F1 scores were consistent with the functional anchor genes, with a few reference packages performing poorly (Supplementary Figs S8 and S9). This could not be completed for DIAMOND and GraftM as one of its dependencies for building reference packages, Taxtastic (https://github.com/fhcrc/taxtastic), would frequently timeout. Classification and JPlace files can be used for purposes beyond taxonomic and functional profiling; TreeSAPP can also discover genes, reactions and pathways associated with organismal genomes, SAGs or MAGs. To evaluate this aspect of the TreeSAPP pipeline, we developed a use case for McrA, because this reference package performed the best during benchmarking. The mcrA gene along with mcrB and mcrG encode the holoenzyme mediating the terminal step in biological methane production, though it is also capable of binding and activating methane in the anaerobic oxidation of methane and more recently may be used in the oxidation of short-chain alkanes (Laso-Pérez et al., 2016). The McrA phylogeny has been shown to be fairly congruent with both small subunit ribosomal RNA (SSU or 16S rRNA) gene and concatenated marker gene phylogenies, with limited recorded instances of lateral gene transfer, and is therefore more likely to perform well as a taxonomic anchor (Evans et al., 2019; Springer et al., 1995). Moreover, methane-metabolism pathway information has been provided for each known clade and expected to be conserved, making accurate metabolic inferences directly from phylogenetic placement possible (Evans et al., 2019). 3.2 Global McrA survey TreeSAPP was used to find and classify all McrABG sequences from the Joint Genome Institute’s IMG/M database image from January 10, 2017. There were 7.715×109 (∼1.2 TB) of putative amino acid sequences included in this analysis and 14 919 McrA, 11 825 McrB and 8609 McrG sequences were taxonomically classified. A total of 816 metagenomes were found to contain at least 1 of the 3 Mcr subunits. Sequences shorter than 84 AA, corresponding to the first quantile of length-sorted sequences, were removed to mitigate incorrect classifications leaving 11 256 McrA. About 58% (6533) of the McrA sequences were classified at the rank of Genus or Species and only 3.6% were classified as Archaea (Supplementary Fig. S10). Methanomicrobia and Methanobacteria were the most sampled classes with 5505 and 1572, respectively. Methanoculleus, Methanobrevibacter, Methanobacterium and Methanoregula were the most common genera accounting for 2891 sequences. Of all the McrA sequences classified 4251 sequences were resolved to at least Phylum (to reduce the chance they were false positives) and no further than Family. Even though 2764 of these were classified to the NCBI’s Phylum rank as either ‘environmental samples’ or ‘metagenomes’, they can still be considered novel with respect to cultured archaea. These novel sequences were from 432 metagenomes, primarily represented by wetland and hydrothermal vent communities. Strikingly, 89% of the 806 McrA sequences from hydrothermal vent metagenomes were considered novel. While the number of uniquely novel sequences was not determined, updating the reference tree after clustering all sequences at 97% similarity added 412 leaves, an increase of 180%. Additionally, the McrA tree was annotated with methane-metabolism pathway information for each known clade (Borrel et al., 2014, 2019; Evans et al., 2019; Whitman et al., 2006) (Supplementary Fig. S11) and this was used to provide metabolic labels for an additional 10 839 sequences. A metabolic label was not assigned to sequences placed at a node where descendants possess multiple metabolic labels. CO2-dependent hydrogenotrophy was the most common metabolism by a large margin and methylotrophy was the least common. There were no signs of mutual exclusion among metabolisms or between metabolisms and broad ecosystem categories but some trends were identified for specific ecosystem sub-types. Hydrothermal vents, oil seeps, thermal springs and an asphaltene lake harboured relatively abundant McrA associated with short-chain alkane-oxidizing and anaerobic methanotrophic Archaea. Only thermal springs also hosted an abundant methanogenic population. McrA associated with CH3-dependent hydrogenotrophic methanogenesis were most prevalent in digestive systems and other environments rich in organic matter. The most aceticlastic McrA were found in soil and freshwater environments as well as anaerobic digesters. Together, using TreeSAPP’s phylogenetically derived taxonomic classifications, these metagenomic data indicate that there are plenty of novel methanogenic and short-chain alkane-oxidizing Archaeal lineages that remain to be described. 4 Discussion TreeSAPP is a functional and taxonomic annotation software that uses phylogenetic placement for accurate classifications. It is readily able to classify sequences derived from organismal genomes, environmental genomes (SAGs, MAGs) and metagenomes—even those that are distantly related to reference genomes present in contemporary databases. Moreover, it capitalizes on the phylogenetic framework to transitively assign taxonomic and functional feature information. In cases of complex evolutionary histories, internalizing these features in the reference package, by annotating clades, ensures that distantly related genes are not mis-annotated, thereby reducing false discovery. TreeSAPP was designed to integrate with iTOL as well as biological sequence databases so sequences can be easily linked to their respective taxonomic lineages. We are looking to expand support for databases beyond EggNOG and Entrez so users can more easily create reference packages. Moreover, to keep reference packages as current as possible, query sequences that meet profile HMM proportion thresholds and are deemed sufficiently divergent from current reference sequences are used to readily rebuild reference MSA, profile HMM and tree using treesapp update. We do not plan to regularly update all reference packages centrally. Rather, in their current state, these are meant to be used as mutable objects that users can update as required to suit their unique efforts (Fig. 4). Fig. 4. Phylogenetic and metabolic analysis of IMG metagenome-derived McrA sequences. (A) All predicted metagenome-derived McrA sequences (14 919) from IMG/M (as of January 10, 2017) were classified using TreeSAPP and visualized in iTOL. The tree shown here contains 228 reference McrA sequences including most newly described lineages from the “divergent McrA” clade hypothesized to be involved in oxidizing higher alkanes. A version of the tree with leaf labels is available as Supplementary Figure S12. (B) Proportion of sequences assigned at each taxonomic rank. (C) Putative methanogenesis and methanotrophic metabolisms supported in each ecosystem category as inferred by their placement on the reference McrA tree. Sequences that mapped deeply and converge across multiple annotated metabolisms were omitted 4.1 Performance Classification performance analyses indicate that TreeSAPP was better at both identifying and assigning taxonomy to query protein sequences than DIAMOND and GraftM, especially when query sequences were novel with respect to the reference sequences. The MCC values generated using EggNOG showed TreeSAPP’s taxonomic classifications were better than both GraftM and DIAMOND, though this was only achieved with taxonomic rank recommendation from linear models. Even still, it is important to point out that the pairwise alignment performance of DIAMOND shown here is not extensible to pairwise alignment in general because databases tend to be more comprehensive than reference packages used in this study. Unfortunately, we were unable to compare our classification results to those of the critical assessment of metagenome interpretation project since their performance summaries were based on classifications of DNA sequences (Sczyrba et al., 2017). A recent meta-analysis of taxonomic classification tools by Ye et al. (2019), though not directly comparable, produced F1 scores similar to our analyses for DIAMOND at the Genus and Species ranks (ranging from 0.1 to 0.4, Supplementary Fig. S7) indicating that both GraftM and TreeSAPP would outperform their tested ‘DNA-to-protein’ classifiers (tools that classify DNA sequences using protein databases). Through these analyses, we also found classification precision to vary by reference package. All classifiers struggled to accurately assign taxonomy with the DsrAB reference package [included reductive and oxidative forms of DsrA and DsrB, and validated with sequences from Müller et al. (2015)]. DsrAB is a composite reference package containing the homologous subunits DsrA and DsrB, anaerobic sulphite reductase subunit C and a nitrite and sulphite reductase clade of Euryarchaeota. Including homologous but functionally diverse genes in a single phylogeny can safeguard from erroneous functional attributions by performing a second classification with additional tree decoration information, as implemented in treesapp layer, but may result in less accurate taxonomic assignments (Supplementary Methods). More research into building reference packages for complex gene families is needed. 4.2 Application TreeSAPP was used to classify all metagenome-derived McrABG subunits in IMG/M. It identified several thousand novel sequences (i.e. not represented by either a Genus or Species) with respect to our contemporary, metabolically annotated reference McrA tree. Specifically, Gulf of California hydrothermal vent samples contained diverse methanotrophic and short-chain alkane-oxidizing Archaea, of which nearly 90% do not have a Genus-level representative. We also estimated the relative abundance of McrA associated with known methanogenic and alkanotrophic metabolisms, identifying CO2-dependent hydrogenotrophy as vastly more common than any other on a per-ORF basis. This use case illustrates the power of TreeSAPP in identifying new lineages and resolving quantitative functional differences between locations. This rich dataset is ripe for further diversity and correlation-based analyses to inform future sequencing and cultivation efforts. 4.3 Future development In the process of developing and testing TreeSAPP several potential areas of improvement were recognized. TreeSAPP is slower and requires more RAM than GraftM (Supplementary Fig. S13) for a number of reasons. Intermediate files are written so runs can be restarted from checkpoints at the cost of more time spent performing I/O operations. TreeSAPP uses Prodigal for ORF prediction and conceptual translation (Hyatt et al., 2010), while GraftM employs the simple and significantly faster OrfM (Woodcroft et al., 2016). TreeSAPP uses one HMM for both search and profile multiple alignments. This is less sensitive than GraftM’s search-specific HMM where sequences were taxonomically deduplicated before building (data not shown). Adopting this strategy could be beneficial if it does not increase the false positive rate. Finally, RAxML’s EPA is used for phylogenetic placement, instead of the faster pplacer. While neither tool scales particularly well on a single compute node (Supplementary Fig. S14) future versions of TreeSAPP will see marked improvements in sequential and parallel computing efficiency by adopting the faster RAxML-NG and EPA-NG (Barbera et al., 2019; Kozlov et al., 2019). Several developments in phylogenetic placement have been published recently that may lead to further improvements in efficiency and accuracy (Barbera et al., 2019; Czech et al., 2019). Among them is hierarchical phylogenetic placement, which involves placing query sequences onto a taxonomically broad and sparse backbone tree with subsequent placement onto high-resolution phylogenies based on the backbone edge position. This method has the potential to increase phylogenetic placement precision while reducing computational requirements (Czech et al., 2019). Moreover, new phylogenetically informed Bacterial and Archaeal taxonomic hierarchies are being introduced (Parks et al., 2018). Leveraging a principled taxonomic framework based on phylogenetic relationships will likely improve the taxonomic classifications of all phylogenetic methods. Modelling the relationship between taxonomic rank and EDR for rank recommendation benefits taxonomic classification performance (Fig. 2). It is most beneficial when classifying distantly related query sequences that map to either a leaf’s edge or a sparse clade resulting in a shallow (Species or Genus) LCA. However, generating the EDR data for training is currently a slow, iterative process. Moreover, a reliable model is not created if there is insufficient taxonomic redundancy within reference sequences (i.e. a single Order is representing a Class, so removing that Order removes the entire Class and next closest ancestor is at Phylum). Using distances directly from the tree, as was used in the GTDB-Tk (Chaumeil et al., 2019) with RED, would accelerate this process and benefit consistency. 5 Conclusion We have developed a functional and taxonomic annotation software, TreeSAPP, with improved classification performance based on regression of evolutionary distances and taxonomic ranks to recommend more accurate taxonomic assignments. TreeSAPP is able to provide quantitative functional and taxonomic information for both assembled and unassembled sequences, classification tables and files supporting interactive iTOL visualizations. Using TreeSAPP, we explored the global distribution of McrA and recovered many new sequences associated with methane metabolizing archaea. With an expanded set of reference packages under development TreeSAPP will support community-driven taxonomic assignment and metabolic reconstruction on a truly global scale. Supplementary Material btaa588_supplementary_data Click here for additional data file. Acknowledgements We would like to thank Tanya Woyke, Natalia Ivanova and Frederik Schulz at the Joint Genome Institute (JGI; Berkeley, California) and Daniel Udwary for providing metagenomic data from IMG/M and assistance with the National Energy Research Scientific Computing (NERSC) Center systems. Many thanks to Simonetta Gribaldo and Guillaume Borrel at the Institut Pasteur (Paris) for validating metabolic annotations and recommending BMGE. We also thank Kishori Konwar for formative conversations and preliminary code used to guide early TreeSAPP development cycles. Funding This work was performed under the auspices of the US Department of Energy (DOE) Joint Genome Institute, an Office of Science User Facility, supported by the Office of Science of the US Department of Energy under Contract DE-AC02-05CH11231 through the Facilities Integrating Collaborations for User Science (FICUS) initiative between JGI and NERSC; the Natural Sciences and Engineering Research Council of Canada; Genome British Columbia; Genome Canada; and Compute Canada through grants awarded to S.J.H. is a co-founder of Koonkie Inc., a bioinformatics consulting company that designs and provides scalable algorithmic and data analytics solutions in the cloud. Conflict of Interest: none declared. ==== Refs References Altschul S.F.  et al (1990) Basic local alignment search tool. J. Mol. Biol., 215 , 403–410.2231712 Barbera P.  et al (2019) EPA-ng: massively parallel evolutionary placement of genetic sequences. Syst. Biol., 68 , 365–369.30165689 Berger S.A. , StamatakisA. (2011) Aligning short reads to reference alignments and trees. Bioinformatics, 27 , 2068–2075.21636595 Borrel G.  et al (2014) Comparative genomics highlights the unique biology of Methanomassiliicoccales, a Thermoplasmatales-related seventh order of methanogenic archaea that encodes pyrrolysine. BMC Genomics, 15 , 679–624.25124552 Borrel G.  et al (2019) Wide diversity of methane and short-chain alkane metabolisms in uncultured archaea. Nat. Microbiol., 4 , 603–613.30833729 Boughorbel S.  et al (2017) Optimal classifier for imbalanced data using Matthews Correlation Coefficient metric. PLoS One, 12 , e0177678.28574989 Boyd J.A.  et al (2018) GraftM: a tool for scalable, phylogenetically informed classification of genes within metagenomes. Nucleic Acids Res., 46 , e59.29562347 Buchfink B.  et al (2015) Fast and sensitive protein alignment using DIAMOND. Nat. Methods, 12 , 59–60.25402007 Canfield D.E.  et al (2010) The evolution and future of earth’s nitrogen cycle. Science, 330 , 192–196.20929768 Chaumeil P.-A.  et al (2019) GTDB-Tk: a toolkit to classify genomes with the Genome Taxonomy Database. Bioinformatics, 36 , 1–3. Ciccarelli F.D.  et al (2006) Toward automatic reconstruction of a highly resolved tree of life. Science, 311, 1283–1287. Cock P.J.  et al (2009) Biopython: freely available Python tools for computational molecular biology and bioinformatics. Bioinformatics, 25 , 1422–1423.19304878 Criscuolo A. , GribaldoS. (2010) BMGE (Block Mapping and Gathering with Entropy): a new software for selection of phylogenetic informative regions from multiple sequence alignments. BMC Evol. Biol., 10 , 210.20626897 Cross K.L.  et al (2019) Targeted isolation and cultivation of uncultivated bacteria by reverse genomics. Nat. Biotechnol., 37 , 1314–1321.31570900 Czech L.  et al (2019) Methods for automatic reference trees and multilevel phylogenetic placement. Bioinformatics, 35 , 1151–1158.30169747 Darling A.E.  et al (2014) PhyloSift: phylogenetic analysis of genomes and metagenomes. PeerJ, 2 , e243.24482762 Eddy S.R. (1998) Profile hidden Markov models. Bioinformatics, 14 , 755–763.9918945 Edgar R.C. (2010) Search and clustering orders of magnitude faster than BLAST. Bioinformatics, 26 , 2460–2461.20709691 Evans P.N.  et al (2019) An evolving view of methane metabolism in the Archaea. Nat. Rev. Microbiol., 17 , 219–232.30664670 Falkowski P.G.  et al (2008) The microbial engines that drive earth’s biogeochemical cycles. Science, 320 , 1034–1039.18497287 Fish J.A.  et al (2013) FunGene: the functional gene pipeline and repository. Front. Microbiol., 4 , 1–14.23346082 Hanson N.W.  et al (2016) LCA*: an entropy-based measure for taxonomic assignment within assembled metagenomes. Bioinformatics, 32 , 3535–3542.27515739 Hua Z-s.  et al (2019) Insights into the ecological roles and evolution of methyl-coenzyme M reductase-containing hot spring Archaea. Nat. Commun., 10 , 4574.31594929 Huerta-Cepas J.  et al (2016a) eggNOG 4.5: a hierarchical orthology framework with improved functional annotations for eukaryotic, prokaryotic and viral sequences. Nucleic Acids Res., 44 , D286–D293.26582926 Huerta-Cepas J.  et al (2016b) ETE 3: reconstruction, analysis, and visualization of phylogenomic data. Mol. Biol. Evol., 33 , 1635–1638.26921390 Hurwitz B.L. et al. (2013) Metabolic reprogramming by viruses in the sunlit and dark ocean. Genome Biol., 14 , R123.24200126 Huson D.H.  et al (2007) MEGAN analysis of metagenomic data. Genome Res., 17 , 377–386.17255551 Hyatt D.  et al (2010) Prodigal: prokaryotic gene recognition and translation initiation site identification. BMC Bioinformatics, 11 , 119.20211023 Jehl P.  et al (2015) OD-seq: outlier detection in multiple sequence alignments. BMC Bioinformatics, 16 , 1–11.25591917 Katoh K. , StandleyD.M. (2013) MAFFT multiple sequence alignment software version 7: improvements in performance and usability. Mol. Biol. Evol., 30 , 772–780.23329690 Kim D.  et al (2016) Centrifuge: rapid and sensitive classification of metagenomic sequences. Genome Res., 26 , 1721–1729.27852649 Konwar K.M.  et al (2013) MetaPathways: a modular pipeline for constructing pathway/genome databases from environmental sequence information. BMC Bioinformatics, 14 , 202.23800136 Kozlov A.M.  et al (2016) Phylogeny-aware identification and correction of taxonomically mislabeled sequences. Nucleic Acids Res., 44 , 5022–5033.27166378 Kozlov A.M.  et al (2019) RAxML-NG: a fast, scalable and user-friendly tool for maximum likelihood phylogenetic inference. Bioinformatics, 35 , 4453–4455.31070718 Laso-Pérez R.  et al (2016) Thermophilic archaea activate butane via alkyl-coenzyme M formation. Nature, 539 , 396–401.27749816 Le S.Q. , GascuelO. (2008) An improved general amino acid replacement matrix. Mol. Biol. Evol., 25 , 1307–1320.18367465 Letunic I. , BorkP. (2016) Interactive tree of life (iTOL) v3: an online tool for the display and annotation of phylogenetic and other trees. Nucleic Acids Res., 44 , W242–W245.27095192 Letunic I. , BorkP. (2019) Interactive Tree Of Life (iTOL) v4: recent updates and new developments. Nucleic Acids Res., 47 , W256–W259.30931475 Matsen F.A.  et al (2010) pplacer: linear time maximum-likelihood and Bayesian phylogenetic placement of sequences onto a fixed reference tree. BMC Bioinformatics, 11 , 538.21034504 Matthews B.W. (1975) Comparison of the predicted and observed secondary structure of T4 phage lysozyme. Biochim. Biophys. Acta, 405 , 442–451.1180967 McKay L.J.  et al (2019) Co-occurring genomic capacity for anaerobic methane and dissimilatory sulfur metabolisms discovered in the Korarchaeota. Nat. Microbiol., 4 , 614–622.30833730 Merchant S.  et al (2014) Unexpected cross-species contamination in genome sequencing projects. PeerJ, 2 , e675.25426337 Müller A.L.  et al (2015) Phylogenetic and environmental diversity of DsrAB-type dissimilatory (bi)sulfite reductases. ISME J., 9 , 1152–1165.25343514 Nasko D.J.  et al (2018) RefSeq database growth influences the accuracy of k-mer-based lowest common ancestor species identification. Genome Biol., 19 , 1–10.29301551 Nichols D.  et al (2010) Use of ichip for high-throughput in situ cultivation of “uncultivable microbial species”. Appl. Environ. Microbiol., 76 , 2445–2450.20173072 Ondov B.D.  et al (2016) Mash: fast genome and metagenome distance estimation using MinHash. Genome Biol., 17 , 132.27323842 Parks D.H.  et al (2018) A standardized bacterial taxonomy based on genome phylogeny substantially revises the tree of life. Nat. Biotechnol., 36 , 996–1004.30148503 Pattengale N.D.  et al (2010) How many bootstrap replicates are necessary?  J. Comput. Biol., 17 , 337–354.20377449 Peabody M.A.  et al (2015) Evaluation of shotgun metagenomics sequence classification methods using in silico and in vitro simulated communities. BMC Bioinformatics, 16 , 363. Price M.N.  et al (2010) FastTree 2 - approximately maximum-likelihood trees for large alignments. PLoS One, 5 , e9490.20224823 Rappé M.S. , GiovannoniS.J. (2003) The uncultured microbial majority. Annu. Rev. Microbiol., 57 , 369–394.14527284 Rinke C.  et al (2013) Insights into the phylogeny and coding potential of microbial dark matter. Nature, 499 , 431–437.23851394 Sczyrba A.  et al (2017) Critical Assessment of Metagenome Interpretation-a benchmark of metagenomics software. Nat. Methods, 14 , 1063–1071.28967888 Seitz K.W.  et al (2019) Asgard archaea capable of anaerobic hydrocarbon cycling. Nat. Commun., 10 , 1822.31015394 Shaiber A. , ErenA.M. (2019) Composite metagenome-assembled genomes reduce the quality of public genome repositories. mBio, 10 , 1–3. Solden L.  et al (2016) The bright side of microbial dark matter: lessons learned from the uncultivated majority. Curr. Opin. Microbiol., 31 , 217–226.27196505 Springer E.  et al (1995) Partial gene sequences for the A subunit of methyl-coenzyme M reductase (mcrI) as a phylogenetic tool for the family Methanosarcinaceae. Int. J. Syst. Bacteriol., 45 , 554–559.8590683 Stamatakis A. (2006) RAxML-VI-HPC: maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics, 22 , 2688–2690.16928733 Stark M.  et al (2010) MLTreeMap - accurate Maximum Likelihood placement of environmental DNA sequences into taxonomic and functional reference phylogenies. BMC Genomics, 11 , 461.20687950 Steen A.D.  et al (2019) High proportions of bacteria and archaea across most biomes remain uncultured. ISME J., 13 , 3126–3130.31388130 Tan G.  et al (2015) Current methods for automated filtering of multiple sequence alignments frequently worsen single-gene phylogenetic inference. Syst. Biol., 64 , 778–791.26031838 Wang Y.  et al (2019) Expanding anaerobic alkane metabolism in the domain of Archaea. Nat. Microbiol., 4 , 595–602.30833728 Whitman W.B.  et al (2006) The methanogenic bacteria. In: Dworkin,M. et al. (eds), Prokaryotes. Springer, New York, NY, pp. 165–207. Woodcroft B.J.  et al (2016) OrfM: A fast open reading frame predictor for metagenomic data. Bioinformatics, 32 , 2702–2703.27153669 Wu D.  et al (2013) TreeOTU: operational taxonomic unit classification based on phylogenetic trees. Preprint at https://arxiv.org/abs/1308.6333. Ye S.H.  et al (2019) Benchmarking metagenomics tools for taxonomic classification. Cell, 178 , 779–794.31398336

PMC007xxxxxx/PMC7750928.txt

==== Front Bioinformatics Bioinformatics bioinformatics Bioinformatics 1367-4803 1367-4811 Oxford University Press 32484876 10.1093/bioinformatics/btaa539 btaa539 Original Papers Genome Analysis AcademicSubjects/SCI01060 LuxUS: DNA methylation analysis using generalized linear mixed model with spatial correlation http://orcid.org/0000-0003-0369-5701 Halla-aho Viivi Department of Computer Science, Aalto University, FI-00076 Aalto, Finland Lähdesmäki Harri Department of Computer Science, Aalto University, FI-00076 Aalto, Finland Birol Inanc Associate Editor To whom correspondence should be addressed. E-mail: viivi.halla-aho@aalto.fi or harri.lahdesmaki@aalto.fi 01 9 2020 02 6 2020 02 6 2020 36 17 45354543 20 9 2019 05 5 2020 27 5 2020 © The Author(s) 2020. Published by Oxford University Press. 2020 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Abstract Motivation DNA methylation is an important epigenetic modification, which has multiple functions. DNA methylation and its connections to diseases have been extensively studied in recent years. It is known that DNA methylation levels of neighboring cytosines are correlated and that differential DNA methylation typically occurs rather as regions instead of individual cytosine level. Results We have developed a generalized linear mixed model, LuxUS, that makes use of the correlation between neighboring cytosines to facilitate analysis of differential methylation. LuxUS implements a likelihood model for bisulfite sequencing data that accounts for experimental variation in underlying biochemistry. LuxUS can model both binary and continuous covariates, and mixed model formulation enables including replicate and cytosine random effects. Spatial correlation is included to the model through a cytosine random effect correlation structure. We show with simulation experiments that using the spatial correlation, we gain more power to the statistical testing of differential DNA methylation. Results with real bisulfite sequencing dataset show that LuxUS is able to detect biologically significant differentially methylated cytosines. Availability and implementation The tool is available at https://github.com/hallav/LuxUS. Supplementary information Supplementary data are available at Bioinformatics online. Academy of Finland 10.13039/501100002341 292660 314445 ==== Body pmc1 Introduction DNA methylation is a widely studied epigenetic modification, which is involved in gene regulation. Aberrant methylation states have been shown to be connected with diseases and cancer. One popular method for quantifying DNA methylation levels is bisulfite sequencing (BS-seq) and its variants. Along with the development of sequencing techniques, several computational tools have been proposed for analysis of differential methylation from BS-seq data. These methods aim to model the underlying methylation proportions and perform statistical tests to assess the statistical significance of the effect of a covariate of interest on the methylation level. Spatial correlation of cytosines’ methylation states is a widely known phenomenon (Eckhardt et al., 2006), and methylation state is generally thought to vary rather as regions instead of individual cytosine level. Consequently, many DNA methylation analysis tools have attempted to capture the phenomenon with different model structures and computational techniques. For example the popular beta-binomial regression-based tool RADMeth by Dolzhenko and Smith (2014) outputs log-likelihood ratio test P-values for each cytosine separately and then uses weighted Z-test to combine the P-value of a single site with the P-values of its neighbors. RADMeth also includes a feature for merging neighboring differentially methylated cytosines into differentially methylated regions (DMRs). Similarly, Wen et al. (2016) first use beta-binomial regression model to calculate P-values for individual CpG sites and then combine them using Getis-Ord statistic. BSmooth tool by Hansen et al. (2012), included in bsseq package, is based on local-likelihood smoothing, which can compensate for low-coverage data and then uses signal-to-noise statistic similar to t-test for identifying differentially methylated regions. bsseq package also includes implementation of Fisher’s exact test. DSS tool includes two versions of a beta-binomial model with hypothesis testing with Wald test, one with two-group comparison (Feng et al., 2014) and another with a general experimental design (Park and Wu, 2016). metilene tool by Jühling et al. (2016) utilizes binary segmentation algorithm with two-dimensional Kolmogorov–Smirnov test for finding differentially methylated regions. Recently published dmrseq tool (Korthauer et al., 2019) combines smoothing and covariance structure with weighted least squares regression. Mayo et al. (2015) proposed a kernel based method, M3D, where statistical testing of putative DMRs is done through maximum mean discrepancy values. Rackham et al. (2017) took a different approach with their tool ABBA, which uses a latent Gaussian model. Another approach without spatial correlation, LuxGLM, was proposed by Äijo et al. (2016), and it enables testing for differential methylation for individual cytosines while considering different methylcytosine species and experimental design through generalized linear model part. LuxGLM can also utilize spike-ins, such as the commonly used (unmethylated) lambda phage genome, to model and estimate the experimental parameters at the same time with the actual model parameters. In the study by Äijo et al. (2016), the LuxGLM tool was shown to be accurate and perform on par or even better than other recent methods in detecting differential methylation. Here, we propose LuxUS, a method where the methylation proportions are modeled with a generalized linear mixed model (GLMM) that can consider both binary and continuous variables and random effects. Instead of analyzing each cytosine separately, we propose to analyze all cytosines in a moderately sized genomic window at a time. The spatial correlation of the neighboring cytosines’ methylation states is included in the model through a covariance structure of the cytosine random effect. To allow individual variation, we also introduce a replicate random effect. After estimating the model parameters, the user can view summary statistics of the posterior distributions of all linear and mixed effect coefficients, allowing for comprehensive evaluation of their significance and proportions. 2 Materials and methods Here, we present the LuxUS model, which consists of two parts: an observation model and GLMM. Observation model attempts to model the bisulfite sequencing count data generation. The linear model part is used for estimating the methylation proportion parameter for the data generation process. The plate diagram for the whole LuxUS model is presented inFigure 1. We use Bayesian approach and probabilistic programming language Stan (Carpenter et al., 2017) for implementing the model and estimating the unknown model parameters. Hypothesis testing of the significance of explanatory variables is done using Bayes factors. To choose the genomic windows for the LuxUS analysis, we also implemented a simple preprocessing step. The workflow of a bisulfite sequencing data analysis with LuxUS is presented in diagram Supplementary Figure S3. Fig. 1. Plate diagram of the LuxUS model. The gray and white circles represent observed and latent variables, respectively. Rectangles represent fixed hyperparameters, design matrices or other type of input data 2.1 An observation model for bisulfite sequencing data We present the LuxUS method for a single genomic window and drop the index denoting a particular window position to simplify notation. The observation model is the same as the one presented in Äijo et al. (2016), which we first review here. From bisulfite sequencing data, we can retrieve, for each cytosine, the total number of reads overlapping the corresponding cytosine in each sample, NBS, out of which NBS,C observations were cytosines. This is demonstrated in Supplementary Figure S2. Both NBS and NBS,C are vectors of length NR·NC, where NR is the number of samples and NC is the number of cytosines in the genomic window of interest. In bisulfite sequencing, the DNA goes first through a bisulfite conversion treatment, which converts unmethylated cytosines to uracils with probability BSEff. The methylated cytosines stay unconverted with probability 1−BSEff*. After the bisulfite treatment, DNA is sequenced and the uraciles are observed as thymines (T) and cytosines are observed as cytosines (C). The probability of sequencing error, i.e. observing uracil as C or cytosine as T, is seqErr. The experimental parameters seqErr,i, BSEff,i and BSEff,i* are used in the model to describe the probabilities of sequencing error, bisulfite conversion and incorrect bisulfite conversion, respectively, for the sample corresponding to the ith observation. The complete probability tree for observing a C or T in a BS-seq experiment is presented in Supplementary Figure S1. The probability of observing NBS,C,i many cytosine reads out of total read count NBS,i is modeled with binomial distribution with success probability parameter pBS,C,i  (1) NBS,C,i∼Binomial(NBS,i,pBS,C,i), where i=1,…,NC·NR and where the elements of pBS,C=(pBS,C,1,…,pBS,C,NC·NR)T are calculated using the experimental parameters described above (2) pBS,C,i=θi((1−BSEff,i)(1−seqErr,i)+BSEff,i·seqErr,i)+(1−θi)((1−BSEff,i*)(1−seqErr,i)+BSEff,i*·seqErr,i), following the probability tree diagram in Supplementary Figure S1. The θi denotes the ith element of methylation proportion vector θ, which is modeled with a generalized linear mixed effect model, as described in Section 2.2. It is often thought that samples with low bisulfite conversion efficiency should be excluded from analyses, as they are considered unreliable. But as the experimental parameters can be considered with LuxUS, one does not have to leave out such samples from the analysis as long as the conversion efficiencies can be reliably estimated. 2.2 A generalized mixed model with spatial correlation As stated above, the methylation proportions θ are estimated using a GLMM, which includes fixed effect covariates and cytosine and replicate random effects. The number of fixed effect covariates, counting in the possible intercept term, is NP. Here, we assume that the effects of the covariates in design matrix X are fixed, but each of the replicates and cytosines have their own random intercept terms. A column vector Y∈RNR·NC can be expressed as a sum of fixed and random effects (3) Y=Xb+ZRuR+ZCuC+e, where design matrix X of shape (NC·NR)×NP and fixed effect coefficient vector b of length NP form the fixed effect term, whereas replicate random effect design matrix ZR of shape (NC·NR)×NR and replicate random effects vector uR of length NR form the replicate random effect term and cytosine random effect design matrix ZC of shape (NC·NR)×NC and cytosine random effects vector uC of length NC form the cytosine random effect term. The last term, vector e of length NR·NC, is the noise term of the model. The rows of Y and design matrices X, ZR and ZR should all be ordered with the same principle, e.g. by first listing the NR replicates of the first cytosine, then the replicates of the second cytosine, etc. Finally, the covariate and random effects are connected to the methylation proportion vector θ=(θ1,…,θNR·NC)T through the sigmoid link function (4) θi=11+exp(−Yi). The methylation proportions θ are used in the calculation of pBS,C,i as described in Equation 2. For the fixed effect, the design matrix X contains the experimental design. The NR×NP design matrix D for one cytosine is used as a block matrix in X which is the design matrix of the experiment and applies to the whole genomic window. The rows of the matrix D correspond to the replicates in the experiment and the columns correspond to the covariates. The matrix D is repeated NC times in the following way to form X  (5) X=(D⋮D). The fixed effect coefficients vector b has a normal prior (6) b∼N(0,σb2I), where the prior variance σb2 should be set to high enough value to enable sufficiently high variation for the fixed effect coefficients. The random effect terms for replicates and cytosines are expressed as vectors uR and uC with normal priors (7) uR∼N(0,σR2I)  (8) uC∼N(0,ΣC), which are multiplied with random effect design matrices ZR and ZC, respectively, in Equation 3. ZC and ZR indicate from which cytosine and replicate each observation is coming from. Random term uR has a diagonal covariance matrix with variance σR2 as the diagonal term, whereas uC has a covariance matrix ΣC which includes the spatial correlation of the model. The covariance matrix ΣC is defined as (9) ΣC=(σC2cov(uC1,uC2)⋯cov(uC1,uCNC)cov(uC1,uC2)σC2⋯cov(uC2,uCNC)⋮⋮⋱⋮cov(uC1,uCNC)cov(uC2,uCNC)⋯σC2), where the covariance terms between the elements of uC are defined as (10) cov(uCi,uCj)=σC2· exp(−|ci−cj|ℓ2), where i=1,…,NC, j=1,…,NC, and the genomic coordinates of the cytosines are stored in vector c=(c1,…,cNC)T. The cytosines in the genomic window are restricted to be located in the same chromosome, as otherwise it would not be possible to calculate genomic distance between the cytosines. The length-scale parameter, ℓ, is set a gamma prior (11) ℓ∼Gamma(αℓ,βℓ). Hyperparameters are set to αℓ=38 and βℓ=1 as they result in covariance that is similar to the methylation correlation plots shown in the studies by Eckhardt et al. (2006) and Song et al. (2017). To demonstrate the shape of the covariance term, the term exp(−|ci−cj|ℓ2) has been plotted as a function of the distance ci−cj in Supplementary Figure S4. The random effect variance parameters σC2 and σR2 have gamma priors (12) σC2∼Gamma(αC,βC)  (13) σR2∼Gamma(αR,βR), where αR, βR, αC and βC are set to make the prior distribution match our prior information about the cytosine and replicate random effects. The residual noise term e is defined similarly (14) e∼N(0,σE2I)  with  σE2∼Gamma(αE,βE), where αE and βE are the shape and rate parameters of gamma distribution, respectively. We have also implemented an alternative version of the model where inverse-gamma distributions are used instead of the gamma priors for the parameters σC2, σR2 and σE2. With the above definitions for the fixed and random effect, the distribution of Y can be expressed as (15) Y∼N(Xb,ΣY), where the elements of the covariance matrix ΣY are defined as ΣYi,j={σC2 exp(−|ci−cj|ℓ2) , if ZRi≠ZRjσR2+σC2 exp(−|ci−cj|ℓ2) , if ZRi=ZRj and ZCi≠ZCjσR2+σC2+σE2 , if i=j, where i=1,…,NR·NC and j=1,…,NR·NC and ZRi and ZCi denote the ith rows of random effect design matrices ZR and ZC, respectively. 2.3 Model estimation For further inference based on the model, we have to estimate the unknown parameters. The variables to be estimated are fixed effect coefficients b, length-scale parameter ℓ, replicate random effect variance σR2, cytosine random effect variance σC2, noise term variance σE2, replicate random effect uR, cytosine random effect uC and linear predictor Y, from which methylation proportion θ can be calculated as described in Equation 4. The model was implemented with probabilistic programming language Stan (Carpenter et al., 2017), and the model parameters are estimated using the no-U-turn sampling algorithm, which is a locally adaptive version of Hamiltonian Monte Carlo (HMC) sampling as implemented in Stan. The PyStan version (Stan Development Team, 2017) of the software was used for the HMC sampling. Stan also has a built-in automatic differentiation variational inference (ADVI) feature (Kucukelbir et al., 2015), which was also tested for estimating the model parameters. The mean-field algorithm, which is the default option in ADVI, was used. A more detailed description of model estimation algorithms can be found from Supplementary Section S1.1. As variational inference approaches are often faster than Markov chain Monte Carlo (MCMC) methods, it could be a more favorable model estimation method especially when analyzing large reduced representation bisulfite sequencing (RRBS-seq) or whole genome bisulfite sequencing (WGBS-seq) datasets. For running ADVI we utilize CmdStan (Stan Development Team, 2018), which is the command line interface to Stan. 2.4 Testing for differential methylation After estimating the model parameters as described above, differential methylation can be tested. With the models described above, it is possible to perform two types of tests. The type 1 test has null hypothesis H0:bi=0 and alternative hypothesis H1:bi≠0, i.e. the statistical significance of the covariate i is tested. The type 2 test has null hypothesis H0:bi−bj=0 and alternative hypothesis H1:bi−bj≠0, i.e. the difference between the effects of covariates i and j is tested for statistical significance. The testing is done using Bayes factors. As calculating the exact values of Bayes factors is often infeasible, instead the Savage–Dickey density ratio estimate of the Bayes factor is used. For the type 1 test, the Savage–Dickey estimate is (16) BF≈p(bi=0|H1)p(bi=0|H1,D), where we denote data with D. The Savage–Dickey estimator for the type 2 test is defined similarly. The numerator of the Savage–Dickey density ratio is analytically solvable as we use normal prior for b. As for the denominator, it can be approximated from the posterior samples for bi by evaluating a kernel density estimate of the posterior at origin. The Gaussian kernel density estimation function from statistical functions module from the SciPy package (Virtanen et al., 2020) is used to calculate the kernel density estimates. Scott’s rule is used as the bandwidth parameter for the kernel density estimation. The Bayes factor is calculated for the whole window being analyzed, as the linear model coefficients b are shared among all cytosines. 2.5 Preanalysis step for determining and filtering the genomic windows To prepare a BS-seq experiment dataset for LuxUS analysis, we have provided a script which also includes a simple preanalysis filtering method. The genomic windows, for which the analysis is performed, can be determined using a fixed window size (in terms of number of nucleotides) or number of cytosines in a window. The prefiltering step is computationally efficient and can be used to filter away regions which e.g. do not contain sufficient number of reads or are unlikely to exhibit differential DNA methylation. A coverage limit can be used to filter out cytosines with low coverages. As estimating the model for Bayes factor calculation can be computationally burdensome in genome-wide studies, we have implemented an F-test for testing the significance of a variable of interest using the logit transformed sample means (calculated over all the cytosines in the window) of the methylation states. The P-value limit for accepting a window for further analysis with the LuxUS model can be set as desired by the user. Each genomic window can then be analyzed further separately, enabling parallelization. 3 Results To demonstrate the performance of the LuxUS model and to compare it to other tools, we applied LuxUS to real and simulated BS-seq datasets. First, the results from analysis of colon cancer WGBS-seq dataset for LuxUS and RADMeth were compared. The performance of LuxUS, RADMeth, M3D, bsseq, DSS and metilene were compared on simulated datasets. For LuxUS, both HMC and ADVI approaches for model fitting were tested. To show the advantage of using spatial correlation structure, each cytosine in the simulated datasets was also analyzed separately. 3.1 Analysis of colon cancer WGBS-seq data The model was tested using a WGBS-seq dataset by Hansen (2018), consisting of matched human colon and colon cancer samples. The processed sequencing data were provided for two chromosomes, 21 and 22. First, the genomic windows for LuxUS analysis were determined using the preanalysis method. The details of the preanalysis and setting the priors can be found in Supplementary Section S2.1. Then Stan was run with HMC sampling to fit the LuxUS model parameters and finally Bayes factors were calculated. The sampling was done with four chains with 1500 samples in each, out of which half were discarded as burn-in. As a result, 1422 genomic windows were discovered with Bayes factor ≥3. These genomic windows covered 21 127 cytosines in total. Kass and Raftery (1995) proposed that Bayes factor values from 3 to 20 would indicate positive evidence against the null hypothesis, whereas values from 20 to 150 indicate strong evidence and values higher than 150 indicate very strong evidence. These limits can be used as approximate guidelines, as LuxUS computes the Savage–Dickey estimates of the Bayes factors instead of the exact Bayes factors. In comparison, we performed RADMeth analysis on the same dataset and discovered 137 142 cytosines with adjusted, FDR-corrected P-value ≤0.05. The overlap of the significant cytosines from LuxUS and RADMeth approaches is 15 008 cytosines, i.e. about three-fourths of the differentially methylated cytosines found by LuxUS are also found by RADMeth. The results for a genomic region in chromosome 21 are shown in Figure 2, which demonstrates that both LuxUS and RADMeth produce similar results, but LuxUS is more conservative. For example near the end of the genomic region in Figure 2, the difference in average methylation state between the cancer and normal colon cells decreases, which RADMeth fails to detect and produces P-values smaller than 0.05. Whereas the said region was filtered from the analysis by LuxUS already in the preanalysis phase. The boxplots of the computation times (on one standard computing node in a cluster) for each number of cytosines in a genomic window for LuxUS are shown in Supplementary Figure S5. The total computation time on a single core for LuxUS for the 9945 genomic windows was 1623.80 h. Note that LuxUS supports full parallelization across all genomic windows, although the computation time here is reported for a single computing core. Running RADMeth for the whole dataset took 31 min and 33 s. Fig. 2. The top panel shows the methylation states for a chosen genomic region in chromosome 21 (from 28 214 962 to 28 217 867 bp) for the six samples in the colon cancer dataset. The colon cancer sample methylation states have been plotted in orange and the normal colon samples in purple. The experiment was paired, and the cancer and colon samples from the same person have been plotted with the same marker symbols. The middle panel shows the LuxUS Bayes factor values with log scaling. The Bayes factor values have been truncated to 1000. The cytosines which were filtered from further analysis in the preanalysis phase have been plotted with value 0 with empty markers. The horizontal red line corresponds to the Bayes factor value 3. The bottom panel shows the RADMeth P-values for the same genomic region. The horizontal red line corresponds to the adjusted P-value of 0.05. The cytosines for which RADMeth produced no P-value are plotted with P-value of 1 with empty markers To investigate the biological significance of the found differentially methylated cytosines, we used the GREAT tool (McLean et al., 2010) for gene set enrichment analysis (online version 3.0.0 with the default parameter settings). The lists of significantly differentially methylated cytosines found by LuxUS and RADMeth were given to the tool as inputs and the unfiltered list of CpG sites for the chromosomes 21 and 22 was used as background regions file. The results from both methods showed enrichment in cancer-related terms in Disease Ontology and Molecular Signatures Database (MSigDB) Perturbation ontology, which indicates that the methods were able to discover truly differentially methylated cytosines. The lists of the enriched terms for LuxUS and RADMeth can be found on Supplementary Figures S6–S10. Supplementary Section S2.3 describes how LuxUS preanalysis step was validated using the colon cancer dataset. In Supplementary Section S2.4, the estimated variance parameters were utilized for setting simulation parameters. To further investigate the effect of the variance term priors, we applied LuxUS to another BS-seq dataset by Hascher et al. (2014). We also experimented using different σb2 values to see how they affect the calculated Bayes factors and consequently the performance of the method with this dataset. These experiments are described in detail in Supplementary Section S3. 3.2 Performance comparisons on simulated BS-seq data Simulated BS-seq data were generated for performance comparison purposes. The simulations were made using the LuxUS model, which considers various mechanisms that affect the real bisulfite sequencing process, such as experimental parameters. First, 100 sets of experimental parameters and cytosine locations were generated. The number of cytosines was 10 and their locations were randomly chosen from range [1,1000]. The choice of cytosine frequency is explained in detail in Supplementary Section S2.4. Two BS-seq datasets were then simulated for each of these 100 sets: one with differential methylation and one with no differential methylation between the case and control samples. These 200 simulated genomic regions form a simulated dataset with 50% proportion of differentially methylated regions. The values for the fixed effect coefficient means, μb=(μB,0,μB,1), for the cases with differential methylation were (−1.4,1), (−1.4,2.3), (1.4,−1) and (1.4,−2.3). These values correspond to approximate values of 0.2, 0.5, –0.2 and –0.5 of Δθ, the difference in methylation proportions between the case and control groups, respectively. The variance for the fixed effect coefficients, σb2, was set to 0.25 for the simulations. The variances for the cytosine and replicate random effects and the noise term, σC2, σR2 and σE2, were set to the means of the gamma distributions which were fitted to the posterior samples from the colon cancer WGBS-seq data analysis. Detailed description of this can be found from Supplementary Section S2.4. The coverage (i.e. the number of reads) as well as the number of replicates were varied as 6, 12 and 24. The priors for the variance terms σC2, σR2 and σE2 were set using the results from the real data analysis and σb2 is set to 15 for the model estimation. The detailed description of the generation of the simulated data can be found from Supplementary Section S4.1. The model parameters were then estimated using the ADVI and HMC methods in Stan. The HMC sampler was run with four chains and the number of posterior samples for each chain was 1000, out of which half were discarded as burn-in. When running ADVI, 2000 samples were generated from the approximate posterior distribution to match the number of samples retrieved from HMC sampler. The number of gradient samples was 10, and number of samples used for evidence lower bound estimation was 200. These parameter values were chosen based on the results presented in Malonzo et al. (2018), where the precision and computation times of ADVI were compared to HMC for the LuxGLM model. After estimating the model parameters, the Bayes factors for the type 1 test were calculated. For comparison, the analysis was run for each cytosine separately to see the difference with the model with added spatial correlation. For this purpose, the cytosine random effect was removed from the model, as there was only one cytosine being analyzed at a time. The resulting model corresponds to the LuxGLM method by Äijo et al. (2016) except that GLM is changed to GLMM by adding the replicate random effect into the model. To compare our method to other published tools, we ran the analysis also with RADMeth, M3D, dmrseq, DSS, metilene and bsseq. RADMeth was run with default parameters and the P-value adjusting was done for each simulated dataset separately. Presumably because of the beta-binomial nature of the RADMeth model, there were some cases where the methylation states of both the case and control groups were exactly 0 or 1 for which RADMeth failed to calculate P-values. The result for these cytosines were removed before ROC and AUROC calculations. dmrseq was run first with default parameters and then with a higher maximum number of permutations and lower cutoff value for the single CpG coefficient, but the AUROC values for the method remained very low (possibly for some technical incompatibility reasons) and were thus left out from the result table. M3D was run with default parameters. For M3D all the simulated datasets (with one simulation setting) were given to M3D at the same time as separate testing regions, each with different dummy chromosome name. For the ROC and AUROC calculations the unadjusted P-values were used, as they gave slightly better result than the FDR-adjusted P-values even though the FDR-correction should not change the ranking of the P-values. metilene, bsseq and DSS were all run to compare the difference between case and control groups for each simulated genomic window separately. metilene was run with de novo DMR finding mode, with maximum distance between CpGs in a DMR set to 1000 and minimum number of CpGs in a DMR set to one. Minimum mean methylation difference for calling DMRs was set to zero. Additionally, we ran metilene with predefined regions mode with the same settings as for the default de novo DMR finding mode, using the start and end coordinates of each simulated genomic window for defining the regions for which the statistical testing is performed. From now on, we will refer to the de novo DMR finding mode as metilene, whereas the predefined regions mode is referred to as metilene mode 2. With bsseq, we first ran the smoothing step with minimum number of cytosines in a smoothing window set to one. Then Fisher’s exact test was run with bsseq. The t-test feature in bsseq was not utilized for comparison, as the tool does not include P-value calculation for the t-test statistics. We ran DSS tool with no smoothing and with smoothing, using two different smoothing window spans, 500 and 1000 bp. The approach with 1000 bp smoothing window span showed the best performance. The most interesting tools to compare LuxUS with are perhaps RADMeth and DSS, which alike LuxUS allow including additional covariates into the analysis. The area under receiver operating characteristic curve (AUROC) values for the different approaches are presented in Tables 1 and 2. The ADVI version of LuxUS has a tendency of outputting very high or infinite valued Bayes factors for easier cases. To avoid problems of comparing infinite values to each other in the ROC calculation, the infinite values have been replaced with the greatest non-infinite Bayes factor value with small constant added. For the cytosines for which RADMeth failed to calculate P-values, the values were removed. If the DMRs found by metilene with de novo DMR finding mode did not cover all the cytosines in a simulated genomic window, the not covered cytosines were given P-value 1 for ROC calculation. The AUROC values in Table 1, presenting results for simulation setting where expected methylation level difference between case and control groups was Δθ≈−0.2, show that metilene with predefined regions mode slightly outperforms LuxUS in most of the cases. While in Table 2, presenting results for simulation setting where Δθ≈−0.5, for the most of the cases the LuxUS model gives the highest AUROC values. The difference between the results from running LuxUS for a genomic window and for each cytosine separately is considerable. For these simulation settings, using variational inference for estimating the model parameters results in nearly as good results as using HMC sampling, while also being significantly faster, as can be seen in Supplementary Table S1 in which the comparison of mean computation times for the HMC and ADVI approaches is presented. RADMeth, M3D, metilene with de novo DMR finding mode and bsseq show better performance than LuxUS run separately for each cytosine, but they do not achieve the same accuracy as LuxUS and metilene mode 2, with one exception where M3D tops all the other methods. DSS performs well, but it has the best AUROC values in only four of the simulation cases shown in the tables. While for LuxUS, RADMeth, metilene, DSS and bsseq performance gets systematically better when the replicate or read count is increases, the performance of M3D fluctuates. LuxUS and metilene mode 2 seem to do equally well in this simulation experiment. However, unlike LuxUS metilene is not able to take into account additional covariates which is often desired in real data analysis. The AUROC values and ROC curves for the simulation settings with different μb can be found in Supplementary Tables S2, S3 and Supplementary Figures S20–S23, respectively. Supplementary Section S4.7 demonstrates how LuxUS can estimate the underlying methylation proportions. To demonstrate that LuxUS performs well even with a smaller, more realistic DMR proportion, we also performed comparisons with a dataset consisting of 1000 simulated regions out of which 50 (5%) were DMRs. The results are presented in the form of precision–recall curves and average precision tables in Supplementary Sections S4.5 and S4.6. The results are fairly similar to the ones with 50% DMR proportion, as LuxUS HMC, metilene and DSS are performing approximately equally well when the methylation state difference between case and control samples is small. When the difference is bigger, LuxUS HMC shows the best performance. RADMeth and M3D perform relatively well, whereas bsseq and LuxUS run separately for each cytosine have the lowest AUROC values. However, this time LuxUS ADVI was not able to reach the same AUROC value levels as LuxUS HMC and its precision–recall curves show unexpected behavior, which might be because of its tendency of returning very high BF approximation values. Table 1. AUROC values for LuxUS with HMC and ADVI, LuxUS for separate cytosines, RADMeth, M3D, metilene, DSS and bsseq for simulated dataset with μb=(1.4,−1), corresponding to Δθ≈−0.2 NBS NR LuxUS LuxUS LuxUS RAD- M3D metilene metilene DSS bsseq HMC ADVI sep. Meth mode 2 6 6 0.734 0.738 0.522 0.607 0.719 0.631 0.767 0.738 0.615 6 12 0.814 0.783 0.606 0.712 0.703 0.658 0.818 0.801 0.643 6 24 0.923 0.909 0.687 0.819 0.682 0.759 0.917 0.904 0.711 12 6 0.687 0.696 0.545 0.632 0.640 0.590 0.698 0.690 0.590 12 12 0.822 0.803 0.625 0.756 0.701 0.678 0.827 0.812 0.654 12 24 0.927 0.899 0.721 0.867 0.755 0.781 0.928 0.924 0.754 24 6 0.667 0.668 0.541 0.612 0.703 0.635 0.686 0.680 0.597 24 12 0.840 0.835 0.637 0.765 0.777 0.735 0.850 0.852 0.712 24 24 0.933 0.886 0.762 0.890 0.732 0.830 0.934 0.916 0.773 Note: The highest AUROC value is bolded for each simulation scenario. NBS denotes the number of sequencing reads overlapping a cytosine and NR denotes the number of samples. Table 2. AUROC values for LuxUS with HMC and ADVI, LuxUS for separate cytosines, RADMeth, M3D, DSS, metilene and bsseq for simulated dataset with μB=(1.4,−2.3), corresponding to Δθ≈−0.5 NBS NR LuxUS LuxUS LuxUS RAD- M3D metilene metilene DSS bsseq HMC ADVI sep. Meth mode 2 6 6 0.934 0.890 0.737 0.824 0.874 0.816 0.926 0.917 0.770 6 12 1.000 0.975 0.889 0.974 0.949 0.923 0.999 0.997 0.912 6 24 1.000 0.960 0.974 0.994 0.969 0.910 1.000 1.000 0.979 12 6 0.960 0.942 0.781 0.881 0.914 0.821 0.962 0.957 0.825 12 12 1.000 0.940 0.913 0.978 0.953 0.950 0.999 0.998 0.926 12 24 1.000 0.940 0.982 0.995 0.959 0.982 1.000 1.000 0.982 24 6 0.946 0.924 0.795 0.871 0.908 0.836 0.939 0.936 0.830 24 12 0.995 0.934 0.918 0.966 0.925 0.931 0.993 0.988 0.923 24 24 1.000 0.945 0.993 0.996 0.972 0.964 1.000 1.000 0.992 Note: The highest AUROC value is bolded for each simulation scenario. NBS denotes the number of sequencing reads overlapping a cytosine and NR denotes the number of samples. To better demonstrate the concept of adding spatial correlation into the model, we simulated data with different numbers of cytosines in a window of 1000 basepairs and calculated the AUROC values for each case. The coefficient mean μb was set to (1.4,−1) and the number of reads and replicates was 12. Using this simulated data, the LuxUS model was estimated with HMC algorithm and Bayes factors were calculated. The results of this experiment can be seen in Figure 3, which shows that analyzing multiple cytosines at a time gives more statistical power and results in higher AUROC values. The turning point for this simulation setting seems to be between six and eight cytosines, after which adding more cytosines does not seem to add more power at least in this specific case. As the number of cytosines in the analysis increases, so does the number of parameters in the model and the size of the covariance matrices. This increases computation time during model estimation and can also lead to estimation inaccuracies or convergence issues during sampling. The increase in computation time for the colon cancer dataset is shown in Supplementary Figure S5, where the computation times for different number of cytosines being analyzed at a time are presented. The results shown in Figure 3 indicate that increasing the number of cytosines in the analysis after some point does not give more power to the analysis, and it is not advisable in the sense of computational efficiency either. Fig. 3. The AUROC value for LuxUS as function of number of cytosines in a 1000 bp window for simulated data and its interpolation with third-degree polynomial. Each AUROC value is estimated using 400 datasets, including 200 datasets with differential methylation and 200 datasets with no differential methylation LuxUS allows a general experimental design to be used in the analysis, and to present the advantages of this feature, we generated a simulated dataset with two confounding covariates. The design matrix consists of an intercept term, a binary covariate distinguishing cases from controls, a confounding binary covariate and a confounding continuous covariate, with corresponding coefficients μb=(1.4,−1,2,−3) used in generating the data. Details of the data generation can be found from Supplementary Section S4.1. The comparisons were conducted for both 50 and 5% DMR proportions. The AUROC values for each method for the comparison with 50% DMR proportion (200 simulated genomic regions in total) are presented in Table 3. Corresponding ROC figures can be found in Supplementary Section S4.8. Based on the AUROC values, LuxUS HMC performs the best, whereas LuxUS ADVI and DSS tools also do well in the comparison. The continuous covariate was transformed into a binary covariate so that it could be utilized by RADMeth, but its AUROC values are lower than for LuxUS and DSS, which both allow continuous covariates. metilene mode 2 is performing approximately equally well as RADMeth. Unlike with the simple experimental design, the AUROC values for metilene mode 2 are considerably lower than for LuxUS and DSS for this simulation setting where the confounding coefficients have significant effects, muddling the difference between cases and controls. The default mode of metilene, bsseq, M3D and LuxUS run separately for each cytosine have the lowest AUROC values, which is expected as metilene, bsseq and M3D tools cannot take confounding covariates into account and bsseq and LuxUS (sep.) cannot utilize spatial correlation. The precision–recall curves and average precision tables for the 5% DMR proportion setting (1000 simulated genomic regions in total) are presented in Supplementary Sections S4.9 and S4.10. Comparing the average precision values, LuxUS HMC shows again the best performance. In this setting, metilene mode 2 seems to be doing slightly better than RADMeth. The average precisions for LuxUS ADVI are again notably lower than for LuxUS HMC in this setting where the DMR proportion is small. Table 3. AUROC values for LuxUS with HMC and ADVI, LuxUS for separate cytosines, RADMeth, M3D, DSS, metilene and bsseq for simulated dataset with confounding covariates NBS NR LuxUS LuxUS LuxUS RAD- M3D metilene metilene DSS bsseq HMC ADVI sep. Meth mode 2 6 12 0.859 0.851 0.605 0.728 0.674 0.625 0.717 0.845 0.614 6 24 0.907 0.883 0.682 0.800 0.618 0.735 0.840 0.870 0.688 12 12 0.809 0.802 0.634 0.702 0.644 0.676 0.712 0.757 0.628 12 24 0.938 0.899 0.748 0.821 0.722 0.772 0.861 0.915 0.737 24 12 0.796 0.738 0.641 0.717 0.658 0.592 0.684 0.750 0.626 24 24 0.915 0.880 0.731 0.827 0.690 0.709 0.836 0.874 0.714 Note: The highest AUROC value is bolded for each simulation scenario. NBS denotes the number of sequencing reads overlapping a cytosine and NR denotes the number of samples. 3.3 Performance comparisons on simulated differentially methylated regions based on real BS-seq data To make further comparisons of the performance of the tools with a model-independent dataset, we applied them on simulated data by Hebestreit et al. (2013). The simulation was based on RRBS-seq data of 12 control samples. Six of the controls were turned to cases by adding 10 000 simulated differentially methylated regions to randomly chosen CpG islands. The size of the DMRs and the methylation difference between the cases and controls was varied. For this comparison, the CpG islands containing a DMR were divided into DMR and non-DMR sets. For both sets, the LuxUS preanalysis step was run to divide them into genomic windows with LuxUS preanalysis method. The details of the preanalysis and data filtering can be found in Supplementary Section S5.1. About 3000 windows with minimum number of two cytosines were chosen randomly from both DMR and non-DMR groups. LuxUS, RADMeth, M3D, bsseq, metilene and DSS analyses were conducted on these 6000 windows. The experimental design consisted of an intercept term and a case–control indicator variable for LuxUS and RADMeth. With M3D, bsseq, DSS and metilene the difference between the case and control groups was tested. For RADMeth, the P-value adjustment step was performed for all the 6000 genomic windows together. LuxUS was run with HMC sampler, using four chains with 1500 samples in each out of which half were discarded as burn-in. With DSS tool, three different smoothing window spans (500, 1000 and 2000 bp) were tested along with approach with no smoothing. The DSS approach that yielded best AUROC value was smoothing with 500 bp window span and was chosen to be presented here. metilene tool was run with de novo DMR finding and predefined regions modes with the same settings, as described in Section 3.2, that the maximum allowed distance between two CpGs in a DMR was increased to 2000 bp. bsseq tool was used to run smoothing and Fisher’s exact test, similarly, as described in Section 3.2. For metilene, DSS and bsseq tools, the picked DMR and non-DMR windows were all analyzed together after sorting them based on their genomic locations. As the differentially methylated regions were known for this simulated dataset, ROC computation could be performed. Figure 4 shows that all methods perform approximately equally well. LuxUS (HMC) has the highest AUROC value of 0.992, LuxUS with ADVI model fitting comes in second with AUROC value of 0.980 and RADMeth is third with AUROC value 0.968. metilene, M3D and DSS with 500 bp smoothing window span all performed well with AUROC values 0.957, 0.938 and 0.937, respectively, whereas bsseq has the lowest AUROC value of 0.909. For this dataset, LuxUS with ADVI produced a great number of very high and infinite-valued Bayes factors. The infinite-valued BFs were again replaced with the highest non-infinite BF value with a small constant added for the ROC calculation. Fig. 4. Receiver operating characteristics curves for LuxUS with HMC and ADVI model estimation, RADMeth, M3D, bsseq, metilene and DSS for the simulated data by Hebestreit et al. (2013). The dashed black line shows the expected ROC curve for random guessing 4 Discussion The LuxUS method was tested on both simulated and real datasets. The results for the simulated data show that including spatial correlation to the model gives clear advantage when compared to the analysis with only one cytosine being analyzed at a time. Comparison with other methods showed that our method performed as well or better than the other methods in almost all simulation settings. In real bisulfite sequencing data analysis, LuxUS was more conservative than RADMeth, as the number of differentially methylated cytosines found by LuxUS was considerably lower than for RADMeth. This of course depends on the chosen significance levels and Bayes factor cutoff-values. There was a clear overlap between the differentially methylated cytosines retrieved from these two methods, as three-fourths of the differentially methylated cytosines found by LuxUS were also found by RADMeth. The follow-up analysis with GREAT tool implied that the both methods were able to find biologically significant DMRs. Based on the results on the simulated datasets, the optimal number of cytosines in a window might be 5–10, as this enhances the statistical power of the statistical testing, whereas the model estimation can be done computationally efficiently. Based on the AUROC values calculated for the simulated data with 50% DMR proportion, both the HMC and ADVI methods for estimating the model seem to perform equally well. However, with 5% DMR proportion the average precision values LuxUS HMC were higher than for LuxUS ADVI. The HMC approach provides better model inference tools for cases where closer investigation on the fitted model is desirable, whereas ADVI is considerably faster. Simple two-group comparison simulation experiments were conducted both with a 50% and a more realistic 5% DMR proportions, which both yielded similar results. Depending on the simulation setting, usually either LuxUS HMC or metilene mode 2 showed best performance. For the simulated dataset by Hebestreit et al. (2013), all of the methods performed nearly equally well, albeit the preprocessing of the data was done with respect to the restrictions of the RADMeth tool. Our method enables including both continuous and binary covariates into the fixed effect design matrix and inference on the estimated model, which is not possible with some other tools such as RADMeth, which cannot handle continuous covariates and does not provide a summary of the fitted models. Some tools, such as metilene, cannot consider any covariates and allow only a comparison between two groups. Accounting for confounding factors is important as several factors, such as age of an individual and smoking history, are known to affect DNA methylation. We demonstrated the advantage of this feature with simulation experiments where two confounding covariates were included. In these experiments, LuxUS had the best AUROC and average precision values. The metilene (mode 2) tool, which performed well with the simple two-group comparison simulation setting, could not reach the same AUROC and AP levels as the tools which could take the confounding effects into account. With LuxUS, the posterior samples for the linear model coefficients and variance parameters can be explored using the summary and plotting utilities provided in Stan. Two questions concerning our model are that does the correlation structure match reality well enough and how the priors for the variance parameters should be set. We have provided a set of default parameters for the tool, but the parameter values can be adjusted by the user to match the problem at hand and to the available computational resources. Similarly, the choice of the parameters used in the preanalysis step and the cutoff-value for the Bayes factors affect the final results. As we use the Savage–Dickey Bayes factor estimate, it may be advisable to empirically calibrate Bayes factor cutoffs for significance e.g. using some known differentially methylated loci. As the ADVI estimates of the posteriors seem to underestimate variance, the Savage–Dickey estimates of the Bayes factors tend to have very high values. To scale down the Bayes factor magnitude, one could opt for using wider bandwidth for the kernel density estimation used in the Savage–Dickey estimation calculation. For example the Scott’s ‘rule of thumb’ bandwidth which is currently used in the kernel density estimation step could be doubled to produce more conservative kernel density estimates. Finally, it is possible to extend our model to incorporate the oxidized methylcytosine species into the analysis as done previously in LuxGLM. 5 Conclusion We have proposed a new tool for detecting differential methylation, which uses the spatial correlation of neighboring cytosines to enhance the accuracy of detecting differential methylation. The presented results show that our tool is able to quantify differential methylation from both simulated and real BS-seq data. Comparisons on simulated data show that our model performs as well as, or even better, than previous methods. The provided preanalysis step can be used to reduce the number of genomic windows for which the Bayesian analysis is done, and the computations can be parallelized for computational efficiency. Opting for variational inference in the model estimation step reduces the needed computation time even further, without having to compromise the accuracy of the fitted model. The tool is available at https://github.com/hallav/LuxUS. Supplementary Material btaa539_Supplementary_File Click here for additional data file. Acknowledgements The authors acknowledge the computational resources provided by the Aalto Science-IT project. Funding This work was supported by the Academy of Finland (project numbers: 292660 and 314445). Conflict of Interest: none declared. ==== Refs References Äijo T.  et al (2016) LuxGLM: a probabilistic covariate model for quantification of DNA methylation modifications with complex experimental designs. Bioinformatics, 32 , i511–i519.27587669 Carpenter B.  et al (2017) Stan: a probabilistic programming language. J. Stat. Softw., 76 , 1–32. Dolzhenko E. , SmithA.D. (2014) Using beta-binomial regression for high-precision differential methylation analysis in multifactor whole-genome bisulfite sequencing experiments. BMC Bioinformatics, 15 , 215.24962134 Eckhardt F.  et al (2006) DNA methylation profiling of human chromosomes 6, 20 and 22. Nat. Genet., 38 , 1378–1385.17072317 Feng H.  et al (2014) A Bayesian hierarchical model to detect differentially methylated loci from single nucleotide resolution sequencing data. Nucleic Acids Res., 42 , e69.24561809 Hansen K.D.  et al (2012) BSmooth: from whole genome bisulfite sequencing reads to differentially methylated regions. Genome Biol., 13 , R83.23034175 Hansen K.D. (2018) bsseqData: Example whole genome bisulfite data for the bsseq package. R package version 0.20.0. http://bioconductor.org/packages/release/data/experiment/html/bsseqData.html. Hascher A.  et al (2014) DNA methyltransferase inhibition reverses epigenetically embedded phenotypes in lung cancer preferentially affecting polycomb target genes. Clin. Cancer Res., 20 , 814–826.24334763 Hebestreit K.  et al (2013) Detection of significantly differentially methylated regions in targeted bisulfite sequencing data. Bioinformatics, 29 , 1647–1653.23658421 Jühling F.  et al (2016) metilene: fast and sensitive calling of differentially methylated regions from bisulfite sequencing data. Genome Res., 26 , 256–262.26631489 Kass R. , RafteryA. (1995) Bayes factors. J. Am. Stat. Assoc., 90 , 773–795. Korthauer K.  et al (2019) Detection and accurate false discovery rate control of differentially methylated regions from whole genome bisulfite sequencing. Biostatistics, 20 , 367–383.29481604 Kucukelbir A.  et al (2015) Automatic variational inference in stan. In: Advances in Neural Information Processing Systems, Curran Associates Inc., NY, US, p. 28. Malonzo M.  et al (2018) LuxRep: a technical replicate-aware method for bisulfite sequencing data analysis. bioRxiv 444711 [preprint]. doi: 10.1101/444711. Mayo T.R.  et al (2015) M3D: a kernel-based test for spatially correlated changes in methylation profiles. Bioinformatics, 31 , 809–816.25398611 McLean C.Y.  et al (2010) GREAT improves functional interpretation of cis-regulatory regions. Nat. Biotechnol., 28 , 495–501.20436461 Park Y. , WuH. (2016) Differential methylation analysis for BS-seq data under general experimental design. Bioinformatics, 32 , 1446–1453.26819470 Rackham O.J.  et al (2017) A Bayesian approach for analysis of whole-Genome bisulfite sequencing data identifies disease-associated changes in DNA methylation. Genetics, 205 , 1443–1458.28213474 Song Y.  et al (2017) Collaborations between CpG sites in DNA methylation. Int. J. Modern Phys. B, 31 , 1750243. Stan Development Team. (2017) PyStan: the Python interface to Stan, Version 2.16.0.0. http://mc-stan.org. interface to Stan, Version 2.18.0. https://mc-stan.org/users/citations/. Virtanen P.  et al (2020) SciPy 1.0: fundamental algorithms for scientific computing in Python. Nat. Methods, 17 , 261–272.32015543 Wen Y.  et al (2016) Detection of differentially methylated regions in whole genome bisulfite sequencing data using local Getis-Ord statistics. Bioinformatics, 32 , 3396–3404.27493194

PMC007xxxxxx/PMC7750933.txt

==== Front Bioinformatics Bioinformatics bioinformatics Bioinformatics 1367-4803 1367-4811 Oxford University Press 32696040 10.1093/bioinformatics/btaa570 btaa570 Applications Notes Genetics and Population Analysis AcademicSubjects/SCI01060 Metasubtract: an R‐package to analytically produce leave‐one‐out meta‐analysis GWAS summary statistics Nolte Ilja M Schwartz Russell Associate Editor To whom correspondence should be addressed. Email: i.m.nolte@umcg.nl 15 8 2020 21 7 2020 21 7 2020 36 16 45214522 29 3 2020 05 6 2020 10 6 2020 © The Author(s) 2020. Published by Oxford University Press. 2020 https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Abstract Summary Summary statistics from a meta‐analysis of genome‐wide association studies (meta-GWAS) can be used for many follow-up analyses. One valuable application is the creation of polygenic scores. However, if polygenic scores are calculated in a validation cohort that was part of the meta-GWAS consortium, this cohort is not independent and analyses will therefore yield inflated results. The R package ‘MetaSubtract’ was developed to subtract the results of the validation cohort from meta‐GWAS summary statistics analytically. The statistical formulas for a meta‐analysis were inverted to compute corrected summary statistics of a meta‐GWAS leaving one (or more) cohort(s) out. These formulas have been implemented in MetaSubtract for different meta‐analyses methods (fixed effects inverse variance or square root sample size weighted z‐score) accounting for no, single or double genomic control correction. Results obtained by MetaSubtract correlate very well to those calculated using the traditional way, i.e. by performing a meta‐analysis leaving out the validation cohort. In conclusion, MetaSubtract allows researchers to compute meta‐GWAS summary statistics that are independent of the GWAS results of the validation cohort without requiring access to the cohort level GWAS results of the corresponding meta-GWAS consortium. Availability and implementation https://cran.r-project.org/web/packages/MetaSubtract. Supplementary information Supplementary data are available at Bioinformatics online. ==== Body pmc1 Introduction Summary statistics from meta‐analyses of genome‐wide association studies (meta-GWAS) have been made freely available by many consortia. These meta‐GWAS summary statistics can, for instance, be used to construct polygenic scores. However, if the summary statistics are used for validation in one of the cohorts that was included in the meta‐analysis, the polygenic score analysis will yield inflated results (Wray et al., 2013). For unbiased results, the validation cohort needs to be independent from the meta‐GWAS results. It is common practice to contact the consortium and ask them to rerun the meta-analysis with the validation cohort left out. As this could be time inefficient, I developed the R package ‘MetaSubtract’ to subtract the results of the validation cohort from the meta‐GWAS results analytically. For this package, it is sufficient to have the meta‐GWAS results and the cohort’s GWAS results that have been contributed. The statistical formulas for a meta‐analysis were inverted to compute corrected summary statistics of a meta‐GWAS leaving one cohort out. These formulas have been implemented in MetaSubtract for different meta‐analyses methods [fixed effects inverse variance or square root (sqrt) sample size weighted z‐score]. It can take into account results from single or double genomic control correction. Finally, it can be used for an entire GWAS, but also for a limited set of genetic markers, e.g. only the tophits from a meta-GWAS. 2 Materials and methods MetaSubtract was built as a package for R (R Development Core Team, 2012). The R platform was chosen because it is operating-system independent, commonly used, freely available, can handle large datasets and is flexible regarding input file format. The main function is meta.subtract(…) with arguments for the filename of the meta-GWAS summary statistics, the filename(s) of the cohort(s) results, the meta-analysis method and the genomic control lambdas for the meta-analysis and the cohorts or whether these should be calculated from the data. The workflow diagram with respect to the genomic control correction is explained in Supplementary Figure S1. 2.1 Statistics In a meta-GWAS results from N different cohorts are combined using meta-analysis. The formulas for a meta-analysis can be inverted to get the meta-GWAS summary statistics of all but one cohort. For example for a fixed effects inverse variance meta-analysis, the effect size of a genetic marker of N-1 cohorts, βN−1¯, can be computed as (1) βN−1¯= (1SEN¯2⋅βN¯)−(1SE12⋅β1)(1SEN¯2−1SE12), where βN¯ and SEN¯ are the effect size and corresponding standard error (SE), respectively, of the marker from the meta-GWAS, and β1 and SE1 those from the validation cohort. The derivation of this formula and for the SE, the allele frequency and the heterogeneity Q value for a fixed effect inverse variance are given in Supplementary Appendix SA in Supplementary Material. In Supplementary Appendix SB the corresponding formulas are given for a sqrt(sample size) weighted z-score meta-analysis. The package also automatically corrects the P-values, z-scores, sample size, number of studies, direction of effects, P-value of Q and the I2 heterogeneity value if available in the meta-GWAS summary statistics. 2.2 Validation To validate the package data from the VgHRV consortium were used (Nolte et al., 2017; Supplementary Table S1). One phenotype was analyzed by the inverse variance meta-analysis using data of 13 cohorts and another by the sqrt(sample size) weighted meta-analysis of z-scores using data from 15 cohorts. Here the GWAS results of the contributing cohorts were meta-analyzed with METAL (Willer et al., 2010). Cohort results were next excluded from the meta-analysis in alphabetical order by METAL or subtracted from the meta-GWAS results using MetaSubtract. METAL and MetaSubtract results of genetic markers that were present in every cohort were compared for the corrected effect size, SE, z-score, -log(P-value), allele frequency and Q statistic using two-way mixed ANOVA intraclass correlation (ICC) coefficients with absolute agreement. The polygenic score calculated from uncorrected and corrected meta-GWAS summary statistics by both MetaSubtract and METAL were associated using linear regression in the TRAILS population cohort. 3 Results Results of MetaSubtract correlated very well with those of METAL for all statistical parameters, for all ranges of effect allele frequencies, and both for the inverse variance and sqrt(sample size) weighted z-score meta-analysis (Fig. 1; Supplementary Figs S2–S7). Even when almost all cohorts were left out, the correlations were mostly still >0.95. Only for the SE in the inverse variance weighted meta-analysis (Fig. 1c), the correlation dropped to 0.7, which is likely caused by the small SE and METAL rounding it to four decimals. The latter also explains the decreasing correlation with increasing minor allele frequencies because for such genetic markers the SE becomes even smaller. Corrected polygenic scores applied in TRAILS showed similar results (Supplementary Fig. S8). Fig. 1. Intraclass correlation coefficients (ICCs) between the meta-GWAS results calculated with METAL and MetaSubtract for an inverse variance meta-analysis (a–e) and a sqrt(sample size) weighted z-score meta-analysis (f–h) both using double genomic control correction. The percentage of remaining samples after exclusion of 1 to 10 (a–e) or 12 (f–h) cohorts is shown on the x-axis. Different forms of the dots indicate different minor allele frequency ranges 4 Discussion The R package MetaSubtract is an efficient and convenient alternative to the leave-one-out meta-GWAS traditionally used to get meta-GWAS summary statistics that are independent from those of a validation cohort. The results of both methods correlate very highly. However, MetaSubtract has the distinct advantage of not requiring access to the cohort level GWAS results of the meta-GWAS consortium. Supplementary Material btaa570_supplementary_data Click here for additional data file. Acknowledgements The author thanks Harold Snieder for critical reading of the manuscript. Financial Support: none declared. Conflict of Interest: none declared. ==== Refs References Nolte I.M.  et al (2017) Genetic loci associated with heart rate variability and their effects on cardiac disease risk. Nat. Commun., 8 , 15805.28613276 R Development Core Team. (2012) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Willer C.J.  et al (2010) METAL: fast and efficient meta-analysis of genomewide association scans. Bioinformatics, 26 , 2190–2191.20616382 Wray N.R.  et al (2013) Pitfalls of predicting complex traits from SNPs. Nat. Rev. Genet., 14 , 507–515.23774735